ML20023D604

Discusses Concerns Re Reactor Safety Study Methodology Applications Program (Rssmap) Evaluation Conducted by Sandia Natl Labs.Excessively Conservative Calculations Used.Related Correspondence Encl

ML20023D604
Person / Time
Site:	Grand Gulf, 05000000
Issue date:	01/23/1981
From:	Dale L MISSISSIPPI POWER & LIGHT CO.
To:	Bernero R NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
Shared Package
ML20023A436	List: ML20023D601 ML20023D604 ML20023D606
References
FOIA-83-123 810723, AECM-81-184, NUDOCS 8305240265
Download: ML20023D604 (49)

Text

_ - -

, . ( '

o ,

I j i.,k' MISSISSIPPI POWER & LIGHT COMPANY

• iL*; !Q Helping Build Mis sis sippi j i'Fh'u b M P. O . B O X 1 G 4 0, JACKSON. talS SIS SIP Pl 3 9 2 05 July 23, 1981 .

NUC4 [ AR N40DUCleoN oLFART ME NT Dr. Robert M. Bernero Director, Division of Risk Analysis Office of Nucicar Regulatory Research U. S. Nucicar Regulatory Commission WashinEton, D. C. 20555

SUBJECT:

Grand Gulf Nuclear Station Units 1 and 2 Docket Nos. 50-416 and 50-417 File: 0260/L-860.0/16684/

L-953.0 f i Reactor Safety Study Methodology Applications Program (RSSMAP) on Grand Gulf Nuclear Station Unit No. 1 AECM-81/184 Mississippi Power & Light.(MP&L) is deeply concerned over the

[ Reactor Safety Study Methodology Applications Program (RSSMAP) evaluation j being conducted by Sandia National Laboratories for the Reactor Risk f, Branch. The analysis as described in the draft report uses excessively j' conservative assumptions and calculations, which in combination with the j simplified approach fundamental to RSSMAP, raises serious questions regarding the validity of explicitly and quantitatively comparing the

, Gaand Gulf Nuclear Station (GGNS) to the reference Boiling Water Reactor e

in the Reactor Safety Study (WASH-1400), wherein realistic as,= ump,tions and. analyses were used. *~

l l As a result, MP&L, Bechtel and General Elect ric met with Sandia Laboratories to express our concerns over the methodology of the report ,

and to offer suggestions on more realistic calculations. Subsequently, MP&L provided Sandia with more realistic data in the areas we believed l to have excessively conservative assumptions or calcylations. In recent conversations, Sandia indicated that the data supplied in several prime i

, contributor areas were being incorporated, where appropriate, in the revised report, including infomation on the Power Conversion System, NIVS assumptions and intersystem LOCA's. Rowever, Sandia stated that i

~the remaining issues would not be incorporated into the revised report due to the scheduled completion date for the report and their opinion that the remaining items lacked significance.

We believe that the core melt probability in the revised report

, will be overstated and not representative of GGNS. Ve also believe most strongly that it is both irresponsible and inequitable tq publish an admittedly conservative NRC sponsored report on GCSS and cospare it to t eqaore realistic Reactor Safety Study.

8305240265 830331

.p] PDR FOIA SHOLLY83-123 PDR ,

.{ lg ,

/

r . . , , AECM-Sl/184

~

MidstSSIPPI POWER & LIGHT COMPAWY Page 2 ,

Furthermore, during a period of tinie that probabili:.t ic risk assessment is of strong interest to both industry and the NRC, it i t. in the best interest of all to make the RSS!!AP report as realistic ar possible.

Such an approach is also snore consistent with the chartes of RSS!!AP to use simplified methods to achieve results which can be emnpared meaningfully to the Reactor Safety Study.

We would like to take this opportunity to offer our nt;sistance in any re-evaluation of the report to support this realistic aisproach. In addition, we would like to meet with you and your staff to discuss this subject in more detail.

Yours truly, L. F. Dale .

Manager of Nuclear Services SIUl/JDR:ad .,

cc: Mr. N. L. Stampley j Mr. G. B. Taylor Mr. R. B. McGehee Mr. T. B. Conner

'l Mr. Victor Stello, Jr. , Director Office of Inspection & Enforcement U.S. Nucicar Regulatory Commission .

. Washington, D.C. 20555 0

t e

i

'J -

f' .

'I . r

't ,

't ee -

4

~

I e

i

, wrsus3

[.

(. #

I i MISSISSIPPI POWER LIGHT COMPANY

) b  !

Helping Build Alississippi 9 P. O. B O X 16 4 0. J A C K S O N MI S SIS SIP PI 3 9 2 05 May 28, 1981

, NUCLEAR PFeOCUCf 60N DEPARTMENT Dr. Steven W. Hatch .

Nuclear Fuel Cycle Division {

Sandia National Laboratory Post Office Box 5800 ' '

Albuquerque, New Mexico 87185 .

SUBJECT:

Grand Gulf' Nuclear Station ,

File: 0260/L-860.0/16684/

L-953.0 RSSMAP Study Additional Comments APO-81/280 Attached are copies of comments which were prepared by D. C. Gibbs of Middle South Energy, Inc. and by Mike Lloyd of Middle South Services.

Portions of t,hese comments were discussed during our meeting on March 5 and 6, 1981; however, Mr. Lloyd's comments were not available until after that meeting. We believe that these comments deserve your consideration. p Yours truly, -

/ A L. F. Dale .

Manager of Nuclear Services '

SIDl/JDR:ad Attachments cc: (See Next Page) , .

k 5'"

Member Middle South Utilities System

" _ _ _ _ _ _ _ _ . _ _ _ _ _ . ____m_-_ __

, [ ( Al'O - b l/ .'S O

, , ' . . ' MISdlSSIPPI POWER & LIGHT COMPANY Page 2 a cc: Dr. D. C. Cibbs Mr. C.'W. Wood Mr. H. H. Weber Mr. Adrian Zaccaria -

Mr. ' R. S. Trickovic -

Mr. J. N. Ward '

e Mr. N. L. Stampley Mr. T. H. Cloninger Mr. J. P. McGaughy -

Mr..T. E. Reaves I

Mr. C. K. McCoy_

i ' Mr. G. B. Rogers

• Mr. Mike Lloyd Mr. A. Smith

, Mr. L. S. Richardson Mr. D. B. Bitter Mr. R. J. Candless Mr. W. P. Sullivan Mr. W. G. Gang 1

Mr. Jim Curry (NRC)

File (All of"above w/ attachment) ,

=

h ,

+

he g 8

_a e

N w em 6

.t g t

s

" e s -

(~' (~E '"] 7} ( ,

u.a *

- ~

MIDDLE SOLITH ENERGY INC/225 BARONNE/NEW ORLEANS. LA 70112/(504)529-5282 MEM0RANDUM -

March 5, 1981 i

TO: John Richardson -

FROM: D. C. Gibbs .

SUBJECT:

Grand Gulf RSSMAP The following-comments were transmitted to you verbally by telecon on March 2, 1981. I expect to be able'to develop additional comments on this document following' review by Mike L1cyd, currently at Charlotte, at work on the Oconee 3 PRA. These vill b~e transmitted to you when available.

1. Pg. 2-4_ The formulation of a single LOCA event. tree is incorrect. RSS methodology should have been retained. The response of the plant to a 2-inch break ~ is c1carly dif ferent f rom its. response 'to a 12-inch break and is much more forgiving to subsequent failures. The grouping of all small break LOCA's into one category excessively. biases the results in

!- an unfavorable direction.

l '2. PA 2-4 PRA cannot be performed without event trees. The " survey and analysis"'techniqua described on this page provides no insight on system failure modes and is not an acceptable methodology. } .

~ .!

3. Pg. 2-7 Exclusion lof epsilon event sequences is not acceptable. These sequences' form the largest contribution to core melt probability, and it is-likely that core melt' probability will become one of the safety goals. . -

4. Pg. 2-8 One wonders to what extent the study was misguided from the onset. Consequence analysis is of little interest to the owner, and judging -

from what is said here, was only incompletely performed. ' It 'seems that a

!4 . better product would ~ have resulted had the authors confined their attention to the fault and event trees leading to core melt and omitted any conse-=

quence activities.

t. : 5. Pg. 3-3 It-is certainly not correct that the SPMS would be called upon for all. LOCA events :as is suggested here..

6. . h 3-4 Failure of.three adjacent rods to enter. the core would constitute

~ RPS unavailability only under extreme. circumstances. PRA should not make

. deterministic assumptions such as this one to evaluate' plant performance under unusual circumstances--doing so invalidates the probabilistic nature

• ~of.the analysis.

