ML20154H151

Difference Between Sandia & NUMARC Analysis of Decay Heat Removal Related Risk for Point Beach

ML20154H151
Person / Time
Site:	Point Beach
Issue date:	03/31/1988
From:	SANDIA NATIONAL LABORATORIES
To:
Shared Package
ML20154G997	List: ML20154G992 ML20154H045 ML20154H151 ML20154H187 ML20154H242 ML20154H257
References
NUDOCS 8805250246
Download: ML20154H151 (71)

Text

-

h t Q o c_ L -, '. C la 2) l: -

s

! 9.

1 i

i DIFFERENCES BETWEEN SANDIA AND NUMARC ANALYSES OF DECAY HEAT REMOVAL RELATED RISK FOR POINT BEACH r

31 MARCH 19S8 i

WHITE FLINT, MD i

l 1

8805250246 880426 PDR ADOCK 05000266 P

PDR GLV:3867N58 L

/

~%,

UNITED STATES

[

.e g

NUCLEAR REGULATORY COMMISSION e

w AsHINo TON. D. C. 20SS5

%,..... /

MAR 2 1988 MEETING NOTICE FOR:

Distribution

"~

FROM:

Roy Woods Task Manager USI A-45

SUBJECT:

DIFFERENCES BETWEEN SANDIA AND NUMARC ANALYSES OF DECAY HEAT REMOVAL RELATED RISK FOR POINT BEACH TIME:

8:30 a.m.

March 31, 1988 (Thursday)

PLACE:

White Flint 1 Building Roo,2F21 PURFOSE OF MEEilhG:

NUMARC will present additional infomation requested by the NRC Staff and by Sandia personnel regarding the NUMARC analysis of decay heat removal related risk at the Point Beach Plant.

Subject areas to be emphasizeo will be those agreed upon at the earlier 2/23/88 public meeting on the same subject with Sandia and NUMARC.

This additional infomation will be utilized by the NRC Staff in preparing an appendix for the USl A-45 Resolution Package.

The appendix will outline the FRA metnods and related assumptions that a licensee should use in a plant-specific PRA of decay heat removal related risk (such plant-specific PRAs will be the proposec resolution of USI A-45).

This meeting was requested by the NRC Staff, and is open to all interested members of the public.

The NRC contact is Roy Woods, telephone (301)492-3568.

()q uM Roy Woods Task Manager USI A-45 m

NSAC-113 MEETING AGENDA l

l 8:30 OPENING REMARKS (G. NEILS) 8:35 BACKGROUND TO NSAC-113 (G. VINE) 8:50 OVERVIEW 0F POINT BEACH (H. HANNEMAN) 9:20 COMPARIS0N OF CASE STUDY AND NSAC-113 (J. HAUGH) 9:50 BREAK DISCUSSION OF NSAC-113 ISSUES 10:00 SMALL LOCA FREQUENCY (W. PARKINSON) 10:15 PORY AND SRV EFFECT ON LOCA P0TENTIAL (D. PADDLEFORD)

i 10:30 CCW SUCCESS CRITERIA FOR HPI (D. PADDLEFORD) 10:45 FIRE ANALYSIS (W. PARKINSON) i o

METHODOLOGY DIFFERENCES o

HALON SYSTEM RELIABILITY 11:15 INTERNAL FLOOD EVALUATION (J. HAUGH) o THOMAS CORRELATION AND SW PUMP HOUSE EVALUATION 11:35 POINT BEACH VISUALS (H. HANNEMAN) 12:00 LUNCH 1:00 COST ESTIMATE DIFFERENCES (H. HANNEMAN) o GENERAL DIFFERENCES o

SPRAY MODIFICATION o

ADDITION OF DD AFW PUMP 3

1:20 RECOVERY ANALYSIS (W. PARKINSON /

E. DOUGHERTY) 2:30 OTHER HRA ISSUES (E. DOUGHERTY) o FEED AND BLEED o

SUMP RECIRCULATION 3:00 OTHER REC 0VERY ACTIONS (H. HANNEMAN) o POINT BEACH PROCEDURES 3:30 SEISMIC HAZARD CURVE (C. STEPP) 4:15 DISCUSSION AND WRAP-UP 5:00 ADJ0 URN

INDUSTRY SPONSORS / PARTICIPANTS NUMARC GERRY NEILS, NSP (W.G. CHAIRMAN)

ROGER HUSTON, NUMARC STAFF EPRI JACK HAUGH GARY VINE CARL STEPP WEP ROGER NEWTON HARV HANNEMAN WDG WARREN ANDREWS SAIC E!LL PARKINSON ED DOUGHERTY h

DON PADDLEFORD GLV:3867NS8 P%

\\

P NUKARC WORKING GROUP ON i

DECAY HEAT REMOVAL CHAIRMAN:

GERRY NEILS

.1 NSSS DESIGNS REPRESENTED CASE STUDY DHRTSG NAME, COMPANY W

GE CE BgW PLANTS MENBER GERRY NEILS, NSP X

X CHAIRMAN JEFF JEFFRIES, CPSL X

X (NSAC T.F. CH. )

ROGER NEWTON, WEP X

POINT BEACH (WO3 CH.)

DAVE HELWIG OR X

GE0i,5E E:Ci. I ;u (BWROS DHR CH.)

AL AN L AD! El' YANJEE X

(WOG ANAL. CH.)

X HlKE MEISNER, LP&L; X

DON JAMES OR X

X TURKEY PolNT/

M!KE SCH0PPMt.N, FPSL ST. LUClE (CE0G REP)

LARRY TAYLOR OR TED ENDS, Aiwm X

X ANO-1 GREGG SWINDLEHURST, DUKE X

X XAVIER POLANSKl. COMMED X

X 0VAD CITIES DON REEVES, NPPD X

COOPER X

GARY VINE, EPR!

(NSAC SIAFF SUPPORT)

BRIEF HISTORY OF NSAC-ll3 DEC 1985 DHRISG MEETING: REVIEW 0F PB & QC DRAFTS FEB 1986 COMMENT LETTERS TO SANDIA FROM EPRI AlF MAR 1986 DHRTSG MEETING. REVIEW 0F PB/0C COMMENTS; DISCUSSION OF TP, COOPER.

NRC RE0 VEST FOR INDUSTRY ANALYSIS JUN 1986 EPRl INITIATED REANALYSIS OF POINT BEACH, SUPPORTED BY WOG AND WEP OCT 1986 FIRST MEETING OF NUMARC WG: ENDORSED PB REANALYSIS EFFORT FE-AII,15E7 ALL SIX CASE STUDIES DISTRIBUTED FOR FINAL REVIEW MAY 19E7 NUMARC (CDUNCll) INCORPORATED AND CHARTERED JUN 1987 EPRI/NUMARC DHR WORKSHOP. NEW ORLEANS.

NUMARC REVIEW COMMENTS ON ALL SIX CASE STUDIES.

PB REANALYSIS RESULTS DISCUSSED.

JUL 1987 NRC (B. SHERON) LETTER TO NUMARC (G. NEILS)

REQUESTING MORE INFORMATION ON PB REANALYSIS: i SUGGESTED MEETING DCT 1987 NSAC-ll3 (DRAFT) FORWARDED TO NRC BY G. NEILS NOV 1987 A-45 PRESENTATION TO ACRS DHR S.C.: NSAC-113 DELIVERED TO ACRS JAN 1988 A-45 PRESENTATION TO ACRS DHR S.C. ON NSAC-ll3 GLV:3867NSS

1. ' f **euq'e, UNITED STATES NUCLE AR REGULATORY COMMISSION

/

n g

{

e m AssiNOTON. D. c. so6ss

(...'.,/

JUL 2 01967 Dr. Gerald Neils, Chairwan NUMARC Working Group on DHR Northern States Power company 414 Nieollet Mall Minneapolis, Minnesota 55401

Dear Dr. Neils:

I received a copy of your June 22, 1987 comunication to Dr. David Ericson on the subject of our USI A-45 Program on Shutdown Decay Heat Removal Sandia National Laboratories (SNL) is in the process of studying Pequirements.

these co ments in detail; however, we are particularly interested in your indication that a separate PRA of one of the SNL Case Studies was sponsored by (PRI, the Westinghouse (Nrners Group, and Wisconsin Electric Power and concludes that core melt risk is about ten times lower than the SNL Case Study for the same plant.

Since we perceive that this industry-sponsored study represents your quantification of the dif fering views outlined in your June 22nd letter, it would assist us in our deliberations to better understand your technical basis for these differences.

Therefore, we ask that you identify the major fic : fr. yer recert PP.'. study Mich centribute to the factor of ten differerte in core melt probability, and present the technical basis for the value(s) selected (e.g., referenceable operating experience data base, human factor studies, component reliability data, external event initiating f requencies, etc).

We also wish to acknowledge the creation of the NUMARC Working Group to study the DHR issue, and we look forward to interfacing with members in the near future.

Since our draft Regulatorv Analysis on USI A-45 is still pre dccisional, perhaps ve can censider a first meeting to focus on our reviss of the items you will identify as key contributors to the factor of ten difference in the core eelt probability between the two PRAs. W6 appreciate the technical attention that NUMARC has apparently devoted to review of the six Cest Studic:, er ir.dicated by Enclosure 1 of your June 22nd letter, and we intend to work with SNL to consider your coments.

Sine rely, 44* d.

A Brian V. Sheron, Director Division of Reactor and Plant Systems Df fice of Nuclear Regulatory Research cc:

D. Ericson 0Q I

e

Northem Statos Power Company 414 N.cohet Mah Mmeapoks.Meesota 55431 Tewphone (610 330 5500 October 28, 1967 Dr. Brian W. Sheron, Director Division of Reactor 6 Plant Systems Office of Nuclear Regulatory Research U.S. Nuclear Regulatory Coc=ission Washington, D.C.

20555

Dear Dr. Sheren:

In respense to your letter dated July 20, 1967, I a= pleased to forvard a draft copy of a docu ent entitled "EPRI/VOC Analysis of Decay Heat Receva:

Risk at Point Beach."

This study, sponsored by EPRI and the Westinghouse Ovners Group, was prepared by Science Applications In:ernational Corporation

)

and Westinghouse Electric Corporation with the assistance of Wisconsin Electric Power Ce=oanv, the evners and operators of Point Beach.

The NW. ARC Working Group on Dhh has followed and endorsed this effort.

The primary purposes of this study were to provide a best-esticate analysis of DHR risk at a selected USl A-45 Case Study plant and to quantify the differen:es discussed in our June 22 ce=:ent letter on the Case Studies.

The results of this Point Beach reanalysis, as they nov stand, indicate an approximate factor of thirty reduction in core-ceit frequency for the sequences included in the scope of the NRC study; an approxicate factor of s e v e r. reduction in the offsite consequences of these sequences, over ar?

above the core-celt f requency reduction; and an approximate 50-400* increase in the esticated cost of the various backfit proposals evaluated in the NRC study.

The EFRl/WOG findings indicate that the core-melt frequency esticate io: Icint beach (1.0 x 3C" pe: reactor year) is a f actor of ten Icst: t h c. n the core-celt frequency target in the NRC's Safety Goal. The EPRl/WOG study, like the NRC study, aise concludes with a very high degree of cerfidence that s r.

add-ot.,

dedicated SDHE syste: vould not be cost-beneficial for Point Beach.

Ve vould be pleased to meet with you and members of your staff, as suggested in your letter, to discuss the methodologies, technical bases and findings contained in both studies.

We have provided you with this draft report prior to publication to allow suf ficient time for your staf f to f amiliarire themselves with the EPRI/WOG reanalysis in advance the meeting.

In antici-pation of that testing, EPRI and WOG are continuing to double-check the codels used in their analysis against the final, as published, numerical values used in the NRC study. Although so=e small changes in the comparative esticates of core celt frequency could occur in so=e instances, ve expect the d-MD h 1

. Dr. Brian Sheron

)

Octcb2r 28, 1987 Page 2 of 2 overall results and ccaciusions of the EPRI/WOG study to remain essentially unchanged.

In the meanwhile, we would be pleased to schedule a meeting for the first mutually convenient opportunity.

Sincerely, G H Neils Chairman NUMARC Working Group on DKR CHN/vf cc:

B. Le e, NW.G C NW.G V:rking Creu; Me:ters T. Speis NEC K. Kneil NEC A. Marchese, NEC D. Ericson, Sandia National Laboratories R. Newton, Wisconsin Electric Power Co.

V. J. Parkinson, SAIC P. T. Paddlefrid, Vestinehouse Electric Corp.

A. Ladieu. Chaircan WOG Analysis Subcom=1ttee J. Tayler. ETRI V. Lay =at, EPEI T. Marsten EFE1 G. Vine, EFF.;

J. Haugh, EFhl l

i t

OBJECTIVE OF NSAC-113 REANALYZE THE DHR RISK AT POINT BEACH, BY BEST ESTIMATE METHODS AND DATA.

USE THE A-45 CASE STUDY OF POINT BEACH AS THE BASELINE,

AND POINT OF COMPARISON.

SPECIFIC OBJECTIVES 1.

QUANTIFY THE CONSERVATISMS IN THE A-45 CASE STUDY, AND DEMONSTRATE THESE CONSERVATISMS AND LIMITATIONS CAN BE CORRECTED BY BEST-ESTIMATE ANALYSIS.

2.

COS.'CT THE REANALYSIS WITH THE SAME SCOPE AND PLANT MODEL USED IN THE CASE STUDY, S0 DIFFERENCES IN RESULTS CAN BE LIMITED TO DlFFERENCES IN INPUT DATA, SUCCESS CRITERIA. AND CONSTPAINTS ON NON-SAFETY E0VlPMENT AND HUMAN PERFORMANCE (PR0/lDE FOR EASY SIDE-BY-SIDE COMPARISON).

3.

DEMONSTRATE QUANTITATIVELY THE PARTICULARLY SIGNIFICANT SHORTCOMINGS IN THE CASE STUDY TREATMENT OF EXTERNAL EVENTS AND THE DEDICATED DHR SYSTEM.

4.

RESPOND TO NRC REQUEST FOR QUANTITATlVE CRITIQUE OF CASE i

STUDIES.

PROVIDE BETTER QUANTITATIVE BASIS FOR NRC REGULATORY ANALYSIS.

5.

PROVIDE QUANTITATIVE BASIS FOR NRC/NUMARC DISCUSSIONS ON A-45 RESOLUTION.

GLV:3867NS8

I POINT BEACH DESIGN 1.

Two UNIT 2-LO0e W PWR (497 MWE NET EACH)

I

- COMMERCIAL OPERATION:

UNIT 1 12/70 (17 YEARS)

UNIT 2 10/72 (15 YEARS) 2.

COMMON CONTROL' ROOM (SEE FIGURE) 3.

COMMON SAFETY SYSTEMS A.

EMERGENCY P0wER (AC AND DC)

O 'VILiteY FEEDWATEP (MOTOR-DRIVEN PUMPS AND CSTs)

C.

SERVICE WATER l

D.

SPENT FUEL POOL COOLING 4.

U,11-SPECIFIC SAFETY SYSTEMS A.

REACTOR PROTECTION B.

SAFETY }NJECTION (HIGH AND low HEAD PUMPS, ACCUMULATORS, RWST) c.

CONTAINMENT ISOLATION 1,.

CONTAINMENT SPRAY f

E.

AFW (TURBINE-DRIVEN PUMPS) 1 F.

CVCS (BAST: CHARGING PUMPS-QA BUT NOT SAFETY-RELATED) c.

COMPONENT COOLING (CAN BE CROSS-CONNECTED)

H.

CONTAINMENT EMERGENCY PAN COOLERS.

i

-.. _. ~ - - -

M 0

t -

t 5

e

~

1 it i n1 I lg."P 1

1 1

l.

C Ce Cs CR cg C

P K

C I-C KL

- n P

1-l ~~.

1 1

1 1-

- y 1;

1

[kgN S

L L

E O

., O

_8, 4

TI R

S R

NL T

Y T.

G. Z v M

l g I I!

T AP S

rf t

f LP O

o

i l. -.

PU C

lX l

.lC3

.S U

CR A

EW 6

7 3

O SP 1

1 C

D D

I

,J I

o,

- s 1

S Y

2 LE O

N Y

1 A

3, 1

P O

'S Y

2IN OP 1

U 4

CS r

B O

IA i

P Y

tO T

1 0 r E

E 0 e T

i 2 t e

C u

• U

- p=I m

J 1

-r o '

N C "

i I

I 5

p C

O C

Y v

1 i

"I i

/

T.

P.

b

~

A F

~

Y 6

r b

r SL T.

R A

i E

f F

e P

~

2 rl O

S I

e o C

U I

t s u n B

0 p o 0

- C T.

2C o f

1 C

I t"

2 U

T.

R y>

T t

2 o

S R

N 0pT T

I f

2II f

2AW p

CS fO Y

3, C

O A

Y V

3 J

?

