ML20030E223

Forwards Addl Info on Probabilistic Risk Assessment Results Re Efficacy of Recirculation Pump Trip at Facility

ML20030E223
Person / Time
Site:	Big Rock Point File:Consumers Energy icon.png
Issue date:	09/10/1981
From:	Bordine T CONSUMERS ENERGY CO. (FORMERLY CONSUMERS POWER CO.)
To:	Crutchfield D Office of Nuclear Reactor Regulation
References
NUDOCS 8109180231
Download: ML20030E223 (20)

Text

.

e e

David P Hoffman Ntacle.ar Licensing AJmunestr.ator Power Company General Offices: 1945 West Pernall Road, Jackson, MI 49201 *(517) 788-1636 September 10, 1981 D

N.

g g 7 7 jggj & g}

Director, Nuclear Reactor Regulation g

7 4

Att Mr Dennis M Crutchfield, Chief i

OperatEng Reactars Branch No 5 9 M $ en T

'7 US Nuclear Regulatory Commission 7e Washington, DC 20555 d,

N DOCKET 50-155 - LICENSE DPR BIG ROCK POINT PLANT - RESPONSE FOR ADDITIONAL INFORMATION REGARDING PROBABILISTIC RISK AS-SESSMENT RECIRCULATION PUMP TRIP ANALYSIS By letter dated August 5,1981, NRC requested additional infornation concerning the Probabilistic Risk Assessment results relating to the erficacy of a recir-culation ptnip trip at Big Rock Point.

The enclosure to this letter provides our specific responses to the seven questions raised in the August 5,1981 letter. provides a model validation test as further clarification to our response to Question #1 and several tables reproduced from Appendix VII of the FRA provide re-analysis as requested by the referenced questiens.

1 rm

<w e/L h Thomas C Bordine St.aff Licensing Engine?

CC JGKeppler, USNRC l

NRC Resident Inspector Attachments - 19 pages 1

9\\

' 0109180231 810910 PDR ADOCK 05000155

. P PDR

r QUESTION C1:

" Provide'information confirming that the RETRAN code provides an appropriate representation of the plant transient behavior as presented in the risk assess-ment analysis for recirculation pump trip. This inforntion should include any available comparisons of code predictions with experimental data and discussions of differences and similarities between RETRAN and other established industry codes. The specific applicability of RETRAN to ATWS analysis should be included."

RESPONSE

As a preface to this response, note that the primary reason for performing the RETRMI analysis of the plant response during ATWS was to estimate the effect of 3PT on the time to RDS actuation. Dep 9 ding on whether RPT is assumed to cecur automatically, manually, or at all, the operator vill have differing amounts of time in the ATWS event in which to successfully inject liquid poison. The rish assessment essumed that liquid poison system injection vould not be success-ful, and therefore that significant core deage as well as containment failure vould occur, if RDS actuation was allowed to occur prior to achieving core sub-criticality. This assumption was based on engineering judg=ent that mixing of the liquid poison vould be very poor following RDS operation. The time to RDS actuation was used in the risk assessment to deter =ine the probability of successful liquid poison system operatie.,n. This probability was assumed to be a function of the time available to the operator for diagnosis of the ATWS and actuation of the liquid poison syctum.

RETRAN VERIFICATION AND QUALIFICATION:

R'effRAN is a computer program for transient thermal-hydraulic analysis of light water reactor systems during postulated accidents and anticipated operational transients. The program was ceveloped from the RELAP series of codes by the Electric Power Research Institute. The program includes proven thermal-hydraulics models fr a RELAP as well as newly developed models which peredt "best-estimate" analysis of accidents and transients; The RETRAN code has been subjected to an extensive verification and qualification program by EPRI, EPRI contractor, end utility users. In this effort code predic-tions have been compared against nuclear plant test data, experiments, and Nuclear Steam Supply System vender calculations. Examples from the RETRAN verification and qual cation program =ay be found in volume h of the RETRAN computer code =anual, e.s well as in reference 2.

Ruference 2 provides 2*e-sults of RETRAN analysis of the Peach Bottom Turbine Trip Tests. This effort is particularly applicable to EWRs during transients including ATWS.