. Memorandum John Richardson March 5, 1981 Page 2'

7. PE . 3-6 The statement regarding VSS failure through suppression pool bypass is not supported in Appendix B4. There appears (o be little if any basis for this statement.

i 8.

Pg. 3-7 Where do the authors get the idea that plant operation would be ... j permitted during HPCS unavailability? Is this contemplated in Tech Specs? I'

9. Pg. 3-8 Same comment as above pertains to RCIC. i.

applies throughout this section. The assumptions This comment in general made by the analysts

{

should be checked against the tech specs for allowable operating modes.

10. _Pg. 3-9 ADS automatically actuates. Operator failure to actuate should not be a consideration except on the small LOCA case.

11. Pg. 3-12 erroneous. The conclusion relative to G.G. RHRS, unavailability is intuitively This should have alerted the authors to the invalidity of the assumptions made in Appendix B10, where gross oversimplifications relative to available operating modes have created spurious results.

12. Pg. 3-12 SSW makeup capability between units is'available at G.G. This seems to have been"omitted from consideration by authors. Also, limited

~

capability for use of PSW appears to be avai_'able in an emergency for une in cooling RHR coolers for example. ,

13. Fig. 3.1 HPCS not shown.

14. Pg. 4-2 Bottom of page - Differ with this procedure. It is not acceptable to group SI , S2, and L into one group.

7 .

15. Pg. 4-3

- The procedure of grouping transients places unacceptably high probabilities on transients which are sensitive to loss of feedwater. The RSS approach should be retained. The basis of this procedure is not adequately described in text. ~

15. Pg. 4-5

}bny transients do not require relief valve ~ actuation.

16. Pg. 4-6 It appears core melt would be unlikely in the presence of HPCS and RCIC regardless of status of RHRS.

4 DCG:eal

- File: 041-01

• s i Q g d y N h at..e w asv.J bsad4-4 7 *'

• C *

. , - . - . ,.,,..w

• ( ', .

w wil . i[i ,

{-

,p, , . . -

9 x ,,f'! '

i -

.-.- - a .

i

) i b hl .

mwm h 2 --

MIDDLE SOUTH ENERGY. lNC/ 225 BARONNE / NEW ORLEANS LA 70112/[504)529 5262 April 28, 1981 #

$ APR 3 01981 l NUCLt AR PROD. Dt?T vast " ....

Mr. Norris L. Stampley

, Vice President-Production Department P.O. Box 1640 Jackson, MS 39205

Dear Mr. Stampley:

' Attached is an assessment prepared by Mike Lloyd of my staf f on the Grand Gulf RSSMAP. Mike is currently assigned to the. Duke-NSAC PRA Project on Oconee-where he is developina_his skills for our futur m as necessary in this area.. I feel his comments are worthy of Project consideration.

Sincerely},ours, D.'C. Gibbs DCC: cal File: OIel-01 Attachment 6

,O r

,NJ s'

e k

.s w 6

.\

L . '_ ':. _; _

-( (, ,

MEM0RANDUM -

. April .13., 1981 .

TO: Dr. D. C.l Gibbs i FROM: M. Lloyd -f)/ . r <O -

SUBJECT:

Review of GGN RSSMAP Draft Report As per your request, attached is a review of the Grand Gulf Nuclear RSSMAP Draft Report. This report was provided to MP8L as an attachment

~ to a letter from Mr. Steven W. Hatch (SNL) to Mr. Larry Cale (MPAL) dated December 12,.1980.

, The review comments and questions concerning the RSSMAP a 'nalysis are organized in' the ' attachment' according to their perceived impact on the re-'

sults: (1) general comments, (2) comments / questions relating to specific event tree _ top events, and (3) textual conneTts'. Incorporation of the- re-

. view comments should, by removing some of the more conservative RSSMAP

. analysis assumotions, result in a more realistic estirate in the Grand Gulf core melt frequency.' A rough assessment of the effect of removing some of-the conservatism from the GGN. RSSMAP analysis is noted in review comments.

In general, a reduction in the overall core melt frequency by an' order of

, magnitude or:more can be expected.

y .

i ML: sam Attachment

s cc:' Dr. T. W. Schnatz

. File: 041-01'

- +

-- e w . . , -

t r e

• g w (

o i

~

.\

s I

l C0tMENTS 'ON THE GRAND GULF NUCLEAR REACTOR SAFETY STUbY METHODOLOGY APPLICATIONS PROGRAM (RSSMAP) DRAFT REP i

f April 13,1981 e

t_ ,

.L. -

a LOCA CATEGORIZATI0ff i

The definitions of LOCA break sizes in the GGN RSStiAP are very coarse. '

Break sizes are divided into on1y two LOCA sizes: large LOCA (designated with the letter A, with diameter >l3.S") and small LOCA (designated with the letter 5, with diameter < 13.5"). This categorization is stated to have been obtained through i

discussions with GE. A revised LOCA break size categdrization is probably now available through GE.

The RSSMAP categorization appears to be based on the requirement for the Auto- [

matic Depressurization System (ADS) before use of the low pressure ECI system is possible. A re-evaluation of the L0CA break size range requiring ADS more rea- l listically accounting for plant equipment / physics phenomena would show that there [

is no need for ADS actuation for much of the break range below 13.5 inches. In- l I

clusion of this fact, in the RSSMAP analysis would reduce the frequency of core melt associated with S LOCA sequences, i

i

" Consideration sh'ould also be given to the possible use of the Control Rod Drive Hydraulic System (CRDHS) as a means of cooling the core.

Incorporation of the CRDHS cooling would define a new small LOCA treak size category. It would have the effect of improving the calculated availability of S LOCA event E and transient event V.

SUCCESS CRITERIA, _,

The GGN RSSMAP system success r c' iteria are sumarized on Tables 4-2 and 4-4.

These success criteria are probably very conservative. The RSSMAP text stat.-

that they were based on discussions with GE and MP&L. A newer, more realistic _

set of success-criteria is probably now available from GE., Application of the new success criteria can be expected to improve most safety system availabilities.

GENERIC EVENT TREES _

Use of single generic event tree for LOCA and another for transients is ac-L ceptable provided the event trees are tailored for the specific LOCA size (either L

Li

(. l. -

A or S LOCA) and for the specific transient type (either Ty or T23). Such tailoring was accomplished in the event trees via specific LOCA size dependent and transient-type dependent success criteria, some of which are "shown on Tables 4-2 and 4-4 t

DATA Pump and MOV maintenance data values used in the GGN RSSMAP are identical to those used in the RSS. All pumps and MOVs were assumed to have a mean outage time of 19h and to have a mean maintenance interval of 4.5 mos. These values re-sult in the mean pump /MOV unavilability of 5.8 x 10-3 The outage time value is probably overestimated by as much as an order of magnitude for power operation

, during emergency conditions. Thus, the pump and MOV unavailability is probably considerably overestimated. Since the GGN RSSMAP found the maintenance unavail-ability to be one of the largest contributors to system unavailabilities, a significant reduction in the pump and MOV maintenance unavailabilities should re-sult in a significant reduction in system unavailabilities and hence a reduction in the frequency of core melt.  ;.

+

SEQUENCE FREQUENCY VALUES -

It appears that arithmetic errors occur in the dominant sequence frequencies quoted in Chapter 6. A comparison of the RSSMAP quoted and the re-calculated frequencies is provided on Table 1 b.elow. The "re-calculated" values iin this table are the sum of the quoted dominant cut set frequencies and, as such, gen-erally represent 90% or more of the total sequence frequency (as noted on RSSMAP .

pg. 6-2). ,

9 0

2

i, . ( o TABLE 1

~

Sequence . , ' Frequency (R-Yr-I). . .

RSSMAP Quoted Re-Talculated

. T1PI 1.4 x 10-5 . 4.2 x 10 -6 4'

T PI-23 4.6 x 10'4 l.7 x 10~4 T PE 3

6.4 x 10 0 3.1 x 10-6 ,

SI 4.6 x 10-6 1.7 x 10 -6 T3Qil 5.9 x 10-6 3.1 x 10-6 T23QW l.1 x 10-5 1.3 x 10 -5

-6 T 3QUV 2.1 x 10 2.7 x 10-7 LOCA EVENT C, TRANSIENT EVENT C The assumption that failure of the reactor protection system (RPS) leads ir -

reconcilably to core melt is not appropriate. Credit should be taken for the PCS.

for small LOCAs and 3T transients and for the Standby Liquid Control System (SLCS) wherever possible.

