ML20092M021

Containment Failure Mode & Fission Product Release Analysis for Limerick Generating Station:Base Case Assessment

ML20092M021
Person / Time
Site:	Limerick
Issue date:	04/30/1984
From:	Ludewig H, Pratt W, Yang J BROOKHAVEN NATIONAL LABORATORY
To:	NRC
Shared Package
ML20092L993	List: ML20092M021
References
CON-FIN-A-3711 BNL-NUREG-33835, NUDOCS 8407020105
Download: ML20092M021 (97)

Text

'

am.-33835 INFORMAL REPORT LIMITED DISTalBUTION ,

m i

n 1

CONTAINMENT FAILURE MODE AND FISSION PRODUCT RELEASE  :

ANALYSIS FOR THE LIMERICK GENERATING STATION:

! BASE CASE ASSESSMENT r

H l H. LUDEWIG, J. W. YANG, AND W. T. PRATT i

DATE PUBLISHED - APRIL 1984 i,

-l I

ACCIDENT ANALYSIS GROUP l .

t

,a 6

DEPARTMENT OF NUCLEAR ENERGY BROOKHAVEN NATIONAL LABORATORY I UPTON. NEW YORK 11973 i

{. .- - _. c a: ,

i Prepared for the U.S. Nweteor Reguia ory Ccmmiss.on

j OMce of Nuclear Regulatory Resecren Contrac? No. OE.ACO2 76CHdC016 i

.8 i

,r:, -.-=x 1 ---:.=--m. = m 1-.= a... .-.< ,- :r -2 - m '., - = .z ~ -- w r. x a :n t 8407020105 840622 <

PDRADOCK05000g A

i BNL-NUREG-33835

. INFORMAL REPORT i LIMITED DISTRIBUTION CONTAINMENT FAILURE MODE AND FISSION PRODUCT RELEASE ANALYSIS FOR THE LIMERICK GENERATING STATION:

BASE CASE ASSESSMENT H. Lu'dewig , J. W. Yang , and W. T. Pratt s

Date Published - April 1984 4

Accident Analysis Group Department of Nuclear Energy Brookhaven National Laboratory Upton ,. New York 11973 Prepared for U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Contract No. DE-AC02-76CH00016 NRC FIN No. A-3711

-1 ABSTRACT The objective of this report is to calculate a range of source terms that will

'be used 'as input to the site consequence analysis to be performed by the NRC staff as part _of. the Draft Environmental Statement (DES) and the Final Envi-ronmental Statement (FES) for the Limerick Generating Station (LGS). . At the direction of NRC staff, a " base case" approach was used in the development of these source terms. This approach relies on the Reactor Safety Study methods

-(or prescriptions) to determine the release of fission products from the dam-

- aged. fuel, primary system hold-up,- suppression pool scrubbing and transporta-

- tion of the fission products in containment. . The report utilizes information in .the LGS-Probabilistic Risk Assessment (internal initiated events) and in

- the LGS-Severe Accident Risk Assessment (external initiated events) and in BNL reviews: of these documents _ to assess the probabilities of potential contain-ment building failure modes and release paths. .However, the use of the " base case"_' approach necessitated extensive reanalysis, using the MARCH / CORRAL sys-tem of codes, to develop appropriate source terms. In all , twenty-seven source terms have been developed to represent all potential failure modes and release paths in the LGS. Each of these source terms provide the fraction of fission product species released to the atmosphere, the characteristics of the release and the frequency of occurrence of the release.

E iii

ACKNOWLEDGMENTS The1 authors wish . to acknowledge R. A. Bari for reviewing this report and making helpful _ comments.- R. A. Bari is responsible for coordinating all BNL staff activities related to the LGS. In addition, we are grateful to K. Shiu

- and ;N. Hanan (Risk Evaluation Group) who helped in the classification of the core-melt sequences.

The ' work is' being performed for the Division of Systems Integration (DSI) at

. '. the U. S. Nuclear Regulatory Commission (NRC). J. Carter is the NRC technical monitor -for .this project, however, J. Meyer had significant input to this re .

port prior to his leaving DSI. J. Rosenthal is the NRC section leader for this project.

We would like to express our appreciation to Theresa Rowland for her excellent

-typing and assembling of this report.

Y i,

~

Y

=

e e

9 iv

a CONTENTS Page ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii ACKNOWLEDGMENTS. . . . . . . . . . . . . . . . . . . . . . . . . . . iv LIST OF FIGURES. . . . . . . . . . . . . . . . . . . . . . . . . . . vii LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . viii

1.0 INTRODUCTION

. . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.1 Background . . . . _. . . . . . . . . . . . . . . . . . . 1-1 1.2 Objectives and Scope of the Analysis . . . . . . . . . . . 1-2 1.3 Organization of Report . . . . . . . . . ......... 1-3 1.4 References to Section 1. . . . . . . . . . . . . . . . . . 1-4 2.0 BINNING 0F ACCIDENT SEQUENCES . . . . . . . . . . . . . . . . . 2-1 2.1 NRC Staff Classification of Core-Melt Sequences. . . . . . 2-2 2.2 Containment Failure Modes and Release Paths. . . . . . . . 2-3 2.3 Sou rce Te rm P rob ab i l i ty . . . . . . . . . . . . . . . . . . 2-3 2.4 References to Section 2. . . . . . . . . . . . . . . . . . 2-4 3.0 CORE MELTDOWN AND CONTAINMENT RESPONSE. . . . . . . . . . . . . 3-1 3.1 Cl ass I Sequences. . . . . . . . . . . . . . . . . . . . . 3-1 3.2 Cl ass I I Sequences . . . . . . . . . . . . . . . . . . . . 3-2 3.3 Cl ass III Sequences. . . . . . . . . . . . . . . . . . . . 3-3 3.4 Cl ass IV Sequences . . . . . . . . . . . . . . . . . . . . 3-3 3.5 Cl ass IV Reanalysi s . . . . . . . . . . . . . . . . . . . . 3-4 3.6 Cl a ss IS Sequences . . . . . . . . . . . . . . . . . . . . 3-5 3.7 Cl ass S Sequences. . . . . . . . . . . . . . . . . . . . . 3-5 3.8 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . 3-6 3.9 References to Section 3. . . . . . . . . . . . . . . . . . 3-6 4.0 FISSION PRODUCT TRANSPORT AND RELEASE . . . . . . . . . . . . . 4-1 4.1 Class 1 (Damage States I-T and I-S) . . . . . . . . . . . . 4-3 4.1.1 Fa i l u re i n t he Drywel l . . . . . . . . . . . . . . . 4-3 4.1.2 Fail ure in the Wetwell . . . . . . . . . . . . . . . 4-4 4.1.3' Failure in the Wetwell with Loss of Suppression Pool. . . . . . . . . . . . . . . . . . . . . . . . 4-4 4.1.4 Class I Sequences Initiated by LOCAs (I-S Damage Stzte); . . . . . . . . . . . . . . . . 4-4 4.2 Cl ass II (Damage State II-T) . . . . . . . . . . . . . . . 4-5 4.3 Cl ass III (Damage State III-T) . . . . . . . . . . . . . . 4-6 4.4 Cl as s IV Sequences . . . . . . . . . . . . . . . . ., . . . 4-7 y

, CONTE'NTS(Cont.)

Page 4.4.1 Class IV Transients (Damage State IV-T) . . . . . . 4-7 4.4.2 Cl ass IV LOCAs (Damage State IV-A) . . . . . . . . . 4-8 4.4.3 Cl ass IV Reanalysi s . . . . . . . . . . . . . . . . 4-8

4. 5 . Cl a ss IS Sequences . . . . . . . . . . . . . . . . . . . . 4-9 4.6 Cl a s s S Se que nce s . . . . . . . . . . . . . . . . . . . . . 4-9 4.7 . S umm a ry . . . . . . . . . . . . . . . . . . . . . . . . . . 4-10 4.8 References to Section 4. . . . . . . . . . . . . . . . . . 4-11 5.0 SOURCE-TERM CHARACTERISTICS . . . . . . . . . . . . . . . . . . 5-1 5.1 LGS-DES Source Te rms . . . . . . . . . . . . . . . . . . . 5-1 5.1.1 Source Terms for Damage State I-T . . . . . . . . . 5-1 5.1.2 Source Terms for Damage State II-T ........ 5-2 5.1.3 Source Terms for Damage State III-T . . . . . . . . 5-3 5.1.4 Source Terms for Damage State IV-T. . . . . . . . . 5-3 5.1.5 Source Terms for Damage States I-S and IV-A . . . . 5-4 5.1.6 Source Terms for Damage States IS-C and IS-Y. . . . 5-4 5.1.7 Source Terms for Damage States S-H20 and S-H20. . . 5-4 5.2 LGS-FES Source Te rms . . . . . . . . . . . . . . . . . . . 5-5 5.3 . References to Section 5. . . . . . . . . . . . . . . . . . 5-5 4

vi

y, f.:.. --

-t . , .. . , .. e. , g. . w,. <y

. =,.-;,;.,....--.., .

,3 7 e pv ._ ,. m . sg.t 2, p.3, .- _ 9;,. ....2,. y. e . .

.s .:

f LIST OF FIGURES Figure. Title Page 4.1 Typical sequence of spike fission product releases for postul ated accidents . . . . . . . . . . . . . . . . . . 4-12 I

vii s

LIST OF TABLES Table Title Page 2.1 Generic. accident-sequence classes . . . . . . . . . . . . 2-5 2.2 Failure modes used in the LGS-PRA-. . . . . . . . . . . . 2-6 2.3 Classification of damage states as used in LGS-DES (core-melt sequences) . . . . . . . . . . . . . . . . . . 2-7 2.4 Classification of damage states as used in LGS-FES (core-melt sequences) . . . . . . . . . . . . . . . . . . 2-12 2.5' Description of damage states. . . . . . . . . . . . . . . 2-17 2.6 Containment failure mode and release path notation. . . . 2-18

'2.7 Assignment of conditional probabilities . . . . . . . . . 2-19 2.8 Damage state probabilitiesaas used in LGS-DES . . . . . . 2-20 2.9 Damage state probabilities as used in LGS-FES . . . . . . 2-21 2.10 Assignment of release categories. . . . . . . . . . . . . 2-22 3.1 Highlights of MARCH analysis. . . . . . . . . . . . . . . 3-7 3.2 Comparison of DNL and Limerick PRA analysis of the Cl ass I sequences (TQUV) . . . . . . . . . . . . . . . . . 3-8 3.3 BNL analysis of S 1 QUV sequence. . . . . . . . . . . . . . 3-9 3.4 Comparison of BNL and Limerick PRA analysis of the Cl ass II sequences (TWLP) . . . . . . . . . . . . . . . . 3-9 3.5 Comparison of BNL and Limerick PRA analysis for Cl ass III, ATWS sequence. . . . . . . . . . . . . . . . . 3-10 3.6 Comparison of BNL and Limerick PRA analysis for Cl a s s I V . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11 3.7 BNL analysis of AC sequence of Class IV accident. .... 3-11 3.8 BNL analysi s _ for Cl ass IS . . . . . . . . . . . . . . . .

3-12 3.9 BNL analysis for Class S S(T RPVRB). . . . . . . . . . . . 3 3.10 Summa ry of MARCH resul ts . . . . . . . . . . . . . . . . . 3-13 viii

LIST OF TABLES

- Table Title Page

-4.1 Fission product release. source summary - best estimate

. total core release fractions. . . . . . . . . . . . . . . . 4-12 4.2 Fission product release fractions for Class I . . . . . . . 4-13 4.3 Fission product release fractions for Class II. . . . . . . 4-14 4.4 Fission product release fractions for Class III . . . . . . 4-15

.4.5 Fission product release fractions for Class IV l ocation DW) . .- . . - . - . . . . . . . . . . . . . ( f ail ure4-16 4.6 Fission product release fractions for Class IV (failure location DW). . . . . . . . . . . . . . . . . . . . . . . . 4-17 14.7 Fission. product release fractions for Class IV (failure location 15T). . . . . . . . . . . . . . . . . . . . . . . . 4-18

. 4.8 - Fission product release fractions for Class IS. . . . . . . 4 19 4.9 Fission product release fractions for Class S . . . . . . . 4-20 4.10' A comparison of fission product release fractions for Class I-sequences initiated by LOCAs and transients . . . . 4-21 4.11 A comparison of fission product ' release fractions for Class IV sequences initiated by LOCAs and transients. . . . 4-22 5.1- Summary of source terms for damage state I-T for in'put to LGS-DES. . . . . . . . . . . . . . . . . . . . . . . . . 5-6 5.2 . Summary of source terms for damage state II-T for input to LGS-3ES. . . . . . . . . . . . . . . . . . . . . . . . . 5-7 5.3 Summary of source terms for damage state III-T for input to LGS-DES. . . . . . . . . . . . . . . . . . . . . . . . . 5-8 5.4- Summary of source terms for damage state IV-T for input

.to LGS-DES. . . . . . . . . . . . . . . . . . . . . . . . . 5-9 5.5 Sunnary of source terms for d6 mage states I-S and IV-A for input to LGS-DES. . . . . . . . . . . . . . . . . . . . . . 5-10 iX

~

~

LIST OF TABLES-(Cont.)

-Table; Title Page 5.6 Summary of source terms for damage states IS-C and IS-C . . 5-11 5.7 Summary of source terms for damage states S-H2O and S-HZ6 . 5-12

'5.8 Summary of source terms for damage state I-T for input to LGS-FES.-. . . . . . . . . . . . . . . . . . . . . . . . 5-13 5.9 Summary of source terms for damage state II-T for input to LGS-FES. . . . . . . . . . . . . . ... . . . . . . . . 5-14 5.10 Summary of source terms for damage state III-T for input to LGS-FES. . . . . . . . . . . . . . . . . . . . . . . . . 5-15 5.11 Summary o'f source terms for damage state IV-T for input to LGS-FES. . . . . . . . . . . . . . . . . . . . . . . . . 5-16

.5.12 Summary of source terms for damage states I-S and IV-A for input to LGS-F ES. . . . . . . . . . . . . . . . . . . . 5-17 o

O b X

c y.