BIG ROK POINT RETRAN MODEL QUALIFICATICN:

In addition to qualifying RETRAN as a code suitable fer evaluation of thermal-hydraulic transients in nuclear power plants, it is also necessary to qualify the

REFERENCES:

1.

RETRAN - A Procram for C*ne-Dimensional Transient "hermal-Hydraalic Analysis of Complex Fluid Flov Systems, EPRI-CCM-5, Volume 4, December 1970.

2.

K Hornyik and J A Naser, RETRAN Analysis of the Turbine Trip Tests at Peach Bottem Atemic Power Station Unit 2 at End of Cycle 2, EPRI-UP-1076-SR, April, 1979 4

g

.,w

-r e

-.--wr.

=

~. - - - - +-

w --

=

2 specific plant model (in this case the BRP RETRAN model) against available test data and analysis. Consumers Power Company has conducted a limited verification effort for the BRP RETRAN model. This eff~ t has consisted of comparisons against l

actual plant data and higher order models. Resu'.ts of comparisons against plant data are summarized in Attachment 1.

A comparison was also made between the core power response of the RETRAN model and of the BRP core simulator model to changes in a key system parameter (ie subcooling). This comparison showed good agreement (15% in core power for complete loss of subcooling) between RETRAN and the higher order model, thus providing added assurance as to the reasonableness of the RETRAN predictions, i

i

.1

\\

{

f I

I i

ATTACPJENT 1 Medel Validaticn Test A validation test of the BRP RETRAN model was conducted to inaure that the RETRAN model vould accurately represent the trar.sient behavior of the Big Rock Point Nuclear: Plant Primary System. This was accomplished by modeling a two-recirer.lation-pump-trip transient and comparing the resulting RETRAN cal-culated flow coastdown with the recorded experimental data. The test chosen for validation was TEST llB as documented in GEAP kh96, " Core Performance and Tran-aient Flev Testing, BRP".

This test cor.sisted of a trip of both recirculation pu=ps from a steady-state power level of 205 MWt. The BHP plant initial condi-tions for this test were taken fram Tables III and VI of GEAP kh96. The resulting plotted experimental test data vere taken from the strip charts of Figure h-16 of GEAP hh96.

The BRP RETRAN odel parameters were changed to the appropriate values for a

' steady-state operation of 205 MWt. To verify that the RETRAN SSI feature had traly provided a steady-state condition, a 60-second null transient was run.

This transient will also verify the operation of all control systems, fills, and time-dependent boundary conditions not related to the SSI operation.

The results of the 60-second null transient were excellent, shoving all system parameters to have negligible deviation from initial values.

Experimental Pump Trip Test llB vas run with a different core configuration than the ERP core nov modeled in the ERP RETRAN deck. Sufficient information to model the core point kinetics and geometry for the test core was not available. Therefore, the Experimental Test llB power response curve was input

-into RETRAN and the present core point kinetics model deleted. Hence, the validation test cannot verify the BRP kinetics model but vill verify the rest of the model. The nor=alized power response curve was scaled from the strip chart results as published in GEAP hh96. The timescale was modified to include a one-second null transient for the beginning of the RETRAN simulation.

Pump Trip Test 123 was run by tripping off both recirculation pumps at t = 1.0 second and using the normalized power curve as the boundary condition. The resulting normalized recirculation flow coastdown is shown in Figure A.1 vith a comparison plot of the experimental data taken as a best estt= ate from the GEAP hh96 strip charts. This shows that the RETRAN model predicts the flow coastdown very well in the 0.0- to 6.0-second time span. However, the discre-pancy between the RETRAN calculated recirculaticn flow and the experimental data from 10 seconds to the end of the transient cannot be accounted for at thia, time. The lack of well-documented test data pertaining to operator actions and to system automatic control behavior makes it impossible to justify further data comparison past the 10-second time span.