I LOCA EVENT E, TRANSIENT EVENT V Is the reduced requirement for low Pressure Coolant Injection (LPCI) after core reflood modeled? b '

The Low Pressure Coolant Injection (LPCI) failure analysis does_ not appear to -

model the failure of RHR heat exchanger and heat exchanger by-pass flow paths; namely, valves F047A-A, F120A, F130A, F003A-A for RHR heat'exchangers Al and A2 a'nd valve F-48A-A, for the RHR heat exchangers A bypass path, and the equivalent flow path associated with RHR-heat exchangers B.

-No cre'dit is g'iven to operator actuation of the ADS following a small LOCA. .

This assumption is very conservative. Removal'of this assumption should reduce -core -

melt frequencies involving SE sequences by at least a factor'of two. ~

W 3

- o

(,  :

e.

No credit is taken for Reactor Core Isolation Cooling (RCIC)' System for LOCAs. The reason given on pg. Al-2 is that GE did not include the RCIC system in its small LOCA plant response analysis. If an updated small LOCA analysis which includes the RCIC System is not yet available, such an analysis should be parformed either by GE or as part of the RSSMAP study. Inclusion of the RCIC in the RSSMAP analysis would reduce the frequency of core nelt for TPE and SE sequences by more than an order of magnitude.

~

LOCA EVENT I, TRANSIENT EVENT W_

The GGN RSSMAP analysis accounted for a number of RHR operational modes fol-lowing and initiating event. For LOCAs, RHR was allowed to operate in the LPCI mode, in the Suppression Pool cooling mode, and in the Containment Spray mode. For transient events, an additional mode, the steam condensing mode, was allowed. Cre-dit should be taken for the RHR steam condensing mode for small LOCAs when contain-ment conditions allow. It is important to note that the steam condensing mode requires RCIC pump operation. After a small LOCA, such operation may require operator action to open the main steam isolation valves (MSIVs), the RCIC turbine steam supply line isolation valves, and the RCIC pump discharge isolation valves.

Accounting for the possible oneration of RHR steam condensing should significantly reduce the RHR unavailability following a small L'OCA. This reduction would have the effect of reducing small break core melt frequencies involving sequence SI by more than two orders of magnitude. Credit should also be taken for a fifth mode of

. RHR operation, normal. shutdown cooling, following both transients and small LOCA events. This mode of operation is applicable only at low RCS pressurepnditions.

Operation in this mode would require operator action to align RHR pumps A and B suction with the Reactor Recirculation loop B following the initiating event. Ac-counting for the possible operation of the RHR normal shutdown cooling should have -

the effect of reducing RHR unavailabilities at low RCS pressure conditions. This reduction would have the effect of reducing core melt frequencies of large LOCAs or transients with successful ADS operation.

Has any credit been taken for the Refueling Water Storage Tank (RHST), the

• Standby Service Water System (SSUS), Fuel Pool (FP) water sources for core and containment injection? The P.WST may be used as a backup to the CST.. The FP and SSWS may be directly connected to the RHR system such that service water is directly-injected to the Suppression Pool (SP). Use of these and additiona,1 water hauled onto .the site can be.used to greatly extend the core melt and containment failure times following a LOCA without RHR.

4

, g- p. u o o TRANSIENT EVENT P The probability of 5/RV failure to reseat has been assigned the conservative value of 0.1 in the GGN RSSMAP analysis. This value is identical to that used in '

the RSS. A more realistic value may be an order of magnitude low'er. This reduction in the value of P would directly reduce by an order of magnitude the frequency of core melt involving sequences Ty P and T P 23 TRANSIENT EVENT Q -

The GGN RSSMAP combines the T2 (.1 ss of PCS) and T3(othertransients)intoa single transient category, given the designation. T This combination apoears

, 23 inappropriate in light of the expected behavior of the power conversion system (PCS).

The RSSMAP study assumes that the PCS is lost for T transients; whereas, the RSS 23 dssumes that the PCS is initially available during the 3T transient. Credit should be taken.for the initial availability of the PCS for T transients in the RSSMAP 3

analysis. The reasoning behind the RSSMAP assumption that PCS is lost for T an-3 sients is either that (1) the main steam isolation valves (MSIVs) close on a small LOCA and the operator is not credited with reopening them, or (2) the PCS is not considered capable.of effectively removing the core decay heat given the_ failure of an S/RV to close. Both of these assumptions aYe conservative. For the fonner, credit should be taken for operator action of opening the MSIVs' on a small LOCA, especially if emergency procedures specify such action. (Do emergency procedures specify such action?) For the latter, the PCS would probably continue to carry off a significant amount of decay heat. The realistic accounting of PCF ::vailabil-ity following T transients 3 would reduce frequency of core melt associated with T 3

sequences involving event I.

CONTAINMENT t

Containment failure is predicted to occur within 27 hours3.125e-4 days <br />0.0075 hours <br />4.464286e-5 weeks <br />1.02735e-5 months <br /> following the ini-tiating event given that no decay heat is removed from the RCS or the Suppression Pool (SP). The 27h containment failure time assessed for the GGN Mark Ill con-tainment is identical to 'the RSS Peach Bottom Mark I failure interval, yet the two containment structures are substantially different from each other. How was the 27h containment failure interval arrived at for GGN?

i. 5

. .. p. . . c.. ..

I l

The containment failure time could be substantially extended by allowing ,

l the' operator to vent sone of the excess pressure to the atmosphere prior to the core melt. The GGN RSSMAP takes no credit for such an emergency operation. -

Consideration should be given to including such an operation into the plant emergency procedures (if it is not already part of them), and into the risk assessment analysis.- 5 i

TEXTUAL ERRORS ,

GGN RSSMAP page B10-2, second paragraph, second sentence should read as follows: Main steam is fed through the RHR heat exchangers and condensed, after which it is supplied to the suction of the RCIC pumps and returned to the reactor.

Valve F005-A should be listed on Table B7-2 as normally closed rather than normally open.

i l -

c.c .

e e

. - 6 '-

. a ' '. . -

j j-* ~

,- , ._ _ . , ,, , , _,_o_ . , _

, ~ f- -- mm rum unu

.f Jb ' '/b

_k MISSISSIPPI POWE & LIGHT COMPANY Helping Build Afississippi 6 ""M wf P. - O . B OX 164 0. J A C K S ON. MIS SIS SIP PI 3 9 2 0 5 June 1, 1981 ,

NUCLEAR PROCUCTION DEPARTMENT Dr. Steven W. Hatch Nuclear Fuel Cycle Division Sandia National Laboratory P.O. Box 5800 Alberquerque, New Mexico 87185

Dear Dr. Hatch:

)

SUBJECT:

Grand Gulf Nuclear Station File 0260/L-860.0/16684/L-953.0 RSSMAP Study APO-81/159 At our meeting on March 5 and 6, 1981, to. discuss the draft NUREG/CR-1659, Volume 4', Reactor Saf2ty Study Methodology Applications Program (RSSMAP), Grand Gulf Unit 1 BWR Power Plant, it was agreed that MP&L would provide information for use in revising the draft RSSMAP voltrne on Grand Gulf.'

Attachment I lists the information and where it is provided.

Attachments II and III contain the remaining information.

v We appreciate the opportunity to provide this information.

Yours truly, e$ N L. F. Dale

. Manager of Nuclear Services SHH/JDR:Im Atta clunents .

cc: (See Next Page) 9 6

-~

~' Member Middle South Utilities System

(;_ _

o MississlPPI POWER & LIGHT COMPANY l 'a ,

APO-81/159 Page 2 cc: Dr. D. C. Gibbs (w/a)

Mr. C. W. Wood (w/a)

Mr. H. H. Weber (w/a)

• Mr. Adrian Zaccaria (w/a)

Mr. R. S. Trickovic (w/a) -

Mr. J. N. Ward (w/a)

Mr. N. L. Stampicy (w/a)

Mr. T. H. Cloninger (w/a)

Mr. J. P. McCaugby (w/a)

Mr. T. E. Reaves (w/a)

Mr. C. K. McCoy (w/a)

. Mr. G. B. Rogers (w/a)

• Mr. H. C. Fron (w/a)

Mr. M. Lloyd (w/a)

Dr. Peter Cybulskis (w/a)

Mr. J. Curry, NRC (w/a)

Mr. L. S. Richardson (w/a)

Mr.lD. B. Bitter (w/a) ,

Mr. R. J. McCandless (w/a)

Mr. W. P. Sullivan (w/a)

Mr. W. C. Gang (w/a)

Mr. A. Smith (GE) (w/a)

Mr. L. S. Richardson-(GE) (w/a)

Mr. D. B.' Bitter (GE)' (w/a) '

Mr. lt. J. McCandless (GE) (w/a)

Mr. W. P. Sullivan (GE) (w/a).