1.0' INTRODUCTION-This section gives the background to our assessment, indicates the objectives of our work, and describes the way in which the report is organized.

1.* Background In' March 1981 a Probabilistic Risk Assessment [13 for the Limerick Generating Station (PRA-LGS) was submitted to NRC. The LGS-PRA considered accident se-quences initiated only by internal events. In Februgry 1983 Brookhaven Na-tional Laboratory (BNL) issued a detailed reviewL2J of the LGS-PRA. In April -1983 contyaptors (NUS) to the Philadelphia Electric Company (PECo) com-pleted a studyL3J which included an evaluation of risk due to seismic ini-tiating events and to fires that might be initiated within the plant. This study, the Severe Accident Risk Assessment for the Limerick Generating Station (LGS-SARA), used generic accident classes developed in the LGS-PRA whenever possible to also represent accident sequences initiated by external events.

However, 'because of the unique characteristics of some of the seismically initiated sequences, additional accident classes were developed in the LGS-SARA.- In addition, the LGS-SARA ingigded a revised analysis of the off-site consequence analysis using the CRAC2L4J Computer Code.

In June 1983 NRC requested that BNL undertake a preliminary, short-term revfew of the LGS-SARA. The review will be contained in a two-volume report.L5J Volume I reports on the review of seismic and fire methodologies as they re-late to the determination of the core melt frequency. Volume II will report on the BNL review of core melt phenomenology, fission product behavior, and off-site consequences will be issued at a later date.

In addition to a review of the LGS-SARA, the NRC also requested that BNL pro-vide inpu$ to the Draft Environmental Statement (DES) for the LGS and a draft versionL6J of this report'was the product of that effort. Specifically, NRC requested that BNL generate a complete set of source terms for core meltdown accidents in the LGS initiated by internal and external events. The DES source terms developed in the draft report were based on information in the LGS-PRA, LGS-SARA and on BNL reviews of These DES source terms were sent via an NRC memorandumL7]these to the Accidentdocuments.

Evaluation Branch (AEB) who uyed them to perform trie LGS site consequence analysis, which was published [8J in the LGS-DES. The only input BNL had to the LGS-DES was the source terms described in Section 5 of this report.

Following the publication of the LGS-DES, BNL was requested to reanalyze the Class IV sequences. This reanalysis was to take into account improved ther-mal / hydraulic modeling and the presence of the reactor building. The recalcu-lated Class IV source terms are included in the present version of this report for input to the LGS Final Environmental Statement (FES).

The approach taken to the development of the so'urce , terms in the LGS-PRA was to utilize the methods (or prescriptions) used in the Reactor Safety Study.L9J Consequently, the release of fission products from the damaged fuel was assumed in the LGS-PRA to follow the Gap, Melt, 0xidation and vapori-zation release phases described in Reference [9] and in Section 4 of this 1-1

j'

~

report. .However, the methods used in the LGS-PRA did differ from the RSS in a number of respects, namely temporary hold-up of fission products released during the gap and melt i release phases in the primary system until after vessel failure  !

" - .a decontamination factor of 10 was used for a saturated suppression  ;

pool (no decontamination was assumed in the RSS for a saturated pool).

The approach taken in the LGS-SARA to develop source terms differed even fur- i ther from- the methods used in the RSS. - The release of fission products (FP)  ;

during the in-vessel spelt release phase was assumed to follow the trends re- '

ported in NUREG-0772.L10J - Thus, significantly higher fractions of all aero-sols are released ex-vessel in the LGS-PRA than in the LGS-SARA. i There has been significant research _ activity in this area since the publi- i cation of the -RSS in 1975. A basis for estimating FP behavior was" pub-lished[10] in 1981 by RES/NRC and h.as been used in the LGS-SARA. In addi-tion updated fission product source _ term methods are currently being devel- '

.oped(llJ and ' are receiving extensive peer _ review. At this stage, BNL is  ;

unable to confirm the validity of the changes made in the LGS-PRA and LGS-SARA f relative to the RSS approach. Hence, at the direction of NRC staff, the ap-proach taken in this report to the development of the source terms is to fol-

' low the RSS prescriptions regarding the release of fission products from the

-damaged fuel, primary . system hold-up, suppression pool scrubbing and transpor-tation of the -fission products in containment. The use of this approach by BNL. staff in .this- particular application does not constitute technical en-

-dorsement of the assumptions, data or techniques that are associated with this  ;

approacn. The aim is simply to calculate a ' consistent set of . source terms

• that are applicable to both internally and externally initiated accidents ,

. based on ' RSS prescriptions. The use of the RSS based source terms is part of the " base-case" approach described in Reference [12] ana in Section 4 of this  ;

. repo rt . . Reference [12] also gives-the justification for such an approach. l c 1.2. Objectives and Scope of the Analysis .

The objective of this report is to provide a complete set of source terms for core meltdown accidents in the LGS initiated by internal and external events. "  !

Each of these source terms will provide the fraction of fission product spe-  !

cies released to the atmosphere, the characteristics of the release and the frequency of occurrence of the release. These source terms will be used as input to the site consequences analysis to be performed by the NRC staff as part of the LGS-DES.and LGS-FES.

The classification of the core-melt accident sequences in-t.lis report has been with References [1]-[3]. The LGS-PRA[1 and the BNL re-made[2$nsistent view e used to classify accidents initiated by internal events. The LGS-SARA was used to classify externally initiated accidents. The proba-I bilit1gs]of of

'reviewl2 'the theinternally LGS-P'RA.initiated The LGS ' SARA 63 acc[dgnt classes were based on -the BNL was used as the basis for de-termining the probabilities of external events. The only changes that were made to the frequencies and accident classifications reported in the LGS-SARA relates to the: influence.of. seismic events on evacuation. In the LGS-SARA it t 1-2 r

., ---.,,-_--.,m-_ .~,.._.,_.,.,,._,,,_,.__,,,_,_,__m__,,.,___m , . _ , . , . _ _ _ , _ , . , , _ . _ . . _ . _ _ , ,

was assumed that evacuation would only be influenced by effective peak accel-erations in excess of 0.619 (refer to Section 10.1.6.5 of Reference 3). This was considered inappropriate by BNL Contractor J. Reed (refer to the BNL re-c view of the LGS-SARA, Reference 5) and accelerations in excess of 0.4g were suggested as having an influence on evacuation. We therefore subdivide the frequencies of all accidents initiated by seismic events into regional disas-ters (RDs) (with accelerations greater than 0.49) and non-regional disasters

-(with accelerations less than 0.49.) This classification is carried through to the final probabilities given for each of the failure modes and release paths in Section 5 of this report.

The conditional probabilities associated with the various failure modes and l release paths were based on the containment event trees in Reference [2] with a number of modifications that are discussed in Section 2 to this report. The failure modes for external events were taken from the LGS-SARA.

The fission product release fra.ctions were calculated using the MARCH / CORRAL system of codes. The MARCH [13J analysis uses the latest decay heat stand-ard, which results in significantly shorter times to major events in the acci-dent progressions titan calculated in References [1] and [2]. The source tepmq used in the CORRAlll43 analysis are based on the Reactor Safety Study.L9J The use of the RSS source terms is part of the " base-case" approach described in Section 1.1.

1.3' Organization of Report The binning of the various accident sequences into representative classes is discussed in Section 2. In addition, the determination of the conditional probabilities of the various containment building failure modes is also de-scribed. The probabilities of all of the failure modes are determined in Sec-tion 2.

In Section' 3, the MARCH [13] analysis of the various representative accident sequences is described. This section provides the timing of major events in the accident progression. In addition, the intercompartmental flow is predic-ted, which in turn determines the movement of fission products (required input to Section 4). Finally, Section 3 provides the basis for the determination of the characteristics of the release from the containment building.

The release of fission products (fps) from the damaged fuel and the movement of these fps through the containment building are calculated in Section 4. In addition, Section 4 calculates the fractions of the airborne fps tnat are re-leased to the environment when the containment building is predicted to fail.

Finally, in Section 5, the representative source terms for the various failure modes and release paths are generated. Section 5 therefore assembles the in-formation contained .in Sections 2.through 4 of this report.

1-3

& =

r . .

1.4 References'to Section 1

1) Philadel phia Electric Company, " Limerick Generating Station, Prot a-bilistic Risk Assessment," March 1981.

2) I. A. Papazoglou, et al ., " Review of the Limerick Generating Station Probabilistic Risk Assessment," NUREG/CR-3028, February 1983.

3) Philadelphia Electric Company, " Limerick Generating Station, Severe Acci-dent Risk Assessment," April 1983.

4) L. T. Ritchie, J. D. Johnson, and R. M. Bland, " Calculations of Reactor Accident Consequences Version 2, CRAC 2: Computer Code - User's Guide,"

NUREG/CR-2326, February 1983.

5) M. A. Azarm, et al ., "A Preliminary Review of the Limerick Generating Station Severe Accident Risk Assessment, Volume 1: Core Melt Frequency,"

NUREG/CR-3493, January 1984.

6) H. Ludewig , J. W. Yang , and W. T. Pratt, " Containment Failure Mode and Fission Product _ Release Analysis for the Limerick Generating Station:

Base Case Assessment," Draft BNL Report dated August 1983.

7) NRC Memorandum from B. Sheron, Branch Chief /RSB to L. Hulman, Brancn Chief /AEB, " Set of Release Categories for Limerick DES." dated August 15, 1983.

8) " DES Related to the Operation of the Limerick Generating Station, Units 1 and 2," NUREG-0974, Supplement No.1, dated December 1983.

9) " Reactor Safety Study: An Assessment of Accident Risk in U. S. Commer-cial Nuclear Power Plants," WASH-1400, NUREG-75/104,1975.

10)- " Technical Bases for Estimating Fission Product Behavior During LWR Acci-dents," USNRC Report NUREG-0772, June 1981.

11) "Radionuclide Release Under Specific LWR Accident Conditions," Draft BMI-2104 Report, 1983.

12) NRC memorandum from B. Sheron, Branch Chief /RS8 to R. Mattson, Director, OSI," Proposed PRA Methodology for Limerick and GESSAR," dated July 22, 1983.

13) R. O. Wooton and H. I. Avci, " MARCH Code Description and User's Manual ,"

NUREG/CR-1711 October 1980.

14) R. J. Burian and P. Cybulskis, " CORRAL 2 User's Manual," BCL report dated January 1977. - -

1-4 L _

4

&. s

, 7') 'g

- v,.

~

^ ,

., .f., '3 .<

4 - .

s , v.g /

2.0 BINNINC OR ACCIDCNT SEQUEtiCES s -

The process of. "binning" is a smeans of reducing a large number of accident sequences,into a smaller number.of." representative" sequences or classes that

, can be analyied, to determiniC potential contain. ment building failure modes.

.s 9, Each of these failure modes will have unique fission product release charac-

' y , , teristics. It is intended that the failure modes and fission product release characteristics associated with a particular accident class will be represen-tative of thq many individual accident sequences " binned" into the class. In the ' LGS-PRA,.13 all accident aequences were binned into four generic acci-dent ' classes' (Classes I through IV in Table 2.1). Note that in the LGS-PRA, only accidents initiated by internal events were considered. Containment event trees were then used to determine potential failure modes. The deval-opment or the trees *and f.he selection"of branch point probabilities depend on 4~ a detailed assessment of core meltdown phenomena and the response of the con-tainment bullilitig. Jeven (reduced from eleven because of similarities) poten.

?

tial failure. modes or fission product release paths were identified (refer to Table 2:2) for each'6f the four classes. The combination of four classes and seven failure modes Fesulted in a total of twenty-eight distinct fission pro-duct release characteris~ tics (source terms). These twenty-eight source-terms were reduced to five for use in the site consequence analysis in the LGS-PRA.

[ The above binning procedure was reviewed extensively BNL and the details of i

our review a.'e given in Section 6 of NUREG/CR-3028. This review will not be repeated here, but our b4' sic conclusions will be used later in this section n.

to establish the'. probabilities of the various failure modes and release paths.

' In 'the LGS-SARAE33 the four generic accident classes (I through IV) were used by PEco whenever possible to represent accident sequences initiated by

'e external

• events. However, because of the unique characteristics of some of the seismicall) initiated sequences, two additional accident classes were

. developed (Classes Ts and S in Tabic 2.1). For Classes IS and S, containment

_ event trees were not necessary (refer to p. 9-4 of Reference 3) because in both cases the containment was Assumed to be open from the start of the acci-dent.

" At this stage'a preliminary review [43 of the LGS-SARA has been performed at BNL which did not, include a detailed re-evaluation of the frequencies of the core mel,. accident sequences. Consequently, it has been necessary to use at

, face value, significant portions of the LGS-SARA directly as input to the probabilities of the various failure modes and release paths associated with externally initiated accident sequences, m In this section we attempt to calculate the probabilities of the various fail-ure modes. The calculations were started by binning the various accident se-quences into representative damage' states. This process was done by NRC staff and is reproduced here in Section 2.1. After the damage states were identi-fled, the, conditional probabilities of the various failure modes and release paths are determined in Section 2.2. Finally, in .Section 2;3, the probabili-

, ties of the representative' damage states (Section 2.1) and the conditional

, probabilities of the failure modes-(Section 2.2) are combined to give the

- 1 source term probabilities. These probabilities are used in Section 5 to fully define'the source terms for use in the NRC site consequence analysis.

k &

~% ,

2-1 im

% .+ .

x ,

/

M ,s - 2.11 NRC' Staff Classification o'f Core-Melt Sequences The classification of core-melt sequences and selection of damage state fre-n ,

quencies_ was _ performed. by NRC: staff and 'is included in this report only for h- . ease , of l reference.. The process is described in Reference [5] and relies

. heavily ,on' References [1] through [3]. Therefore, the classifications and

' frequencies contained -in this.section and the associated tables should not be regarded as y BNL' estimates. They were selected by NRC s staff with input from

' LBNL reports'and other,information sources as discussed below, a - A complete listing of the' higher frequency accident. sequences is given in Ta-bl e 2.3.' The accideat sequences or damage states are identified by the ini-

tiating event (internal, seismic or fire).and by the classes described in Ta-bl e ' 2.1. . The accident sequences and ' the ' determination of their associated probabilities -are1 described in detail in References -[1] through [3]. Basi-cally -the n" internal" damage states are those evaluated by .BNL in NUREG/CR-

' 3028,(2J' while the " external" damage states are those described in' the LGS-SARA.t Note that since Chapters 5, 6, 7 and 8 of.the LGS-SARA were not part of the Apr11'1983 version of the documents, contributions to the external events

, from the initiators in these chapters are not included in Table 2.3.