20/05/81 R1A PT205 BtG ROCK POINT i

i i

i i

i El RETRAN Calculated Recircul tion Flow, Loop 1 i

N O RETRAN Calculated Recirculation Flce, Loop 2

~

~

l nR e

N to n

ON 11.1 *

(n N

E' aa i

d, a

3o

-I ll-j i

Experimental Recirculation 9

flow from Test llB, GEAP-i196 1

I I

I I

I

'b l o 0

10 20 30 40 50 60 i

TIME AFTER TRIP (SEC)

Figure A.1 Nonnalized Recirculation flow Coastdown and Experimental Data from GEAP-4496 Pump Trip lest Illi Versus Time

]

QUESTION 2:

" Compare the current risk analysis for recirculation pump trip (References 1 and

2) with the earlier assessment described in Reference 3.

In particular, discuss the difference in predicted steady state power level following a turbine trip-110%

in the earlier analysis vs 80% in the current analysis using RETRAN."

RESPONSE

The earlier ATWS analysis (NEDE - 21065) vss based on a plant transient analysis performed prior to initial plant criticality (ref: R J Romeo and E C Eckert, Transient Analysis Consumers Power Company Big Rock Point Plant, APED

h093, October, 1962). This early transient analysis was performed on an analog computer, and was done assuming vorst case core conditiors (ie, most negative void coefficient), a reactor operating pressure of 1500 psia vice the current operating pressure of 1350 psia, and a safety valve setpoint of 1750 psia vice the current 1550-1600 psis setpoint range.

(Note that BRP was originally intended to be operated at 1500 psia, but that it has operated at 1350 psia since completion of the High Power Density Demonstration Progrms.)

The RETRAN snalysis, on the other hand, was meant to be a best estimate analysis of plant response during ATWS events. Nominal plant and core conditions were assumed in the analysis as deceribed in our suhcittal of 2/26/81. The differ-ences between the earlier ATWS analysis and the analysis in question here are judged to be so significant that a detaileu comparison of the differences in predicted plant response is not possible. It would be expected however that the more realistic treatment of void feedback by RETRAN vould result in a lover steady state power level as compared to the earlier ATWS analysis.

QUESTION 3:

"The ATWS analysis presented in References 1 and 2 did not include the case of ATWS for inadvertent control rod withdrawal. Provide an ATWS analysis for the case or provide a discussion justifying its emiction."

RESPONSE

An explicit analysis of control rod withdrawal ATWS event was not considered necessary for assessment of ATWS risk for two reasons. First, the frequency of inadvertant control rod withdrawal events is extremely lov (no such event have en cecurred based on EPRI Report #NP-801, July 1978) as compared to other ivortant ATWS events like turbine trip and loss of feedvater. Second, the evntrol rod withdrawal ATWS event in very similar in terms of plant response to ATWS events already analyzed for the risk assessment. This' conclusion is supporte0 *cy the following discussion of the plant response to a control rod withdrawal ATWS event.

PLANT RESPONSE TO INADVERTANT ROD WITHDRAWAL ATWS:

The plant response to an inadvertant rod withdrawal assuming no reactor scram vill be as follows: The rod withdrawal vill first cause a rise in core power and then in steam pressure. At 10 psi over nor=al pressure the steam bypass valve vill begin to open to relieve excess steam to the condensor. Continued rod withdrawal vill cause power to continue to increase until the rod is completely withdrawn. A resulting steady state power of more than 300 myth is possible based on analysis with the BRP 3-D core simulator. Plant trip prior to full rod withdrawal is possible en either low condensor vacuum or low steam drum level (if the feedvater system in incapable of maintaining steam drum vater level at the higher power level). If a plant trip setpoint is reached, but the reactor fails to scram, the turbine vill trip and the bypass valve vill open fully to control system pressure.

(A safety valve may also open at this point if the bypass valve cannot pass sufficient steam.) At this point feed-water vill be lost due to instabilities in the condensate system, and the resulting plant response vill be similar to that analyzed for the lost of feedvatec ATWS event. If the operator fails to shut down using the liquid poison system, the RDS will eventually actuate. The time for operator action to prevent RDS will be somewhat less in this case than in the loss of feedvater case due to the higher core power level attained as a result of the rod with-drawal.