I O

O C

.S Y

E 2,

2 rI I

y vL O

4 AP L

Y.

e^0r o0 O

TP O

2f J

C PU R

Y 8

III t

S.

2

.S rr 7.

?

CR O

I M r-EW C

U r

SF A o

. C 7

4 3

i 5

4 7

)

l 0

0 0

N cC0 0

0 0

1 y

L 2

gI 2A3 T

3. 1 1

1C 13 1

I I

~

t

e e

e e go O

e e

e A

g c

3 e

5 2

a E

A C

2 U

b o

5 O

ti O

l J'

3 e

e n

x 5

.c hO

• -C k

1 a

o O"

O E

go 0 3 0 2

O n

8 O90

.?

~

s g

C o

.cu Oc:

e w

e E

y O G O i

y b

a s

e m

m m

0 8

o e

e m

o c.

e e

g l

$O 6 0 0

3 0 !

l

• 6 3 { j

1. O C

,s.-

e e

o m

~

,n.

o u

f" C

~-

g mm U

U 5

m E

~

5 O EO s.

e s

II.maeumm O

n-0 3 0 0 50 O ? O O

o e

o P

MMMNEW

-O-0

• • -O B

M M

n======

EO c'$

e a

PENP OPERATOR STAFFING DUTY AND CALL SHIFT SUPERINTENDENT SUPERINTENDENT (1 -S.R0)

OPERATING SUPERVISOR' DUTY IECHNICAL (1-SRO)

ADVISOR (DTA)

I CONTROL OPERATORS l

(3-ROS) l AUX 1LIARY OPERATORS (3-A0s)

NottS:

1.

ABOVE STAFFING LEVELS ARE FOR TWO-UNIT OPERATION.

2.

ONE SR0 (SS OR OS) AND ONE R0 PER UNIT ARE IN THE CONTROL ROOM AT ALL TIMES.

3.

PTA IS INSIDE SITE' BOUNDARY AND WITHIN 10 MINUTES OF THE CONTROL ROOM.

4.

DUTY AND CALL SUPERINTENDENT IS ON CALL AND WITHIN 30 Mit:UTE S OF THE PLANT.

5.

FIVE-MAN FIRE BRIGADE CONSISTS OF ONE SR0 (OP SUP)

AS LEADER AND FOUR OTHER OPERATORS (1R0 AND 3 A0S).

(IWO CREEKS FIRE DEPARTMENT IS IWO MILES FROM SITE 7 6.

IHERE IS NORMALLY ONE A0 IN AuxlLIARY BLDG. AND TWO A0S IN THE IURBINE HALL (ONE FOR EACH UNIT).

7.

EMERGENCY PLAN IMPLEMENTATION CAN AUGMENT ABOVE STA/F WITHIN 30 MINUTES (DCS, HP, 1&C, RAD CHEM, TSC MANAGER, BACKUP OPERATORS) AND ONE HOUR (ISC AND EOF)

9 W

y g

'n

.s C

E

~

5=

d W

5

~0 8

g 5

. F' A-

~

O i

8, w5 R.

g$

p 2

e5 g,

t

$2 j

E h5 b y E

~$

5 v

az v.

=

=*

~

l

~2

___ I I

I I

I E

Ie I

ss s

c s:

6 r

-=

Es g5 s

s

• a m

v w

la

=

-C

[

6l fl r

25 5

c r

zo 5

gg

-r i

ic_,

s I

g p

i 5

Ig$

E l

g "P

ERE i

J EE

• i E

EEI l

E 8 E-E 5

c s[m

~*

I wo e

g Em i EE I

E i-p, E9 EEE55 l

5 13 aaz=m e

l E

g i

i i

ii e

e i

E I

SE g

I eV aR i g Mr 5

I Es J

6-i

~e 5

5w

=

[5** gig i

.a ge c

w gw e

r ro a p E "$WE ts C E

55 e

i *v ega E g E n

.c g

a nr e

h-ae si d s l i !!Io w.

s

"1 i

'5 i

i, i

i i

3 g

l k

m-A

/,

r e

.P^1NT BEACH OPERATING PROCEDl!RES 1.

OPERATING iLOCEDURES (0PS) - NORMAL PLANT OPERATING PROCEDUPES

/"

EXAMPLES:

e. Pl. ANT STARTUP. HEATUP, POWER OPERATION,

~*

SHUTDOWN, AND COOLDOWN i

s. RCP OPERATION f

f J

f

'f C. EMERGENCY DIESEL GENERATOR

n. GAS IURBINE OPERATION J

2.

REFUELING PROCEDURES (RPS) - NOT APPLICABLE TO DESIGN BASIS EVENTS s

3.

Gelksi n.e INsikuCT101.s (OIs)

ExAMFLES:

A.

CHARGING PUMP LOCAL CONTROL B.

HALON FIRE PROTECTION SYSTEM C. McT0a-DRIVEN AFW SYSTEM

/'

D.

!URBINE-DRIVEN AFW SYSTEM i

E. SERVICE WATER SYSTEM 4.

ABNORMAL OPERATING PROCEDURES (A0PS)

EXAMPLES:

A. AuxlLIARY FEED PUMP STEAM BINDING OR OVERHEA11NG B.

EMERGENCY BORAT10N C. SERVICE WATER SYSTEM MALFUNCTION D.,

LOSS OF COMPONENT COOLING E. CONTROL ROOM INACCESSIBILITY F. SAFE TO COLD SHSUTDOWN IN LOCAL CONTROL o

5.

FIRE PROTECTION MANUAL A.

FIRE ATTACK PLANS (GENERAL, A, B, C FIRES, DIFFERENT EQUIPMENT)

B.

FIRE EMERGENCY PLANS (DIFFERENT AREAS OF THE PLANT)

C.

TRANSIENT COMBUSTIBLE CONTROL PROCEDURE D.

IGNITION CONTROL PROCEDURE J

POINT BEACH EMERGENCY OPERATING PROCEDURES E0PS WERE DEVELOPED FROM WOG EMERGENCY RESPONSE GUIDELINES,,,

REv. 1 l

ENTRY PROCEDURES:

E0P-0 REACTOR IRIP OR SAFETY INJECTION ECA-0.0 LOS$ OF Att AC POWER SUPPLEMENTARY AND EMERGENCY CONTINGENCY ACTIONS (ECAS)

EXAMPLES:

A.

E0P-1 LOSS CF REACTOR OF SECONDARY COOLANT F0P-1.2 - SMALL-BREAK LOCA C00LDOWN AND DEPRESSURIZATION E0P-1,3 - TRANSFER TO CONTAINMENT SUMP RECIRCULATION B. E0P-2 FAULTED STEAM GENERATOR ISOLATION C. E0P-3 STEAM 6ENERATOR IUBE RUPTURE CRITICAL SAFETY FUNCTION STATUS IREES (HARDCOPY AND SAS COMPUTER)

SUBCRITICALITY CORE COOLING SECONDARY HEAT llNK REACTOR VESSEL INTEGRITY CONTAINMENT INTEGRITY REACTOR COOLANT INVENTORY CRITICAL SAFETY PROCEDURES

PBNP D.C POWER SYSTEM DC SYSTEM IS SHARED BETWEEN UNIT 1 AND UNIT 2

... ~

D.C. SYSTEM CONSISTS OF 4 MAIN DISTRIBUT!QN BUSES EACH MAIN DISTRIBUTION bus CONTINUOUSLY POWERED FROM IWO SOURCES DOS}ORIGINALSTATIONBAT

1. BATTERIES:

D06)

D105 D106 ) NEW STATION BATTERIES (1985)

2. BATTERY CHARGERS-SUPPLIED FROM 480V SAFEGUARDS BUSES E.- rrv DESIGN - DOS & 6 - 60 CELL (NOM) 125 VOLTS 950 AMP-HR (1 HR. RATE)

SEISMICALLY ANALYZED (SSE)

DESIGNED FOR SB0 LOADS FOR 1 HOUR D105 & 6 -

60 CELL (NOM) 125 VOLTS 795 AMP-HR (1 HR. RATE) i SEISMICALLY QUALIFIED-lEEE 344 DESIGNED FOR SB0 LOADS FOR 1 HOUR 1

BATTERY MAJOR LOADS:

DOS & 6 (0RIGINAL BATTERIES)

ESTIMATED DEMAND:

400 AMPS DC CONTROL POWER (BKRS, SOLEN 0ID VALVES, RELAYS, ANNUNCIATORS, ETC.)

MOVS FOR TD AFW PUMPS INVERTERS FOR RED AND BLUE INSTRUMENT BUSES EMERGENCY DC LIGHTING EDG STARTING CIRCUITS & FIELD FLASHING VARIOUS EMERGENCY OIL PUMPS (TURBINE & FEEDPUMP)

D105 & 6 (NEW BATTERIES)

ESTIMATED DEMAND:

180 AMPS BACKUP FOR EDG STARTING CIRCUITS / FIELD FLASHING ALTERNATE SHUTDOWN INSTRUMENTS INVERTERS FOR WHITE & YELLOW INSTRUMENT BUSES SAS (SPDS) AND PPCS COMPUTERS

l PQINT BEACH DIESEL GENERATOR Two EMERGENCY DIESEL GENERATORS - SHARED BETWEEN UNITS EACH EDG DESIGNED TO POWER ALL SAFEGUARDS LOADS NECESSARY FOR DESIGN-BASIS ACCIDENT MITIGATION IN ONE UNIT AND ALL NECESSARY SAFE SHUTDOWN LOADS IN OTHER UNIT 1

DESIGN: GENERAL MOTORS ELECTRO-MOTIVE DIVISION MODEL 999-20 (20 CYLINDERS)

RATING: 2850 Kw (CONT.)

3050 KW (30 MINUTES) 4160 VOLTS, 30, 60 Hz.

FUEL OIL USAGE: 205 GAL /HR AT RATED LOAD Fun Olt AVAILABLE: 550 GALLON BASE IANK 550 GALLON "DAY IANK" FOR EACH DG t

(CAN BE CROSS-CONNECTED) 12,000 GALLON UNDERGROUND EMEP.GENCY Futt Oil TANK 60,000 GALLON ABOVE-GROUND STORAGE IANKS (1 0F 2 NORMALLY FULL)

EDG STARTING CIRCUlTS REQUIREMENTS:

SOLENOID VALVES FOR AIR START SUPPLY i

FIELD FLASHING CONTROL / ALARM CIRCulTS POWER:

1. DC FROM OLD STATION BATTERIES 2.' ALTERNATE FROM NEw BATTERIES (SWITCHED LOCALLY)

l I

l EDG RELIABILITY AT POINT BEACH G01 G02 BOTH l

NO OF DEMANDS 174 173 347 1

(1983 - 1987)

1. OF DEMAND FAILURES 1

0 1

PROBABILITY OF START FAILURE - ACTUAL (1983-87) 2.9E-3 HOURS OUT OF SERVICE 529 360 889 FOR ANNUAL MAINTENANCE (1985-1987)

PROBABILITY OF ONE EDG 2.0E-2 1.4E-2 1.7E-2(Avo)

FOR MAINTENTANCE TOTAL PR03 ABILITY OF ONE EDG OuT Or SERVICE OR FAILURE TO START ON DEMAND -

ACTUAL PBNP DATA

- 2.0E-2 (1983/85-1987)

EPRl/WOG STUDY

- 2.8E-2 SNL STUDY

- 4.4E-2

-___u_-_ - - _ _ _ _ _ - -, -

PBNP RWSTS/CSTS REFUELING WATER STORAGE IANKS (RWSTS)

ONE PER UNIT (285,000 GALLONS) 1/4".32" WELDED STAINLESS STEEL SHELL 1/4" STAINLESS STEEL BOTTOM

'7' lb x 70' HIGH (ASPECTS RATIO = 2.6) t ANCHORAGE - 27 BOLTS (1 1/4" - DIA. x 4' LONG)

DESIGNED AS SEISMIC CLASS 1 FOR SSE OF.12G PGA MOUNTED ON CONCRETE PAD CONDENSATE STORAGE IANKS (CSIS)

Two TOTAL (SHARED AND NORMALLY X-CONNECTED-40,000 GAL. EA) 1/4' WELDED CARBON STEEL SHELL AND BOTTOM 20' ID x 24' HIGH (ASPECT RATIO = 1.2)

ANCHORAGE - 8 EOLTS (3/4" DIA. X 1 1/2' LONG)

SrIselC CLASS 111 (NON-SEISMIC)

MOUNTED ON ROO: c

'r.iSMIC CLASS I CONTROL BUILDING c

INSIDE IURBINE EU.LDING

i l

t l

l l

COMPONENT COOLING WATER 1

PURPOSE:

INTERMEDIATE COOLING SYSTEM TO SEPARATE RADIOACTIVE,

REACTOR COOLANT FROM SERVICE WATER (LAKE MICHIGAN) 1 TO MINIMlZE POTENTIAL FOR RADIOACTIVE RELEASES 1

l l

ONE COMPONENT COOLING LOOP FOR EACH UNIT ONE PUMP AND ONE HEAT EXCHANGER REQUIRED / UNIT (NORMAL OPERATION OR DESIGN BASIS ACCIDENT)

~TWO PUMPS AND ONE UNIT-SPECIFIC HX/ UNIT WITH IMO SHARED BAcxue HXS (EACH UNITS PUMPS CAN ALSO BE CROSS-CONNECTED)

PUMPS POWERED FROM SAFEGUARDS 480V BUSES (1.E., 0FF-SITE POWER, ON-SITE GAS IURBINE GENERATOR, OR EMERGENCY DIESEL GENERATORS) 1 IMPORTANT LOADS FROM Risx PERSPECTIVE l

1. RHR HXs (COOLING FOR ECCS RECIRCULATION-OR RHR)

2. RHR (LOW-HEAD SI) PUMPS - SEAL COOLING ONLY

3. S1 PUMPS - SEAL COOLING ONLY

4. CS PUMPS - SEAL COOLING ONLY S. RCP THERMAL BARRIER (SEAL COOLING-BACKUP TO SEAL INJECTION) l l

l l

I

r POINT BEACH PORVS, BLOCK VALVES, AND ALTERNATE VENTS

~

" ' ~

PORVS -

AIR OPERATED VALVES 1.

PNUEMATIC SUPPLY: INSTRUMENT AIR TO CONTAINMENT (N

BOTTLE BACKUP - LTOP ONLY) 2 2.

SOLENOID VALVE POWER: 125 V. D.C.

BLOCK VALVES - MOTOR-0PERATED VALVES 1.

F0WERED FROM SAFEGUARDS 480 V MCC 2.

BL0ct VALVE STATISTICS:

1986 SEPT-DEC 1987 PERCENTAGE

(# or SHIFTS)

(# OF SHIFTS)

Or TIME UNIT 1 UNIT 1l UNIT 2 B:Te Cto:E:

307 0

1 43 22%

ONE CLOSED 465 0

l 5

30%

BOTH OPEN 272 294 l

192 48%

ALTERNATE VENT PATHS

1. REACTOR COOLANT GAS VENT SYSTEM - DC POWER (REACTOR VESSEL AND PRESSURIZER VENTS-7/32" ORIFICE)

2. LOOP DRAINS - NON-SAFEGUARDS AC POWER

3. LETDOWN & EXCESS LETDOWN - DC & SAFEGUARDS AC POWER

4. SAMPtlNo - DC POWER

EPRI/WOG Study Results Core Melt Frequency per Year Reduction Source of Risk E

EPRUWOG Eg.Lg.t Intern!

1.4 E-4 2.6E 6 54 Seist:0:

6.1E 5 7.4E 6 8

Fue 3.2E 5 6.3E 8 500 Internal Fic<d 7.7E 5

<l.0E 8

>7700 Extema) Flmd 1.9E 8

<l.0E 8 *

(>>2) u.;~

4.0E 6

<1.0E 8*

(>>400)

Lightning 3

c)Op.g.

g TeR' 3.lE.

1.0E 5 31 Ce:e m:h hequ:n:) redu::4 to <1.0E S by explanation of errors in NRC Case Study witho RMQS requantification.

1

INTERNAL EVENT S2 M H1' H2':

SBLOCA + MFW FAILURE + HPRS & LPRS FAILURES W/0 RHR H/X ESTIMATES:

4.70E-5 CASE STUDY 5.80E-7 NSAC-113 PRINCIPAL DIFFERENCES:

1.

SBLOCA FREQUENCY CASE STUDY: 2.0E-2 BASED ON LEAKS <2-IN. DIA; DERIVED FROM SNL/IREP ANO-1 AND MURLEY MEM0: RCP SEAL LOCAS DOMINATE NSAC-113:

3.0E-3 BASED ON LEAKS <2.0IN, DIA; DERIVED FROM OCONEE PRA AND INDUSTRY EXPERIENCE.