!n response to an NRC memorandum [63 of June 15, 1983, we have subdivided the seismic ~ events _ damage. states into two categories; those that are not classi-fled as regional disasters.

(RDs) ("g" level less than 0.40) and those classi-fled as' R0s ("g". level greater than 0.40). This subdivision is then carried through to the listing of the release categories for incorporation into source term characterization in Section 5.

Based .on the damage states listed in Table 2.3, we developed .10 damage state surrogates also listed in Table 2.3. The reduction from 67 damage states to

'10 damage states is made possible because many of the original damage states

1. are very similar _in terms of the core-melt accident progression and contain-ment - failure characteristics. Table 2.4 gives a brief description of each of the surrogate damage states. It should be noted that although LOCAs are con-sidered very low probability events, we have included two in the ten damage states, namely, a small-break, Class. I LOCA labeled as . I-5 and a large-break ATWS-type Class ;IV LOCA, labeled as IV-A. These are included because they.

l result in sufficiently different release. categories to warrant separate con- .

sideration. - ~ For the "!S". Class accidents, two damage states were chosen (in a similar manner to the LGS-SARA), namely, an ATWS-event, IS-C, and a . non-lATWS-event,^IS-T. For the "S" Class accidents, two damage states again were chosen with the distinguishing ' feature being whether the vessel failure

-drained'thelowerplenumwater(S-M)orwhetheritdidnot(S-H20).

All ten of the damage states in Table 2.3 have been analyzed using the MARCH (Section 3) and CORRAL (Section 4) computer codes. Each damage. state has the

_ potential to fall tho' containment by a number of failure modes and release

• z paths.. In the following section we calculate the conditional probabilities of achieving the various failure modes for each damage state.

2-2 t _ _ . - _ _ _ _ _ _ _

, o , , ,

x ,

t j

Note: that the information in Table 2.3 was used- as input to the LGS-DES and I

--based on .information in References [1] through [3]. _However, for input to the  !

' LGS-FES it' was decided by NRC staff to revise the probabilities. in Table 2.3.  :

The revised-numbers quency are includedsuggested

- for ' loss-of-offsite-ppwgr in Table 2.4 byand BNL. ref{egt 7J aInchange addition, Table in the fre- -!

.2.4 also reflects the revisedL8J frequency for sequences initiated by fire.

2.2: Containment Failure Modes and Release Paths

--In :this section we define the conditional probabilities associated with the various containment failure modes and release paths for the ten damage states defined'in Table 2.5. For damage states I-S, (.-h II-T, III-T, IV-T and IV-A, f the containment' event trees developed at 8NLLsJ were 'used to calculate the i probabilities of the failure modes.- However, the BNL event trees were modi-  !

-fied for the - present analysis. The modifications are described in detail  ;

later in this section.  ;

Containment' event trees simply relate a given damage state to a number of po-

- tential failure ' modes 'or release paths. In the present analysis we have de-

. fined seven failure modes and include a description of them in Table 2.5. For ,

a given damage state.-the containment event trees in Reference [2] can be used l

to calculate the conditional probabilities of achieving one or more of the l failure modes in Table 2.6. The damage ~ states and the conditional probabili- ,

ties of the failure modes can be summarized in a containment matrix (refer to {

-Table ' 2.7) . A-damage state together .with one of the failure modes in Table t (2.7 defines a unique fission product release path. From an inspection of Ta-  !

ble 2.7, it is clear that forty release paths have been defined. j

- l

- The containment event trees used for damage states I-S, I-T, II-T, III-T, IV-T l and -IV-A are basically those of Reference [2] with the following modifica-  ?

tions. The probability of a steam explosion ~ induced . failure of ~ the contain-  !

- ment building has been -reduced from 10-3 to 10-4, which is more consistent [

with the current trends _and beliefs regarding this' failure mode. Al so, the l conditional probability of a failure ~in the wetwell, which results in suppres- l

!sion pool . drainage, has been reduced from the value (CP=0.25) used in Refer-

~

ence [2] to a CP=0.05 :as given in Reference [1]. This change is' based on a i structural assessment of the . LGS containment by NRC staff. This assessment .

found that if a crack occurs in theiwetwell, it will tend to prcpagate up- ,

wards, hence the probability of suppression pool drainage should be given a  !

relatively low probability.

l

. 2.3 Source-Te~rm Probability- 's i In this section we simply indica'te how the probabilities of the damage states  :

were combined with the conditional probabilities of the failure modes to de -  !

termine the frequency of occufrence of the source terms. Tables 2.3 and 2.4 include the frequencies of each damage state and also indicates .the frequen-

~ cies of accident ' sequences that result from regional disasters. These prob- .

abfl.ities are summarized in Table 2.8 for input of the LGS-DES and in Table [

5 2.9 for. input to the LGS-FES. y i k'

N f

i 2-3

--,w.- . . , -s. -,nn.---_c.n- ,v,.-.,,n nw,,--,-.~.,mn,-,.~r--,------,~,--,-.--.,-,'

I I

In Table 2.10 we indicated the assignment of release categories to the condi- ,

tional ' probabilities in Table 2.7. Note that the forty potential failure modes have been reduced to twenty-seven release categories Decause of similar-ities; The rational - for combining the failure modes is discussed further in

' Sections 4 : and 5.

The final step is simply to multiply the probabilities of '

the various damage states in Table 2.8 and Table 2.9 and the corresponding '

conditional. probabilities -. in Table 2.7 to calculate the frequencies of the twenty-seven - release categories in Table 2.10. These frequencies are included ,

.in the source term characterization in Section 5.

'2.4 References to Section 2 1)i Philadelphia Electric Company, " Limerick, Generating Station, Probabilistic

~ Risk Assessment," March 1981.

'2): I.- ~A. Papazoglou, et al . , " Review of the Limerick Generating Station Prob- i abilistic Risk Assessment," NUREG/CR-3028, February 1983. l

3) Philadelphia Electric Company, " Limerick Generating Station, Severe Acci- '

Tdent Risk Assessment," April 1983. ,.

t

- 4) M. A. Azarm, et al., "A Preliminary Review of the Limerick Generating Sta- I tion '. Severe Accident Risk Assessment, Volume 1: Core Melt Frequency," t

-Draft BNL report dated August 15, 1983. ,

5) NRC memorandum from B. .Sheron,- Branch Chief /RSB to L. Hulman, Branch  ;

Chief /AEB, " Set of Release Categories for Limerick DES," dated August 15, c1983. t

6) NRC ' memorandum from L. J. Hulman, ' Branch Chief /AEB to A. Thadani, Branch  !

._ Chief /RRAB and B. Sheron, Branch Chief /RSB, " Technical Assistance Request 6 for: Severe Accident Analysis for Limerick DES Supplement-RE: SARA," June  !

-15, 1983. ,

t 7): I. ' A. Papazoglou, " Frequency.of Loss-of-Offsite-Power in the BNL Review of the Limerick PRA," BNL Memorandum to W. Y. Kato, dated July 19, 1983, i

8)-_ Philadelphia Electric Company, " Limerick Generating Station, Severe Acci-  !

' dent Risk Assessment," Supplement .No. _2, dated November 1983. [

i

> i i

I

.$(

h 2-4 i

Table 2.1* Generic accident-sequence classes T

Generic, Accident- Physical Basis System-Level Contributing

-Sequence Designator for Classification '

Event Sequence Class I'(C1)- Relatively fast core- Transients involving loss of melt; . containment inventory make-up, small LOCA intact at core melt events involving-loss of in-

'and at low pressure inventory make-up Class.II(C2) Relatively slow core Transients or LOCAs involving melt due to lower loss of heat removal, inadver-decay heat power; tent SRV opening accidents containment- failed with inadequate heat-removal before core melt capability Class III (C3) Relatively fast core Transients involving loss of melt; containment scram function and inability intact at core melt to provide coolant make-up, but a high internal large LOCAs with . insufficient

, pressure coolant make-up, transient with loss of heat removal and long-term loss of inventory.

make-up Class IV (C4)- Relatively fast : Orc Transients that involve loss e melt; containment of scram function and a loss fails before core of containment heat removal melt because of or all reactivity control overpressure but have coolant make-up capability

' Class IS Relatively fast core Seismically induced sequences melt; containment that lead to failure of the fails before core inventory make-up systems and melt because RHR a breach of wetwell integrity, suction lines are with the reactor scranned severed Class S . Relatively fast core Reactor-vessel failure with melt in an open ves- immediate containment failure sel and failed containment 9

• Reproduced from Table 12-4 of Reference 3.

2-5 u

Table 2.2 Failure modes used in the LGS-PRA Designator Description

'a Steam explosion in vessel-6, u ' Steam explosion in containment and H2 explosion induced containment failure

- Y, u Overpressure failure - release through drywell and H2 deflagration sufficient to cause containment overpressure failure

~Y' -Overpressure failure -' release through wetwell break Y" ' Overpressure failure - wetwell pool' drained (c, ec. Overpressure, large leak and small leak both with SGTS failure

(, 6 Overpressure, large leak and small leak both with SGTS

. operating D

9

+

9 2-6

A Table 2.3 Classification of damage states as used in LGS-DES ,

(core-mcit sequences)

~ .

Annual Number of Frequency Damage

+

Initiating- Damage Cross (point State

-- Event Type State Reference Damage States estimate) Surrogates CLASS I Internal .1 Table 5.22,23 S1QUV 7.6-8a 13 (Ref.12) 2 Table 5.26 Tp0VX 3.7-5 (7.6-8)

(Ref. 2) 3 TEUV 3.2-5

'4 TEUX 8.6-6 ,

5 T TQUX 8.0-6 6

Trux 4.0-6  !

7. T(DC)

T 2.0-6 8 TpQUV. 1.1-6 9 T(AC)

T 6.1-7 I-T 10 TT(WSW) 6.1-7 (1.0-4) 11 T IC'UX 5.0-7 12 T IUV 3.6-7 13- TgQUX' 3.6-7 14 Tr(DC) 3.1-7 15 Tr(AC) 9.2-8 b

Seismic 16a 'Tabl e 12.5 TSS E UX 9.0-7

,(Ref. 3) .

16b Table 12.5 TSS E UX(RD)b '2.27-6 (Ref. 3)

-Fire 17 Table 12.5 TpVV 2.3-6 (Ref. 3)  :

i 2-7

Table'2.3 Classification of damage states as used in LGS-DES (core-melt sequences)-(Cont.)

Annual Number of Frequency Damage

- Initiating- Damage Cross (point State ,

_ event - type : State Reference Damage States estimate)- Surrogates CLASS II Internal- 18 Tabl e ' 5.27 Tp0W 1.3-6 (Ref. 2)

_ 19 TTPW 7.7-7 20 TWE 6.4-7 21 TT(WSW) 5.9-7 II-T

~ 22 .T IW 4.3-7 (4.1-6) 23 TpPW 1.2-7 24 Tr(WSW) 1.1 " "

, 25 T TQW 9.4-8 Seismic 26a Table 12.5 TEW 33 1.0-8 (Ref. 3)_

'26b Table-12.5 TEW-(RD) 33 4.0-8 (Ref.-3).

l 2-8 L-

Table 2.3 Classification of damage states as used in LGS-DES (core-melt sequences) (Cont.)

Annual Number of Frequency Damage

. Initiating-- Damage Cross (point State

, event' type' -State Reference Damage States estimate) Surrogates CLASS III Internal' 27. Table 5.28 TTgI PU C 8.7-7 (Ref. 2) 28 TC pgUUR 5.3-7 29 TECWg 12 4.3-7 30 Ty CUg 2.9-7

'. 31 Tg4CC 2.6-7 M 12 )

32 Tp2CgPU 2.4-7 33 TFCWg 12 1.6-7 34 TT CCM2 1.6-7 35 TEC MUUR 1.1-7 III-T 36 TTC PW M 2 6.6-8 (4.1-6) 37 Tp2CC M2 4.3-8 38 T 4C PU 3.2-8 7M 39 TECMPU ,

2.4-8 40 TE CC 2.1-8 g2 Seismic- 41a Table 12.5 T RPV RgE 5 '9-0 S

(Ref. 3) 41b Table 12.5 T3 RPV IB E(RD) 3.5-7 (Ref. 3) 42a Table 12.5 TECC 33M2 4.3-8 -

(Ref. 3) 42b Table 12.5 T33g2 E C C (RD) 3.9-7 (Ref. 3) 2-9

f.

p 4 --Table 2.3 - Classification of damage states as used in LGS-DES (core-melt sequences) (Cont.)

Annual

. . Number.of Frequency Damage Initiating- Damage Cross (point State event type State Reference Damage States estimate) Surrogates CLASS IV

Internal- 43 Table 5.21,22, AC 5.0-9 IV-A 23-(Ref. 2) 44

" " 1 TT gD 1.4 -7 (5.0-9) 45 T C UD 4.0-8 FM 46 TF CDg 3.3-8

"- " 4 47 TyC gPW 2 2.0-8 48 Ty 1CR E 1.5-8 49 TT CUgH 1.4 -8 50 TTCgPD 1.3-8 IV-T

-" " I TTCM 51 g 7.5-9 (4.2-7) 52 TT CRg 7.4-9

" " 4 53 Tg CDM 3.6-9 54 TFCMPD 3.6-9 55 TF CU 3H 3.3-9 .