If the steam bypass valve fails to open the plant response vould be somewhat different. A plant trip would occur much earlier on high reactor pressure (normal pressun plus 50 psi). Following turbine trip the plant response vill be much like that analyzed for the turbine trip without bypass ATWS event although a somewhat higher steady - state core power level vill be reached due to the rod withdrawal. Thus feedvater vill be lost, and RDS vill r.ctuate, somewhat sooner in this case.

Thus the control rod withdraval ATWS event leads to similar censequences, although on a somewhat accelerated time scale, as the loss of feedvater and turbine trip without bypass ATWS events already analyzed for the risk assessment.

QUESTION h:

"Your analyses assumed nominal values for the volumes of the condenser hotvell and steam drum. Provide sensitivity analyses or discussion of the sensitivity of the risk analyses to the use of minimum and maximum valves for these two volumes."

RESPONSE

The steam drun (feedvater flow) and hotvell level control systems at Big Rock Point are both automatic systems. The feedvater control system is a three element system whose inputs include ; team flow and feedvater flow as well as steam drum level. This system maintains level at the steam drum centerline during power operation. The hotwell level control is a single element system consisting of a level transmitter which regulates the position of air controlled valves in makeup and reject lines connecting the condenser to the condensate storage tank. Steady state hotvell level is maintained at approximately three feet above the bottom of the condenser. Charts of steam drun and hotvell levels indicate that these systems maintain drun and hotvell inventory within fractions of an inch of their level setpoints during steady state power operation.

Calibration checks of these instruments are performed periodically. Calibra-tion data indicates that the instruments are per.'orming accurately if the drum level element indicates level within i 1 inch of actual level over its range and if hotvell level is regulated i 1 inches of its setpoint. The uncertainties casociated 'cith drum and hotvell inventories are believed to be dominated by calibration acceptance criteria. As a result of these uncertainties drum in-ventory is 2h,000 930 lbm and hotvell inventory is 25,000 1 2500 lbm during steady state power uperation.

Variation in the amount of coolant available during an ATWS can be translated into a variation in the amount of time to RDS actuation and ultimately the amount of time available to the operator to initiate poison injection.

Un-certainties associated with hotvell level affect the timing of caly those transients in which feedvater is available. The primary effect of hotvell variations, therefore, is to change the amount of time from transient initia-tion to feedvater pump trip due to loss of suction pressure. Drum level uncertainties affect all transients in which primary coolant high pressure or low level occurs. For these transients, drun level variations affect the length of time from loss of feedvater to RDS actuation. Those transients in which neither a low level or high pressure condition occurs (infinite feed-water transients) are unaffected by these variations in drum and hotvell level. For these transients, coolant inventory is maintained by normal steam flow to the condenser and feedvater flow to the drum.

Tables in the RPT submittal of February 26, 1981, and in Appendix VII of the PRA presented plant and operator response tires for che various ATWS transient categories. This table is reproduced here showing the effects of hotwell and steam drum inventory variations on plant and operator actions. Effects

Plsnt end Ope,dter Rup:nco Time for ATWS Transient Categorico Initial Power Power Power Time to Time to Operator time Transient

~

after Drum Mass after loss Drum Mass RDS mix to inject Power prior to Btu /hrX10-6 RPT RPT at RPT of fu

@ loss of seconds LPS poison Btu /hrX10-6 Btu /hrX10-6 X10-6 fu seconds seconds Lou Level 450 24,000(i9T.145 US.)

41 104[r6) 1;o RPI -

820 450 EPT G 80 seconds 820 450 225 5,500(19M) 450 24,000(1909)268 (?8) 75 193(f6 RPT G 60 secords 820 450 225 10,100(I9CC) 450 24,000(39Cx487(182 75 212(fN RPT C 35 seconds 820 450 225 15,900(r900) 450 24,000(1900)12/1 8) 75 231 (f Ed iiigh Pressure with Feedwater s

bd 41 226/11)8N

.24',000(f 500)267(M)f 73 No RPT 820 738 406 RPT G 60 seconds 820 738 442 24,000(t 900) 243 24 000(f M')430(1b 75 355(fF85 RPT G 8 seconds 820 738 442 24,000(19Co) 243 24,000 (f 9C<l530('8)f*72 75 455(1 6273 H1,h_ Pressure No Fw 3

RPT G 60 seconds 820 406 243 24,000(122) 406 24,000(# D l3 9(1 8) 75 234[f8) i RPT G 0 seconds 820 820 243 24,000(f5tD) 243 24,000(f 900350(18). 75 275(18) i I

4 W

D W

G 1

of hotwell inventory variations are shown in brackets [ 3, drum inventory variations in parenthesis (). Available operator response time to initiate poison injection varies by no more than eight seccnds as a result of drum level uncertainty and eleven seconds as a result of hotwell level uncertainty.