Mr. W. G. Gang (GE) (w/a) Q

' File (w/a) s a "$

e

+; .

-__m . __

. T

ATTACEMENT I . +-

Itcm- .

Number. Item Response .'

1. Technical' Specifications *

1. Provided on March'24, 1981 in APO-81/129.

2. - Emergency Procedures '

2. Provided on March 27, 1981 in APO-81/140.

3. Estimate of containment ultimate pressure capability. 3. Attachment III.

4 Low pressure operating capability for the power conversion system. 4. Attachment II.

5. Reactor water cleanup heat removal capability for emergency conditions. 5. Attachment II.

6. Alternative water sources to the suppression pool such as fuel pool. 6. Preliminary responses in Attachment II.

7. Estimate of the loss of offsite power frequency and' recovery probability 7. Not provided, see reasons in Attachment II.

8.

~

Motor operated valves permissive pressore involved in the inter-system 8. Attachment II.

LOCA.

9. Parametric calculations to show how long it . 11 take the suppression 9. Attachment II.

pool temperature to reach 212 F from various initial pool temperatures with a stuck open relief valve and the main sCeam lines cpen to the condenser.

210. Estimates based on operator inputs of how long it would take to 10. Attachment II.

reopen the main steam isolation valves following a loss of reedwater event with a stuck open relief valve and loss of residual heat removal. The estimates should be consistent with previous ,

estimates of .5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> for the Limerick PRA.

~

'l l .

Estimates, based on operating plant data, of how many transients plus 11. Attachment II.

loss of offsite power events per year are expected to require plant shutdown for BWR/6 plants. r3 79-

<d (1) e

g. . . . _ _-. .. .

' ATTACHMENT I ,o

. It@ .

Number. Item Response .'

12. Estimates, based on operating plant data of the safety relief valve

• 12. Not provided, see ' reasons in opening frequency for BWR/6 plants (i.e., safety' relief valve challenge ' Attachment II.

• frequency). '

' i-

13. Review NUREG-0626, Page F-7 and provide an explanation of.the estimate 13. Attachment II.

of .16 safety relief valve failures per challenge provided there and explain how it relates to General Electric estimates for the probability-of stuck open relief valves for BWR/6 plants (Dikkers and Crosby Valves).

Relate the General Electric' estimates _for a stuck open relief' valve

probability for BWR/6.to all available' operating plants and test data by showing the-data, analyzing failure modes, and discussing design

' differences between the-various. safety relief valves designed.

14.. Estimate of the' stuck'open relief valve recovery probability based on 14. Attachment II.

operating plant experience._ ,

15. - - Estimates'of the standoy liquid control reliability for ATWS 3A 15. Attachment II.

implementation at GGNS. * '

16. ATWS success criteria' estimates for GGNS. "" '

16. Attachment II.

17. Available operating plant data for ECCS.and RHR system maintenance

' 17. Attachment II. ,

availability versus technical' specification allowables.

18. Estimates.of the needed time to repair the BWR/6 RER system. 18. Attachment II.

19'. BWR/6 RHR reliability estimates applicable to GGNS. 19. Attachment II.

20. Basislkor revising the Sandia T QW and T QW event' probabilities. 20. Attachment II.

Theseeventsare'transientswitkandwitkhut'lossofoffsitepower ,

.' events followed by loss of power conversion system and RHR system.

a a -

+

• 4 (2) i 1._ - -

. . . :2 ,

GRAND GULF IhTOPJ1ATION PROVIDED FOR RSSMAP STUDY Item M

4 1 5 2 6 3 7 4 8 5 9 6 10 8 11 9 12 10 13 11 14 16 15 17 16 18 17 20 18 22 19 23 '

20 24 f n *

)

, i e

e

', . I ten . 4' Provida information on. the low pressure operating capability

.. . for.the power conversion system.

Response .

The' concern is the ability to maintain condenser vacuum at low main steam pressures. Condenser vacuum can be maintained as long as:

4 (1) Circulating Water is available.

(2) The Air Ejectors are operable.

The Air Ejectors require 100 psig steam to operate. This is normally supplied from the main steam system,' but auxiliary steam can be used if main steam pressure is low (e.g., during startup or during an event where main steam pressure is less

, , than 100 psig). Using auxiliary steam, condenser vacuum can be maintained indefinitely without regard to low pressure in the main steam lines.

In the event that condenser vacuum must be established, the mechanical vacuum pumps are used. These are motor driven pumps and can be operated independent of the availability of main steam.

9

. h-e

,$ m

.g -

J i

-* -5 g ,-

~

(1)z

~

y ,

~

L'_ __  ; Q.) ' W - -

- = 1

.. . . 2.

9

.. Iten 5 - Provide an estimate 'of-the Reactor LWater Cleanup (RWCU) System

, , . capability for heat removal during emergency conditions, Response .

' The. Emergency, Procedures discuss how the RWCU System can be used for heat removal during emergency conditons. *

-The two non-regenerative heat exchangers would be used for

~

rejecting heat' to the Component Cooling Water System. Design information for each heat exchanger follows:

Design Pressure, psig

-Shell 150 Tub'e 1420 Design Temperature, F Shell 370' Tube 575 Normal Operation Flow Rate,.lbs/hr Shell 405,000 Tube 178,000 Heat Flow Rate, Btu /hr 6 19.7 x 10 h

3

=

e 4 '

$. Y

~

i.

.(2)

3 i

L._ -.

g

s. A

• 1 ,

(-$ , .

g

A A A Acnnt.h1 11

' ' Itea 6 - Provida information on alternative water sources to th2

, suppression pool such as the fuel pool, standby service water, etc.

Response

The plant operations staff is thoroughly trained in the details of the-plant systems. As a result, there is an awareness of the system interconnections which are normally used for suppression pool makeup or cooling as well as system interconnections not normally used which could be used if neces sa ry. Normal makeup sources are:

Upper Pool Dump Suppression Pool Makeup (about 500,000 gallons available from the Condensate Storage Tank and the Refueling Water Storage Tank)

Standby Service Water (Containment Flood Connections emergency use only)

' Fire Protection (Suction available from two tanks of 300,000 gallons each with automatic makeup to each tank from the Plant Service Water System - emergency use only).

^

Other non-normal Makeup sources are:

Fuel Pool (if no spent fuel is stored)

Condensate System (lwater available from condenser hotwells through Suppression Pool cleanup.

connections to the Condensate Cleanup System)

The Emergency Procedures at present do not list such alternative water soarces or appropriate system lineups, but the necessity of monitoring and controlling suppression pool temperature is emphasized very strongly in the Emergency Procedures.

9 9

.O a

(3)

Ite: 7- Previde en estimrte of the loss of offsite power frequency and recovery probability versus time for GGNS.

Response

No response will be provided at this time due to the facts .

that:

1) Reduction in core melt probability would be '

relatively small; and

2) Information currently in the FSAR is in the form of transmission Ifne reliability per unit length and considerable additional analysis is required to obtain offsite power frequency and recovery probability as a function of time.

Should this information be needed, it will be provided.

4 e

k W

9 4

e (4)

-' - tu . '.cn;;:.:o 12

. e- ,

., , Ites 8 - Provida the low pressure permissive setpoints for the motor

, operated valves involved in the inter-system LOCA.

Response .

Section 7.6.1.3 of the FSAR discusses high pressure / low pressure interlock protection for LPCS, RHR and the Feedwater

. Leakage Control System. The setpoints for the valves listed in FSAR Section 7.6.1.3.3 are:

Valve Function Setroint 490 psig

'LPCS Injection (Operation)

LPCS Injection (Testing) 50 psig Recirculation Suction (RHR) 135 psig*

Feedwater Discharge (RNR) 490 psig*

Head Spray (RER) 490 psig*

Steam Condensing (RHR) **

FW Leakage Control :23 psig*

~

• Preliminary response.

• • Response will be provided.

.. 6-b J

7..

4 4

5 a

^

l(5)

- s

' 61 . '

-m -

~ --

i,. .