56 TECMPD 3.0-9 E

57 TE QUD 2.7-9 58 TE QD 2.7-9 59 TyCM g ' 2.1-9 60 TTCgPU M 1.3-9 Seismic . 61a , Table 12.5 TECC 33g2 1.5-8 (Ref. 3) 61b Table 12.5 T33g2 E C C (RD) 9.5-8 (Ref. 3)

+- 2-10

J.-

e Table 2.3 Classification of damage states as used in LGS-DES (core-melt sequences) (Cont.)

Annual Number of Frequency Damage Initiating-- Damage- Cross (point State event type State Reference Damage States estimate) Surrogates CLASS IS Seismic 62a Table 12.5 TR 3B 1.0-7 IS-f 23 (Ref. 3) 62b T3BR (RD) 9.0-7 (1.0-6) 63a- "

TRC 3Bg 1.4 -8 IS-C j 63b TR3 B M(RD) 1.3-7 ( 1.4-7 )

l CLASS S Internal 64 Table 12.5 R 2.7-8 S-H2O (Ref. 3)

Seismic 65b T3RPVRB (RD) (5.5-8) 66b T3RPHE ( ) -H 0 (3.8-7) 67b T3RPVY TI(RD)

Seismic 4.1-7 Total a' 7.6-8 = 7.6x10-8

.b The seismic damage states,have been subdivided i.nto those which also repre- ,

sent " regional disasters" (RD) and those which do not. This separation is for the DES site-consequence analysis, n

2-11

r Table 2.4 Classification of damage states as used in LGS-FES (core-melt sequences)

~

Annual Number of Frequency Damage Initiating- Damage Cross (point State

. Event Type State Reference Damage States estimate) Surrogates CLASS I Internal - 1- Table 5.22,23 5 1QUV 7.6-8a I_3 (Ref. 2) 2 Table 5.26 TpQUX 3.7-5 (7.6-8)

(Ref. 2)

~

3 Reference [6] TEUV 1.8-5 4

TEUX 4.9-6 5 Table 5.26 TTQUX 8.0-6 (Ref. 2) 6 TguX 4.0-6 7

T(DC)

T 2.0-6 8 TpQUV 1.1-6

. 9 T(AC)

T 6.1-7 I-T 10 T(WSW)

T 6.1-7 (8.31-5) 11 TIC'UX 5.0-7 12 T IUV 3.6-7 13 TgQUX 3.6-7 14 Ty(DC) 3.1-7 15 Tr(AC) 9.2-8 Seismic' 16a Table 12.5 T33 E UX(RD)b 9.0-7 (Ref. 3)

'. 16b Table 12.5 TE0X(RO)b'.

33 2.27-6 (Ref. 3)

Fire 17 Reference [7] TruV 3.1-6 2-12

e a .

Table 2.4 Classification of damage states as used in LGS-FES (core-melt sequences) (Cont.)

Annual Number of Frequency Damage Initiating-- Damage Cross (point State event type State Reference Damage States estimate) Surrogates CLASS II Internal 18- Table 5.27 T FQW 1.3-6 (Ref. 2) 19 TTPW 7.7-7 20 Reference [6] TWE 3.7-7 21 Table 5.27 T(WSW) i 5.9-7 II-T (Ref. 2) l 22 TgW 4.3-7 (3.8-6) 23 TpPW 1.2-7 24 -

Tp(WSW) 1.1-7 25 T TQW 9.4-8 Seismic 26a Table 12.5 TEW 1.0-8 33 (Ref. 3) 26b Table 12.5 TEW(RD) 33 4.0-8 (Ref. 3) 2-13

Table 2.4 Classification of damage states as used in LGS-FES (core-melt sequences) (Cont.)

Annual Number of Frequency Damage Initiating- Damage Cross (point State event type State Reference Damage States estimate) Surrogates

, CLASS III Table 5.28 TTCI PU 8.7-7 Internal 27 g (Ref.~2)

" " U

-28 TCpg UR 5.3-7 29 Reference [6] T E CWg 12 2.5-7 30 Table 5.28 ,,

Ty4CU M 2.9-7 (Ref. 2) 31 Tg4 CCg 12 2.6-7

" " 2 32 TpC MPU 2.4-7

" " 2 33 T p CWg 12 1.6-7 34 TT CgC2 1.6-7 35 Reference.[6] TE CguuR 6.3-8 III-T 36 Table 5.28 T TI Cg PW 2 6.6-8 (3.9-6)

(Ref. 2) 37 T p CCg2 4.3-8

"

• 4 38 TgC gPU 3.2-8 39 Reference [6] T E CgPU 1.4-8 40 TE CC g2 1.2-8 Selsmic 41a Table 12.5 T RPV F 3.9-8 3 BE (Ref. 3) 41b Table 12.5 T3 RPV RgEIi(RD) 3.5-7 (Ref. 3)

TECC 4'3-8 Table 12.5 -

42a 33g2 (Ref. 3) 42b Table 12.5 T33g2 E C C (RO) 3.9-7 (Ref. 3) 2-14

Table 2,4 Classification of damage states as used in LGS-FES (core-melt sequences) (Cont.)

Annual Number of Frequency Damage Initiating- Damage Cross (point State event type State Reference Damage States estimate) Surrogates CLASS IV Internal 43 Tables 5.21,22 AC 5.0-9 IV-A 23 (Ref. 2) 44 TT CD 1.4 -7 g (5.0-9) 45 TpCguD 4.0-8 46 Ty CD g 3.3-8

" " 4 47 Tg C gPW2 2.0-8 48 " "

TTI CRE 1.5-8 49 " "

T CUI 1.4-8 T gg 50 TTCI PD 1.3-8 -

IV-T g

51 TT ICMg 7.5-9 (4.2-7) 52

" " I 7.4-9 TT gR 53 Ty 4CD g 3.6-9

" " 2 54 TpC gPD 3.6-9 55 Tp2CU gH 3.3-9 56 Reference [6] TECgPD 1.7-9 TE3CguD 57 1.5-9 58 Tg3CD M 1.5-9 59 Tables 5.21.22 Tp2CM g 2.1-9 23 (Ref. 2) 60 " "

.T T I CM PUg 1.3-9 Seismic 61a i 1 1 5 TECC 1.5-8 33g2 61b Table 12.5 T33M2 E C C (RD) 9.5-8 (Ref. 3) 2-15

Table 2.4 Classification of damage statas as used in LGS-FES (core-melt sequences) (Cont.)

Annual Number of Frequency Damage Initiating- Damage Cros: (point State event type State Reference Damage States estimate) Surrogates CLASS IS Seismic 62a Table 12.5 TR 3g 1.0-7 IS-C 23 (Ref. 3) 62b TR(RD) 3B 3.0-7 (1.0-6) 63a TRC 3Bg 1.4-8 IS-C 63b T3Bg R C (RD) 1.3-7 ( 1.4-7 )

CLASS S Internal 64 Table 12.5 R 2.7-8 S-H2O (Ref. 3)

Seismic 65b TRPVR(RD) (5.5-8) 3 B 66b T3RPHE (RD) S-H2O (3.8-7) 67b TRPVR'R(RD) 3 Seismic 4.1-7 Total a 7.6-8 = 7.6x10-8 b The seismic damage states have been subdivided into those which also repre-sent " regional disasters" (RD) and those which do not. This separation is for the DES site ' consequence analysis.

2-16

t Table 2.5 ' Description of damage states n

Designator Description I-S These are LOCA initiated sequences (medium and small breaks only) involving loss of inventory make-up. They result in a relatively fast core melt and the containment is intact at the time of core mel .  ;

I-T These are sequences initiated by transients again involving -

loss of inventory make-up. Core melt is relatively fast and the containment is intact at the time of core melt. _

II-T These are transient or i.0CA initiated sequences involving loss of containment heat removal or inao.>ertent SRV opening acci- -

dents with inadequate heat removal capability. Ocre melt is _

relatively slow due to lower decay power level and the contain- '

ment has failed prior to core melt.

• III-T Transients involving loss of scram function and inability to provide coolant make-up, large LOCAs with insufficient coolant .--

make-up, transients with loss of heat removal and long-term  ;

loss of inventory make-up. Core melt is relatively fast and a the containment is intact at core melt.  ;

i IV-T Transients that involve loss of scram function and a loss of g containment heat removal or all reactivity control but have 3 coolant make-up capability. Core melt is relatively fast and the containment fails prior to core melt because of over- _

pressure. -

( IV-A As above but initiated by large LOCAs. "

1 IS-! Seismically induced sequences that lead to failure of the in-ventory/make-up systems and a breach of wetwell integrity, with -

the reactor scrammed. Core melt is fast and the containment fails prior to core melt because the RHR suction lines are  :

severed.

! _]

l IS-C As above but coupled with a loss of the scram function.

} .

S-H2O Seismically induced reactor-vessel failure (plus random reac- )

tor-ves el failure) coupled with immediate containment failure. J Core melt is fast and the vessel and containment are both failed at the time.of. core melt. This sequence assumes.the vessel b.reak is high, which allows water to be retained in the' '

b'ottom of the vessel prior to core slump.

S-H2O As above but with a vessel failure location that results in complete draining of the water from the vessel.

2-17 j m

__ . , . , , . . , . . _ . - . _ . , , . _ _ _ , _ - _ . , . . . . ~ . .

Table 2.6 Containment failure mode and release path notation Designator Description DW Containment Failure via overpressurization. Failure location in the drywell.

WW Containment Failure via overpressurization. Failure location in the wetwell above the suppression pool.

157 Containment Failure via overpressurization. Failure location in the wetwell below the suppression pool resulting in loss of suppression pool water.

SE Failure via in-vessel steam explosion generated missiles.

HB Failure via H2 burning during the periods when the contain-ment atmosphere is deinerted. This failure mode also includes H2 detonation and ex-vessel steam explosion failure modes, which ara of very low frequency. ,

LGT Containment leakage rates sufficiently low to allow the stand-by gas treatment system (SGTS) to operate effectively.

L37 Contair. ment leakage rates so high that the SGTS is ineffec-tive.

e 2-18

E l

=

Table 2.7 Assignment of conditional probabilities Containment Failure Modes and Release Paths Damage States No Core DW WW I5i SE HB LGT LGT Mel t I-S 0.247 0.223 0.025 0.0001 0.01 0.222 0.273 0 I-T O.247 0.223 0.025 0.0001 0.01 0.222 0.273 0  ;

II-T 0.250 0.225 0.025 0.0001 0 0 0 0.5 III-T 0.247 0.223 0.025 0.0001 0.01 0.222 0.273 0 IV-A 0.500 0.45 0.05 0.0001 0 0 0 0 l

7 IV-T 0.500 0.45 0.05 0.0001 0 0 0 0 l

IS-C 1* 0 0 0.0001 0 0 0 0

~

F IS-C 1* 0 0 0.0001 0 0 0 0 l 4

S-H2O O 0 1** 0.0001 0 0 0 0 S -112 0 0 0 1** 0 0 0 0 0

• In the LGS-SARA, this failare mode was considered similar to a drywell (DW) failure mode, however, this should not be interpreted as a failure location in the drywell. Class IS sequences result in failure of the RilR suction lines, which partially drains the suppression pool exposing the downcomers but leav-ing the quenchers submerged. Thus, the melt release will be scrubbed by the pool (similar to WW fail- =

ure mode) and the veporization release will not be scrubbed by the pool (similar to DW failure mode).

h

• • Again, assigning the IN failure mode to Class S sequences relates to the tission product release path (and lack of suppression pool scrubbing) rather than to the failure location.

L T

Table 2.8 Damage state probabilities as used in LGS-DES Damage Total Probability Probability Non-State Probability Regional Disasters Regional Disasters I-S 7.6(-8)* -

7.6(-8)

I-T 1.0(-4) 2.27(-6) 9.8(-5)

II-T 4.1(-6) 4.0(-8) 4.06(-6)

III-T 4.1(-6) 7.4(-7) 3.36(-6) 3 IV-A 5.0(-9) -

5.0(-9)

IV-T 4.2(-7) 9.5(-8) 3.25(-7)

IS-C 1.44(-7) 1.3(-7) 1.4 (-8)

IS-Y 1.0(-6) 9.0(-7) 1.0(-7)

S-H2O 5.45(-8) 4.1(-8) 1.35(-8)

S-HZ6 3.83(-7) 3.79(-7) 1.35(-8)

• 7.6(-8) = 7.6 x 10-8 i

2-20

Table 2.9 Damage' state probabilities as used in LGS-FES

).

Damage Total Probability Probability Non-State Probability Regional Disasters Regional Disasters I-S 7.6(-8)* -

7.6(-8)

I-T 8.31(-5) 2.27(-6) 8.1(-5)

.II-T 3.8(-6) 4.0(-8) 3.8(-6)

III-T- 3.9(-6) 7.4(-7) 3.2(-6) l IV-A 5.0(-9) -

5.0(-9)

IV-T 4.2(-7) 9.5(-8) 3.25(-7)

IS-C 1.44(-7) 1.3(-7) 1.4(-8)

IS-Y 1.0(-6) 9.0(-7) 1.0(-7)

S.H2O 5.45(-8) 4.1(-8) 1.35(-8)

S- M 3.83(-7) 3.79(-7) 1.35(-8)

• 7.6(-8) = 7.6.x 10-8 2-21

, e e f

G G L L M 4 4 OM g e6 re g ru g <q:

- N N O- N N N N N N  %

g C C g C C C C C C C 4"

r 4,

b. , . - ,. . . , ,. ,,

lC 4

% i N N N N N N C ** C C C C C C 1 I w w w ee 2

en >

C a > H ,

1. 45 C, C  %

A W .- J 70 H tu g 9 m en re W f=l3  %  % N I N N N N N l}'

u G

4 w

C ""* C C C C C N

C O @

u W: W CC

-as- CI3 V V Z C 2 Cl3  %

G @ Z Z @ > tu to 90 m n W

M W g N W

N W

N C

I N

C C

C on C

N C

C

@ _ W_ 9 1 m-e

. ** w e.=

W .: .'

6-

% N.