CREDITS SHUTDOWN PRIOR TO RECIRCULATION ~ 20 hrs; SUMP RECIRCULATION NOT REQUIRED FOR LOCAS EXPERIENCED S0 FAR CRITIQUES OF NSAC-113:

GENERALLY ASSESS NSAC-113 ESTIMATE AS REASONABLE CONTINGENT ON SUPPORTING INFORMATION

INTERNAL EVENT (CONT'D) 2.

OPERATORS FAIL TO IMPLEMENT SUMP RECIRCULATION CASE STUDY: 1E-3 NSAC-113:

1E-4 BASED ON RULE VS DIAGNOSIS AND LONG TIME TO DEPLETE RWST CRITIQUES OF NSAC-113:

NEED ADDITIONAL EXPLANATION 3.

REC 0VERY FROM RECIRCULATION FAULTS CASE STUDY: NO CREDIT NSAC-113:

SE-2 BASED ON REFILLING RWST OR USING CVCS; NOT APPLIED TO OPERATOR FAILURE TO IMPLEMENT RECIRCULATION CRITIQUE 5 0F NSAC-113:

NEED BACKUP DN PROCEDURES / TRAINING 4.

CCW SUCCESS CRITERIA APPLIED ONLY TO HPI (I.E., INJECTION FAULTS)

INCORRECT ENTRY IN TABLES 1-8, 8-3, 0F NSAC-113

INTERNAL EVENT S2 MD1D2:

SBLOCA + MFW FAILURE + HPIS & LPIS FAILURE ESTIMATES:

8.7E-6 CASE STUDY 9.5E-8 NSAC-113 PRINCIPLE DIFFERENCES:

1.

SBLOCA FREQUENCY - SAME COMMENTS AS S2 MH1' H2' 2.

CCW SUCCESS CRITERIA -

CASE STUnv. HPI DEPENDENT ON CCW AND SW AVAIL-ABILITY NSAC-113:

CCW NOT REQUIRED FOR PT. BEACH HPI (INJECTION MODE).

REC 0VERY MUST OCCUR PRIOR TO SUMP RECIRCULATION

i INTERNAL EVENT TIMLE:

LOSP + MFW FAILURE + AFW FAILURE +

F&B FAILURE ESTIMATES:

6.7 E-6 CASE STUDY 7.7 E-7 NSAC-113 PRINCIPLE DIFFERENCES 1.

USE OF NEW STATION BATTERIES CASE STUDY: ANALYSIS DID HOT INCLUDE NSAC-113:

INCLUDED OPERATOR ACTION TO USE NEW BATTERIES TO START DIESELS AFTER COMMON CAUSE FAILURE OF STATION BATTERIES CRITIQUES OF NSAC-113:

CONCUR WITN CREDITING "NEW" BATTERIES CONTINGENT ON SUPPORTING INFORMATION NOTE:

LOSP DATA WAS OF MINOR IMPORTANCE CASE STUDY: 8.4 E-2 BASED ON HRC GENERIC ESTIMATE (NUREG-1032)

INTERNAL EVENT (CONT'D)

HSAC-113:

6.2 E-2 BASED ON PT. BEACH SPECIFIC DATA J

y.-w-

-"mT*7

CRITIQUES OF NSAC-113:

GENERALLY AGREE WITH NSAC-113 APPLICATION OF PLANT SPECIFIC. DATA p

INTERNAL EVENT T3 0 H1' H2' AND T3 0 D1 D2:

TRANSIENT (MFW UNAVAILABLE) + SRVs FAIL TO CLOSE

+ EITHER (HPRS & LPRS FAILURE W/0 RHR H/X) OR (HPIS & LPIS FAILURE)

ESTIMATES:

2.5 E-5 + 4.6 E-6 = 3 E-5 CASE STUDY N/A NSAC-113 PRINCIPLE DIFFERENCES CASE STUDY: SRVs ASSUMED TO OPEN -- FAILURE TO RECLOSE (EVENT 0) RESULTS IN TRANSIENT INDUCED LOCA NSAC-113:

EVENT 0 SEQUENCES DO NOT EXIST FOR REACTOR OR TURBINE TRIPS AT PT. BEACH.

NEITHER PORVs NOR SRVs WILL BE CHALLENGED BASED ON WESTINGHOUSE OPERATING EXPERIENCE.

EVENT 0 SEQUENCES CONSERVATIVELY MODELED FOR LOSS OF 0FFSITE POWER AND LOSS OF MAIN FEEDWATER t

INTERNAL EVENT (CONT'D)

CRITIQUES OF NSAC-113:

GENERALLY AGREE WITH NSAC-113 CONCLUSION CONTINGENT ON THERMAL / HYDRAULICS ANALYSES l

l e

i I

l

T2 MON'1H2':

LOSS OF PCS + MFW FAILURE + SRVs/PORVs FAIL TO CLOSE + HPRS & LPRS FAILURE W/0 RHR H/X PRINCIPAL DIFFERENCES:

1.

SAME AS FOR S2 INIATOR 2.

STUCK OPEN PORY -

CASE STUDY: 1.4E-3 PER DEMAND.

ASSUMES A PORV STICKS OPEN 7% OF THE TIME AND BLOCK VALVE FAILURE TO ISOLATE IS 1.0E-2 PER VALVE NSAC-113:

CONSERVATIVELY ASSUMES BOTH PORVs DEMAND OPEN, AND 1% OF THE TIME ONE STICKS OPER.

ALLOWS 30 MIN FOR BLOCK VALVE CLOSURE CRITIQUES OF HSAC-113:

ACCEPTANCE OF NSAC-113 IS CONTINGENT ON VERIFICATION OF BLOCK VALVE CLOSED-CLOSED FREQUENCY i

I

INTERNAL FLOOD SEQUENCES i

1 ESTIMATES FOR ALL SEQUENCES:

j 7.7E-5 CASE STUDY

<1.0E-8 NSAC-113 PRINCIPAL DIFFERENCES:

SW PUMP HOUSE - FIRE MAIN RUPTURE CASE STUDY: BASED ON GENERIC AUX BLDG FLOOD DATA (2.2E-2 FOR MODERATE FLOOD); 1.0E-1 l

ASSIGNED TO SIMULTANEOUS FAILURE OF SIX SW PUMPS; 3.48E-2 TDAFW PUMP FAILURE.

TOTAL = 7.66E-5 NSAC-113:

BASED ON APPLICATION OF THOMAS PIPE RUPTURE CORRELATION SIMILAR TO OCONEE PRA TOTAL ESTIMATED WAS 9.8E-5 FOR LEAKS SUFFICIENT TO DAMAGE SW PUMPS.

DID NOT APPORTION FREQUENCY ACCORDING TO BREAK SIZE AS IN OCONEE PRA.

DID NOT DIRECTLY APPLY 1.0E-1 FACTOR FOR SW PUMP FAILURE AS IN CASE STUDY; ASSUMED BREAK LOCATION INSTEAD.

INTERNAL FLOOD SEQUENCES (CONT'D)

CRITIQUES OF NSAC-113:

FURTHER JUSTIFICATION OF NSAC-113 METHODOLOGY IS REQUIRED.

l l

1 M

SEISMIC EVALUATIONS ESTIHATES FOR ALL SEQUENCES:

6.1E-5 CASE STUDY 7.4E-6 NSAC-113 PRINCIPAL DIFFERENCES:

CASE STUDY: o GENERATED A SEISMIC HAZARD CURVE BASED OH ZION (SSMRP),

o CALCULATED RWST FAILURE DUE TO BUCKLING AND ANCHOR PULLOUT.

i o

NO CREDIT GIVEN FOR SEISMIC RECOVERY ACTIONS, o

DID NOT INCLUDE NEW STATION BATTERIES.

NSAC-113:

o SEISMIC HAZARD CURVE WAS REDUCED FROM CASE STUDY VALUES BY FACTOR OF TWO FOR 1-3XSSE AND FIVE FOR

>3XSSE.

o RWST NOT EXPECTED TO FAIL i

CATASTROPHICALLY (INSTANTANEOUSLY)

AT LOW ACCELERATIONS: ALLOWS ADDITIONAL TIME FOR RECOVERY.

4 i

(

SEISMIC EVALUATIONS (CONT'D) o ALLOWED CREDIT FOR RECOVERY ACTIONS o

E.G.,

USE OF ALTERNATE WATER SUPPLIES IN PLACE OF CST AND RWST o

INCLUDED NEW SEISMIC I BATTERIES.

CRITIQUES OF NSAC-113 o

UNABLE TO EVALUATE ADEQUACY OF RWST AND NSAC-113 MODIFIED HAZARD CURVE WITHOUT MORE DETAILS.

o GENERALLY CONCUR WITH ALLOWING CREDIT FOR REC 0VERY PROVIDED THOSE ACTIONS CAN BE SUBSTANTIATED.

o CONCUR WITH CREDITING NEW BATTERIES. CONTINGENT ON REVIEW 0F BATTERY DESIGN DETAILS.

---_--------2

i l..

i FIRE EVALUATION i

I i

ESTIMATES FOR ALL SEQUENCES j

3.2E-5 CASE STUDY I

6.3E-8 NSAC-113 PRINCIPAL DIFFERENCES:

CASE STUDY: o CREDITS TWO TRAIN HALON SYSTEM IN SWITCHGEAR ROOM FIRE o

CREDITS 1 TRAIN HALON SYSTEM IN A FW PUMP ROOM FIRE j

i o

HALON SYSTEM FAILURE ESTIMATED i

AT 0.2 PER DEMAND BASED ON DATA REPORTED IN MILLSTONE PRA ET AL.

o TDAFW PUMP FAILURE ESTIMATED AT 1

0.1 DURING AFW PUMP ROOM FIRE.

NO CREDIT GIVEN FOR 4160V j

SWITCHGEAR ROOM FIRE.

o AUX BLDG GENERIC FIRE FREQUENCY l

DATA RATIONED BY THE AMOUNT OF j

COMBUSTIBLES IN THE AFW PUMP ROOM TO THAT IN ENTIRE AUX BLDG.

TRANSIENT COMBUSTIBLE FIRE EVALUATED USING UCLA COMPBURN

)

(1983) CODE.

)

o DID NOT INCLUDE FIRES STARTING IN CABLE TRAYS AND ELECTRICAL PANELS i

~

FIRE EVALUATION (CONT'D)

NSAC-113:

o CREDITS 2 TRAIN HALON SYSTEM FOR ALL SEQUENCES.

o HALON SYSTEM FAILURE ESTIMATED AT 0.06 DERIVED FROM DOE HALON SYSTEM RELIABILITY DATA BASED ON ACTUAL FIRES.

o FREQUENCY OF TRANSIENT COMBUSTIBLE FIRE ESTIMATED TO BE OF LESSER IMPORTANCE.

DID NOT USE COMPBURN.

o CONSIDERED FIRES STARTIES IN CABLE TRAYS AND ELECTRICAL PANELS BASED ON GENERIC PLANT DATA.

CRITIQUES OF HSAC-113:

ENDORSED GIVING CREDIT FOR SECOND HALON TRAIN.

GENERALLY PREFERRED THE CASE STUDY APPROACH FOR INITIATING EVENT DATA; QUESTIONS ASKED ABOUT USE OF DOE HALON RELIABILITY DATA.

i t

SMALL LOCA FREQUENCY j

l i

i 1

l

]

1 I

i 1

1 1

NRC Case Study Assumptions l

1 l

4 d

EPRl/WOG Assumptions l

l l

Need for Recirculation t

l 1

1 I

t i

Summty of Other PRAs i

j h

e i

Small LOCA Classes I

i i*

!.4

)

i l

C0 r.:! actor.

i 1

i t

i i

l 1

1

I r

\\

b NRC CASE STUDY ASSUMPTIONS t

All Small LOCAn Require Recirculation Murley to Eisenhut Memo on RCP Seal Failure Charging System Not Available l

i O

5 i

EPRIAV0G ASSUMPTIONS 1

i I

Experience Justifies a Rare Event Probability for Small LOCAs Requiring Recirculation

]

Data is Adequate to Model Likellhoed Of Reaching RHR or j

isolating the Break Before Recirculation is Required i

Small LOCA Frequency is Val!d for Injection Faults 4

l 1

i 1

l 1

i i

j 1

NEED FOR REC!RCULATION d i 1

Plant Experience Indicates that Shutdown Can Occur Prior to Emptying the RWST Generic Thermal Hydraulles Analysis Shows RHR Entry C,an be Reacned Prior to Sump Recirculation

/

NRC Case Study Makes Similar Assumption Regarding More Stressful Case, l.E., No injection

(

f l

4 i

y

SUMMARY

OF OTHER PRAS Most Reference Murie/ Merno Directly or Indirectly Pipe Breaks Generally Assessed With Lower Frequency Than RCP Seal LOCA Categorization Scheme Might Be Best l

I l

l COMPA RISON OF EPRl/WOG AND NRC CASE STUDY :

SM TLL LOCA FREQUENCY TO OTHER PRAS 1,

PRA Study Value

_ Comments / Refer _ence_s.

I NRC Case Study 2.0E-2 IREP Procedures Guide which references ANO IREP which references Murley memo i

EPRl/WOG Study 3.0E-3 Oconee PRA which references isolable small LOCA at Zion as one event in total PWR experier;ce as of early 1980s WASH-1400 1.0E-3 Based on nuclear and non-nuclear experience ANO-1 IREP 2.0E-2 Both studies reference memo, identify the value as an RCP seal LOCA value Calvert Cliffs IREP and identify a pipe break value of 1.0E-3 j

Oconee PRA 3.0E-3 Based on one rceent in approximately 160 years of PWR experience and l

updated with rio events at Oconee 4

i l

Indian Point 2.0E-2 Based on three events, the Isolable Zion LOCA in 1975, an RTD blowout j

at Surry in 1972, and an RCP seal failure at Indian Point in 1977 I

l Seabrook PRA 2.8E-2 Basis, i.e., data, is proprietary. Event broken down further to yleid:

LOCAs requiring recirculation - 1.0E-3 non-isolable LOCAs not requiring recirculation - 5.8E-3, and I

isolable LOCAs not requiring recirculation - 2.3E-2 ASEP 2.0E-2 Basis is an average of many of above PRAs. Dismisses Murley rnemo, but (NUREG-1150) includes indirectly by referencing other PRAs which in turn reference the memo i

4 I

SMALL LOCA CATEGORIES

\\

\\

t Small LOCAs Requiring Recirculation isolable Small LOCAs Not Requiring Recirculation Non-Isolable Small LOCAs Not Requiring Recirculation f

-l CONCLUSION EPRl/WOG May Overestimate Small LOCAs Requiring Recirculation (Dominant Accident)

EPRl/WOG May Underestimate Small LOCAs Requiring injection (20 /o of Total Small LOCA Contribution) l

VERY St%LL LOCA REC 0'ERY WO SUf? RECIRCULATION 0

NRC CAE STUDY INCLUES VERY SMALL LOCA CATEGORY - FRE0 0F 0.02/YR WITH SUBSEQUENT FAILURE OF SUtP RECIRCULATION AS SIGNIFICNiT O

BJT LO"A WITH FREQ D0h'i TO AT LEAST 0.002/YR ARE REC 0ERED WITH0JT SUP RECIRCULATION 0

C0fFIRFED BY EX;ERIEfCE SINI NO SUFP RECIRCULATION EENTS IN OVER 503 P A T YEARS 0

W)3 DEVELOPED EFERGENCY RESPONSE GUIELifE ES-1.2, "POST LOCA C00LE0kN NO IE0;ES9JD!?tT!0N" TO DIFJT RCC T P ND PPISS ELOW 230F ND 400 PS!G RH?