Itca 9 - ~ Previda parzmetric calculationa. to show how long it will take a;

th2 suppression pool temperature to reach critical point where E

containment fails given a stuck open relief valve (SORV), loss of residual heat removal (RHR), and the main steam lines open f _;, . -

to the condenser.

4<;,. , ,

Response

y  ;. The following assumptions were made:

1. - j -

.l t i a) Initiating transient leading to successful scram.

');- 'y '

.b) Main steam line isolation, p,

s- ,' '

c) SORV

, e..

3*

[f

, d) Loss of RER.

• t j e). Variable times for reopening of all four MSIVs and

{.1 restoring main condenser heat removal.

,, f) Containment failure pressure of 50 psia. '

p, . , jg ,

, g) Upper pool dumped to suppression pool.

, !jf C y .

+ 3. l 'h) Suppression pool temperature is 100 F at start of j

~

/ A

/

, transient.

, s 5

/i) Manual ADS raises suppression pool temperature to 185 F.

cf ,

This is maximum temperature which would be reached by

'/ -# 9 following Operator Guidelines.

g / i m j)- Vessel water level maintained such that steam flows

,'4 , through relief valve's and MSIVs.

4n 1 M Y

/k) Water makeup sources to the core are,available.

d.

1 F

1) ' The. increase in water-level in the suppression pool

1. because of the flow from the reactor pressure vest.el *

, -(RPV) is neglecte'd.

- , 4 s y ,f . m) '

The sensible heat of the RPV at scram is neglected.

.n) . The heat capacity of the containment walls and residual water in the upper pool is neglected.

j , / ' y ,. - Results.

s "

se i '; y The following are the calculated times to containment failure g% - ( y for, various times -: for. reopening the ' MSIVs . (main condenser

/; . available'.as heat sink).

s er f 'f ,Q,;[v -f k

3. .

'+,

.)] _

L-fe .

m (6)-

+

>._I,

l. "

_. l_ l' ( l l

p .

.__u. = i l Itea 9 - Response (Continued)

Time to Time to Reach Reopen Containment Failure

, MSIVs Pressure (Days)  :

1/2 Hr. 44.0 1-. Hr. 34.7 3 Hr. 11.6

• 5 Hr. 2.8 '

It' can be concluded from above that there is more than enough I time to restore the RHR. Realistically, if the initiating transient causes isolation from the condenser (for the majority of transients no isolation will occur), the MSIVs can be reopened in 20 minutes to.1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br />. If the higher 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> value for MSIV opening is assumed, the time to containment.

failure is 35 days. This more realistic failure time period

, is significantly' longer than the 33 hours3.819444e-4 days <br />0.00917 hours <br />5.456349e-5 weeks <br />1.25565e-5 months <br /> assumed in the 1 Sandia Reactor Safety Study Methodology Applications Program (RSSMAP) for Grand Gulf 1. The use of a more realistic failure time (~ 35 days) provides the basis for restoring the RHR or taking other actions to prevent overall containment failure.

l

+

4

-4

?*O e

B O-k 9

-t..

O (7)~

IT _

n. 7. ' ' , 4 . ,, _. . .. , .. . _ .

. . , . , . .,. .. . ___l_________. _ _ _ _ _ _ _

. =

p -s-

, . Iten 10 - Provida estimates,. based on operator inputs, of how long it

,; .. . would take to reopen the MSIVs following a loss of seedwater event;or other transient requiring MSIV closure with an SORV-and loss of RHR.

Response 1 Based on a survey of operations's people, the time to open the HS1Vs is generally between 20 minutes to 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br />, but can take e as long as 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> in some cases. Given a SORV event occurs following a MSIV closure, the operator first makes a determination as to whether the MSIVs can be opened (i.e.,

. determines radiation levels are within specified limits).

MSIVs can be opened by pressurizing the area between the MSIV and the stop valve with high pressure steam from the reactor side through the main steam drain line.

- Operator 1 training emphasizes the use of the main condenser as a heat sink if at all possible.

9

=

.g 9

• '6 sg e

• 4 I

e

,  : (8)

N 5'

, , .a m -e-- e

, , . . - , 4 _ . __

Iten 11 -- Provids estimates, based on operating plant data, of how many transient's plus loss of offsite power events per. year are expected to require plant shutdown for BWR/6 plants. >

Response .

NEBG-25202( ) provides an estimate of the total transients (in

• terms of scrams per year) based on BWR plant operating experience (21 domestic BWR plants from first generator synchronization to the end of'1978, approximately 121 total reactor years of operation). After two years of operation the average number of transients is 7.4 per year. If all improve-ments identified in the report are incorporated, the transient frequency is reduced to 4.0 per year.

The BWR/6. plant design (Grand Gulf) can be expected to have-a lower transient frequency then current operating plants based on several incorporated design improvement;. Our best

, estimate of. the transient frequency for a BWR/6 Grand Gulf type plant based on, design improvements is 5 to 6 transients per year.

. (1) NEBG-25202, " Analysis of Scrams and Forced Outages at Boiling Water Reactors."-May 1980, Sandia Laboratories Contract 13-6441.

l,

'G 9

l

[ ~ c.

. - ~

l

[

e t

. . . . (9)-

• - e'- -e. 4

, a_ - 2 Iten 12 - Provida~ estimates, based on operating plant data of the SRV opening. frequency for BWR/6 plants (i.e., what i~s the SRV challenge frequency expected to be for BWR/6 plants).

Response

Our estimates are not different than that implied in the RSSMAP results. ,

t e

P 4

r.~ .

. W O

O (10)_

Q_ m __ _ _. _ _

. Itca 13 .R;vicw NUREG-0626 pag 2 F7 and provide cn explanttien of the estimato of "0.16 SRV feilure per challenge" provided there, and explain how it relates to GE estimates for the probability of SORV for BWR/6 plants (Dikkers and Crosby valves). Relate the GE estimates for SORV probability for BWR/6 to all available operating plants and test data by showing the data, analyzing failure modes, and discussing design differences between Dikkers, Crosby and Target Rock SRV designs.

Response

NUREG-0626 page F7 states that "the record of relief valve failures to close for all BWR plants in the past three years of operation is approximately 30 in 73 reactor years (0.41 failures / reactor year)." The report further states that failure rate per challenge as 0.16 failures per challenge and 0.04 failures per individual valve challenge (assumes 4 valves are challenged).

The GE experience base for domestic BWR/2, 3, 4 operating plants indicates 57 failures in 920 individual safety relief valve years (0.062 failures per S/RV year) or 57 failures in 120 reactor years (0.48 failures per reactor year). This failure rate is approximately equal to the NUREG-0626 value.

The following is a discussion of the analysis of the various types of safety / relief valves in operating and future plants.

Using the current Three Stage Target Rock Valve as a benchmark, the improvement factor for the Grand Gulf Dikkers type safety / relief valve is estimated to be 0.125. Applying this factor to the current Three Stage Target Rock Valve failure rate the failure probability of the Grand Gulf Dikkers Safety / Relief Valve design is 0.01/ challenge (0.002 per individual valve challenge).

RELATIVE PROBABILITY OF A SAFETY / RELIEF VALVE TO STICK OPEN Introduction p.

The purpose of this analysis is to estimate for various types of Safety / Relief Valvre (S/RVs) the relative probability to stick open following normal valve opening. Three Stage Target Rock, Two Stage Target Rock, Dresser Electromatic, Crosby, and -

Dikkers Valves are considered. The Three Stage Target Rock Valve was taken as the benchmark valve an~d assigned a factor of 1.0 for probability to stick open. The factors were then estimated for various types of valves as described in the following sections.

Two. Stage Target Rock Valve

'~

Following the series of Stuck Open Relief Valve (SORV) events associated with Three Stage Target Rock Valves in operating plants, the valve topworks have been redesigned. The valve equipped with the modified topworks is referred to as the Two Stage Target Rock Valve.

(11) l

}

r 1

. Itea 13 - Response (Continued) i In this study, the improvements are quantified by calculating the percentage of the failure modes which contributed to SORV '

incidents in the old design and which will be eliminated with the two stage design. ,

The 57* events that occurred in domestic BWRs were studied and the failure modes are tabulated in Table I. In'the case of 20 of the 57 events, the failure modes were not identified. Of the remaining 37 events, 31 (i.e., 84% of 37) would have been avoided with the two stage design. The failure modes whose climination results in th(s improvement have also been identified in Table I. An assumption was made that the failure modes for the 20 events with unknown causes and the 37 events with known causes were similarly distributed amongst -

. the various failure modes. With the assumption, 84% of all - ,. I(,/ #

3 the 57 events would have been avoided with the two stage v

, valves. It is thus projected that the probability of Two p#9 ,f g,/

Stage valve to stick open is (100-84) or 16% of that of the }7 Three Stage valve. In :her words the relative SORV probability factor'for the Two Stage valve - 0.16. To account for the un'ecrtainties introduced by the assumptions made in the analysis, the relative SORV probability factor for the Two Stage valve is conservatively taken to be 0.25.