V 6 W W 2 W W W W W W M M - W W W  % M M M M N g

- M M N M N N N C N 9 W N N H I H H lN N C

'u. W s s i w i i U. U. z a e-e I

w 4 w w > > i

.* e.= em ee M.

e- Me M

.s=

W e . E E =

o 5 $s $N C

O E

3%  %

H I (

e N H  %

, o N

e N

N 'O -

M H t ** e I C C Z Z e e e-o

• • === > > a e

'y

. == e-o e sw w w Yj 4

3 3 a.m 3

=

e 2 2

C2 2  % H  %  % g eg rg g 3  %  % H e aC H  %  %  %  %

3 M W I == t C C C C I

e-o we I ew em

=

sw m.e 2 2 3 E E E N  %-

e E c e N N W  %

(  %  %  % . .

I O lW  %  %

E M - i = . H. i . .C C

, I e-e

- 0 me m*

m

==

M

==

M . .

=

• e D

$$ N Ll9 G & O C

& W l 4 W U lQ N N

a. M. i e w i i i i Z z W ** ** ** *=e > > M M I i M me me ==. me == == M M 2-22 '

p 3.0 CORE MELTDOWN AND CONTAINMENT RESPONSE In this section, MARCHE 13 analyses of core meltdown and containment response for the various representative accident sequences are presented. The MARCH computer ANS-5.1-1979L cod;2Jhas been modified to include a new decay heat model based on the standard. The new MARCH computer code model (programmed by C. Shaffer of Sandia National Laboratories) includes the decay of U239 and NP239 and the effect of neutron capture in fission products which was not in-cluded in the original MARCH model. The new model produces an integrated decay heat over the first hour after shutdown about 20% greateg than the orig-inal MRCH model, which was used in the previous BNL reviewL3J of the LGS-PRA.[4 The main effect of the new decay heat model is the change in timing of major events. The time to core meltdown, reactor pressure vessel (RPV) failure and containment failure predicted using the new decay heat model are significantly earlier than that in NUREG/CR-3028. The highlights of the basic assumption of the- MARCH analysis are summarized in Table 3.1. This table is similar to Table 7.12 in NUREG/CR-3028. The major difference relates to the assumed decontamination factor (input parameter DCF in MARCH) for a saturated pool. In NUREG/CR-3028 a DCF=10 was assumed for a saturated pool, which is consistent with the LGS-PRA. However, in the present study, a DCF=1 is as-sumed, which is consistent with the CORRAL analysis in Section 4 and our " base case technology" approach (refer to Section 1.2).

3.1 Class ! Sequences The Limerick PRA describes this Class as follows:

"The Class I (C1) events can be characterized as transients involving loss-of-coolant make-up to the reactor core. For the Limerick analy-sis, these events are found to have the highest calculated frequency of occurrence. They involve successful control rod insertion; how-ever, there is postulated to be a loss of both high pressure and low pressure injection. The physics model used in the consequence calcu-lation represents the sequence designated TQUV."

The TQUV sequence is analyzed for three containment failure modes, namely, structural failure in the drywell, in the airspace of the wetwell, and below the pool waterline in the wetwell (which drains the suppression pool). Due to the loss-of-coolant make-up to the reactor core, the MARCH code predicts the start of core melting at 90 minutes and slumping into the lower vessel head at about 145 minutes. The molten core is discharged onto the diaphragm floor at 174 minutes. Containment failure is rssumed to occur as the core debris pene-trates 70 cm of the diaphragm floor. Table 3.2 shows a comparison of the pre-

-sent BNL analysis with the analyses of the Class I (TQUV) sequence in the LGS-PRA and in NUREG/CR-3028 of the Class I (TQUV) sequence. The earlier timing

.of the major events predicted by this study is caused by using the 1979 decay power standar4. During the transient, the suppression pool was predicted to remain subcooled. Hence, 'a DCF=100 is used in the MARCH analysis. It is noted that the vessel head failure occurs prior to the containment failure.

Any rapid debris / water interaction or steam explosion in the pedestal region takes place in an inerted atmosphere. It was assumed that the inerted atmos-phere prevents the oxidation release associated with steam explosions (refer to Section 4),

3-1 u - - _ - - _ - - - _ - _ _ _ _ _ - - _ - - _ _ - - _ - - - _ - -

To sinulate a failure in the wetwell in which the suppression pool is drained into the reactor enclosure, the MARCH input gqrameter NT is specified as -7.

Using this option, the MARCH code assumesE J complete f ailure of the sup-pression pool. The intercompartment transfers from the drywell to the wetwell and direct blowdown from the primary system through SRV's are assumed to by-pass the suppression pool.

In addition to the transient events discussed above, the LGS-PRA also con-sidered accidents initiated by loss-of-coolant accidents (LOCAs). For Class I sequences, an S1 0VV (nedium LOCA) accident was identified as one of the more probable LOCA-initiated sequgnces. A medium LOCA event is defined as a break of between 0.044 and 0.1 f t in a liquid line, and between 0.016 and 0.08 ft2 in a steam line. The 51 QUV sequence is defined as a failure of the condensate and feedwater systen, and the failure of both high pressure and low pressure injection systemg. A MARCH analysis of the S iOUV has been per-forned assuning a 0.08 ft break in a steam line. The results are shown in Table 3.3. The tining of the major events are similar to that of the TQUV se-quence, which is also included in Table 3.3.

3.2 Class II Sequences This class is described in Chapter 3 of the LGS-PRA as follows:

"For Class II (C2), tne sequence modeled is a transient with long-term loss of heat removal (TW). For Limerick, this sequence involves the failure of the power conversion system and of the RHR systen.

Also included in this class are other sequences, such as LOCA accon-panied by a failure of the containment heat removal systems, and in-advertently open relief valves with failure to remove heat from con-tainment. The key feature in this class is that the containment is assuned to fail prior to core nelt, but af ter a lengthy period of time into the accident. Postulated core melt begins with a relatively low decay heat source, leading to a slower core melt than anticipated for Classes I, !!!, or IV, but with a failed containment."

In our modeling of TW, the high pressure ECC switches its suction from the condensate storage tank (CST) to the suppression pool at N20 ninutes because of a high suppression pool level and fails at 372 min because of pool heatup (temperature > 200 F). The low pressure pumps start at 372 min (TW=TWLP).

ECC pumps are throttled back when the total water mass in the pressure vessel exceeds 600,000 lb, hence keeping the water level steady and considerably above the core. Eventually, the containment pressure exceeds 155 psia as steam passed t.hrough the SRVs heats up the pool to saturation at about 380 min and then passes the decay heat mostly to the containment atmosphere. Injec-tion is assumed to fail when containment fails. The pool DF is assumed to be 1, because the core nelt begins after the pool has been heated to Saturation.

Table 3.4 compares the results of the .present BNL calculat. ions and with the

' LGS-pRA and NUREG/CR-3028 results. The timing of the major events predicted by this study are much f aster than the predictions of both the LGS-PRA and NUREG/CR-3028. This is because of the pronounced effect of the 1979 decay heat correlation. The increased decay heat causes the reactor pressure to fail at 26.8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br /> compared with 38.7 hours8.101852e-5 days <br />0.00194 hours <br />1.157407e-5 weeks <br />2.6635e-6 months <br /> in NUREG/CR-3028.

3-2

c

( - . . .

a 3.3. Class III Sequences In this transient initiated accident sequence, the control rods fail to insert followed by poison injection failure. The recirculation pumps trip and the feedwater flow is stopped, which rapidly brings the power level down to an assumed 30*.. The high pressure injection systems are modeled to come on with

- flows of 600 gpm (RCIC) and 5600 gpm (HPCI). At 4.5 minutes, the high pres-sure pump suction is automatically switched from the CST to the suppression pool because of high water level in the suppression pool (ECCRC=0.850).

These high pressure systems will subsequently fail either because of high suppression pool pressure (Limerick PRA assur.ption) or high (200*F) suppres-sian pool temperatures (BNL assumption). The high pool tenperature causes the ECC turbine lubricating oil, which is cooled by the suppression pool water, to break down causing the turbine to seize. The present analysis has the ECC pumps failing at 14 minutes because of this loss of lube oil cooling. Other important MARCH input data for the ATUS-III sequence are:

Fourteen safety relief valves operate compared to four for Class I.

MARCH modeling simulates 307. power for the fraction of the core that is covered and ANS-1979 standard decay power for the uncovered part.

Ex-vessel core debris / water interactions are included but the amount of water on the diaphragm floor is so small that essentially no time delay is observed.

A comparison of the present calculations for the above sequence with the LGS-PRA and NUREG/CR-3028 results is given in Table 3.5.

3.4 Class IV Sequences The MARCH modeling for the ATWS Class IV sequence differs from the ATWS Class III modeling only in the following respects:

a) The high pressure injection is allowed to stay on even after the sup-pression pool temperature exceeds 200*F [the ECC turns off when the containment fails - a MARCH parameter option (ICBRK=0)].

b) The choice of containment break area is 5 ft2 in order to prevent the containment from appreciably overshooting the 155 psia failure pressure. The Limerick PRA used a 3.14 ft2 hole size.

The ATWS-IV sequence was analyzed for the three containment failure locations considered in Section 3.1. The timing of the major events as predicted in the present analysis, are compared with the LGS-PRA and NUREG/CR-3028 results in Table 3.6. ,

In addition to a transient initiated event, a large LOCA initiated event (AC) was also analyzed. The sequence AC was identified in the LGS-PRA and is de-fined as a pipe break in either the recirculation line (water break) or nain stean line (stean break), coupled with the failure of the control rods to in-sert. For our analysis of this sequence, blowdown data for a main steam line 3-3

. . . o .

. break'-(given in Table 6-2-11 of 'the LGS-FSAR) were used as input data to the MARCH code. TM FSAR blowdown data indicates that the steam and liquid flow rates : approach zero in approximately 60 seconds and do not change signifi-cantly during the remainder of the sequence. The containment is assumed to fail in the drywell . The results are shown in Table 3.7 and were used as in-put:to the fission product release calculations (Section 4.4.2) for the LGS-z0ES.- However, we noted in Section 1.0 that BNL was requested to reanalyze the Class IV sequences for input to the LGS-FES. The reanalysis is described in

, the following section.

3.5: Class IV Reanalysis

-In the MARCH code, the containment feJponse after the failure of reactor ves-sel is analyzed by using the INTERLU code as a subroutine to model corium/

concrete interactions. The INTER code was developed in 1977 to model small-scale experiments in which melts with a generally high metallic content were poured into a concrete crucible. In these experiments, the principal attack on the concrete was from the metallic layer that did not wet the concrete and resulted in a purely thermal attack. This is not necessarily the case with oxide melts that do wet the concrete. The applicability of tne INTER model to melts having a high oxide content was questioned by the developer of the code, W. B. Murfin. Murfin also stressed that- the model represented only a first stage in the modeling project, and he cautioned against applying the model to prototypical containment ouilding mej t- An improved core /

concrete ' interaction model, CORCON,L6]through has been analysis.

issued by Sandia National Laboratories. The CORCON model roves upon the preliminary INTER model be-cause -it is intended to provide quantitative estimates of full-scale reactor fuel-melt accidents. While it is outside the scope of this report to replace the INTER model in MARCH with CORCON, it is possible to run MARCH and CURCON concurrently.

The MARCH /CORCON technique was applied to the four ATWS-IV sequences. The initial conditions for corium/ concrete interactions obtained from the MARCH

~

calculations were input. to CORCON. The outputs from CORCON, involving the flow rates and temperatures of the steam. hydrogen, carbon dioxide, and carbon monoxide, were fitted as polynominal equations by the least-square method.

.These polynominal equations were incorporated into the MARCH code which by-passed the INTER model . The :CORCON code predicts significantly slower con-

. crete erosion velocities and gas generation rates than the INTER code. Hence, the predicted containment pressure, temperature, and leakage rates are lower than the predictions given by the MARCH / INTER analyses. However, the timing of the major events for the ATWS-IV sequences is not influenced by using the MARCH /CORCON approach. In addition, the following two improvements were added

'in the revised calculation:

. a) . Containment break area was reduced from 5 ft2 to 3 ft2 to be con-sistent with the LGS-PRA analysis.

.b) 'Conta'inment heat sinks were increased from 8 to 15 to' provide a more realistic representation of all structures in tne LGS containment.

.The impact of these improvements in the fission products transport and release will be discussed in Section 4.4.3.

o 3-4

- :.~ .. .' .

[

l r

3i6 . Class IS Sequences f

[The LGS-SARAU3 ' describes this class as follows:

I

" Class TIS' includes earthquake-initiated transients witn a loss of  !

the ability to maintain core-coolant inventory. The reactor-coolant i system.and the containment structure are otherwise intact. The reac-  !

tor enclosure is severely damaged by the earthquake. The earthquake l also causes the 24-in. RHR suction line from the suppression pool .to [

shear outside the containment. This allows the suppression pool to  :

drain to-the level of the RHR suction line. The depth of water above i the. quenchers is reduced from 18 ft.11 in. (minimum water level) to I 7 ft. -The. sheared line also provides a direct flow' path from the con- I tainment to the environment." i f

- The typical sequence for this class was defined in the LGS-SARA as T Rg. S [

-For this sequence, coolant. inventory make-up is lost due to loss of all ac and l de power ~and the RHR heat-exchanger lines are assumed to be severed. Anotner  ;

sequence.ST RgCM is similar.to the TS Rg sequences, but with the reac- [

. tor also failing to scram. For this class, the suppression pool is availaole  !

for scrubbing the melt release since the quenchers are below the level of the  !

RHR suction line. The pool DF is 100 as the pool temperature remains sub- F cooled during the transient. However, as the suppression pool nas drained to i below the level of the downcomers, the vaporization release is not subjected  :

-to pool scrubbing.

l Botn the TRS8 and TRCS8M sequences have .been analyzed and the re. I suits are summarized in Table 3.8. No comparisons with LGS results and the ,

previous BNL results are included in Table 3.5. The results of a core ' melt- *

' 'down analysis for this class 'are not presented in the LGS-SARA report and  !