EtiTRY C0iDIT10NS ACOElSiEL BY CD'_DM WITH A SG NiD REDUCIN3 S1 FLOW NiD ESTABLISHIN3 0

J fte'G'. CWGIN3 kHEri MINltUM SUSCODLIN3 NO PRESSURIZER LEVEL C0folT10NS E.' C 0

BA%GROJND DXUSENT FOR ES-1.2 INCLUES GEIERIC APPLICATION (TREAT T/H CODE) TO A OtE IfCH C0_D LIG BREAK CASE AfD TO A STUCK OPEN PORV CASE.

lll B)TH, RHR EtiTRY C0!OlT10NS REACHED ELL EFORE RWST DRAlfED 0

EF MS E6EOFED PT. BE/Cn SPECIFIC PROIDURE CONSISTENT WITH ES-1.2 i

FRE0'JENCY OF SAFETY VALVE OPENlt6 IN W PWR'S 0

NRC CASE STUDY PESUIES 7 TRANSIENT /RY WITH 7% CfMNT OF OPENING bRV - OR ABOUT 0,5/RY

'~

0 E SUR'EY RD AVLYSIS If01 CATES PORV OPEN!fGS FROM OPERATIOWL TRNGlENTS AT FRE0 0F A30iJT 0.23/RY PRE TMI AND 0,12/RY POST TMl 0

E SVETY VR.VE CPEN!!GS EXTRE.'ELY RARE - 0!E THE ORDER 0.01/RY OR LESS 0

TPR;SIENTS CDEf;1tG PORV RELATIVELY INFREQUEf;T TFJ.';I:E';T: 'l:T:: ::T::T Or, R;TICIPATORY REA:iOR RD TdEliE TRIT D: 'CT OPEN PORV'S TPMS!ENTS WITH LAR3E PR!PARY SEC0tDARY MIS"ATCH RD fD IffEDI ATE TRIP i

CA!! CAUSE PORV'S TO OPEN (LARGE LOSS OF STEA'i LOAD, MSiv CLOSUE),

REDUID CCO_1iG (LOO'ED ROTOR), REACTIVITY ADDIT 10N (ROD W1TFEPML),

i A!O f%SS ADDITION (S1 OR STE#i BREAK WITH SI) - ALL OF WHICH INFRE0JENT 0

Is_0X D PORV DIS fDT CAUSE SRV OPENltG ECAUSE:

(1) SET POINT 150 PSI HIC 4ER, (2) HIGH PRESSUE TRIP AT 2400 PSIA l

0 EVENTS OPENltG SRV ARE COPP00NDED TRAf1SIENTS SUCH AS LOAD REICT10fi WITHOUT STE#1 DUFP OR DIRECT REACTOR TRIP UNDER AIhERSE C0f01T10f6, OR POSTULATED ACCIDENTS SJCH AS LOSS OF P'AIN AND AUXILIARY FW, OR ATWS, OR RCCA EJECT 10f1

,,WHICH ARE FUCH LESS FRE0JEf(T

i DIVERSE MECM'ilS".S FOR PRESSURE CONTROL EC%NISi SET PESSURE ESIGN TRANSIENT KATERS 2250 PSIA NORML OPERATION SPRAY 2275-2325 10% ET LOAD CM!E PORV 2350 50% STEP LOAD REDUCTION EA:T0; TRIP 20]

TOTAL LOAD REJECT 10i S' tTV VALVES 2503 LOAD REJECT 10f1 WO E

If?EDIATE TRIP O

EA:4 E:C 'FF DESIG'E3 TO PREENT OPERATION OF fEXT EC%ilSM FOR ITS ISIGN RA43E 0: TRA"SIENTS C

TrE Ecra',:52 ARE ESIG'ED Wliti REDUhLAlCY ATO m'E SUE.STANTIAL,

OVERCADA:lTY (Ifi STARTUP TEST SPRAY WAS ADEQUATE TO PREVENT PORV OPENIfG FOR 25i LOA 3 EDJ: TION, O'E PORV AIQ'JATE TO PREVENT HIGH PRESSURE TRIP,

~ETC.)

PRESSUR12ER PRESSURE.

PSIG 2410 23g5 l

~

g i

l f

2335 - -

1 I

I I

l t

I i

i

[

l 8

I i

i 2;0 4.0

,C.0 l E.,0 T!P.E. SEC.

..y

(---

e i

i i

7.1 SEC (FAX, PRESSURIZERPRESSURE) 5.7 SEC (HIGH PRESSURE REACTOR TRIP DEMANDED) 3.7SEC(PORVSETFOINT)

Figure 1 PRESSURIZER PRESSURE RESPONSE FOR NET LOAD TRIP FULL POWER WITH NO PRESSURIZER PORVS OPERATIONAL

6 PT. EA04 - FACETY INJECTION PUMP PUFP TYPE:

BYRD!l JkOCSON M) DEL 4x6x9C, 8-STAGE, Ihm PU!P BEA,1fGS:

ANT 1-FRICT10'; RifG OlLED RADIAL AfD THRUST BEARifGS.

CollfG VIA AIR FAfiS LOCATED ON EACH SKACT Ef0 TO PROVIDE BEAO,1fG FDJSifG C0llfG.

PECHANICR Sets STA';DARD EriD FAI RUBBlfG SEALS 0:E SEAL Oii EACH SrR:T END fECWA';1CE SEE COOLERS UTILIZE CCW SEIC. TYPE:

J^E CRA'E Ej<_ TYPE 1 3 IfC.H SEAL L

AF:CT OF LOSS 0; CCW ON FtTY INICTIO'i PU'P C0JLD ltPACT ONLY TFE FEC%'ilCR. SEALS SINE ONLY SEAL CODERS UTILIZE CCW i

SEAL COylRS EASIC4.LY EFPLOYED Ofi PutP APPLICATIONS WHERE FLUID TEtP EX Tr's 16:? FCc, A SJSTAlfED PERIOD OF TlfE JRi CRRd C0!FR;Y TESTS - CRA'E REPORT "SEAL PERFORFW11 TESTlfG FOR 3

fults FJJii x :.n liiJECiiO:i SiSTEfF, Ed.LETiti fD. 342 0

JXi CPA'E TYPE 1 SEALS TESTED FRDM 0-00 PSIG NiD 14>300? WITH TO SE _ CX :'G EE 'd EtF_00 0

FO:,150: TrE PRDICTEL SEAL LIFE W)JLD BE GREATER TE'i 3 YEA?.S CONTlfGJS OPERAil0'i (WO EXTERNAL SEAL COD.ItG)

FO'. INICTION C0'OIT10'G 0: SHMT PERIOD 0: INICTION FROM BEIC ACID STORE TNM ( 170F) FOLL0kED Bf INICTION FROM RWST, OPERATION 0: S1 PU5PS FO: 24 H3JRS WILL fDi RESJLT IN PLYE TA'i EFECTED NDRPAL W%R j

- CONSEQUENCES OF SEAL FAILURE - LEAKAGE, BUT NOT CATASTROPHIC FAILURE OF PUMP (FLOOR DRAINS DESIGNED FOR EXPECTED LEAK RATES:

BACKUP PACKING RING ALSO AVAILABLE)

0 e

FIRE CASE ANALYSIS COMPARIS0N OF NRC CASE STUDY AND EPRI/WOG INITIATING EVENT FREQUENCY DISCUSSION BASIS FOR HALON SYSTEM RELIABILITY

COMPARISON OF NRC CASE STUDY AND EPRl/WOG Plant Design Differences New Batteries Redundant Halon System in AFW Pump Room i

New Source of Halon System Reliability Data l

Differing Application of Fire PRAs for Initiating Event Frequency Human Reliability Analysis Of AFW Operation

(

INITIATING EVENT FREQUENCY Divide Frequency into Component Parts i

Development Similar to Limerick Analysis Similar Totals For Each Room

,_,,,,,,.,,,_n.,_,

l l

DIVIDE FREQUENCY INTO COMPONENT PARTS i

Frequency for a Room Location Within the Room i

Intensity i

DEVELOPMENT SIMILAR TO LIMERICK ANALYSIS Cable Trey Fires Electrical Panel Fires Laroc Transient Combustible Fires i

l

SIMILAR TOTALS FOR EACH ROOM..

30% of NRC Case Study for AFW Pump Room 70% of NRC Case Study for Switchgear Room l

i

)

e

i l

HALON SYSTEM RELIABILITY Source - Summary of Fire Protection Programs of USDOE, Calendar Year 1986 Uses Actual Fire Experience Recent Fire Experience Versus EPRl/WOG Assessment Comparison to Millstone 3 PRA Data Source i

l

I NSAC-113 SW PUMP HOUSE FLOODING EVALUATION o THOMAS PIPE RUPTURE CORRELATION IS BASED ON LEAK BEFORE BREAK CONCEPT Pc = [;. PL = };. P. QE

}{. P.

[0P + A0w]. B. F. S. E

=

WHERE Pc = PROBABILITY OF RUPTURE Pc/PL = FRACTION OF LEAKS THAT RESULT IN RUPTURES P = PROBABILITY OF LEAKAGE PER QE-YR Ot = SIZE AND SHAPE FACTOR FOR PARENT MATERIAL AND WELDS QP = SIZE AND SHAPE FACTOR FOR PARENT MATERIAL Qw = SIZE AND SHAPE FACTOR FOR WELD MATERIAL A = WELD PENALTY FACTOR (=50)

B = SYSTEM DESIGN AGE FACTOR F = PLANT AGE FACTOR S = MODIFIER FOR QUALITY OF MATERIALS E = MODIFIER FOR SPECIFICS OF STRESS AND FATIGUE

i RANGES AND RECOMMENDED VALUES GIVEN BY THOMAS:

0.05<Pc/PL<0.10, Pc/PL~0.06 1E-9<P<1E-7/QE-YR P-1E-8/QE-YR A=50 3 lei B-1 BEYOND 10 YEARS F-1.7 EXTRAPOLATED TO 40 YEARS SIZE AND SHAPE FACTORS QP =.

P Qw = Q Dw lu = 3.75 TP (DEFINED BY THOMAS)

Dw ~ DP TW ~ TP l

I HENCE Qw = 1.75 >

Dw Tw WHERE LP = PIPE LENGTH Lw = WELD LENGTH DP = DIAMETER OF PIPE Dw = DIAMETER OF WELD TP = THICKNESS OF PIPE Tw = THICKNESS OF WELD N = NUMBER OF WELDS FORMULATION IN NSAC-113 Pc = P (};.)

[QP + A*S*Qw] B*F NOTE THAT QP SHOULD BE MULTIPLIED BY S BUT FOR S~1, AS IS THE CASE FOR "AVERAGE" COMMERCIAL GRADL FIPE, THERE IS NO LOSS OF GENERALITY AS SHOWN.

POINT BEACH 10-IN. DIAMETER, CARBON STEEL FIRE MAIN AT 125 PSIG.

A BREAK IN THE 3-FT. CENTER OF THE PIPE SPAN ABOUT THE T-JUNCTION WITH THOMAS NORMAL VALUES AND F~2 4:

Pc = 1E-8 (0.06) [1440 + 50 (1)(70)]

(2) (1)

Pc = 5.93E-6 P

IF LEAK IS SUFFICIENT TO DAMAGE SW PUMPS, LET Pc/PL ~ 1.0 THEN Pc = 9.88E-5 4

p

WEPCO COST ESTIMATE FOR INTAKE STRUCTURE SHIELD WALL EXTENSION (SPRAY) MODIFICATION ARCHITECT / ENGINEER COST:

DESIGN - 500 MAN-HOURS AT $110/HR

$ 55,000 INSTALLATION / REMOVAL-630 MAN-HOURS AT $110/HR 69,000

$124,000 IN:LuDES: 1. 0A DESIGN CALCULATIONS FOR SEISMIC CLASS } STRUCTURE

2. INDEPENDENT REVIEW OF CALCULATIONS

3. ONSITE INSPECTION AFTER INITIAL DESIGN

4. FLOOR LOADING ANALYSIS

5. INSTALLATION PROCEDURES 6

CONSTRUCTION DRAWINGS / SPECIFICATIONS

7. WORK PLANNING INCLUDING SEQUENCE

{

8. REVISION OF DESIGN DURING CONSTRUCTION

{

9. FINAL AS-BulLT DRAWINGS i

MATERIAL:

STRUCTURAL STEEL /0THER CONSTRUCTION MATERIALS $ 32,000 (h0TE: NON-STANDARD STEEL REQUIRED-20FT STANDARD)

WEPCO COST:

400 MAN-HOURS AT $50/ HOUR f 20,000 WRITING SPECIFICATIONS, PURCHASE ORDERS, ENGINEERING AND SUPERVISORY REV15W

CONTRACTOR COST:

REMOVAL - 1000 MAN-HOURS AT $22.50/ HOUR

$ 22,500 INSTALLATION - 4000 MAN-HOURS AT $22.50/ HOUR 90,000

$112,500 INCLUDES: 1. 2 DAYS OF PLANT ACCESS TRAINING FOR 8 PEOPLE

2. 1 DAY OF WELDER QUALIFICATION FOR 2 PEOPLE

3. AWS D1.1 WELDING COMPLIANCE (REQUIRED BY WISCONSIN)

4. INTERFERENCE BY ERECTING NEW WALL PRIOR TO REMOVING EXISTING WALL (OTHERWISE 2-UNIT SHUTDOWN REQUIRED)

5. REWORK FOR f! ELD INTERFERENCE WITH DEAD IlME FOR ENGINEERING DESIGN AND REVIEW

6. SETUP AND IEARDOWN IlME TOTAL COST ESTIMATE

$288,000*

A 2-UNIT SHUTDOWN MAY BE REQUIRED FOR THIS MODIFICATION AT ABOUT $450,000/ DAY FOR REPLACEMENT POWER

.w

WEPCO COST ESTIMATE FOR

'j INSTALLATION OF DIE 5EL-DRIVEN AUXILIARY FEEDWATER PUMP (INTERNAL 9)

WEPCO ESTIMATE:

$ 18,000,000 INCLUDES: 1. IWO DIESEL DRIVEN AFW PUMPS (ONE/ UNIT)

2. ASSOCIATED INSTRUMENTATION AND CONTROLS, STARTING SYSTEM, COOLING SYSTiM, FUEL OIL SYSTEM

3. tie-IN TO EXISTING AFW SUCTION AND DISCHARGE LINES WITH SEISMIC CLASS I PIPING

4. CONSTRUCTION OF SEISMIC CLASS I AND IORNADO-MISSILE RESISTANT DESIGN BulLDING 8 ASIS:

1. A Two-LOOP W PLANT INSTALLED TWO ADDITIONAL MOTOR ~ DRIVEN AFW PUMPS IN ExtSTING SEISMIC Bu!LDING IN 1979.

COST WAS $16,000,000

2. WEPCO RECENTLY ESTIMATED COST OF INSTALLATION 0F 3RD EMERGENCY DIESEL GENERATOR IN A SEISMIC CLASS I, IORNADO-MISSILE RESISTANT BUILDING f

FOR ~ $7,000,000

~

I

3. ANOTHER IWO-LOOP )! PLANT IS INSTALLING IWO NEW EMERGENCY DIESEL GENERATORS IN A NEW BUILDING - ESTIMATED COST IS ABOUT $20,000,000

[

4 r

RECOVERY ANALYSIS e

i

.s Methodology 3

Basis for Quantification t

i Example - Manual Operation of TD AFW Pump i

N 4

NRC CASE STUDY METHODOLOGY I

DETERMINE TIME AVAILABLE IDENTIFY RECOVERABLE FAILURES QUANTIFY NON-REC 0VERY EVENT REQUANTIFY DOMINANT ACCIDENT SEQUENCES u

r P

METHODOLOGY t

'N' 1

e f

y/

Identify Potential Recovery Actions Operator interviews Past PRAs i

Consider Feasibility for Accident Scenarios Dasatification Cut Set Review and Implernentation i

Verification by Walkdown l

i l

f 1

BASIS FOR QUANTIFICATION d

i l

l l

Simulator Data 4

I Comparison of EPRl/WOG and Other Studies

\\

I s

4 I

l 1

1 l

l 1

1 f

i j

1 l;

/

.i

)

1 1

i

DATA FOR CREW RESPONSES Taken from the LaSalle Nuclear Power Plant Simulator i

f 1

10

.,0...

i 4

.c k

C.1 O.

0.05 8

i 11

=

c.01 I

j L

2

< 1,3,

,o B

k 10 0

,,5 i c

\\

0.01 0.1 1.0 10 100 Minutes j

i i

I

TRCS IN COMPARISON WITH SIMULATOR DATA E

i i

l

.x

'= 0.5

. ~

j

'M TRCs with hesitation

\\

co s._..

g

r. un~y..

- mysan l

\\

. % s... w +.. -

s..#,.

.f wi,

1 T L,.

~ ':s.T"

^'

Q.

i

~'*cw w,e..

I

,,x.

C i Sim ula to r._~tz.'. %'.

'r;:;m i

~

_.. ~ ~,g._

.x.

.'.xm curves w

_3 m

Sme w.e + - w v.v-u 4 mv3,%,0..yv. % e. v y,,we e. --

EN. 3

'N '

42 w.

y.

l l

1

($if'n.O~iTRCs without

~~ #

7:

m_wgu,<..x. w +.y 1E 6

' * *

• _ " ' ~ + ~

- ' +*^'--~,s._~,

.n i

Available Time, min l

,_,,1.w..-,w-ges-3--=w*T

-~""'"'Jw'W-*

NUMBER COMPARISON I

EPRI/WOGi ASEP:

Time SL1=0.5 SL1=0.7

mir, no H' H

no II H

lower upper lo 0.01 0.1 15 0.006 0.05 0.002 0.02 (0.003)

(0.03) 20 0.001 0.01 30 0.0003 0.01 0.00006 0.006 0.0001 0.001, l

Page 19. NS AC/l13.

2 L ii1CR-4772, A. D. Swain, ASEP HRA Procedurc, Sandia National Laboratories, February 1987.

H stands for hesitation, or some source decisional burden.

3 1

NOTES 1.

ASEP includes another TRC; SAIC TRC system another two.

2.

match between two systems is not qualitatively accurate.

1 3.

the distinction between ASEP curves is not well-specified.

4.