Dresser Electromatic Valve-The 57 SORV events associated with~ the Three Stage Target Rock valves occurred _ in nearly 1000 S/RV years of operating history. There have been five ~SDRV incidents associated with the Dresser Electromatic Valve in'about 290 S/RV years of operation. An estimate of the relative-SORV probability factor for the Dresser Electromatic Valve is made as follows:

SORV Relative Probability- * ** * ** #'

Factor for Electromatic Valve *f,#*{**[*" y 3, , f Three Stage Target-Rock Valve

\- 6.346**

'2_x_29.0 57 s 0.13 - -

920 To compensate for the limited experience,-_ this factor is conservatively taken to be 0.25.

~* Includes spurious-blowdowns

• • Using Chi-Squared dis.tribution,'50% confidence level w

(12) 4 Et. . ..

qy.. , ,

Ite:2-13 .Responsa-(Continued). t' FAILURE MODES FOR SORV INCIDENTS .

i

' IN ALL DOMESTIC BWR 2/3/4 PLANTS Is This Failure .

Mode Eliminated Number of by the Two-Stage Cause Events Design?

Electrical' Ground 2 Yes Pilot Valve Leakage 22 Yes

. Unknown 20 *

- Topworks- (includes bellows) 4 Yes .

- Insulation 1 Yes 2nd Stage Piston 1 Yes

~ (stem t outs)

Solenoid Valve ,3. No

~_ Human Error 2. . No Circuit Error 1 ' Yes p.

. Air Actuator-

'l ^

s lNo.

~ TOTAL- 57 Assumed that- same percentage: of the failure ' modes were elirc;- sted as in the case of events with known causes.

e k

e O

s .

(13)

+

%. % 3 ' -f '

...?. - s-.-

n.?na...- G '11 '

\

. 1 Iten 13 -LResponse (Continued)_ I' Crosby Valve Crosby Valves have only limited experience in the' field.

However, extensive production tests have been conducted as -

part of .the valve qualification program. Six valves were subjected to 180 cycles each in the relief mode and 109 cycles each in the' safety mode, which adds up to a total of 1734 cycles. During these tests only one SORV event was experienced, and a design modification has been made to preclude such an occurrence in the future.

The SORV rate can be estimated using the chi-square distribution at 50% confidence level using the above test data to estimate the relative SORV probability factor for the Crosby valve.

Using zero failures in the 1734 test cases:

1.386*

i SORV Probability _ Factor =.2 x 1734 0.04 0.01 (The 0.01 SORV per demand for Three Stage Target Rock valve estimated based on operating experience of 57 SORV events in 730 transients.)

Using as an upper bound one failure in 1734 test cases:

e

2.366*

. Relative SORV Probability Factor = 2 x 1734 = 0.07 0.01 Based on engineering judgement, to account for other modes of failure that could occur during operation, the relative SORV probability factor is conservatively _ taken to be = 0.1pf_r ,

, Dikkers Valve The relative SORV probability fact'or for the Dikkers valve is .

calculated in the.same manner as was done for Crosby valves.

There were 2350 cycles of operation involving 200 valves and no SORV were experienced during these tests:

1.386 Relative SORV Probability Factor = = 0.03-0.01

.. Based on engin'eering judgment' and considering the fact that Crosby and Dikkers valves .are very. similar, the relative SORV

. probability factor. for the ~Dikkers -valve is also .taken to be 0.125.

• Chi-square' distribution ~at 50% confidence level.

'(14)j

-9

i t oa 13 - 7<.;en:.e (Cent naco) r

, Conclusion Using the Three Stage Target Rock Valve as a benchmark with the relative SORV probability arbitrarily assigned = 1.0, the corresponding factors for other types of S/RVs are estimated as follows:

Two Stage Target Rock Valves = 0.25 Dresser Electromatic Valves = 0.25 -

Crosby Valves = 0.125 Dikkers Valves = 0.125 O

O e

4 7-..

E e

4

,G .

9 (15) .

e

A.-...

. . /. t . 1 .1 ,

Iten 14 - Provide an estimate of the SORV reco'ery probability based on

, operating plant experience.

Response ,

Seventy-five of all SORV events have not resulted in a -

complete depressurization (i.e., valve has reclosed at a lower pressure). This assessment is based on a survey of SORV events recorded in the GE Component Information Retrieval System.

D t

6.

e4 3~.,

m O

8 (16) v

iu i t. i . . Zu 2I 4

, .. . Item 15. - Provide estimates of the SLC reliability for ATWS 3A

, impicmentation at Grand Gulf.

Response

A reliability assessment of the SLC (Standby Liquid Control) -

System for ATWS 3A implementation has been completed for the BWR Limerick plant. This same assessment should apply to the Grand Gulf plant since there are no identified design differences significantly affecting the reliability of the SLC system. The as'sessed religbility of the SLC system for ATWS

.3A implementation is 4x10 / demand. A description of the ATWS 3A SLC system design is provided in~ the Limerick analysis.

g 1

e E'*

4 .

i +

h +

A c o

r -

. (17)_. ,

W.~ ,

4)% ,

e i wr v- -

T g-

A1 TACH ENT II

'

• Item 16 - Provide ATWS success criteria estimates for the Grand Gulf

, plant.

-Response- .

Table II provides ATVS success criteria estimates for the Grand Gulf plant.

t e

O O

e (38)

p ' --

g.

ATTACHMENT II Item 16 - Response (Continued).

TABLE II

SUMMARY

OF B'-1t/6 SYSTEM CAPABILITT TOR ATWS EVENTS (With Alternate 3A Modification Incorporated)

EFFECT OF POTENTIAL ADDITIONAL TEATURI.S (tfUTUALLY EXCLUSIVE TAILURE C,0MBINATICFS) e Inad rt at TRANSIENT I'f,"j','fgf[f'

• Reduced Coolant Injection p,f{cedSp.

,g g ATVS Mitigsting Features.

I SLC .I SLC 1 SLC 1 SLC FW + 1 SLC FW + 1 RHR I SLC ADS HPCS ARI TEED- RPI MSIV PUNP' PUMP + . + FW + FV HPCS + FW llPCS + FAILS PUNP + ACTUATES CON- TAILS VATER TAIL TAIL EVENT TAILS 1 RHR TAIL + liPCS TAIL + RCIC RCIC 2 RHR TINUES RUN- AT (1 of 2) TAIL FAIL FAIL FAIL FAIL TO RUN BACK LEVEL I DACK A A

• A .1 N A CLOSURE .

"80INE g A A A N A* A N .A A N N 6* A A N A

'IORV A .N A N N A N A N N N** A A A A LOSS Of Off-SITE N N N N A* N N A N N N ** A A A A POWER Legend:

. A = Acceptable (successful); N = Not Acceptable (not successful)

• This assumes that the operator is able to keep the drywell pressure signal from actuating the ADS, or that the operator will manually .,

inhibit ADS if the loss of drywell cooling results in a signal to initiate ADS on low level and high dryvell pressure.

• • These evaluations neglect operator action to stop the HPCS from overfilling the vessel. If such action were taken in 10 minutes af ter level recovery was apparent to the operator, successful shutdown would probably be maintained since the excess boron provided would be greater than the potential dilution. .

• f l

+

r Ites PrNida sva

~

operating plant data- for ECCS and RHR systes  :

,5 .

maintenance unavailability -

versus tech spec alloyables.

., Response -

F Table III provides estimates of system unavailability caused

. by-on-line. maintenance for ECCS'and RHR systems. This data was ba' sed on a-survey covering Il plant years.

a Y

-W e n

~

w 9

4 6

-'**7 g

, W e

b

]

9 1

8

/

f 9 .

k

- S.

+

t _* 4 _ ,

- -(20) c- ,.

> j

~

r t._ s

]

' : =m - se e o <

t y-N '- _

f n

_, 17 *,..; #

s

'. . :.. +

'. s' ',

o

, Iten.17 - Response (Continued) -

TABLE III .