NUREG/CR-3028 did not, of course, consider any externally initiated accidents.  ;

The TsRgCM sequence is similar to the TRSB sequence but with the t reactor failing to ' scram, the major events of the TSRgCg sequence' occur  ;

sooner than the T RS 8 sequence.  ;

3.7 Class S Sequences The LGS-SARA describes this class as follows:

" Class S has two components. The first is an earthquake-induced failure of the reactor-vessel lateral supports, leading to a failure of the main steam lines and a simultaneous containment failure, either i directly, through a failure of the RHR suction lines, as described for ]

Class IS, or indirectly, through an overpressurization or a mechanical j impact resulting from the vessel losing its lateral restraint. The ,

second is a random reactor-vessel failure accompanied by a containment  ;

failure." . .

t

. .. .  ? i There 'are two source terms appropriate for this class of acciifent. The two'  !

source terms are associated with two cases, one in which vessel failure leads I*

to a loss of all water from the vessel and one in wnich some water remains in the botten of the vessel. The source terms for the cases with and without i

f f

3-5 u.. _ _ _ _ ___ _ _ __ _ _ _______ _ ________.__._________.__________._______ _ __ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ . . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

e

. o .

water in the lower plenum are labeled S-H2O and S-H20, respectively (refer to Section 2). A major contributor to tnis class is iJentified as the T R?VRB S sequence.

To simulate the T SRPVRB sequence, the failure of the main steam lines are considered equivalent to a large LOCA, whicn occurs simultaneously witn tne containment failure at the initiation of the accident. The blowdown data for a main steam line break given in Table 6.2-11 of the LGS-FSAR were used as the input data for the MARCH code. The cases associated with the two source terms, S-H2O and S-E, are modeled in the MARCH code by varying the input parameter WATBH, wnich is defined as the mass of water tnat can be stored in tne bottom vessel head. The parameter WATBH is taken as 182710 and 100 for the two source terms, S-H20 and S-H20, respectively. ( A zero-mass in the bot-tom nead will lead to an overflow condition for tne MARCH numerical computa-tion precedure; hence. it is necessary to specify 100 lb of water for tne S- E case). The results of TS RPVRB sequence are snown in Tabl e 3.9.

Again, we are unable to compare tne results in Table 3.9 witn the LGS-SARA.

3.8 Summary MARCH analyses for the various representative accident sequences have been performed. The analyses cover both internal and external initiated accidents and tnree potential containment failure modes. Using tne 1979 decay neat standard, the MARCH predicted timing of major events in the accident progres-sions are significantly earlier than those reported in NUREG/CR-3028. Tne MARCH results are summarized in Table 3.10.

3.9 References to Section 3

1) R. O. Wooton and H. I. Avci, " MARCH Code Desc. iption and User's Manual ,"

Battelle Columbus Laboratories /USNRC Report ;lVREG/CR-1711, Uctober 1980.

2) ANSI /ANS-5.1, " Decay Heat Power in Light Water Reactors," August 1979.

3) 1. A. Papazoglou, et al., "A Review of the Limerick Generating Station Probabilistic Risk Assessment," Brookhaven National Laboratory /USNRC Re-port NUREG/CR-3028, February 1983.

4) Philadelphia Electric Company, " Limerick Generating Station Probabilistic Risk Assessment," Maren 1981.

5) S. R. Greene " Undocumented MARCH BWR Containment Modeling Feature," ORNL memorandum dated January 21, 1983.

J. F. Muir et al .', "CORCON-M0'01: An Improved Model for Molten-Core /

6)

Concrete Interactions," NUREG/CR-2142, July 1981.

7) Philadelphia Electric Company, " Limerick Generating Station, Severe Acci-dent Risk Assessment," April 1983.

3-6

a , , , s-

'y f: ~

g j .

t  ; ..

, y . >

Nable 3.1. til0hlights cf MARCH analysis u . n.. . -

r , <. - y 4 . '.-

5 c ,'

~ '

n '

1. OnIly 2 compartments modeled (wetwell and drywell).

2. Failure of coolant injection due to overheating of luce oil when the suppressioq pool temperature is greater than 200 F.

3. No H2 burning or detonation.

4. 8 heat sinks used* in MARCH instead of the 17 used in If4COR.

1

~

5. Heat transfer coefficient between steel and concrete = 2 Btu /nr/ft , 2

6. Pool decontamination factor, DCF = 100 for subcooled water and 1 for saturated water or W fatture mode.  ;

l

7. Wetwellcompartmentvolumeisairspaceonly(VC(2)=155,000ft).

8. Containment failure. occurs when penetration of diapnragm floor ),70 cm orcontainmentpressure1155 psia.

9. Containment leakage taken as 1/21 volume / day.

10. Equivalont clad thickness includes zirconium from fuel channels.

(~

, 11. Core slumps when 80% of. core is melted.

12. HOTOROP subrouttne made loactive ny using MARCH options.

13. Core dehtis assumed retilned on diaphragm floor inside the pedestal well.

• 8 heat sinks were utilized in the MARCH analysis for the LGS DES.

However, the heat sinks were increased to 15 for the four additional Class IV sequences reanalyzed for tne LGS FES (refer to Section 3.5). ,

37 l

gs

t I

Table 3.2 Comparison of BNL and Limerick PRA analysis of the Class I sequences (TQUV) l r

F Key Analysis in BNL Analysis Events Limerick PRA This NUREG/CR-3028 Work Start of core 1.3 1.65 1.50 .

melt (hours) ,

Core slump 2.5 3.08 2.42 (hours) ,

t Vessel head 4.3 3.71 2.90  ;

failure  !

(hours) .

1 Start of core / 4.3 3.71 2.90 concrete inter- '

r actions (hours)

I r Time (hours) core 6.5 6.12 5.17 P

debris penetrates 70 cm of diaphragm floor causing col-lapse of floor and dr containment failure -

Pressure at con- 88 113 118 tainment failure (psia) i 3-8

Table 3.3 BNL analysis of S QUV 1 sequence Key Events (hours) .S 1QUV TQUV Start o'f core melt 1.'35 1.50 ,

i Core' slump . 2 . 24 -

2.42 Vessel head. failure 2.83' 2.90 t

^

Containment faiiure*- 6.0 5.17 N

. I

• Containment failure caused by 70-cm penetration of the floor, the containment pressure at floor failure is 122 psia for, the S QUV 1 sequence and.118 for the TQUV sequence. y s ,' ,

c, .

Table 3.4' Comparison. of BNLA and Limerick PRA analysis of the. Class II sequences (TWLP) ,

i

's

., Analysis _

e BNL Analysis x ' in < limerick Key Even'ts _

PRA Ref. [3]* This Work j

. , f~ ,

e

. Containment failure '(hr)

'.. 30 29.2 19.5 Core melt begins (hr) 36.6 ,,$6.0 24.9 Core melt-ends (hr) 39.0 38.6 26.8 Vessel head: fails (hr) . 40.8 38.7 26.8

-Z = 70 cm penetration.'.(hr), ;

43.3 47 33.2

-* Containment failure mode is a WW and the break ar'ea is 0.208 ft 2,  ;

3-9

i

' Table 3.5. Comparison of BNL and Limerick PRA analysis  ;

for Class III, ATWS sequence j .

i BNL Analysis Analysis in 0F=10 DF=1 DF=1 i

, Key Events Limerick:PRA Ref[3] Ref[3] This Wnrk

- Core melt begins'(hr)' O.85 0.78 0.76 0.5 Core melt ends (hr) 2.5 2.30 2.22 1.80 v

- Vessel head failure (br). 4.3 2.55 2.47 2.05' INTERbegins'(hr) 4.3 2.55. 2.47 2.05 Containment. failure (hr) 6.5 4.45*- 3.83** 2.67* t

• Containment fails beca s 'u e of overpressure (155' psia) before 70-cm of  !

concrete pene,tration is reached.

~ 'r

'** Time that the core debris penetrates 70-cm of diaphragm floor causing floor collapse and containment failure. '

. \'

s 9

'* A O

v

- Y 3-10 l

.- +---r--

Table 3.6 Comparison of BNL and Limerick PRA analysis for Class IV BNL Analysis Analysis in DF=10 DF=1 0F=1 Key Events Limerick PRA Ref[3] Ref[3] This Work Containment fails (br) 0.67' O.67 0.67 0.67 Core melt begins (hr) 1.2 .1.25 1.25 1.16 Core melt ends (hr) 2.2 2.7 2.7 2.24  :

Vessel head fails (br) 4.0 2.97 2.95 2.47 Time for 70-cm penetration 6.5 6.97 7.03 6.16 of floor .(hr)*

• Limerick assumed the molten core to spread over the entire diaphragm floor.

BNL assumed the core materials to be confined to the pedestal region.

Table 3.7 BNL analysis of AC sequence of Class IV accident Key Events (hours) AC ATWS Containment failure 0.65 0.67 Start of core melt 1,17 1.16 Core slump 1.58 2.24 Vessel - hea,d . fail ure 2.20 2.47 ,

70-cm penetration of floor 5.43 6.16 3-11

I. .

. O e Table 3.8 BNL analyses for Class IS

' Key Events (hours) TRSB TRCSBM l -Containment fails 0 0 Core melt begins 1.47 0.37  !

' Core melt ends 2.32 1.28

-Vessel head fails 2.37 1.53

.70-cm penetration of floor- 6.46 5.02 h

k Table 3.9 BNL analyses of Class S (T RPVRB)

S

. Key Events (hours)

S-H2O S-H26 ,

^ Containment fails- 0 0 ,

-Core' melt begins 2.67 2.83 l Core melt ends 3.65 3.85 f Vessel head fails 5.23 4.38 '

i cm penetration of floor 8.82 7.22 i

3-12

.Q.

Table 3.10 Summary of MARCH 'results . . .

Occurrence of Major. Events (hr) .

Cont &inment Sequence Pool- . Pool Break Area Start Core Vessel Containment 70-cm Floor Temperature

• DF (ft 2)-

Melt' Slump Failure Failure ~ Penetration TQUV 1.5 2.42 2.90. 5.17 5.17 Subcool 100 5 4

S1QUV 1.35 2.24 2.83 6.0 6.0- Subcool 100 5 TWLP 24 .9 26.8 26.8 19.6 33.2 Saturation 1 0.208

~

4 ATWS 0.5 -1.80 2.05 2.67 5.13 Saturation 1 2 ATWS 1.16 .2.24 2.47 0.67' 6.16 Saturation 1 5**

AC 1.17 1.58 2.20 0.65 5.43 Saturation 1 5 TSRB 1.47 2.32 2.37 0 6.30 Subcool 100 5

! TSRBCM 0.37 1.28 1.53 0 4.81 Subcool 100 5 1 TSRPVRB 2.67 3.65 5.23 0 8.82 Subcool 1 5

(S-H20)

? TSRPVRB 2.83 3.85 4.38 0 7.22 Subcool 1 5 i

l (S-H20) l j

• Pool temperature during core meltdown.

l **A containment break area of 3 ft 2.was used in the Class IV reanalyses.

i (refer to Section 3.5).

I i

_2

.1 .. .

4.0 FISSION PRODUCT TRANSPORT AND RELEASE Due to current activities in _ the field of fission product chemistry and transport, a two-pronged approach is being pursued in detemining the fission product release for core meltdown accidents in the LGS. The two approaches are described in Reference 1, which also gives the justification for the ap- '

proaches. . The two approaches are briefly described below: '

1) Base page technology. This approach is based on the Reactor Safety StudyL2J (RSS) methods regarding the fission product source tems, pool scrubbing and . fission product transport.

2) Advance 3fechnology. This approach will be based on the current methods -and data generated by the Accident Source Tem Project Office (ASTFG) of NRC/RES.

Briefly, these two approaches will affect the following four areas related to the detemination of fission product release:

1) -Fission product released from core material. In the base case tech-

~

nology, these quantities are based on the four release periods used in the RSS (gap, melt, oxidation, and vaporization). Furthemore, the timing of the releases will follow the same prescription outlined in the RSS. In the advanced technology case, the release fraction in the reactor pressure vessel (RPV) will be based on the core heatup

' history and the latest data on fission product release from heated core material. In the ex-RPV release phase, the prediction of fission product release will be detemined by models based on the latest core / concrete interaction data.

2) . Release' of fission products from the RPV to containment building.

In the base case technology, no attenuation of the fission products is allowed in the - primary system. Thus, all the fission products released during the Gap and Melt release phases enter the containnent building. In the advanced ' technology. case, cn attempt will be made to detemine the fraction of the fission products.which after release from the fuel, either plate out or chemically afix themselves to structures in the primary system. This detemination will al so include - that fraction' of the ' retained fission products which are re-emitted,-and the timing of the re-emission.

3) Fission product attenuation in the suppression pool. In the base  !

case technology, the suppression pool attenuation will be detemired by RSS suggested methods, .i.e., a decontamination factor (DF) of 100

. is uced for the subcooled pools and a DF of 1 is used for the sa.tura-ted pools. Noble gases and organic iodine' are not subj ect to pool

^

scrubbing. In the. advanced technology case, the DF will be deter-mined by a model which will account for parameters such as aerosol particle diameter and density, bubble size and velocity, pool temper-i ature and carrier gas, 4-1

o t

4) Fission product transport and- atmospheric release. In the base case technology, the fission product transport within th containment building volumes is predicted using the CORRAL-IIE4,e J code. This codes'crubbing

-pool is used inmodel conjunction -with t e]5 fission and the MARCH code product release as described model, in Section

~3.- In _ the. advanced technology case, an upgraded code for fission product transport within the containment will be used. This code will interface with fission product sources from the in-vessel nelt release phase l and the ex-vessel core / concrete interactions. In ad-dition, mechanistically determined pool DF's will be used. 'All these quantities - will be consistent with the latest methods con-cerning _ these phenomena.