An evaluation of the effect of coolant inventory v;riations on ATWS core damage probability is also attached. Note that only low pressure loss of feedwater transients are affected by these uncertainties. The reason for this is a result of the lack of environmental qualification of the liquid poison system and shutdown systems for which all high pressure ATWS transients are assumed to lead to core damage situations. Inventory variations in the hotwell have no effect on core damage probability as e result. Combining uncertainties using SRSS, it can be seen that normal variations in steady state reactor coolant inventory affect the Big Rock Point core damage probability associated with ATWS less than 2%.

BIC M iK IMENT 1*l.Affr AWS CONC 81AfoQ: OtAwrtrirATION (Tutal-2.7 e tC' /Yrl [ 4 # /C Sequence Transient i

Core Demage Core Damage Frequgncy Frequgacy Transient Sequence G4antification (Yr I (Yr i IPR /PR railure 2.7 x 10~

d./

/S#

T AY L L g g j

T AY L,L, 2 43

\\

g g i.18)(3.5 e 10,511.86 3 8 8.0 ) t.9 )(.28 )

1.4 x le d. / / /o T AY L, y 3,i r i g g gg 1

,3 T AY L,C IPR /PR railure Closud (T l

.!:/Yr g g g

railure to SCRAM (Al 3.5 m 10' Successful Turbine Snuse Valve

.46 i

railure of reeduater Given Bypass Valve 1.0 l

is Open (Y l g

l Manual RCPT Wathan let 60 Seconds

.9 Failure to Segin Posson In) action Wathin

.24 212 seconds (L )

T AY CL,L g g T AY'OL L

.4af A

(.la) 3.5 m 10~5)s.asic!.0)s. )(.in 3.gu,le-7 y, f xfo#

l T AY OL,

,a,y< g gri

.W 3 g T AY %,LC raalure to Trap RCP Wathin 1st 60

.1 g g SeconJs (0) raiture to Inject misen civen SCP Not

.7C Tripped (L,1 4

4

=

i 2

Sequence Traneaent Core Lkanage Core Daeage Frasignch Freqqncy Transient Sequenro Quantificatinn (Yr )

(Yr )

T A3 L g

T AD OL

,e g

T All,Y L, g

g T AS I o( r

(.18)(3.5 m to li.14)(1.0) 4.4 x 10*I g

• j T Au L g

YE*

T AB OL a

s-I o T AB,Y L Failure to inject toison Civen no gnviron-g g

mental Qualification.of system 1.0 9 3, y g 1 oi 26 T "#(L 4.1 (3.5 m 10~ l(.8)(1.0)(.9)(.28) 4.4 x 10* k,(o 3.3 m 10 10 I

spursous 2

3 I

w rbir.e Sypasa Valve Spurious Opuning of h rbine Bypass

.1 Wlve (7 2 Successful Dypass - Operator Fails to

.1 Terminate Slow 1uwn Dolore,RDS

. Ca$

T AY OL 4.1)(3.5 m to-5 ) (.1 )( 1.0 ) (.1 ) (. D) 3.3 m 10~ 3.3 I #

79 i

<t AB

(.1)(3.5 m 10" 3 8.9 )(1.0 )

3.2 m le

I 2 o 1

Failure of Dypass - Operator Terminates Blowlown j

.SG

-6 1.2 m le

!. / 2.4 a 10~4 *,,./.

f T Av L, 4.16)(3.5 m to-I(.86 )(1.0 )(.9 )(.29)

Loss of tw 3 g

. 5/

loss of fw (T 3

.16 3

4 Failure of hkeup Civen Transient (Y )

g eloQ i

(.16)(3.5 m 10 5)(.e6)(1.0)(.1)(.76) 31 a 14*I 2. 3 l

T Av r4.,

3 g 79

(.16)(3.5 m 10"3 (.14) 7.4 m 14*I 3

T Aa, 3

e i

e e

.~

3 1

Sequence Transient Core Dmage Core Dm age Frequgncy Frequ,jacy Transient Sequence Quantificat6on (Yr )

(Yr )

2.9 x 10~I Ioss of 1 FP T AD,

(.06)(3.5x10bt.84)

?