ASEP does not consider decision making significant.

5.

ASEP does not, thus, consider hesitation or burden.

EXAMPLE - MANUAL OPERATION OF TD AFW PUMP

~

Two Actions Manual Start Alternate Water Supply Manual Start Fire in AFW Pump Room Fire in 4160 Volt Switchgear Room Alternate Water Supply Long-Term Station Blackout i

Seismically induced CST Failure

r A

FEED AND BLEED ISSUES c

Factors That Enhance Reliability Detract From Reliability cue is simple goal conflict action set is simple competing recovery action in control room uncertainty in phenomena easily simulable unanticipatable system status incredulity in losing all feedwater i

Assessed Failure Probabilities Sandia 0.003 SAIC TRC System 0.05-0.000006 NUREG/CR-1278 (20-60 min) 0.01-0.0001 Other PRAs 0.05 0.0001

'?.....,.. - Ov :rs' Group 0.01 i

i Note Both hesitation and possible over use of feed and bleed have been noted in different PWRs.

\\

J

)

ECCS RECIRCULATION ISSUES o

Factors That Enhance Reliability Detract From Reliability cue is simple step order is significant action set is simple multiple mission (flow & cooling) action in control room never performed i

prior success in injection DHR can avoid recire in some cases procedure has contingencies easily simulable i

l i

Assessed Failure Probabilities Sandia 0.003/0.001 SAIC TRC System 0.05 0.000006 l

NUREG/CR-1278 (20-60 min) 0.01-0.0001 Other PRAs 0.003-0.0001 Note No evidence of conflict or unusual burden associated with '

recirculation-should be little decisional stress

OTHER KEY REC 0VERY ACTIONS STARTING EDGS ON LOSS OF STATION BATTERIES (SEISMIC AND SB0)

DC CROSS CONNECT ALTERNATE SOURCE TO RWST (SEISMIC AND RECIRCULATION RECOVERY)

COMP 0NENT COOLING CROSS CONNECT

)

COMMENTS FOR CONSIDERATION IN NRC'S A-45 REGULATORY ANALYSIS BASED ON NUMARC REVIEW LETTER /NSAC-113

'~

1.

DEDICATED SDHR FAILS ALL COST BENEFIT MEASURES BY A WIDE HARGIN 2.

MANY OF THE REASONS FOR HUCH LOWER RISK AT PB GENERALLY APPLY TO OTHER CASE STUDY PLANTS.

3.

WITH POSSIBLE EXCEPTION OF SEISMIC, ALL OTHER "EXTERNAL RISK" FACTORS AS ANALYZED BY A-45 CASE STUDIES SHOWN TO BE INSIGNIFICANT.

4.

A-45 HAS NOT YET ACCOUNTED FOR SAFETY IMPROVEMENTS FROM OTHER NRC AND INDUSTRY PROGRAMS (IH PARTICULAR, SB0 AND SEISMIC MARGINS).

5.

BEST-ESTIMATE ANALYSIS IS ESSENTIAL FOR CREDIBLE USEFUL RESULTS.

ANY ADDITIONAL MARGIN (IF NEEDED) SHOULD BE ADDED AT EHD OF ANALYSIS.

IMPORTANT LESSON FOR IPE PROCESS.

6.

U.S. NUCLEAR POWER PLANT OPERATING EXPERIENCE IS THE BEST SOURCE OF CREDIBLE DATA FOR BEST-ESTIMATE ANALYSIS, AND BEST A ION Fog ngEFINING THE PROBLEM."

gg P

e I

}

3 A-4S RESOLUTION VIA OTHER PROGRAMS MAJOR SOURCES OF RISK IN PROGRAMS ADDRESSING THESE SANDIA STUDIES RISK SOURCES STATION BLACK 0UT A-44, NUMARC CONTAINMENT REllABILITY, NUMARC, IDCOR, NUREG-ll50 FAILURE MODES, CONSEQUENCES PWR AFW REllABillTY, BWR INPO, NSSS OGs H.P. CORE COOLING SBLOCA EPRI, INPO, NSSS OGs PCS LOSSES (SCRAMS)

INPO, NSSS OGs, NSAC, AEOD DE F0WEP RE'.1 AE!LITY NSAC, INP0 VALVE PERFORMANEE EPRI, INPO, AE0D FIRE PROTECTION APP. R, INP0 INTERNAL FLOODlNG, LIGHTNING NSAC, INPO, A-17 SEISMIC RISK EPRI, SQUG, A-46, A-17 l

SAB0TAGE UTILITY SECURITY PROGRAMS, FITNESS FOR DUTY RULE 3895NS7

GIIGCUdENI C 1911 "EPRI/WOG ANALYSIS OF DECAY HEAT REMOVAL RISK AT POINT BEACH", NSAC/113, March, 1988.

This report was prepared by Scienca Applications Inter national Corporation and Westinghouse Electric Ccrporation.

The report is copyrighted and so cannot be processed into the NRC document control center withou* written permission from the copyright holder.

This was not known to the autho-of this meeting summary in time to request and obtain the necessary permission.

Therefore a copy cannot be included here.

The author is sorry for the inconvenience.

However, the report can be ordered from Research Reports Center (RRC)

Box 50490 Palo Alto, CA 94303 Phone (415) 965-4081 2

l

oc te e >-

d' EPRI Ele tne Powe' Resea'ch InstiMe March 30, 1988 Ms. Elaine Gorham-Bergeron Division 6413 Sandia National Laboratories P.O. Box 5800 Albuquerque, NM 87185

Reference:

Letter from E. Gorham-Bergeron, SNLA to G. Vine, NUMARC dated March 8, 1988

Dear Ms. Bergeron:

Your recent request for additional information regarding certain details of the EPRI/WOG analysis of DHR at Point Beach was provided to me by Gary Vine.

The enclosed material is provided in response to that request.

Although the time available to us did not permit the preparation of a full response to all of your questions, each of the topics identified in your letter will be addressed at the NUMARC-NRC meeting on March 31, 1988.

Our presentation material at that meeting will constitute our responses to the several open items identified in the enclosure.

With regard to the application of SAIC's Risk Management Query System (RMQS) in the EPRI/WOG study, we previously extended an invitation through Ken Adams (SNLA) that you, Ken and Dave Ericson (ERCI) visit Bill Parkinson at SAIC's Los Altos, CA office.

Although a meeting could not be scheduled prior to the March 31, 1988 review meeting, due to constraints on Dave Ericson's availability, we would be pleased to reschedule a meeting at your convenience.

Sincerely yours, um 4

John

. Haugh Project Manager Nuclear Safety Analysis Center Nuclear Power Division JJH/jph 3970NS8

Enclosure:

"Responses to SNL Questions on EPRI/WOG Analysis of DHR at Point Beach" 34 2 H NeA AveNo. P 51 C".;e B:n 10412 Pa o A1: CA 94M3 Te'epoae (415) ti$ 2003 wtsegm CW<e 1019 h notee-tm St'ett. M Sa te 13X ansvg's DC 2xn (222 6 72 g222

Ms. Elaino Gorhtm-B2rgsron March 30, 1988 Page 2 1

cc:

R. Woods, NRC/RES K. Adams, SNLA/6413 D. Ericson, Jr., ERCI G. Neils, NSP (Chairman, NUMARC Working Group on DHR)

R. Newton, WEP (Chairnan, Ucstinghouse Owners Group)

A. Ladieu, YAEC (Chairman, WOG Analysis Subcommittee)

W. Andrews, Southern Co. Services (Vice Chairman, WOG Analysis Subcommittee)

W. Parkinson, SAIC W. Layman, EPRI l

T. Marston, EPRI l

C. Stepp, EPRI G. Vine, NUMARC l

l l

l i

I l

l l

l

r

?

ENCLOSURE Responses to SNL Questions on EPRI/WOG Analysis of DHR at Point Beach March 30, 1988 l

I

s I.

Plant Systems I.1 The contention the HPI does not require CCW for cooling of bearings.

Is there engineering data to support this contention? How long can HPI operate without cooling.

Response

i 1.

Pump Description a.

SI Pump Type:

Byron Jackson Model 4x6x9C, 8-stage, DVMX B-J Drawing No. 2E-2002, Rev. B b.

Pump bearings:

anti-friction ring oiled radial and thrust bearings. Cooling is via air fans located on each shaft end to provide bearing housing cooling.

c.

Mechanical seals:

standard end face rubbing seals. One seal on each shaft end. Mechanical seal coolers are employed which utilize component cooling water.

John Crane Seal Type 1-3 in. seal Crane Drawing No. F-SP-13257 2.

Loss of CCW Event Affect i

The loss of CCW event is assumed to be applicable to the injection phase of the accident and not the recirculation phase such that the pumped water from the RWST typically is about 100'F. Note that the SI pumps initially take suction from the boric acid storage tanks which are maintained at approximately 170*F. This water volume is small (<5000 gals) so that the time the SI pumps are subjected to this temperature is short. The loss of CCW event will only impact on the pump mechanical seals since only the seal coolers' utilize CCW.

Ho.ever, seal coolers are basically employed on pump applications where the fluid temperature exceeds 160'F for a sustained period of f

1.1-1

time. When temperatures are below 160*F, seal cooling has little '

effect on increasing seal life.

John Crane Seal Co. performed a series of tests in the late 1960's to test the performance and estimate the life of mechanical seals used in nuclear applications. These tests were documented in a Crane Report, entitled, "Seal Performance Testing for Nuclear Power Plant Safety Injection Systems," Bulletin No. 3472. John Crane Type 1 seal tests were performed on seals at pressures from 0-400 psig, and temperatures of 140-300'F. with no external seal cooling being employed. Seal life and performance is a function of temperature and pressure. The test results show that for a defined "normal condition" of 160 degrees F and 400 psig the projected seal life would be greater than 3 years of continuous operation (without external seal cooling).

Operation of the SI pumps without seal cooling for 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> will not result in more than expected normal wear for this seal under the maximum temperature and pressure conditions stated, lastly, although no cooling water is circulating through the seal coolers, the pumped water circulating through these coolers still obtains some cooling by virtue of the fact that the pumped water in the seal chamber is now circulating through an air to water heat exchanger.

t I

I.1-2 i

Where are I.2 Capabilities of new batteries and details of installation.

these new batteries located? What systems are served? Are these batteries "on-line" like the station batteries or are they isolated?

The information requested will be presented at the March 31

Response

1988 meeting.

l l

i t

l l

1.2-1 m

How are these Venting capability option as alternative to one PORV.

I.3 vents actuated?

Although the Venting capability was not credited in NSAC-113.

Response

DC vent valves appear sufficiently large to compensate for a failed PORV The small size of the or block valve, the lines are orificed downstream.

orifice precluded the crediting of the vent valves as an alt (rnative for The feed and bleed without more detailed thermal hydraulics analysis.

vent valves are actuated by DC power. The configuration of the vent

+

paths is such that either train of DC power can be used.

It should be noted that significant conservatism exists in the thermal hydraulics analysis which indicates that both PORVs are required for feed It is believed that the additional DC vent capacity could be and bleed.

The most significant beneficial when combined with these conservatisms.

impact would be to eliminate feed and bleed failures caused by loss of DC bus initiating events and those events where a single PORV or block valve fails to open.

1.3-1

11. Failure Data i

11.1 Halen system reliability.

Is there a more complete reference than the Where were the data collected 7 What is personnel communication cited 7the justification for applying it to comm The reference for v.alon system reliability is Summary of

Response

This Fire Protection Programs of USDOE Calendar year 1986, pp. 29-32.

report is updated annually.

The reference indicated that for 17 fires in which automatic Halen The systems were involved, all fires were extinguished.

suppressior:

EPRl/WOG study conservatively assumed one failure thereby yielding a failure probability of 1/17.

In a follow-up phone call with Mr. Walter Maybee of DOE (301-353-5609),

Complete we have learned that 2 failures have occurred recently.

documentation is not yet available to DOE, but will be included in the 1957 update of the above reference.

(Publication will be some months since annual fire reports from individual DOE facilities are not due until April 15.) Consequently, information on recent successes is not documented and no new failure probability could be generated.

Of the two reported failures, one failure occurred at Richland and one The Richland failure did include failure of the syste-at Brookhaven.

The Brookhaven failure included failure of to automatically initiate.

the system to manually initiate. The automatic system was determined to be operable.

(The fire did not reach sufficient intensity to initiate the automatic system.)

Because of the nature of the Brookhaven f ailure and the lack of cred given in the EPRI/WOG study to manual initiation of Halen, the Halon system reliability estimate pre:ented in NSAC-113, i.e., 1/17, remains Including the second fault in the data base, would double reasonable.

Halon unreliability and increase fire risk by a f actor cf 4 in the EPRl/WOG estimate (since redundant systems are credited).

!!.1-1

Through discussions cith Mro Maybee, we were able also to explain the apparent difference between these Halon reliability estimates and those quoted in the Hillstone 3 PRA. The American Nuclear Insurers, the source for the Halon reliability estimates in the Millstone 3 FRA, generally quote Halon reliability estimates using acceptance test data rather than actual experience in extinguishing fires.

Acceptance test data is not an adequate basis for predictinj Halon system reliability in the event of a fire at a nuclear facility for two reasons. First, an acceptance test is part of the design checkout phase of system design and implementation.

If the Halon system fails to meet its acceptance test, the system is modified and retested until the required concentrations are delivered and maintained for the required time interval. Usually an acceptance test failure is a small variation from the criteria and only minor modifications are required.

Second, the acceptance test criteria are conservative. Mr. Maybee noted that whereas most fires are extinguished (according to research data) by a 3% concentration, the acceptance tests generally require a 5% concentration to be held for 10 minutes. Further, the experience quoted in the above reference indicates that Halen systems are more capable than their design bases suggest.

In one case a Halen system put out a so-called "deep seated" fire, e.g., a fire starting at the bottom of a trash container. According to Mr. Maybee the research data suggests that Halen would not have pit out such a fire, in conclusion, the Halon system reliability estimate quoted in NSAC-113 is reasonable for automatic system operation. Furthermore, that estimate appears to be more appropriate than the estimate provided in t

the Hillstone 3 PRA.

l

!!.1-2 I

G 9

11.2 Diesel generator f ailure rates. What run times are assumed /used in the data base?

The diesel generator failure probabilities used in the

Response

EPRI/WOG study were taken directly from industry-wide experience collected for NSAC-108. These f ailure probabilities are detemined on a mission basis where the mission of the diesel is to respond to an ur. planned event or to meet the requirements of a monthly or annual No specific f ailure rate, i.e., failures per hour, is calculated test.

in NSAC-108; consequently, no specific run time is used in the EPRI/WD3 analysis.

Before applying the NSAC-108 data, a judgment was made as to its applicability to a PRA analysis.

It was judged that the EPRI/WD3 mission was similar enough to the mission defined in NSAC-108 to justify its application in NSAC-113.

NSAC-108 describes certain criteria for the "load run" portion of the test and for unplanned event data.

If in a test or unplanned event, the diesel generator ran for significantly less than one hour, the In the typical monthly event was not counted as a "load run" success.

test, the largest portion of the data base, the diesel is run for at In the annual test, about 1% of the data base, the least one hour.

diesel is run typically for 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />, it is likely that the average mission of nuclear plant diesel generators is greater than one hour.

Point Beach specific information submitted to EPRI for NSAC-108 indicates an average run time of 3.3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> per start based on 1158.7 hours8.101852e-5 days <br />0.00194 hours <br />1.157407e-5 weeks <br />2.6635e-6 months <br /> accumulated in 347 starts.

l The NRC Case Study assumed a mission time for the diesels of 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br />.

This assumption is in conflict with the assumption inherent in the However, as mentioned in NSAC-113, a Level 1 PRA EPRI/WO3 data base.

would generally consider a time dependent analysis for diesel generator A time dependent analysis would indicate the run f ault sequences.

average necessary mission time for the diesels.

11.2-1

Such an analysis in the Florida Power Corporation Crystal River Unit 3 PRA found that af ter about 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />, continued successful diesel operation was risk insignificant. This finding occurs because af ter about two hours the chance of recovering offsite power increases substantially and the importance of continued diesel operation decreases proportionately.

The NRC Case Study did not give credit for offsite power recovery after 30 minutes, even for the so-called "long term station blackout" This conservatism, together with the 8 hour9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br /> run time, sequences.

This level of significantly influences the NRC Case Study results.

conservatism is not consistent with the state of the art in Level 1 The EPRl/WDG analysis provides a more realistic basis for diesel PRAs.

generator mission time. Based on the Point Beach plant specific mission times, the NSAC-10B data base appears to be consistent with that basis.

l 4

i I

11.2-2 i

11.3 PORV block valve positicning data. What is the source of this data?

How as it substantiated? Are records kept?

The NRC Case Study assumed the following probabilities for

Response

block valve position:

both closed 1.0 one closed, one open 0.0 0.0 both open No basis was provided for this assessment.