In the analysis to be presented in this report, only the base case technology will be used. Thus, the CORRAL-II code with its four distinct core release

' mechani sms (Gap, Melt, 0xidation, and Vaporization) together with the RSS source model and pool DF model, will be used to_ determine the fission product transport within the containment. The four release mechanisms are shown schematically in Figure 4.1 (note that the oxidation release was assumed to result from -a steam explosion in the RSS.) The gap release is modeled as a single' event and is assumed to occur at accident initiation. The melt release is divided into 10 equally sized releases evenly spaced between the time of

-core ' melt to the time of core slump. The timing of core melt and slumping were taken directly from the . MARCH analysis. The _ oxidation release is modeled as a single event and chosen to occur at RPV head failure to model the.oxida-tion of that fraction of the core debris assumed to interact with water on the diaphragm floor or to fall into the suppression pool. The vaporization re-lease is divided into 20 parts,10 releases of exponentially decreasing magni-tude Lin the first.1/2 hour, followed by 10 more releases during the next 1-1/2 hours, al_so of. exponentially decreasing magnitude. The vaporization release is assumed to start after vessel failure when core / concrete interactions be-gin. The -core release fractions for input to CORRAL were obtained from the ~

RSS. Table 4.1 is reproduced from the RSS and indicates the fraction of fi s-sion products released corresponding to the release mechanisms noted above.

The fractional release of fission products indicated in Table 4.1 would be input to CORRAL using the schematic indicated in Figure 4.1.

. For our purposes we use the oxidation release to model the oxidation release when a fraction of the core is assumed to drop into the suppression pool, and the containment building is assumed to be failed at the time of RPV failure.

The failed containment building ensures the presence of oxygen which is neces-sary for the oxidation release to occur'. The oxidation release affects only the Kr, Xe,1, Te,2 and Ru releases as assumed in WASH-1400.

In 'the LGS-PRA,[6] an oxidation release was allowed for at the time of RPV

- failure for all sequences, and additionally 15". of the suppression pool water was assumed to flash at the time of containment failure. The flash release af-fects all-the isotopes in the suppression pool equally for Classes I and III.

Since the fission products,in.the suppression pool will primarily be from the melt and gap release, the flash release will affect elemental iodine, cesium

'and barium, more than the renaining fission product groups. Finally, in the

. LGS-PRA, the RPV was modeled as a separate volume in CORRAL, and thus 4-2

7 ._

.w a . .

temporary holdup of the fission products released during the melt phase was calculated. These fission products were released to the drywell after vessel failure. This was not the case in the current calculation which assumed no prida'ry system hoidup.

Another important aspect of the model relates to pool decontamination factors.

In CORRAL, if flow between compartments goes via the suppression pool, the ef-fect of pool scrubbing can be calculated directly by subjecting the' flowing fission products -to an appropriate pool decontamination factor. However, as the primary system is not modeled as a volume in the CORRAL model, fission products released during the Gap and Melt stages have to be input directly

into appropriate containment volumes. For LOCAs, the release is directly to

.the drywell airspace so that -the core release fraction in Table 4.1 can be used'directly. However, for transients, the release is via the SRVs through the suppression pool and into the wetwell airspace. Thus, the Gap and Melt releases may be subject to pool scrubbing. This pool scrubbing is modeled in ,

. the CORRAL model by simply dividing the core release fractions in Table 4.1 by the appropriate pool 0F.

Each of _the ten damage states identified in Section 2 have been analyzed. For two damage states (I-T and IV-T) three potential containment building failure locations - (DW,. WW . and 15) were _ considered while in the remaining sequences,

'only'one failure mode was treated. A discussion for each CORRAL-II calcula- i tion follows.

4.1 Class I (Damage States I-T an'd I-S)

• qSince this accident class has a relatively- high frequency of occurrence, a complete series of calculations was carried out for all three failure modes.

. The LGS-PRA and -the 1 previous BNL analysis only considered the C 1 Y failure node (equivalent to I-T/DW in the present analysis) because this failure node results in-the: largest release of fission products. Since the' RPV fails prior to the containment failure, and since the containment is inerted, it is as-sumed that no oxidation release occurs in this class. This assumption is con-sistent with the base case technology.- The calculated release fractions for each failure mode will now be discussed separately assuming that the accident is initiated by a transient event (damage state I-T). - Release fractions based on LOCA initiated sequences are discussed in Section 4.1.4 4.1.1 _ Failure in Drywell (DW)

In this sequence, the containment failure is assumed to occur in the drywell wall. This implies that any activity airborne in the drywell atmosphere at the time of failure can enter the environment without first passing through '

,the suporession pool.

~ '

Thus, nuclides which are emitted during the vaporization release phase, and which. are not carried down into the suppression pool, or agglomerate and set-tie on the drywell floor can be. expected to be released. An inspection of Table 4.2 indicates that the release fraction of Te, Ru, and La are approxi-mately _8.4%, 6.2%, and 7.6% of the possible release, respectively. For I, Cs, P

4-3

and Ba. .these values are .18%,1.8%, and .9%, respectively. This is because the forner three groups are released.primarily during tne vaporization release (particularly Te) and the latter three are released primarily during the gap ar.d melt release phases. and thus are subject to a pool DF, wnicn is 100 in this sequence.

A comparison between the current analyses and the LGS-PRA and NUREG/CR-3028 analysis is also shown in Table 4.2. In this case. .tne suppression pool DF for all cases is 100. The two BNL calculations predict approximately the same

' fission product. release- fractions for those nuclides released during the gap and melt release phase. The nuclides ' emitted during the vaporization release phase are' predicted to be higher in the current calculation. The difference

'in. the 'latter case can be attributed, partially, to an improved t he nnal -

hydraulic representation in the current calculation. The inclusion of a _ flash

. release at the time of containment failure-is the primary reason why there is

'in increased release for .I 2

, Cs, and Ba for the LGS-PRA. Other modeling l differences such as temporary (until vessel failure) primary system holdup and di f ferent thermal hydraulic representation account for the remaining di f ferences.

4.1.2 Failure in the Wetwell (WW)

.In this case the containment failure occurs in the wetwell airspace. Thus any fission products which enter the outside environment have- to pass through the suppression pool. No distinction .is made in the DF to whicn the aerosols are subjected, whether they are released through the SRVs at the base of the sup-pression pool or through the downcomers at a shallower level From Table 4.2, it can be seen that in.this case the release fractions are substantially lower than in the I-T/DW release path. This is due to the above-mentioned reasons, and in addition, that the wetwell airspace is also available for particle ag-glomeration and settling.

4.1.3. Failure in the Wetwell with Loss of Suppression Pool (UU)

In this case the suppression pool is assumed to drain away at the time of con-tainment ~ fail ure.- Thus, although the melt and gap release fractions are fully scrubbed, that portion of the vaporization release, which does not flow down into the wetwell via the suppression pool before containment failure will not

~be scrubbed. This portion of the vaporization release is assumed to be air-borne in the drywell at the time of failure and will thus flow down into the wetwell, where it will be subject to settling. Thus agglomeration and . set-tling is the only attenuation mechanism acting on this portion of the vapori-

, .zation release. Table 4.2 shows that the release fractions for the I-T/VR failure mode are approximately twice as large as these for the I-T/WW failure mode but still substantially below those of the I-T/0W release patn.

4.1.4 Class 1 Sequences Initiated by LOCAs (I-S Damage State) ,

Riis sequence has been described above. However, two differences have to be accounted for to model a LOCA initiating sequence. First, a small break LOCA is assumed to initiate the accident sequence, ratner than a transient event.

4-4

F' Second, since the primary system is open from the start of the transient, the

~

gap and melt. release is not through the SRV's, into the suppression pool and fran there into the wetwell airspace, but rather directly into the drywell airspace. Thus, in this sequence, only a portion of the gap and melt release is subject .to suppression pool scrubbing. Only those fission products that are swept from the drywell into the suppression pool via the downcomers are subjected to pool scrubbing. The remainder of the gap and melt release will

.be subject to attenuation by the process of agglomeration and settling. Thi s removal mechanism has a substantial affect on the aerosol fission product

~ groups. Elemental iodine will be-subject to plate out on surfaces.

Table 4.10 shows a comparison of a treasient initiated I-T/DW failure mode and a LOCA initiated I-S/0W failure mode. It is seen that the noble gases and or-ganic iodine release fraction are essentially identical. The elemental iodine release fraction is higher in the case of the LOCA scenario. This indicates that the pool scrubbing, assumed in the transient event, is more efficient than the plate out removal mechanisms which daninates the LOCA event. A com-parison of the aerosol fission product groups shows that the fractions re-leased are lower for the LOCA event. In this case the agglomeration and set-tling of the-gap and melt release outweighs the suppression pool scrubbing of the transient event. The time available for agglomeration and settling in this sequence is at least 3.5 hrs.

4.2 Class II (Damage State II-T)-

Class II sequences are characterized by long-term overpressurization of the containment building due to steam generation. The RPV failure occurs after the containment has failed and (since the pool 'is saturated) the DF is 1 in the current. " base case" calculation. . In NUREG/CR-3028, a DF of 1 was assumed for the aerosol relea'se fractions (Cs-La) and a DF of 10 was assumed for ele-

.menta1>1odine. In the LGS-PRA, a DF- of 10 was assumed for all releases, ex-

- cept for the noble gases and organic iodine.

This release is characterized by a rather small containment failure area.

Thus, the' blowdown to ambient . pressure is slow. In view of the slow depres-surization in this sequence, it was not clear whether sufficient oxygen would enter into the containment building atmosphere to ensure an oxidation release at the time of RPV failure. This failure occurs approximately 7 hours8.101852e-5 days <br />0.00194 hours <br />1.157407e-5 weeks <br />2.6635e-6 months <br /> after the containment has failed. Thus, the fission product transport calculations were carried out with and without the oxidation release.

A direct comparison between the current BNL calculation and the other calcula-

- tions (NUREG/CR-3028 -and LGS-PRA) is difficult since in the current calcula-tion, a failure in the wetwell airspace is assumed, while in the other two

-calculations, a failure in the drywell was assumed. The largest difference between these two sets of calculations occurs for elemental iodine. The bulk of this difference is directly. attributible .to the . change in pool DF from 10 to 1. Furthermore, it will be noted from Table 4.3 that the species, emitted primarily during the melt release phase (Cs-Ba), are higher in the current calculations, whereas the species emitted during the vaporization phase (Te,

' Ru, and La) are lower in the current calculation.

4-5

a- .

s

[Thisi difference is; partially due to the location of the failure.

~

In the cur-  :

rent calculation, the failure is in the wetwell .above the suppression pool and I m

~

LthusJnuclidestrelease to the1wetwell airspace- (melt and gap release) escape ldirectlyt to the- environment at the time. of: containment fail ure. - In the case lof_ the vaporization release,ithe release path is not as direct, since the va-porization- release will :be -airborne in the drywell. The drywell atmosphere n ;L has'to-' pass through the~downcomersuto the'wetwell airspace and then escapes to

. the ; envi ronment. It .is'.then subject to attenuation due to settling in both

' volumes ~,;which tends to reduce the vaporization release for the current calcu- }

• lation.? .Onlygone failure location was considered in the current calculation

becauseTattenuation due to thei suppression

! pool has .been 1 eliminated (DF=1),  ;

cand thus it wasifelt that the release fraction would not be a strong function i of the failure = location. This assumption'will be discussed further in Section u

~

- [

4.4  !

G+ J AL comparison. between the release fractions with and without oxidation re-leass inDTable 4.3 shows _ very little change, except for the Ru release, which  !

is almost doubled -if an. oxidation release is assuned. This release is a di-

~ '

rect--reflection of the " base case technology" assumptions.

14.31 Class III (Damage State-III-T)  ;

.This is' an ATWS - accident in which the containment fails a little after the

time of RPV failure, due -to" steam pressurization. Thus the suppression pool i isjsaturated throughout- all:of ~ the fission product release periods and conse-f quently the'. suppression pool DF .is lE for all fission products, except noble

gases : and organic iodine. . Since the containment _is still intact at the time .;

- of RPV failure,ino " oxidation' release -was assumed. . . In the . previous BNL cal-

~

- ' culation, it was assumed that .the DF was:10 for. elemental iodine' and unity for iallE aerosol groups. The LGS-PRA assumed :a .0F of 10 for allL _ fission product groups. except1the: noble gas and organic -iodine.- Since in the current BNL ,

calculation theivaporization release phase starts approximately. 36 minutes l

- before containment failure.and then proceeds for 1.4 hrs with a failed contain-ment. The nuclides released during this phase.-are again prominant contribu-

• 1 tors _ to the fission: product - release (Te, Ru, and La). Results for these cal-culations are shown in' Table 4.4. Only one failure: location was considered-in the current calculationLbecaus; attenuation due to the suppression pool does not. exist '(DFel), and thus it was felt that the' release fractions would not be a strong function of the -failure location. .This dependency will be discussed

'in-Section 4.4..

JComparison between the current calculation and NUREG/CR-3028 and the LGS-PRA

are difficult since in this-case, a failure in the wetwell above the suppres-

'sion pool was assumed. The-~ previous calculations (LGS-PRA and NUREG/CR-3028) assumed -a failure in the drywell.- An inspection of Table 4.4 indicates that

' 'in.all cases-the current BNL predictions are substantially higher than the two previous calculations. This is partially due to .the different 0F's used and partially due to the different failure modes assumed.- This latter difference allows for more :settli'ng and plate' out in the current calculation for those

-nuclides released-during the vaporization release phase (Te, Ru, and La) but less for the melt and gap release. However, the greatest difference between

. these. calculations, other than the pool DF's for the LGS-PRA, is the timing.

4-6

e s .

From Table 4.4 it is .seen that in the current calculation, the vaporization release phase starts at 2 hrs and ends at 4 hrs. The containment building fails at 2.67 hrs. this allows for approximately 1.4 hrs (or 70%) of the va-porization release to be emitted .into an open containment. An inspection of the tining for the previous BNL calculation shows that approximately 36 nin-utes. (or only 30%) of the vaporization release is emitted into an open con-tainment, and in the LGS-PRA all the release is enitted into a closed contain-ment. .Therefore, one would expect that the opportunity for fission products to leak into the envirqnment would be largest for the current BNL calculation.

This. is borne out by the results, and this effect is greater than the in-fluence of the failure location in this sequence.

4.4 Class IV Sequences For this accident class, sequences initiated by transients and LOCAs were con-sidered and are discussed separately in the following sections.