2.9mid l

3 toss of I reed Pump T

.14 4

toad Rejection 7.0 m 10

%4

..L T ay,t.,

(.59)(1.5 x 10-5)( 86)(1.0)(1.0)(.23) 4.1 x 10 f",4 %

3

• .2 b~

toal sejection(7 )

.59/Yr 5

RCPT Civen toss of taa! 0 25 Seconds 1.0 j

T AD,

(.59)(3.5 x 10-5)(.14) 2.9 m 10' g

j toss of Main 4.3 x 10' Cond. (.06)

(.18)(3.5 x 10'5)(1.04 6.3 m 10-6 T "'o Loss of Inst."

6 A4r (.06)

I"** 'I C"

"""*' T'***E****

IY I

• I'!Y#

MSlv Closure (.06) 6 Byp se/Condenmar railure Given 1.0 j

j Transient IMP "4.6 m 10 i

-(.13)(3.5 m 10' 3(1,0) 4.6 a 10' T)AB, LOSP (T l

.13/Vr y

3 Bypaus/G esensar ratture Given 14se 1.0 of Power 7

Misc. Scrarna 1.7 m 197 tat.,

(.56)(3.5 x 10' )E.a6)(.0!!

1.7 m 10'

~

i Misco!!aneous Scrane

.56/Yr 4

e*

9 e

o h

QUESTION 5:

"The Big Rock Point liquid poison system (LPS) is presently not qualified.

Describe any progran planned for qualifying this system, or, if no such program exists, discuss the possible effects of the expected environment on the operation of the LPS during the A'IVS transients included in the risk assessment. In particular, address the effect of the steam released from the pressure relief valves on the LPS."

RESPONSE

The Big Rock Point liqu'.t poison system is presently not qualified for a LOCA environment. Neither Consumers Power Company nor the NRC identified this system as essential in developing the safety related electrical equipment list for the environmental qualification program. In this program, emphasis was placed on qualification of reactor protection system and control rod drive system equipment to assure reactor shutdown by control rod insertion during a LOCA. The LOCA simultaneous wi n a failure to scram was not considered to be of sufficient probability to warrant consideration. There exists no pro-gram for qualifying equipment associated with the liquid poison system for the LOCA environment.

Liquid poison system electrical equipment subject to the containment environ-ment includes seven de actuated explosive squib valves, e. de operated mult, contact relay and ascociated electrical cable. The squib valves Lre manu-factured by Conax Corporation and ir.clude a DDNP (Diazodinitrophenol) primer.

The Copax Corporation indicates that DDNP is a lov temperature explosive.

That is, prolonged exposure to elevated temperatures can significantly degrade the reliability of the primer (ie, at temperatures greater than 1750 F the recommended shelf life of the primer is less than one day). Conax further indicates, however, that at tempela.tures above 2000F the primers may even self detonate.

The enviroraent in which the equipment associated with the liquid poison system must operate is dependent on the type of transient and the time at which poison injection is initiated. Approximately 75% of all ATWS transient sequences do not result in a significant pressure rise in the primary coolant system or release of steam to the containment through the safety relief valves.

Tha Containment environment during these transients doer not degrade unless RDS actuation occurs. On reactor depressurization the containment environment vculd become similar to that for a DBA (224 F, 20 psig). Un until the time of RDS actuation the containment environment is considered to be normal and the liquid poison system availatle for injection. For the purpose of the PRA the poison systcm was considered to became ineffective after RDS actuation due to environmental considerations and for reasons associated with poison mixing described in branch points 15 and 16 of the ATWS event tree.