In contrast, the EPRI/WO3 study used the following estimates of block valve position:

both closed 0.01 '

one closed, one open 0.50 0.49 both open As This assignment was based on initial estimates offered by WEP.

discussed eelow, the safety impact of these numbers was judged to be relatively insignificant in the EPRI/W3G model.

In an atterpt to veelfy the accuracy of those initial estimates, WEP recently conducted an in-depth survey of Point Beach Unit 1 PORV block valve positioning data for the entire year 1986.

In addition, a more limited review covering only the last quarter of 1987 was conducted for both Unit 1 and Unit 2.

This involved tallying the changes in PORV block valve positions as recorded in the shift 109 books.

(The positions of the block valves are recorded once per shif t.)

The percentage of time that the block valves were in each position during the periods investigated was as follows:

September - December 1987. Unit 1:

both closed 0.0 one closed, one open 0.0 II.3-1 4

6

-~-

both open 1.0 t

September - December 1987. Unit 2:

'I l

both closed 0.18 one closed, one open 0.02 t

both open 0.80 Full year 1986 Unit 1:

4 l

i both closed 0.29 one closed, one open 0.45 both open 0.26 i

,j Total for above periods, both units:

j both closed 0.22 i

one closed, one open 0.30 l

both open 0.48 The Unit 1 valve position data for the last quarter of 1987 clearly is in conformance with the initial estimate for both block valves being closed. The Unit 2 data for the same time frame and the Unit 1 data for 1986, ho.ever, suggest a much higher incidence of both valves being i

clese' then was previously estimated. Taken in the aggregate, these data would suggest a frequency of 22% for both units during the time periods investigated. Although this value is considerably greater than the 1% estimate used in NSAC-113, it remains substantially less than the 100% estimate used in the NRC Case Study.

i Given that the plant datt suggests that the frequency of both block valves being closed may be higher than previously thought, it is appropriate to re-examine the relative risk significance of this f

j parameter. As applied to Point Beach, the block valve positioning data l

can enter the DHR risk evaluation in three instances. These include situations where:

i j

!!.3-2 l

l i

a closed block valve prevents its PORV from opening and sticking o

open, both block valves being closed leaves the SRVs as the only pressure relief devices, and any closed block valves syst be opened for feed and bleed.

Each of these situations is examined in turn.

if either PORV block valve is closed, that PORV cannot open spuriously to cause a LOCA, not can it stick open after a loss of offsite power The EPRI/WOG study transient, or a loss of main feedwater transient.

did n:t credit this beneficial effect of either or both PORV block valves being closed.

If both PORV block valves are closed, it appears that the NRC Case Study considered that the SRVs might be susceptible to opening during'a Based on experience with previous primary system pressure transient.

PRAs and thermal hydraulics analyses, the EPRl/WOG study assumed that the SRVs wesid not open on loss o' offsite power or loss of main Hence, the change in block valve positioning data has no feedaater.

ircact on the transient induced LOCA frequency calculated in NSAC-113.

J lf any block valve is closed at event initiation and feed and bleed is l

required, the block valve or valves must be opened for feed and bleed to f

The EPRl/WD3 analysis included the ef fect of PORV block valve I

succeed.

The effect of the p:sition in the f ailure probability for the valve.

new PORV positioning data is being evaluated and will be presented in the meeting.

l

\\

11.3-3 I

J

?

m

!!!. Analytical Methodo13 y i

III.1 SAIC's Risk Mangement Query System (RMQS). What are the constraints and limitations of this system?

Is this sof tware generally available? What documentation is available?

The data bases RMQS contains a series of linked data bases.

Response

include:

initiating events accident sequence cut sets 4

system or super component cut sets components Component types risk measures RMQS Each data base contains a numerical estimate and descriptors.

allows the user to change any entry and determine its propagating For example, the reliability of a diesel generator can be effect.

changed and the impact on core melt frequency can be determined.

t The constraints and limitations of RMOS are p*edo?.inantly affected by its application by the analyst.

For exam;1e, the user may also add By adding events to the accident sequence cut events to the data base.

set, recovery can be credited.

In this activity, the user must be careful to avoid adding a recovery which would be inapplicable due to This limitation is true regardless of other failures in the cut set.

how recovery is applied, i.e., whether done for the NRC Case Study or for NSAC-113.

I Another constraint is the ability to load the model correctly even before recovery is assessed. Of ten PRAs are not as traceable as even the authors would prefer. Semetimes inconsistencies are identified.

As mentioned in NSAC-113 the Point Beach model in RMQS was ben These results differed by only a against the kRC Case Study results.

l few percent from the total core melt frequency, j

l I

i 1!!.1-1

RMQS can be purchased from SAIC subje:t to license agreement. The code has been subject to a Quality Assurance progras and has passed two l

audits / surveys by utilities. Like the code, documentation such as the RHQS Users Manuti is proprietary.

I i[

l 1

l I

I I

I i

6

!!!.1-2 9

e w----

, [

-,.<w

-,-r--

e---

---

t--w

• i k

. t PORV performance and recirculation for small LOCAs. What' III.2.

analysis / experiments support assumptions about PORVs performance?

Is there documented analysis to support the claim that recirculation will not be required for small LOCA?

Response (a): Openinc of pressurizer safety valves during powet operation is a very rare event on Westinghouse PWks.

The reasoas are (1) Westinghouse used a very conservative approach for primary pressurerelief;and(2)transientsrequiring're1jefaremuchless frequent than commonly assumed. Basedonenginadingjudgmentand estimates of knowledgeable Westinghouse engineers, we egtimate the rate of pressurizer safety valve opening (durirg) power operation) to This is in sharp be roughly 0.01 safety valve openings / reactor year.

contrast to the NRC Case Study assumption of approximately 0.5 safety valve openings per reactor year due to an assumed 7 pressure transients per reactor year with a 7% chance of opening the safety valves in each. This difference of a factor of 50 can be very important in arriving at an unisolatable small LOCA frequency (should a safety valve fail to reclose there are no back-up block valves as in the case of the PORVs) and is due to several causes as follows:

The NRC Case Study assumes 0.5 transients per year (7 transients

)

per year times 7% probability) reaching the PORV setpoint.

Westinghouse survey and analysis results indicate about 0.23 pre TMI and 0.12 post TMI PORV openings per reactor pear from an operational transient.

The NRC Case Study assumes Point Beach PORVs are both blocked 100%

of the time.

Point Beach plant records indicate both are blocked aoproximately 20 percent of the time on average.

The NRC Case Study assumes that a transient reaching the PORV setpoint would rehch the safety valve setpoint if the PORV block valves are closed. The vast majority of such transients would still not reach the safety valve setpoint because the safety valve setpcAnt is much higher (2500 ?sia compared to the PORV 2350 psi setpoint). Additionally a hign pressure reactor trip would occur i

111.2-1 y'

i w

y c.----r

---

n

-v-

=+

--v-

.-T-

~ - -- - - ~*

t

at 2400 psia to relieve the mismatch if there had not been a direct or earlier reactor trip. This is supported by the infrequency of safety valve openings in practice.

The Westinghouse conservative approach to pressure relief consists of diversity, redundancy, and overcapacity.

Diversity: Four mechanisms are provided to control rapid increases in pressure: Spray; PORVs, Reactor Trip; and Safety Valves.

(Normal pressure control is provided by regulating pressurizer heaters. Only fairly rapid changes in pressure require spray.)

Each of these four mechanisms is designed to prevent operation of the next mechanism for its design range of transients. The follo<ing set points illustrate this point:

Mechanism Set Pressure Design Transient Heaters 2250 psia Normal operation Spray 2275-2325 10% step load change PORY 2350 50% step load reduction Reactor Trip 2400 Total load rejection Safety Valves 2500 Load rejection without immediate trip Recundancy: Two (or three) spray valves, PORVs, reactor trip

=

circuits, and safety valves are provided.

Overcapacity:

Plant startup tests at Mihama unit 1 (1971) drAonstrated the very conservative sizing basis. One test performed a 25% step load reduction (versus a 10% step used in the designbasis): Spray was adequate to prevent PORV opening.

Another test performed a 100% step load reduction (versus 50% in thedesignbasis): One PORV (of the two installed) proved more than adequate to prevent high pressure reactor trip (it cycled open and shut several times, requir:ng about half of its full relieving capacity.)

III.2-2 y_

,-c e---g-

9 9

In addition, large load rejections or other transients that would require pressure relief occur infrequently. Normal load changes are 1% per minute or less. Large load changes requiring pressure relief occur only as a test or a fault (on the power grid).

Following the THI event, Westinghouse made a survey of operating plants. The results showed, that as a result of these reasons, PORV operation due to at-power transients occurred at a frequency of 0.12 per reactor-year.

(Another approximate 0.1 PORV openings /R-Y was the result of testing or other I&C error or because spray had been blocked, and a third approximate 0.1 PORV opening /R-Y occurred at cold shutdown.

(See Reference 1.)

j Even if PORVs are not available (blocked out), sa/ety valves would j

seldom open. The reactor trip (set 100 psi below the safety valves) typically begins reducing core power within one second of pressure reaching the set point. Thus, only the most extreme transient can f

reach the safety valve set pressure.

Indeed, one cost-saving idea was once submitted to eliminate the safety valves as unnecessary for overpressure protection.

(The idea was supported by analyses showing The no overpressurization for any PWR Condition 2 design basis event.

idea was rejected since safety valves are desirable for hypothetical ew.ints such as control rod ejection and ATWT, and also Westinghouse's desire to preserve its conservative design basis.)

An informal survey was made of experienced Westinghouse engineers knowledgeable of PWR operating experience. They could recall only one instance of a transient opening a pressurizer safety valve.

(Atotal turbine load rejection without steam dump or direct reactor trip under adverse conditions.)

Humerous PRAs, including those by Westinghouse, have assumed higher frequencies than the values estimated above. This is largely because safety valve openings are generally not part of the dominant contributors to risk. Therefore, over-estimation is conservative, but III.2-3

h does not cause a significant increase in the total risk. 'Other PRAs have not considered it worth the affort to prove safety valve openings are less frequent than assumed from a simplistic model.

One such simplistic model, for instance, is NUREG/CR-1363 (Reference

2) Statistics derived from Licensee' Event Reports _(LERs).' This reference tabulates both safety' valve "demands" and "failures".

However, essentially all the "demands".are safety valve test (to check the opening pressure), and the "failures" are failures to open within a prescribed tolerance of the intended set pmssure (e.g., 2485 psig 1 25 psi).

Another example of such cost-effective conservatism is WCAP-9804 (Reference 3). This reference lists all design transients and assigns to each a frequency of occurrence and a probability of. safety valve opening.

(Both the frequency and the probability were conservatively over-estimated.) The sum of the probabilities for safety valve opening for all individual transients suggests a total frequency cf safety valve opening of about 0.1/R-Y, even though the same refererce reviews PWR operational experience and "concludes that no operational openings or failures of pressurizer safety valves have occurred domestically during approximately 181 reactor years of operation and specifically, 2,9493,324 hours0.00375 days <br />0.09 hours <br />5.357143e-4 weeks <br />1.23282e-4 months <br /> of safety valve operation."

i In conclusion, the estimated frequency in which operational transients cause PORVs and pressurizer safety valves to open are of the order of 0.1 and 0.01 per reactor year respectively. Most transients could not cause PORV opening, e.g., as transients with direct or anticipatory reactor and turbine trip. Transients which could open the PORVs are those tyhich exhibit a strong mismatch between primary power generation and secondary neat removal (loss of steam load, MSIV closure), reduced j

cooling transients locked RCP rotor), reactivity addition transients causing core power increase (rod. withdrawal) and RCS mass addition transients (SI or steam break with SI if the plant has a high head design SI system) all of which are relatively infrequent. Transients which could open safety valves are compounded transients such as load i

Ill.2-4

)

s

)

rejection without steam dump or direct reactor trip under adverse conditions, or loss of main an' auxiliary feedwater, or postulated 1

accidents such as RCCA ejection, which are much less frequent.

In NSAC-113 we have conservatively assumed that NRC Case Study transient categories T1 (loss of offsite power) and T2 (complete PCS interrupticas) can lead to PORV opening but that the more frequent T3 (reactor trip / turbine trip transients) do not. However, we do not assume these transients cause safety valve opening because it is contrary to experience.

Rather, safety valve opening has been considered for transients such as a total loss of main and auxiliary feedwater, or unspecified transients with frequency 0.01 per year.

REFERENCES 1.

Letter, T. M. Anderson to NRC, NS-TMA-2078, May 1, 1979.

2.

NUREG/CR-1363, "Data Summaries of Licensee Event Reports of Valves at U.S. Commercial Nuclear Power Plants," 1980; and Rev. 1, 1982.

3.

WCAP-9804, "Probabilistic Analysis and Operational Data in Response to NUREG-0737, item II.K.3.2 for Westinghouse NSSS Plants," D. C. Wood, 1981.

Response (b): The EPRI/WOG study concludes that small LOCAs requiring sump recirculation are unlikely. This conclusion is supported by the fact that no small LOCA requir.ing recirculation has occurred in 500 reactor years of Westinghouse PWR experience (greater than 500 reactor years when all PWR experience is considered).

Besides this experience which identifies that small LOCAs will be terminated prior to recirculation, thermal hydraulics analysis and plant procedures support this assessment. The WOG-developed Emergency Response Guidelines include guideline ES-1.2, POST LOCA COOLDOWN AND DEPRESSURIZATION, providing actions to reduce the RCS temperature and pressure to 200*F and 400 psig for small LOCAs where SI can keep up with break flow at pressures above the shutoff head of the low-nead SI pumps. The supporting BACKGROUND document includes a generic 111.2-5

application of guideline ES-1.2 to a one inch cold leg break case and to a stuck open PORV case.

In both cases RHR conditions are shown to be reached r: ell before the RWST is drained"for both low pressure and high pressure ECCS designs.

The background analyses were performed with the Westinghouse TREAT T/H network code. The ERGS and BACKGROUND document have been submitted to and reviewed by the NRC. Wisconsin Electric Power has developed a plant specific procedure consistent with this guideline.

Finally, it should be noted that the NRC Case Study credits similar response to a small LOCA followed by HPSI failure. The NRC Case Study credits the operators with depressurizing the primary system to below the shutoff head of the low-head SI pumps (i.e., below RHR entry conditions) so that LPSI can be used for injection.

The EPRI/WOG analysis credits this as well, but also credits that the operators will perform the similar actions under the less stressful case where HPSI works and significantly more time is available.

III.2-6

Revision of seismic hazard curve. What is the rationale / analysis, III.3 other than that cited, for EPRl/WOG's revision of seismic hazard curve.

The EPRI/WOG comment on the Point Beach seismic hazard

Response

curve accepts the method used to prescribe the curve as reasonable, it was suggested that the prescribed curve is conservative by a factor of two to five based on recent computations at the Braidwood site by LLNL and EPRI. The Braidwood site is in the same tectonic region with Zion and Point Beach and also is a reasonable basis for comparison.

The comparison shows a factor of about 10 at low acceleration level (0.1g) increasing to more than a factor of 40 at very high acceleration level (1.0 g). A basic assumption made in prescribing the Point Beach seismic hazard curve is that its slope is the same as the Zion hazard curve. The Braidwood comparison shows divergence between the LLNL and EPRI results with increasing acceleration level. Moreover, the probability of exceeding the Braidwood SSE is about a f actor of ten lower than the assumed 2.5E-4 per year assured for Poirt Beach. Thus considering local site amplification, the prescribed Point Beach hazard curve should be considered conservative, i

l 111.3-1 O

III.4 RWST failure mode. What data supports the assertion that RWST will not fail catastrophically? Was a fragility analysis performed?

Response: The supporting evidence for the assertion that the RWST will not fail catastrophically is based on the attached letter. That basis did not include a fragility analysis for the Point Beach RWST. While the letter does argue that experience indicates that tanks have stronger capacity than the theoretical calculations performed in the NRC Case Study, additional capacity was not credited in NSAC-113.

111.4-1

~ - -

_ -. ~

- m m m - - cc. - - s January 27,1987 Mr. R. K. Hanneman WISCONSIN ELECTRIC POWER COMPANY Point Beach Nuclear Plant 6610 Nuclear Road Two Rivers, WI 54241

Dear Harv:

At your request, we have reviewed briefly the performance of anchored larger storage ta n ks in recent carthquakes.

We have, in particular, reviewed the performance of tanks that are in facilities that are included in the SQUG data base.

Further, we reviewed other available data on tanks from earthquakes and sites not coverec' by our investigations for SQUG.

The goal of this review was to (1) determine whether we have experience data with tanks similar to the Point Beach RWST and CST; and (2) to briefly review the performance of such tar.ks in past earthquakes and study their damage and failure modes. These data would then be used to evaulate the fragilities and failure modes of tanks such as the RWST and CST.