ESTIMATED SYSTEM UNAVAILABILITY (A) CAUSED BY ON-LINE MAINTENANCE BASED ON FIELD DATA FOR 11 PLANT YEARS AS REPORTED BY AN OPERATING BWR/4 b "

NO. OF - TOTAL AVERAGE REPORTED EVENTS HOURS TIME (HRS) A/ YEAR

, llPCI 22 507.0 23.5 .69

' RCIC 20 605.5 30.3 .82 CS 7 86.1 12.3 .06 (PER LOOP)

LPCI 213.2 14' 15.2 .15 (PER LOOP)

RHR 11 213.0 26.4 .20.

, (PER LOOP) 3 "

D/G 14 111.5 8.00 .05 (PER_D/G).

IIPCS NOT AVAILABLE ,

NOTES: *; -

1. - With average plant availability of .762/ year

2. Two'1 oops / system

3. Four.D/G's- - -

4

,,8, t

-(21)-

4 e

~

s :. - .. .

iir1ACh:lthi 11 Item 18 - Provide estimates of the mean time to repair for the BWR/6 PJIR system.

Response

The RHR system mean time to repair during operation is .

estimated to be 26.4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />. (See Table III of Question 17.

4 o

g a 94 9 m en e

S 9

O S

'(22) .

_ , . - . _. . ~ .

~

. 1 tera 19. - Provide EG/6 DiR reliability estimates applicable to Grand.. _

- Gulf..

s i?

Response

p.

. The.following is our BWR/6 reliability assessment for the RER

, . (Containment Heat Removal):

I-

' e Probability of not starting the RHR within 15 h,urs o of the

- accident start is estimated as:

-2 o Primary Side 0.7 x 10 f o Secondary Side 1.3 x 10" l: .

Total per RHR Loop 2 x 10~

' -4 o Both lgogs Fail. 4 x 10 (2x10 )

. o: Operator fails to < 1 x 10~ '

_.,. start RHR

\:

-4 '

Total probability 5 x 10 l -both RHR loops not started within 15 hrs.

l i.

l - * ,.

1 .gs'

-s l- '

I ;- .

i e

W D

• l

j. .

t 5

I '? . '

D I l-e (23)

[.

' k

, g ..

b-L

- , ( g , ' 9 [,

s *, , * .,. . .4 MJ

' ' , , , , , , , . ~ . e -~

' ~

m . ._ . 11:

• lItea 20 '- Provid2 o basis for revising the Sandia T)QW and T23 9 "#"

probabilities. -

. Response The T QW events are loss of offsite power (T ) and other)QW and T23 (T23) ini ia ng events f transients wed by a oss of the power. conversion system (Q) and the residual heat removal system (W). These events are assumed to ultimately-lead to core melt. For the BWR/6 Mark III design, containment failure due to overpressure (caused by loss of power conversion and

. heat removal systems) does not necessarily lead to core damage. The Reactor Core Isolation Coolant (RCIC) System can be switched back to the Condensate Storage Tank, and the BWR/6 ECCS pumps are rated at 212 F at 15 psia. Thus, cavitation is not a limiting-factor for the ECCS function. Also,

' containment failure due-to static overpressure does not necessarily ~ sever the ECCS -injection line or damage the ECCS equipment. Thus, it~has been assessed.that core damage occurs 7 '

, in no more . than twenty-five percent of the _ accidents where ,

heat removal capability is lost (T QW and T 2QW). The event probabilities for' T QW and T 9 * " * ' *"8' accordingly.

3 23 7 ..

t 4

9 4

-.d 4

(24)

,4

' ULTIMATE COSIAIN!If.ET CAPACITY ANALYSIS l

, . i This attachment is Section 5 from the " Preliminary Evaluation of Additional liydrogen Control Measures for the Grand Gulf Nuclear Station" submitted to the Nuclear Regulatory Commission by letter AECM-81/139 on April 9, 1981.

4 5

e i

I-i-

f h

gb e

e O

N_6 Waupe cumpe

E _

..J ]

-l u-;.- .

, 5.0 containment-Ultimate Capecity , j 5.1 Introduction I l

In a letter dated December 19, 1980, the NRC requested that MP&L perform an ultimate capacity analysis for the CGNS containment. -

Details of the analysis and results are described in the following sections.

5.2 Design Pressure The containment design pressure is 15 psig as stated in FSAR subsection 3.8.1:3.5b-1. The drywell internal. design pressure is - ,.

,30 psid as stated in FSAR subsection 3.8.3.3.1.5b-3.

5.3 Calculated Static. Pressure Capacity

~

Results of the analysis indicates that the ultimate capacity of the containment is calculated to be 47 psig. The ultimate capacity is defined as that pressure at which a general yield state is reached at critical structure sections. This number is based on the specified strength of the materials used for reinforcement. The actual material strengths are-being tabulated and are expected to be-higher than the specified values. Therefore, the actual calculated ultimate capacity will be proportionally increased and will be provided in the final report.

The ultimate capacities of large penetrations such as the personnel locks and equipment hatch will be furnished by the vendors at a later date. Based on information f:om other Bechtel projects, the capacities of these large penetrations are expected to be higher than the containment shell.

The ultimate capacities of components which form part of the containment. boundary are also being evaluated. The results will be

, provided in the final report. ,

,c..

The ultimate capacity of the drywell will also be provided in the final report.

5.4 Calculated Dynamic Pressure Strength l, ,

As stated in a MP&L letter (AECM-81/38) to R., L. Tedesco of the NRC, dated January 21, 1981, the Grand Gulf containment analysis does not consider dynamic pressure effects.- In selecting a hydrogen control system, design' criteria prohibit conditions leading to a hydrogen detonation. .This eliminates the narrow pressure spike which accompanies the detonation. MP&L.'

investigations of hydrogen combustion indicate .that approximately

, 20-30 seconds are required for a' flame ~ front to propagate through the containment. The pressure increases-associated with hydrogen combustion take place too slowly for dynamic effects to be of concern. Pressure decreases due to the effects of such things as heat sinks are even slower, on the order of several minutes.

• A

- dyn mic pressure ennlysis, therefore, provides littic additional J . infermatien and is nst likely to affect the conclusions of the r  ?. entlysis..

5.5 Failure Modes The evaluation of the finite element analysis indicates that the hoop reinforcement in the containment cylinder is the highest stressed element. A general yeild state is reached.when the inner and outer hoop reinforcement, as well as the liner plate, have yielded. As stated in Section 5.3, this state occurs at a pressure of 47 psig if containment material specifications are used in the calculation. <

- - . 'A review of penetration closure plates shows that plate yielding begins to occur at 60 psig. This is well above the calculated ultimate capacity of the containment shell.

As noted previously, pressure capacities of the personnel air locks and equipment batch are not yet available from the vendors, but based on information from other plants, the air locks and hatches ,

are not expected to be more limiting than the containment shell.

t Preliminary evaluations of the drywell and bulkhead indicate that their ultimate capacities are higher than the containment shell.

Therefore, the ultimate capacity of the containment shell is expected to. be limiting, which makes the containment the most likely failure point.

-5.6 Original Design Criteria Original containment analyses and design methods are specified in FSAR subsect' ions 3.8.1.4.1.1 and 3.8.1.4.1.2,- with applicable design codes listed in subsection 3.8.1.2. A further discussion of design criteria is found in the response to NRC question 130.29.

Original drywell analyses and design methods are specified in FSAR subsection 3.8.3.4.1, with applicable design codes listpd in subsection 3.8.3.2. A further discussion of design criter?c is found-in the response to NRC question 130.33.

. r Original containment liner plate analyses and design me'thods are specified in FSAR subsection 3.8.1.4.2. . .

5.7 Analysis Details .

.The analysis was performed using a finite element model (as shown in Figure 5-1) Land the Bechtel inhouse computer program FINEL.

This program has the capability of modeling concrete cracking in tension and calculating the redistribution of forces and moments for the statically indeterminate structure. The finite element.

o. . model-consists of the containment dome and a portion of the containment. cylinder. This portion of the containment' structure-was selected because this section of the' cylinder has the leest' amount of hoop-reinforcement, and when the general yield state is

~

reached,~the hoop reinforcement is the limiting element.

~

? .6- . .; :

5.8 Varificatien Drawings

^ '

f r. - . $4 . .

The drawings needed to allow verification of the.modeling used and

'to evaluate the analyses employed for penetrations are Figure 5-1

of:this ' report and' Figure 3.8-2 through 3.8-9 and 3.8-58 through 3.8-60 of the FSAR.

e e

4 . . .* ...., +

_ 4

=

b 4

e s

6 e

5 4

f . $ ?