The remaining 25% of the ATWS sequences are high pressure sequences requiring the safety relief valves. Steam release to the containment is expected to re-sult in a rise in containment pressure to 12 psig prior to RDS actuation after which containment con 2itions approach 33 psig,~ 2500F vhich exceed containment

design (27 psig, 235 F). Due to the apparent rapid degradation of the reliability of the squib valves in a high temperature environment, and the 0

unknown reliability of auto-detonation of the squib valves above 200 F, the poison system was considered ineffective even during the early stages of high pressure ATWS transients. Once RDS actuation occurs, the poison system is considered ineffective for the same reasons described for ncs-SRV ATWS transients.

QUESTION 6:

"Your analysis for loss of feedvater appears to neglect the time delay for operator response until the operator is made aware that an automatic reactor trip has not actuated. Justify and discuss the effect of this time delay on your analysis."

RESPONSE

Your comment is valid for all transients in which a loss of the feed pumps occurs vithout release of steam through the ufety relief valves. Procedures exist which instruct the operator to trip a recirculation pump anytime feed-water is lost. With the power and loss from the primary system reduced in this manner, the operator can attempt to restart a feed pump or perfom whatever other corrective action is necessary to regain primary ecolant level. The human error probability used for failure to reduce power by manual recirculating pump trip, therefore, is basically valid. However, there is a short period of time in which the drum level falls to the reactor protection setpoint of eight inches below centerline which will delay operator response to attempt to trip the reactor manually and inject the poison.

RETRAN analysis indicates lov drum level occurs approximately 35 seconds following a loss of feedvater makeup to the primary system. The failure probability to initiate liquid poison injection shown in Figure III-3 of our February 26, 1981 submittal should be delayed 35 seconds for these sequences.

A sensitivity study was performed to determine the effect of the uncertainties associated with operator action to inject liquid poison on the core damage fr3quency. This study is presented in Attachment VII.T to the FRA and bounds the 35 second delay associated with invent.ay depletion to the low drum level trip setpoint. A re-analysis of the core damage frequency including this 35 second delay is attached. The total core damage probability due to ATWS increases less than 7% due to this delay.

m N

D eG 8dM'E lo tf47 l'I.At47 AWS CORE DAMACE eJtlANTIFI'*4TIOf, (Total-2.9 x 4G"$/Yr)

Sequence Tra,nstant i

Core bmage Core braage Frequgncy Frequgncy Transient S.rquence Osantification (Yr *

(Yr )

1 Ira /Pa Failure 3 2 x 30*'

j 1 ( r g

T AY L,L g g a

4 T AY L

(.38)(3.5 m 10-5 ) (. 86 ) ( 1.0. (.9 ) (.57) 1.8 x la g gr TMLL 1 fr1 T AY L,C IPR /PR Failure ClosG (T )

.!B/Yr g g g

Failure to SCRAM (A) 3.5 x 10' Successfu' 'nsrlain Dypase valve

.86 Failure of Feedwater Given aypass valve 1.0 is Open (Y )

g Manual RCPT Wtthan ist 60 Seconde

.9 Failure to segin Poison twjection Wathin 3/

/T/ Seconda (L,)

Tj AY CL,'.

g 7 AY OL L

1. [

(.18)(3.5 x 10-5)(.86)(1,0)(.1)(.92) s.O m' 10" T AY oc g g 7

1 f r1 T AY OL,LC Failure to Trip BCP Within let 64

.1 g g SeconJs (0) 4 i

Failure to In)*ct Poison Givea acP Not

. 9.2.

Trippent (L,) [(p9.5GC) e 1

2 Sequwece Transint Core Da.aage Core Damage Fregjgncy Frutgncy ransient Somwnr*

Osantification (vr I (Yr )

f T AD,t, g

7 An OL

?