The RWST is 70' high and 26' in diameter. It is anchored with 27 bolts using welded saddles to the bottom tank rung. The bottom rung steel plate thickness is 0.33 in.

The CST is 30' high and 25' in diameter. It is also anchored with bolts. The thickest piste is 0.33 in.

In general, large vertical tanks have performed poorly in strong carthauskes compared to other equipment such as pumps and motors. In particular, numerous gross failures are known; we have documented many such cases.

Typically, the failed tanks are large unsnchored tanks. They were usually subjected to acceleratior.s well above 0.25g; most often they were located in the most intensely shaken area of the earthquakes or were subjected to long duration shaking; i.e., in excess of 40 seconds. We doubt that gross failures of heavy industrial tanks can be found in areas with secclerations less than about 0.20g PGA in the free field.

Because the tanks of interest are anchored, we then narrowed our scarch to anchored ta n k s.

We are currently unswsre of a gross failure of an anchored industrial tank which has equal to or better anchorage than the Point Beach RWST and CST. Our data base contains more than 20 anchored tanks for which we have detailed data.

The aspect ratios of the tanks envelope those of the Point Beach tanks. The free fictd PGAs execed 0.20g. All of the tanks were subjected to ground shaking with PGAs greater than to several times greater than the static lateral coefficients for which they were designed. Further, we reviewed the reported damage to several anchored tanks in recent strong events (California and Japan). We found several tot hcor:co'ed k er s*.e te%cr 3xctw A-ves'* x 1.e Ame. ur, s'. tot w % e c4 oss:e w w cA92eo scn o -m c4 e4m o

tec NSW(4 % t?b SE (714) W W (415) 6 172

~

Nr. L2.lX.1A!a nne ma n January 27,1987 Page3 instances of damage to anchorages or steel plates. Gross failure did not occur. None had tears in the steel plate that led to evacuation of contents. One poorly anchored tank in Chile had a slow leak that was attributed more to corrosion than the earthquake. This type of damage is not' relevant for the R' ST since it is constructed of stainless steel. The scismic literature may contain evidence of gross failures of reasonably well anchored industrial grade tanks. Currently we are unaware of any such failures. I personally have not witnessed any in the more than 20 earthquakes that I have investigated.

Based on our brief review, as outlined in the above abbreviated summary, we believe that significant margins against rapid loss of contents of the RWST and CST exist.

Specifically, even if damage occurs, such as plate bulging or anchorage pullout, the tanks will not lose their contents quickly. Higher margin would be provided by flexible inflow piping. At the Point Beach SSE level, leakage is highly unlikely; if some develop 3, it is likely to be very slow and the tanks should be available for at least 30 minutes at a withdrawal rate of, say 200 GPM.

The above erguments are qualitative and based on experience and many analyses.

They can be quantified. Experience data exist and our analytical methods can be tested against such dam.

Very truly yours, KM Peter 1. Yanev a

EQE Incorporated San Francisco PY/ maw x c:

Mr. Bill Parkinson, SAIC mk l 7/h a n n e.it r

=

f

III.5 Thomas pipe break correlation. What is the basis fcr applying this correlaticn in this study? What is the rationale for changes / revisions to Thomas' original work? What is the basis of data used in numerical calculation?

The Thomas pipe break correlation (1) was proposed in

Response

1981.

It represents an approximation strategy to estimate the probability of catastrophic pipe rupture (Pc) which is related to the The probability of pipe f ailure resulting in leakage (P1).

genertlized approach was based on "analysis of actual service f ailure statistics (1)." The approach considers the pipe geometric factors and the number of weldments as the most important factors in determining the failure probability; it also makes allowance for aging effects.

Basis for Application in NSAC-113: The Thomas correlation has been utilized in develop 1.ig pipe f ailure probabilities in the Oconee PRA (2). That application was reviewed by Brookhaven National The Laboratories and was judged to be an acceptable methodology (3).

Thomas correlation, as utilized in the Oconee PRA, was applied in the EPR1/WOG analysis (NSAC-113).

In this case it was used to estimate the failure probability for a low pressure (125 psig), carbon steel That 10-in. diameter pipe runs ovarhead the six service fire main.

water (SW) and two fire pumps in the Point Beach SW pump room.

i Unlike the Oconee PRA, however, NSAC-113 did not credit the apportioning the computed probability of occurrence according to break In addition, NSAC-113 differed from the Case Study analysis by size.

not crediting a multiplier (=0.1) to accommodate the ability of the SW pumps to withstand the spray impingement from the ruptured fire main.

Use of the Thomas Correlation in NSAC-113): NSAC-113 uses a formulation of the Thomas correlation that directly results from the derivation in the original work presented in Reference 1.

The derivation of the Thomas correlation follows.

Thomas relates the probability of catastrophic pipe rupture (Pc) to the probability of pipe f ailure resulting in leakage (P1) as 111.5-1

-,,-,n

1 PceP1oP1xhy 1

l Where Pc/P1 is estimated from actual pipe rupture statistics, i.e.,

i the fraction of pipe leaks resulting in pipe rupture.

l l

Thomas suggests that, in general, P1 is related to:

Qe, the quantification of pipe geometry factors (e.g., size and shape) and the number of weldments F, an age factor B, a design experience-related, learning-curve factor 5, a measure for quality differences a

. a summation of factors for failure causes.

+

Thus F1 o Qe. F B. S I

In this formulation, the term groupings can be considered as:

Qe. F represents a global estimate for P1 B.S represents plant specific modifiers to the global estimate l

I represents a modifier accounting for fatigue and other factors.

In estimating the individual terms, the following relationship is used:

Qe = Qp + A Qw 111.5-2 7

y 4+

J where Qp is the size and shape factor for the parent caterial Qw is the size and shape factor for weld material A is a penalty factor applied to weldments.

In general Q=

t where D is the pipe diameter L is the pipe length t is the pipe wall thickness Qp = 0 p for the pipe parent material Thus tp QW = 0 w for a single weldment tw Dw w and hence Qw = H for N weldments.

tw Thomas notes that because the length of the weld is defined arbitrarily as i

LW = 1.75 tw, a penalty factor (A=50) should be applied to Qw.

For full penetration weld, tw = tp.

Thomas also notes that leakage failure rates are typically in the range of 10-7 to 10'9/Qe-yr, i.e.,

P ~ 10-8/Qe-yr which would imply a constant failure rate with age.

To account for age related effects, the factor F is applied according 4

I!!.5-3 l

I

to Figure 1 which is reproduced from Reference 1.

es

..=**

/

ee

/

A u

/

/

t" f

l.e f

n..

/

e.

e,

'e a

e ii se e,

se se

-se aos.vsame g

Fig j A smyk ers6 mets averest of as the various ey Immeers Extrapolation of Figure 1 to 40 years, yields F ~2.

To account for design experience and other learning-curve effects.

Thomas suggests that f actors related to the technology as a whole may be neglected entirely, or they may be assumed to be incorporated within the global statistics for P1 (i.e., within the product QeF).

To account for the age of the design of the specific plant, the f actor B is applied according to Figure 2 which is reproduced from Reference 1.

,_s!_

s ru i

t..

i*

r 1

g.

is I 35 a

b 90 tt e

l e

e e

a e

o e

e o

a

, J Ard Of M Sarde 94 YT AAS

, n. m s., ~. -,.,,

111.5-4 9

Note that the curve asymptotically approaches unity, where age is measured from the start of service. Because the curve is hypothetical in nature, Thomas notes that it must be used with caution.

Nevertheless, it appears f air to say that older, established designs should be expected to have failure rates that are average as compared

/

to newer untried designs. After 10 years service, F 1.

Since the primary intent of the Thomas paper was the prediction of j

failure in nuclear systems, an allowance was made for the higher f

The quality of nuclear versus commercial grade piping and components.

quality factor (designated herein as 5) is intended to account for better design, manufacturing, operation, and in-service inspection Hence, for practices for nuclear grade piping and components.

nonnuclear grade carbon steel piping, the logical inference is to assume a quality equal to that of average commercial grade installations, i.e., S'1.

The last factor in the Thomas correlation (designated herein as I) is intended to account for plant specific details of stress, fatigue, Thomas notes, however, that "there is no need to environment, etc.

factor for any detailed causes of failure when the component is being subjected to average conditions" (1). Hence, for the carbon steel fire main seeing ambient conditions over essentially its entire service life, it appears warranted to assume I 1.

As an overall statistic, Thomas suggests that 5-10% of all leaks are

ruptures, i.e.,

h<0.10 l

0.05 <

Based on detailed review of the four data sources cited in Reference 1, Thomas suggests a nominal value k-0.06 but cautions that this value may be slightly optimistic especially for 111.5-5

pressure vessels, as opposed to scall pipes. He also suggests that

.l this value should be augmented by more detailed fracture mechanics modeling. For the low pressure fire main, however, the nominal value appears reasonable for a first approximation.

1 Thus the full formulation of the Thomas correlation is Pc = P [Qp + AQw] F B

S I

fy where 10-7 < P1 < 10-9 hy<0.10 0.05 <

and A=50.

Note that the statement of the correlation in NSAC-113 is Pc = P (Qp + A *S* Qw] (Pc/P1) where I is taken as unity. To be precise, the term Qp also should be multiplied by S.

For S=1, however, there is no loss of generality in the formulation used.

Application of Data in NSAC-113: To calculate a best estimate value for Pc using the Thomas correlation, the followirg numerical values were used for a 3-ft length of carbon steel fire main:

P = 10-8 (10-7 < P < 10-9) fy=0.06(0.05<{E<0.10 A = 50 Op = 10 in III.5-6 e

6

tp a 0.5 in Lp = 36 in Dw = 10 in tw = 0.5 in

\\

N: 2 1

\\

F: 2 (=1.735 9 40 yrs)

B: 1 5: 1 I: 1 Thus (k ) (Qp + AQw) F B

S I

I Pc = P

= 10-8 (0.06) (Qp + 50 QW) (2) (1) (1) (1)/ year l

For a three foot pipe with two welds:

)

i Qp = Op p, 10 (36) = 1440 tp (0,5)

)

4 Qw = Ow w,1.75 N Da tw and tw tw 1.75 (2) (10) = 70 0.5

!!!.5-7

Thus Pc = 10-8 (0.06) (1440 + 50 (70)) (2) (1) (1)/yr

= 5.93 X 10-6/yr As an added measure of conservatism, it can be postulated that the leak is sufficient to damage the pumps in the SW pump room, such that

~

Pc/P1 = 1.0.

This results in Pc = 9.88 X 10-5/yr.

Since there has been no leak in this system during 16 years of plant operation, a best estimate approach shvuld discount the calculated value.

i Sensitivity Calculation Let P = 10-7 Thomas upper bound k=0.1Thomasupperbound Then Pc = 10-7 (0.1) (1440 + 50 (70)) (2)

= 9.88 X 10-5/yr upper bound Then applying the conservatism that Pc/P1 = 1.0 yields Pc = 9.88 X 10-4/yr upper bound. This value also should be discounted for the fe:t that no leaks have occurred in 16 years.

Aeolication in the Oconee PRA The Oconee PRA used the following form of the Thomas correlation:

Pc=F(h)(Qp+AQw)BF where the terms are defined as previously by and the same nominal values as those in NSAC-113 were used for F Pc/P1, A, B and F.

111.5-8 O

O

l-It was also assumed that catastrophic ruptures could be distributed as follows:

l P maximum, DE guillotine break 0.1 Pc P large rupture 0.3 Pc P medium rupture 0.6 Pc When the above methodology was applied in the turbine building flood i

analysis, the total mean annual frequency was calculated to be 2.9 X 10-2/yr. An appraisal of historical data on turbine building floods presented elsewhere in the Oconee PRA indicated a historical frequency of 1.6 X 10-2/yr. Thus, the use of the Thomas correlation along with discounting for break size appears to generate a reasonable estimate for that specific application.

Overall, the Brookhaven evaluation (3) of the Oconee (OPRA) analysis concluded that:

The above approach yields higher pipe-break frequencies than could

'e obtained from the use of the mean rupture rates given by the RSS (Reacter Safety Study, WASH-1400) for pipes larger than 3 inches in i

diameter. The pipe rupture rates used by Thomas are based on an appraisal of the data in References 5 to 8 (of the paper).

The reviewers do agree that, overall, the Thomas methodology as modified by OPRA to include the break-size-frequency distribution represents a realistic model. While the rupture rates derived by Thomas seem to be on the high side, they are used in the OPRA for the piping of the secondary system which can be anticipated to have rupture rates somewhat higher than those of the primary system.

Alternate Analysis: At the February 23, 1987 meeting with the NRC, it was suggested by D. Ericson, consultant to the NRC, that a 15-ft run of pipe with four welds would be a more appropriate basis for computing the failure probability.

111.5-9

a Usi e Thomas' nominal values yields Pc = 10-8 (0.06) [Qp + 50 Qwl (2) (1) (1) (1) 10 (15 x 12) where Qp =

(0.5)#

p, 4 (1.75)(10) 0.05 and Pc = 10-8 (0.06) [7200 + 50 ('140)) (2)

Pc = 1.70 X 10-5/yr If Pc/P1 = 1.0, then Pc = 2.84 X 10'4/yr Using Thomas' upper bounds F = 10-7 Pc/P1 = 0.1 Pc = 10-7 (0,1) (14200] (2)

= 2.84 X 10~4/yr If Pc/P1 = 1.0, Pc = 2.84 X 10-3/yr.

References 1.

H. M. Thomas, "Pipe and Vessel failure Probability," Reliability Engineering, 2 (1981), p. 83ff.

2.

Oconee PRA, NSAC-60, June 1984, p. 9-183ff.

3.

N. A. Hanan et al., "A Review of the Oconse 3 Probabilistic Risk Assessment," NUREG/CR-4374, BNL-NUREG-5197, Vol. 2, p. 2-6ff.

i 111.5-10

)

l l

l-IV., Plant Operations (Recovery)

IV.1 Recovery options. What are the specific procedures and training which I

support the recovery estimates? How was the Success Likelihood Index l

(SLI) established? Have the Time Reliability Correlations been tested or verified in simulator exercises?

Response: The specific procedures and training will be discussed at the.

March 31, 1988 meeting.

SLIs were not specifically established for operator actions in NSAC-113. Probabilities for operator failure to feed and bleed were presented in terms of various SLI values as a sensitivity study.

The Time Reliability Correlations (TRCs) used in NSAC-113 and referenced from NUREG/CR-1278 have been confirmed by simulator data. More details on simulator data and TRCs will be available at the meeting.

IV.1-1

IV.2 Plant and personnel conditions at recovery point in accident sequence.

What assumptions are made about personnel location and functionality under earthquake conditions? How close to fire are staff assumed to function?

Response

The principal recovery actions subsequent to an earthquake are:

providing AFW water supply from the service water system if a loss of

+

main feedwater occurs coincident with CST failure.

starting the diesel generators after station battery failure during a station blackout, providing HPSI or changing water supply from a source other than the RWST after a small-small LOCA.

The first recovery action can be implemented from the control room.

For the other recovery actions, the EPRI/WOG study assumes access to equipment is not restricted by jammed doors or damaged equipment, etc.

Operators expected to perform the actions have keys for access to the room in case the plant security system has failed.

In the event of RWST failure, the effects of flooding have been considered.

Regarding fires, two scenarios must be considered--fire in the 4160 volt switchgear or the AFW pump room.

In either case, the steam admission valves for the turbine-driven AFW pump could be opened from the control room, if DC power is available, or opened locally, if DC power is unavailable. These valves are located in a separate building. No immediate operator action is required in the AFW pump room, since the discharge valves are throttled and locked, and the turbine starts automatically, when steam is supplied.

1 IV.2-1

~

)

IV.3 Tools and equipment availability. Where are tools and extra equipment stored? How fast can these items be retrieved under adverse conditions, e.g., fire, earthquake aftershock, etc.?

Response: The only recovery action credited in NSAC-113 which requires "tools or equipment" is refilling the CST using the fire water system for a long term station blackout scenario. This recovery action can be performed in any one of four ways. Two of the means for refill require i

l the use of a connector which makes the fire water system fittings compatible with the fitting at the base of the CST. Both of these also require fire hoses. The preferred means is to use the connector to attach to a fire hose drawn from a hose station located about 40 feet away. The other means is to connect to fire hoses drawn from a hydrant outside. There may be insufficient hoses nearby for this latter method.

Both the connector and additional hoses can be obtained from the nonnuclear room or fire hose storage lockers located outside the turbine hall. This room is a few minutes walk from the CST.

The other two means for refilling the CST require inserting a fire hose in the top of the CST. This action requires unbolting the top of the CST manways or vent and using either of the two fire water sources quoted above.

Unbolting the manways or vents of the CST requires a wrench, available from turbine operators work station on the floor beneath.

The Two Creeks Fire Station is less than two miles from the plant and could easily connect a pumper from the Pump House Forebay (i.ake i

Michigan) to the CSTs.