. Kb i

o" l

-W-Y T

i 4

, . g ..' , .7

.I' 9

• O I

!. . [

. L

+

i 2 ,.

y. N

. g-l

, a '- ^-

'k' i j w Y j .

$ _ =. '_ .yt- ', g

. e ,s..",.... ., , , , , ,. _ , . . .

_),,..,_

, (.s Finite Element Madel I,

-.C- Figure 5-1 -

3 ,, -

l 6ee g & -3.-

3> .,-

~  !

G._. e'- 'yy9 I

e ,~

g6 5

.W_ ~,, . . . . . ..

\., >.

.L.

~

. . . . '7I TI'L 2_ m ,.. a j .

.. )"

fJf o'e' C' g 3b 3e -

At i

- s ,,

a 6 . G ..

e 6<\

~

t .

$ h*

Y b .?

N *Ast 1 II V

(' I

. -4 e.-

0 c N,;;

1 og ,isas1*** /$ M t** . - .

.f( 95x,1c

, 4

'= -

I e

. 4 "

- *q. , d' ' ,..

.,T @, ,,

ilL ei .> k, 9 9 ..' a c ,, ,,, s y ,,, ,  ;,

.' g

• =  :

I 3il -

Zoo

. .t nMC -

L. 1 ,;a i _

l *

. 32C 393

s. .u-au, a est - s.r t 3'* g,s  ?  ;< 4 h3 b b gg 6 6 6 G l .

ee e e ,

)! . 6 6 6 6d 2, C' 6' 8' 45

~tas I

• L,'

6,Q @ C.'Q 0 d

• q.s.y

(*.

1

' 21 *s* e.

% U'_- "'u- .a.. .u., ,u.n-, dese t17 A et l M _

6

'. e

%e

, ,(. .

(

i '

MISSISSIPPI POWER & LIGHT COMPANY

~#

0lTr N

Helping Build Mississippi P. O. B O X 16 4 0. J A C K S O N. MISSISSIPPI 39205 uuctt An enoot>ctioN DCPAPTMENT (U

tM C f w September 3, 1981

,Mr. Frank Rowsome, Acting Chief f4N L mml Reactor Risk Branch X 4':.'.[ J

. Division'of Risk Assessment Office of Research U. S. Nuc1 car Regulatory Co raission Washington, D. C. 20555

SUBJECT:

Grand Gulf Nuclear Station Units 1 and 2 Docket Nos. 50-416 and 50-417 FilD0260/L-860.0/16666/L-9.53.d RSSMAP Study -

e AECM-81/346 Mississippi Power 6_ Light Company has reviewed Volume 4 of NUREG/CR-1659

-" Reactor Safety Study. Methodology Applications Program: Grand Gulf #1 BWR Power ,Plant" and prepared the attached comments for your consideration. ,

Very uly yours, p ,,

L. F. Dale Manager of Nuclear Services SIDi/JDR: dn - r.

Attachment cc: Mr. N. L. Stampley .

Mr. R. B. McGehee Mr. T. B. Conner ~

Mr. C. B. Taylor .

. Mr. Victor Stello, Jr., Director -

, Office of Inspection & Enforcement

. U.- S. Nuclear Regulatory Commission Washington, D. C. 20555 ,

. ~

Mr. Harold R. Denton, Director Of fice 'of Nuc1 car Reactor Regulation U. S. Nuclear Regulatory Consission

-Washington, D.'C. 20555 fDp.J

~

hy .

Member Middle SUuth Utilities System -

(,

- (. '

Attachment to

- AECM-81/346 Co=ments On Volume 4 of' NUREG/CR-1659 RSSMAP: Grand Gulf

~

. General Comments:

1. In many instances, insufficient infor=ation is provided to adequately l review sequence evaluations. For example , for some cases the cut sets presented represent less than 25% of the failure probability.

. 2. A costainment failure pressure of twice design was assumed. Analyses

-done in support of the Grand Gulf H2 control system design have shown .

that the f ailure pressure would be a factor of 3 to 4 times design. ~

In' many sequences this would significantly impact the time to core melt and thus the recovery factor.

-3. . Pressure / temperature curves for many sequences were not provided. Therefore, a complete review was not possible. In at Icast one instance (sequ'ence TPQE),

the curves presented contradicted assumptions maab (see specific comments on T 1PQE). ,

J

4. Although noted in the report, it is misleading to present results for Grand Gulf and try to compare. them to WASH 1400 when dif ferent assumptions were used. In particular,-the assumptions for the TPQE sequence.regarding the PCS, will make a diff'erence of one order of magnitude. ,

5.- No credit in the report. was given to recovery, of the diesels. Recovery of of f site power was included and credit for hardware repair was.. given in some cases. WASH 1400 uses a value of 0.1 for failure to ' recover diesels in the long term. .This could make a dif ference of an order of magnitude in many cut sets. m -4/ Ae*9 e% G (W li~ M " O/, M ' L

6. 'Results presented in Appendix C show a spike in the fraction of cladding reacted (FCR) at the time of core slump. There is no physical bas 1; for

.this model. It arbitrarily increases the amount of hydrogen in the-atmosphere at the-time a burn is assumed. (Note that all sequences resulting in on H2 burn.will be changed significantly by the. proposed hydrogen. ignition system.) -

Comments on Specific Accident Sequences:

• TPQ1 - it is not credible to believe that .the operators Eould stand by and watch the containment fail due to overpressurization over a 28 hour3.240741e-4 days <br />0.00778 hours <br />4.62963e-5 weeks <br />1.0654e-5 months <br /> period (or longer). Numerous : options are available to the operator to prevent over-pressure from breaching-the containment. Among these are:

o Drywell air' coolers could be made available in this time to remove

~

z

' heat.

The f ans can be manually loaded on the diesel.and cocling

• vater is supplied by the plant service water system.

9

~

I r

.( , (.

o Containment air coolers may be available for the T 23 PQI sequence (they are powered by offsite power). Cooling water is supplied by the plant chilled water system.

o In the long term, numerous options are open to the operators to supply additional water to the suppression pool.

.The dominant cut set in the Ty PQI sequence involves loss of all power.

However, a recovery f actor of 0. 23 is applied. This f actor is apparently sed on a time reriod of 28 hours3.240741e-4 days <br />0.00778 hours <br />4.62963e-5 weeks <br />1.0654e-5 months <br /> and a mean time to repair of 19 hours2.199074e-4 days <br />0.00528 hours <br />3.141534e-5 weeks <br />7.2295e-6 months <br /> (R = e gg/19).

In fact, this value is applicable to hardware failures but not to diesel recovery.

WASH-1400 assumed a value of 01 for failure of long term recovery diesels. This f actor should be used in place of the 0.23 in the dominant cut set and in addition to the 0.23 in the other cut sets in which both diesel failure and hardware failures are considered.

T1 PQE - This sequence assumes the manual initiation of ADS is required for LPCI operation. In f act , as shown in Figure C-9 of this report , at' 70 min. (the onset of co're melt) the system pressure is well below the set point for LPCI and the head available via the condensate booster pumps. ,' Consideration of automatic e

operation of LPCI wpuld significantly reduce core melt probabilities.

T PQE - Again it was assumed that automatic ADS does not occur but no RCS 23 pressures are provided in the report to demonstrate that manual ADS operation -

. is required.

= '

SI - No recovery factor is included despite the fact that this is apparen'tly a long term event (no time scales or heat up curve'i. were included in the report).

TQW - As described earlier for sequences TPQI, recovery of diesel failures was not properly considered. This could reduce individual cut set frequencies by factors of 2 to 10. Also, as mentioned earlier, the containment failure pressure is quite pessimistic. Finally, although a conservative recovery factor is included, it only covers repair times for failed hardware. No recovery is considered for alternate actions by operators (i.e. , initiating air coiler's, supplying additions cooling water, etc.). It is totally unreasonable not to assume some corrective actions by the operators during the minimum of 30 hours3.472222e-4 days <br />0.00833 hours <br />4.960317e-5 weeks <br />1.1415e-5 months <br /> of the pressure rise.

T23C - The assumption of containment failure ignores alternate containment cooling options (e.g. air. coolers). ,It also ignores the f act that f or s ome initiating transients it may be possible to continue plant operations at a 30%

power level.

Ty QUV - Manual ADS actuation is assumed to be required although no analysis to support this assumption is provided. As noted earlier (sequence TPQE), this assumption may not be justified. Also , as' decribed earlier, recovery from .

diesel failure has not been properly considered.

9 O