. T AB,Y, L g

r

(.18)(3.5 x 10 II.44)st.0) 3.3 x Ig*l T AB I oi r 7 AD L Turbine Bypass Valve ra. lure (n,)

.14 3,

4 T A8 Y L Failure to Inlact Ponson C6ven no Environ-I

• mental Osatification of Systies 1.0

{

7 3,,i

& o

(.4)(3.5 x 10'$l t. I l t l.0 ) (.Si t.5M

f. A x 10*#

3.3 x 10 L

T "*i r Spuraous d

2 4

hrt.ine J

Bypasa Valve Spurious Opening of habine Bypass

.1 valve (T i 2

I Successfiel pypass - Operator rails to

.1 Terminate Bloudecn Iksfore RDS T "# 0'r

(.1)(3.5 x 10' li. 8 )(1.0 )(.1)(.9%

3.2 m 10-0 2 f T # 'o

(.1)(3.5 x 10~ 3(.9)(1.01 3.2,a 10*'

2 raiture of Bypass - Operstor 'Nrminates

.9 Blowlown 1.6 m 2.8xId I.96)(1.0)(.9)(.5 M T Av L,

(.16)(3.5 x 10' t

Loss c.f FW 3 g 1

loss of EM (T )

.16 3

i Failure of Makeup Civen Transient (F )

g T Av oL,

(.16)(3.5 x 10* ) (.06 )(1,0 ) (.1 ) (.92) 4.4 x 10" 3 g

,(.16)( 3.5 x 10-5)(.14 )

7.8 x 10*I T As 3

9 e

i I

e

3 Sequence Trarotent Core Dmage Core D4 mage j

Freqqncy Frequgacy Translant Sequence Quantification (Yr n (Yr i tose of 1 FP 2.9 x la' T Aa, t.06)(3.5 x 10 1(.14)

?

2.9 x 10 3

Loss % I reed Pump T,

.14

~

Load Rejection 6.0 x 10

? Aa,, j

(.59)(3.5 x IO* I t.86 )(1.0 ) (1.0 )(.298 5 1 x It

3 Road sa}oction(7 )

.59/Yr 5

RCPT Civen toss of tomt 9 25 seconde 1.0 T AB,

8.59)(3.5 x 10' li.143 2.9 x is g

toss of Hein 6.2 x 10 tt,' ;6L.

T.A.,

t.i.I n.5 x 10-s)(i.0 )

6.2 x 0-'

Air (.06) se of CoMeneer Transiente (T I

• I'###

MSIV Closure (.06) 6 3

l Transient pypass/ Condenser Failure Given 1.0 4

LOSP 4.6 x 10 T ts,

(.13)(3.5 x 10' )(1.0) 4.6 x 10 y

l&P (T l

.13/Yr y

Bypase/Candenser raiture Given taas 1.0 of Pouer Misc. Scrase 1.7 x 10*7 tat.

(.54,14 3.5 x 10-5)(.86)(.08 )

1.7 x 10'I c

Miscellaneous Scraes

.56/Yr 4

}

QUFSTION 7:

"Your analysis states that a delay of LPS injection until after PDS actuation -

is assumed to cause limited core damage with releases similar to the TMI ac-cident. It would seem that such a failure, in conjunction with predicted con-tainment failure, would cause a significant contributian to risk but that is has not been included in your risk calculations. Address the contribution for this case."

i Ri'SPONSE You have correctly surmised that limited core damage in an ATWS event in which LPS injection is delr.yed until after RDS actuation has been predicted to ul-timately lead to containment failure and significant releases of radionucliae to the environment. Our risk calculations have, however, included such se-quencas in the estimated risk from the Big Rock Point plant as indicated in the last paragraph in NOTE 10 on page 89 of the main report.

j In our analysis these limited core damage sequences have been treated as ccre j

melt sequences, and radionuclide releases for these sequences have been judged to be similar to those in release category BR"-3.

These fractional releases from containment are presented in Table 5.5 on page 97 of the main report, and are listed below:

RELEASE CATEGORY BRP-3 RELEASE FRACTIONS X,-K 0.89 T,

0.14 r

I org.

0.0069 B,-S 0.039 I -3 0.083 R

0.12 2 r u

2 C,-Rg 0.30 L,

0.0017 The approach used to include the subject accident sequences in the BRP risk evaluation is discussed in somewhat more detail in Table V.5 h starting on page V-123 of Appendix V.

Of special interes ; is note 1 to this table on page V-131.

Evaluation of the information in Tat.e V.5-h indicates that ATWS se-quences contribute slightly more than 77, of.he total probability of release category BRP-3.

-.