IV.3-1

IV.4 Further information concerning the value-tepact analysis of two of the suggested alternatives will be necessary. /4 agreed to at the February meeting, more details on two alternatives, ine Diesel-Driven Auxiliary Feedwater Pump and the Intake Structure Shield Wall Extension, are desired.

Particularly more details on the assumptions made and the costs of specific equipment and activities associated with the installation of these alternatives would be helpful.

We are also attempting to provide more definitive documentation on these two modifications.

Response: The details of the WEP cost analysis of the two alternatives identified will be presented at the March 31, 1988 meeting.

1 IV.4-1

-.,., - -.l

l 1

<U

) m "f f N { +

d 1

S R

E B

MU M

X E

S D

I N

R E

I R

E Ao K

E M

T McEoE R E

M S

ME A N

MMoS ER A

U I

S S

O E

SCOwO mC E

Y St T I RoRRF 0

OM As TrSE

~

T t ACEL ToL M e

~

E 1

u*qAivO AoL h

I f

A I

YlLSsL RWWvM M t

g E

n M 1 234 s 8 F. s90 i

1 s

e

[

t i

1 S

2*

e K

l p

C i O

O m

Mm a

O S

O D

I e

R N

N h

A E

t L

D

• D S

7 f

i I

E

+

M o

R D

n o

2 E

~

i N-L t

T H O

L a

A c

N C IZ A

S S

o, IOE 4(

A I

S PB D

L E

AI 2

UT OI 4

C e

4 r

6 u

g i

F n',.*

i

\\

l I

f

~

Pei or e ces Seise HA+Ae.b Cuevc

~

O Peescret e er enev ca :

S L. OP E c r~

e:.Io M MM2.4# n COe ve" 7A// cal ff0M

.TS#fA'A

/ ot ur BEACM D

&&t ED Tb SE 7 SSS Co I2g )

AL7-ExcEEDMkt pro GAB /L/ry a2.s-E - $ l Lf f dtM VE Cet'#EC 7^ED Fe d' S}rE A97ffPL. / ftc A-7~/oAJ kS~JUM97?6Al '

Seissuc #Maev cueves fea s?Tus

/W SiMis ne 7Ec rc Dic ggisa s H&z/E Si#1/t M SLOPES.

T8 Rhit weot

(@ A4b GuevEE PGA gpxx g L gge g;,,o sat-r S.ve-Y

.? n O. \\o q D ZG}

$ S'E-fo 5ZE-f 3

I2 0

S~o 3.5E-7 52E-4 3 15 l < oc a 7 2 E-9 3.oE-7

$" <l-2.

G.

y Octo-2s yz/, Vol. l.

A t t > < L

-L o

ce i)

Y APPENDIX D:

Insights Gained From I ndu s t r y--Sp on sor ed Study of Point Beach As part of a nuclear-industry-sponscred effort regarding DHR-related r i s}:, a reanalysis was performed for one of the limited-scope PRAs considered in the A--45 c ase studies (Reference 1).

Discussions held between industry representatives and the NRC staff regarding similarities and differences between the two analyses are summarized in Reference 2.

Considerable detail regarding those di sci,si ons is presented in the several enclosures to Ref.srence 2.

This appendix nummarices the results of thcee d '. c c u s s i o n s.

In terms of predicted DHR-related core damage frequency, the A-45 case study for Point Beach calculated a value of ;E -G4 per reactor ysar, whereas the i ndustr y-sponsored study j

calculeted 1 E--05 p er -r eac t or year ( 1. c., a factor af _. O l owc r T.

The fcllowing summary identifies the mejar j

differences in assumptions and methods and their resulting contributions to t h e_

total difference.

In addition, arcas

)

where agreement was pc zible are indicated.

The NRC staff bc11even t h :4 t the approximate core damage frequence that would r m;u l t f r orn use at these agreementa in a revised 3? 34 i - = p on s ta "d analys15 would be aliout 9E-05 per reector year.

M given below, e small fraction (l ess than D )

of the r evi si on it due to changes that have been mede in the plant, and the remainder is due to change.s in the

,e t h o d s, assunptions, and d '.4 t a.

i It should be cautioned that the revised values quoted below and the -;p ec i f i c methods and correlations discussed were e> anined only to the context of the Point Death analyses discussed in thi c Append 1:

Their applicability to other ir l a n t, would have to be det ermined by speciftc erml,ses of the other plants since dominant sequences, plant equipment.

and operating pectedures could be different, and the "resired" values quoted and methods discussed may not be directly 3pplicable.

The b e s e :- for thic re.ised result of 9E-05 per reactor year are given below.

In summery, it was considered reasonable tc accept a 1ower frequenc/ for the SBLOCA and to al1ow more credit for the presence of new batteries and the 17ck of d e p sg ri d e n c e of the SI pumps on the avallobility of the CCW s, stem (the SBLOCA frequency change is the dominant ene)

It was not considered prudent to al l ow mor e credit for many of the operator recovery actions as proposed in the EPRI/WOG s t tid v.

Appendi: D W 990RetiS90 of Eo!Ot Ejggc h Stud [g3 gegugege*

Ggte Melt Eceguency net Bggetgr vest Eccet USG Gese budy "EBILWOG Study Beviggd NSG ygl_ygo S MH H, 4.7E-OS 5.GE-07 7.OE-06 7

The staff accepted the 3E-03 per reactor-vear value proposed by EPRI/WOG for the initiating SBLOCA event after considering operational data pre sented by EFRl/WOG.

However, the staff believes that the IE-04 preposed for cperator a: tion failure per demand is too optimistic since operatin' data (though limited) do not appear to supoort the lowcr value, and the NRC value of IE-03 was not changed.

T MLE

6. 7E--06

'/.7E-07 7.7E-07 The staff accepted the initlating event f requeric i proposed by EPR1/ WOO based on plant specific data.

The staff also t e nt a t i vel y accepted credit for new batteries (nince they are now installed and operational), but information is needed to vsr1 4v the quanti t Ati ve credi t q)ven.

T.C H H,,

2. ',E - 0 5 0

?. t., h -- v e 1

~

The staff used a new value cf O O.01, which is belcw the a

D20.07 value previ ousl y used in the NHC studies but till above the (bolieved op ti mi st i c ) value of zero proposed by EPRl/WOG (EPRl/WOG contends that for t<ansient T.

rem. tor / turbine trip, it is not possible to causs opening of a PORV, therefore O is zero.

The staff believes the probcb2 '. 1 t,,

s smalI but non-zero).

s T_NOH H 3.5E-06 1.9E-07 L.OE-07 1

E.ecauce of uncertainty in the operational data presented (i t varied greatly from year to year), the staff d o e.s not re comend the EFRI/WOG proposed c r ed 11. for PORVs being available (t.e.,

unblocked) a portion of the time.

The staff therefore continues to endar w the c on s ter va t 4 se assumption that FORVs are not av'tlable to prevent SRV optning.

Th9 staff did howe.cr

yree with a reduction in the probabil t t Inadsertent opening of an SRV from the pre <1ous)y useo s.07 to 0.01 por demand (based on aperational data) with a 0,01 per demand probability that an CRV will fall to re close once open I 'St$U$ticoE~5 C descrlhed in narrative form on the last oage of than Append 1

• • Thlu repr esen ts the 116ely value that would be used if a "revised" NR C-s p on s or'e d study were to be conducted for Foint I<each m of the date of this writing.

The norretive below each entry summarl.es the bases.

Appendix D )

(Table continued from previous page):

E29990G9 C9CS UHlh ECggugggy ggt Beggggt yggt Ecgm USC C339 Sigdy EP61/WOG Study Bgviggd USC yglug S MD D 8.7E-06 9.5E-08 9.2E-07 2

7 The staff agreed t.ith the event initiating f reqt uncy of 3E-03 as previously discussed.

Based on pump manufacturer's data, the staff also agreed with removal of the SI pumps' dependency on component cooling water, but additional information is needed to quantitatively confirm the risk change due to that removal.

The source of the remaining difference could not be identified; therefore, additional information is also needed to consider ar.cepting the remaining difference.

T,DD D 4.6E-06 0

1.GE-07 1 7 The note for a previous "T,

." sequence also applies here.

In addition to that' note. more Jaformation would oe needed to identify the source of and to consider accepting any part of the remaining difference (1.8E-07 vs.

O).

T,MLE 6.4E-07

1. OE--07 6.cE-07 No changes.

The A-45 cane study initiating event frequency of 1.0 per reactor year is not significantly different from the 0.91 proposed by EPRI/WOG: and it is not clear hcw MFW recovery differs in the EPRI/WOG study.

slso, additional information would be needed to identify the source of and to c on si der accepting the remaining difforence (6.6E-07 m.

1.OE-07).

T MOD D

6. tE- 07 4.1E-03 4.1L-OG 7

The staff used a new value of D= 0.01, which is below th9 D

= 0.07 value used in the A-45 studies but above the 1_wr (not directly specified) value used by EPRI/WOG.

The staf+

recommends recognizing the low ibut non-Jero) probability that a SRV wil1 be !ifted during this event.

Tht staff agrees with removal of SI dependency on CCW (per manufacturer't data),

t, u t did not change the event frequency, as the EPRI/WOG-suggested frequency oi 0.91 la not significantly difierant from the 1 u used in the case Itudies.

The u ts o v e changest resulied in a value that um Inaer than the EPRI/WOG result.

The staff therefore agrees with the EPRI/WOG result.

Appendi> D *

(Taole continued frcm orevious page):

Ed99e0gg Cptg Mgli Ecggggggy get Bggetgt yggt Ecos:

NBG Case Study EPSifbOG Study Bevised USC velge 5,MXD 5.7E-07 1.0E-08 1.0E-08 e

1 As alet.ady discussed, the staff, based on data, agroed with the 3.OE-03 initiating event frequency proposed by EPRI/WOG and with removal of the dependency of the SI pumps on the component cooling water system, but information is needed to justify ths quantitative credit given.

T,MLE 9.1E-07 1.3E-08 9.1E-07 T"MLE 6.2E-07 0

6.2E-07 T,4MLH 2.0E-OS 1.0E-07 2.oE-nB T~1OD b, t.OE-OG 1.0E-07 1.0L a

s 1

No changes were made by the staff for these four sequences.

EPRl/WOG credits considerable additional ecovery in the form of operator actions that have not been adequately justified.

The A-45 study assumgd that loss of an AC bus would either trip the plant or lead to a manual trip, end the staff h es s elected to retain the readest conservatism sEsociated with that a u qudip t i on (no significant impact).

LT5D 3.6E-05 S.4E-07 9.9E-06 The staff agreed to plant-specific (lower) valu en for the I fr.quency and diesel gener ator local faults, but has not taken additional credit for CST refill and other long-term recosery actiona proposed by EPRI/WOG because of untertninty regarding the operator's abilltv to recogni:'e the need and perfor,a the ections in the tiine available.

It is considered 116 el y that further reduction could rnasonably be j u =,t 1 f i ed provided the baaea for awuming offstte power recovery u t t h i r.

a few hour 5 are suffIclent.

IOIOL 1.3E-04 2.5E-06 2.bE-OS (Internal E.ents only)

The above represents the total for all significent "internal" events as listed above.

"E;: t er n a l " events are listed below.

Appendix D 's (Table continued from previous page):

E99990E9 99C2 U91$ EC999205Y REC BenGI9C YSfC ECgms UE9 9ese SIWdy EEBifWQG Qtydy Egyiggd UBQ yglug S MD D 8.7E-06 9.5E-08 9.2E-07 2

g g The staff agreed with the event initiating frequency of 3E-03 as previously discussed.

Based on pump manufacturer's data, the staff also agreed with removal of the SI pumps' dependency on component cooling water, but additional i nf or mation is needed to quantitatively confirm the risk change due to that removal.

The source of the remaining difference could not be identified; therefore, additional l

information is also needed to consider accepting the remaining difference.

T,DD D, 4.6E-06 0

1.GE-07 1

v The note for a prevxous "T,

." sequence also applies 3

here.

In addition to that* note, more information would be neuded to identify the source of and to consider accepting any part of the remaining difference (1.8E-07 vs.

O).

T MLE 6.6E-07 1.OE-07 6.oE-07 j

l No changes.

The A-45.ase ctudy initiatina event frequency of 1.0 per reactor year is not significantly different fron the 0.91 proposed by EPRI/WOG; and it is not clear hcw MFW recovery di f f ers in the EPRI/WOG study.

Al so, additional information would be needud to identify the scurce of and to i

c on si d er accepting the remaining difference (6.6E-07 vs.

1.OE-07).

T MOD D 6.6E-07 4.lE-08 4.1E-08 The ctaff used a now value of O = 0.01, Nhich is below the O

= 0.07 value used in the A-45 studies but above the lower (not direct 1/ speci fied) value used by EPPI/WOG.

The staff recommends recognicing the low (but non-coro) probabi.11ty j

that a ERV will be lifted during thin event.

The staff

~,

agrees with removal of SI dependency on CCW (per j

manufacturer's data), but did not change the event I

frequency, as the EPRI/WOG-suggested frequency of 0.91 is not significantly,different from the 1.0 used in the case studiec.

The.uovo changes resulted in a value that vas j

i Igwet than the EPRI/UOG esult.

The staff therefore agrees with the EP7I/WOG result.

3 l

l i

i i

1

.,.~.-,,-,-->--~v.-

l Appendix D l l

(Table continued from previous page):

l SS9990G2 99C9 d211 EC99990GY P9C 62BGt9C Y9BC EC998 NBG 92S9 Study EESifBOG Stygy Beylggd U6g yglyg i

l S,MVD 5.7E-07 1.0E-OB 1.0E-00 g

l j

As already discussed, the staff, based on data, agreed with the 3.0E-03 initiating event frequency proposed by EPRI/ WOO l

and with removal of the dependency of the SI pumps on the component cooling water system, but information is needed to l

justify the quantitative credit given.

l T,MLE 9.1E-07 1.3E-08 9.1E-07 T{MLE 6.2E-07 0

6.2E-07 T,MLH 2.0E-09 1.0E-07 2.OE-09 T'QD b 1.0E-00 1.0E-07 1.0E-00 l

l 2 No changes were made by the staff for these four sequences.

EPRl/WOG credits considerable additional recovery in the form of operator actions that have not been adequately r

justified.

The A-45 study aggymed that loss of an AC bus i

would either trip the plant gt lead to a manual trip, and the staff has elected to retain the modest conservatism associated with that a s surnp t i on (no significant impact).

LTSD 3.6E-05 5.4E-07 9.9E-06 The staff agreed to plant-specific (lower) values for the T frequency and diesel generator local faults, but has not talen additional credit for CST refill and other long-term recovery actions proposed by EPRI/WOG because of uncertainty i

regarding the operator's ability to recognt:e the need and perform the actions in the tiine avai l abl e.

It is considered l i l:el y that further reduction could reasonably be justified provided the baaes for assuming offsite power recovery within a few hour s are sufficient.

i TOTOL 1.3E-04 2.5E-06 2.5E-05 (Internal Events only) 1 c

l i

The above represents the total for all signi ficant i

"internal" events as listed above.

"En ternal " events are listed below.

1 i

~"

Doninant Scquanco Definitionc S HH Ht 2 - A small break LOCA with subsequent less of main 2

feedwater and failure of emergency core cooling in recirculation.

T MLZ - A loss of offsite power transient with failure of t

auxiliary feedwater and feed and blsad.

)

T QH H13 - A transient followed by stuck open relief valve 3

(translent induced LocA) and failure of emergency core cooling in recirculation, T M0H H

- Loss of feedvater transient followed by a stuck 2

t open re$ief valvo (transient induced LOCA) and failure of 1

emergency core cooling in the recirculation mode.

S MD D1 2 - Small break LOCA with loss of main feedwater and 2failure of e=ergency core cooling in the injection mode.

J T QD 01 2 - A transient followed by 6 stuck open relief valve 3(translesnt induced LOCA) and failure of the emergency core coolingintheinjectionmode.

T HLE - A less of foedvater transient with failure of auxiliary 2feedwater and feed and bleed.

T MQD 012 - A loss of feedwater transient followed by a 2stuck open relief valve (transiant induced LOCA) and failure of the emergency core cooling in tna injection mode.

S MX01 - Small break LOCA with failure of energency core 2coolingininjectioncodeandfailuretoachievesecondary blevdown.

T MLE - Loss of DC bus transient with failure of auxiliary 5

feedwater and feed and bleed.

T HLE - Loss of AC bus transient with failure of auxiliary 4

feedvater and feed and bleed.

i T MLN1 - Loss of feedvater transient with failure of 2auxillary feedvater and failure of emergency core cooling in recirculation.

T QD 012 - Loss of offsite power transient follcwed by stuck 1

open relief valve (transient induced LocA) and failure of emergency core cooling in injection mode.

LTSB - Long term station blackout caused by loss of offsite power transient and failure to recover offeite power with subsequent h8K65$

i A.

A

..