4.4.1 -Class IV' Transients (Danage State IV-T)

Since this accident sequence is a major contributor to risk at the LGS, all three failure modes were analyzed separately. Furthernore, this sequence rep-resents an' ATWS sequence in which the power- is maintained at 30% of rated power by coolant injection. This results in rapid pressurization and leads to containment. failure in approximately 40 ninutes. Since the suppression pool

~is saturated, the OF is assumed to be 1 for all fission product groups except for the noble gases and- organic iodine. The LGS-PRA used a value fo 10 for the .DF, except when the containment failure occurred in the wetwell below the suppression pool where a DF of 1 was used, since the pool was assumed to have drained away-before fission product release. In NUREG/CR-3028, a DF of 10 was used for elemental. iodine and 1 for all the aerosol species. The exception

' again being the wetwell failure location when a DF of 1 was used because of loss 'of the suppression pool.

The containment failure time in the sequence is early, and the blowdown es-sentially complete by the time the core-starts to melt and the release of fis-

'sion products commences. - Thus, the release fractions for all three failure modes in the current BNL calculations are expected to be of similar magnitude.

The presence or' absence of the suppression pool plays no role (because the pool DF = 1) except to change the airspace in the wetwell. By comparing the three current BNL calculations shown on Tables 4.5, 4.6, and 4.7, it is seen

.that the release fractions are indeed quite close.

For the C4y (equivalent to .the current IV-T/DW failure mode) vdtich was the only failure location rigorously analyzed in NUREG/CR-3028, the release frac-tions are slighly higher for the aerosols. This is prinarily due to the dif-ferent thermohydraulic - representation used. The large difference for elenen-tal iodine 115 due to the different OF used. A comparison of the C4 Y" re-lease, shown in- Ta.ble 4.7, in which a DF of 1 was used for all cases; shows similar results for all cases. There is particularly good agreement between

~the LGS-PRA release fraction and those predicted by the current BNL calcula-tion (IV-T/WR failure mode).

4-7

5 4.4.2 Class IV LOCAs (Damage State IV-A)

~

-This sequence is identical to the -Class IV sequences outlined in Section 4.4 i:

'with the exception ' that a large LOCA event is additionally imposed at the l start of the accident. The different location of the melt and gap release (drywell, rather than the wetwell) has a small effect in this case since the pool DF is 1 (saturated pool). The difference between these two calculations is.that in the transient, the melt and gap release to the environment is con-trolled by flow from the wetwell back into the drywell and then out, while in the' LOCA case, the release is direct, and only controlled by the flow out of the rupture. .However, since the suppression pool is saturated in this case, it has no effect on the fission product release fractions.

L Table 4.11 shows a comparison between the transient and the LOCA release frac-tions. It is 'seen that they are very similar, with the LOCA only slightly higher. This difference can be ascribed to the longer path reouired by the melt and gap release in the transient case.

4.4.3 Class IV Reanalysis The calculational procedure described in Sections 4.4.1 and 4.4.2 results in very little fission product retention and extremely high release to the envi-ronment. - After review of a draft version of this report we were requested by NRC staff to revise the Class IV calculations. These revised calculations are described in this section. The containment response calculations described in Section 3.5 were used as input to the revised fission product transport calcu-lations. The reanalyzed fission product releases are shown on Tables 4.5, 4.6, 4.7, and 4.11.

In the revised calculations the in-vessel release was divided into two phases.

The first- phase involves 70% of the release and was released via the SRVs to t the wetwell volume. .The second phase. (involving 30%) is added to the oxida- ,

tion release and is thus released as a puff at the time of primary system "

failure. This splitting of the in-vessel release is consistent with the anal-ysis ' carried out in WASH-1400. However, this division of the in-vessel re-lease only applies to the transient sequences. For the sequence initiated by a large break LOCA, all of the in-vessel release was assumed to be released directly to the - drywell . In addition, we also added a third volume to the analysis of fission product transport. This volume represents the reactor building, which was neglected in the analyses described in Sections 4.4.1 and '

4.4.2. The inclusion of a reactor building volume in this analysis was also made consistent with the approach taken in WASH-1400.

Both of the above mentioned revisions increase the retention of fission pro-ducts. In the first assumption, part of the enhanced oxidation release is  !

passed into the wetwell at the time of vessel failure. It is thus subject to dgglomeration and settling in both the wetwell and drywell'. The flow into the '

reactor building during this pha.se of the accident is choked and thus pressure, changes'in the drywell due to vessel failure have only a slight effect on the-fl ow . rate. The addition of a third volume enhances the fission product reten-tion by increasing the volume available for agglomeration and settling before 4-8

1_  %.

l- .

i the fission-products leik into the environment. . From the above discussion it U

can' be concluded that if .these methods were also applied to any of the other

. sequences, the Srelease fractions would be reduced. However, for the other  :

. accident sequences,: the suppression pool is . subcooled and the containment  !

fails ' late so -that fission product attenuation is dominated by these mecha- i

' nisms. : . Consequently, it .was.not considered necessary to also revise the fis- i sion product release calculations for the:other sequences.  ;

i The results of this re-evaluation-are shown on the last: column of Tables 4.5, 4.6, 4.7,1 and 4.11. It is seen that for noble gases and organic iodine,- there  ;

is' essentially' no change in the fraction released.- However, for elemental i

' iodine'and the. particulate ' species, the release fractions have been reduced by l a factor of approximately 1.5 to 2. The timing of the release is not appre-ciably affected by these changes.-

l

. 4.5 Cjass-15 Sequences  !

These sequences are based on 'the LGS-SARA [73 and on the descriptions of -the h accidents in- Section 2 and ' the MARCH analyses in Section 3.5. Two sequences j were :modeled, namely .TSRB and TSRBCM. The major difference relates to the

+ failure to scram for the TSRBCM sequence. In these cases the suppression pool

is subcooled, thus the DF is 100, except for noble -gases and organic iodine.

For these sequences,j,he RHR suction lines are assumed to fail at the start of -;

'the accident. Failure of the RHR suction lines results in partial. draining of- l

' the suppression - pool, which leaves the SRV submerged out exposes the down-  !

comers. ' Thus, for transients, the gap 'and . melt releases are scrubbed by the pool but the oxidation-and vaporization releases do not pass through the pool.  !

r Byl inspection of the release fractions in Table 4.8 ! tris seen that those fis-  !

sion product groups with a high ~ release during the ' vaporization;or oxidation i

~

. release.-phase (Te,- Ru, and La) are major contributors to the release frac-

~

tions.- The addition -of an oxidation release at the time of RPV failure, into .  :

anlopen containment enhances the Ru' release even more. Those fission product i

. groups. which are : released primarily during the melt release ' (I, Cs, and Ba)

-are'quite ' low,'especially. Ba, which is essentially only released during the

, melt 1 release phase. - Thus,' the release fractions for these sequences are es-sentially proportional to their vaporization release fraction, except Ru,

- which Jis_ enhanced = by- an oxidation release,- and ' the noble gases ' and organic - ,

t iodine both of which are entirely. released. ,

4.6 . Class S Sequences q~ .

' For'this class -the containment and the vessel fail at the start of the acci-dent.. Thus, all fission product: release bypasses the suppression pool. How-ever, -Athough the . Various releases take place. into a failed containment '

building with 'no suppression pool DF, it 'is also evident that. the flow rate  :

tout of the . building at the time of release will be comparatively low. Thus,  !

. the 1 fission' products 1.n the aerosol group are subject to attenuation .by ag- I glomeration .and . settling. - An- inspection of Table 4.9 indicates that for the S-FM sequences for those species dominated by melt release (I, Cs, and Ba) approximately 251, of the total - release fraction escapes to the environment, while ..for those. nuclides released primarily during the vaporization release j k

i

4-9 7 y P W u --,.~,,-,vw .

w- w e -wp, , . - - *--e, -r-.,--eseo --..---m--wy--$------,r---gro-- -y.--=----- we r .s-,-.-en.-,=--,3 m..-~,,%,--y-e.

. phase (Te and ' La)' approximately 38% are released to the environment. The Ru release fraction is enhanced -by the oxidation release at the point of RPV

~ failure. It is seen that the release fractions for the S-H2O sequences are slightly higher than for. the S-H2O sequences. Th.is is particularly due to the delayed ' start-of the vaporization release phase in the S-H20 case.

In the latter sequence the vaporization release starts approximately one hour

,1 ;later. Since the flow out of the containment building is dropping off with ff -time, the leakage - to the environment from- the vaporization release becomes a smaller contributor to the overall release.

A comparison between the BNL release fractions and the LGS-SARA rel ease

' fractions for the VR sequence (equivalent to S-H20/WW) shows good agreement.

The.only difference being due to the use i the LGS-SARA of fission product release coefficients based on NUREG-0772. 3 Thus, the organic iodine is lower (approximately a factor 22) and the barium release is higher. A sim-

~ilar comparison for the VRH2O sequence (equivalent to S-H20/WW) does not show i the same level of agreement. This difference can only be attributed to the large release of fission products during the melt release in the LGS-SARA be-cause the NUREG-0772 fission product release coefficients are used. The large melt release is assumed to be airborne in the RPV and is expelled at the time of core slump. In the BNL approach, this release fraction is lower. Further-more, the release is deposited in the containment building and not held up in  ;

the RPV since the latter is not explicitly modeled. These differences in fis-  ;

sion product transport and thermal-hydraulic modeling account for the lower release fraction in the BNL case.

4.7 Summary In this section the fission product release fractions and the associated tim-ing is presented. These determinations are based on the base case technology as outlined above, and the release fractions are summarized in Tables 4.2-4.9.

The time of release is defined as the time of containment failure for those cases in which the meltdown takes place in an intact containment building.

For those cases, when the containment building fails prior to core damage, the time of releases is defined as the start of core melting. The duration of re-lease will be defined as the time for the containment building to blow down to atmospheric pressure.- However, if the building fails first (meltdown into a failed containment building) the duration of release will be from the start of core melting to the completion of the vaporization release. The warning time is defined as the time period .between the start of core melt and the time of containment failure. If the containment building fail s . fi rst , the warning time is defined as the difference between the start of core melt and the time of containment failure.

The e.nergy of release is the en,ergy release rate associated with the plume at the time of failure. This Value i's extracted ' fron the MARCH cal.culation (refer to Section 3). In those cases where the release is spread out over many hours, the energy of release is very low. The height of release is chosen to be 25 m (82 feet) in all cases. The information in Tables 4.2 4-10

a  ;. . . .-

c through.4.9 is used in the following section lo generate tne source terms for  ;

used-in the DES'for the LGS.

4;8 References to Section 4

1) NRC Memorandum - from B. Sheron, Branch Chief /RSB to R. Mattson, Director, OSI," Proposed PRA Methodology for Limerick and GESSAR," dated July 22, 1983.  !

~ 2)~ Reactor Safety Study, "An Assessment of Accident Risks in V. S. Commercial Nuclear Power Plants," WASH-1400, NUREG/75-014,1975.  ;

i-

' 3) ."Radionuclide Release Under Specific LWR Accident Conditions, Volume 2:

BWR, Mark - I Design and Volume 3: BWR, Mark III Design," BMI-2104 draft  ;

report.

4) R. J. Burian and P. Cybulskis, " CORRAL 2 User's Manual ," BCL report dated January 1977.

i

5) LR. O. Wooton and H. I. Avci, " MARCH Code Description and User's Manual,"

.NUREG/CR-1711, October 1980. -

l

'6) Philadelphia Electric Company, " Limerick Generating Station, Probabilistic i

Risk Assessment," March 1981.

• I

7) Philadelphia' Electric Company, " Limerick Generating Station, Severe Acci- i dent Risk Assessment," April 1983. '

8)_ " Technical Bases for Estimating Fission Product Behavior During LWR Acci-dents," USNRC Report NUREG-0772, June-1981.

i a

e b

i i

g e e

4-11

. -- _ -_._ _ _ -. ._- .-_._ . - - . _ _ _._ _ _ _ ~c

Ta bl e 4.1 Fission Product Release Source Summary -

Best Estimate Total Core Release Fractions.

Fission Gap Relea.se .Meltdova Release vaporization Release Pr:due: Fraction Steam Exolosion Fraction Fraction (d) Fractiente)

Xe, Kr 0.030 0.870 0.100 (X) (Y) 0.30

• I, 3r 0.017 0.883 ,

0.100 (x) (Y) . 0.30 C3, Ra 0.050 0.760 0.190 --

YeI *I 0.0001 0.1s0 0.850 (X) (Y) (0.60)

Se, sa 0.000001 0.100

• 0.010 --

Ru # -

0.030 0.050 (X) (Y) (0.90)

LaI *I --

0.003 3.010 --

(a) Includes Se, Sb (b) Includes Mo, Pd, Rh, Yc (c)  :..cicdes Nd, Eu, Y, Ce, PT, Pm, Sm, Np, Pu, Ir, Nb '

(d) Exponential loss over 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> with half time of 30 minutes. :" a steam expleston ec urs prior to tais, caly cae core fra,etica not involved : n :te steam explesson esa excerience vapori:stien.

(el X = Traction of core involved La :he steam explosica. Y = react:.:n o' -.ven-tory remainin:; for release by exadatien.

..~_

Steam

1. -

i 7

e i

$cJo -

i 1

~

1

• C.5 3 03 lillllillll Irini,,,,i r i i i , , , ,

t '

t e O  !

1 2 '

3 rm. ~ '

Figure 4.1 Typical Sequence of Spike Fission Product '

Releases for Postulated Accidents. '

4-12

mga

~-

3-  :. . .

Table 4.2 . Fission product release fractions for Class 1 I

LGS - PRA ASSESSMENT NUREG-3028 DES CALCULATION l

FAILURE MODE CY Cyy I-T/0W I-T/WW l I-T/W 1

l l l l OXIDATION RELEASE Yes Yes No No No l l

I- l l Xe - Kr 1.0 '

.939(-1) l 1.0 l 1.0 l 1.0 I .

l l

. Organic: Iodine l --- ---

6.99(-3) 6.99(-3) 6.99(-3) l l l .

l

.I2' 1.1(-1) 9.3(-3) 1.78(-3) 1.48(-4) l 2.09(-4)

I Cs 9(-2) 1 2.0(-2) ,

1.88(-2) l 3.11(-4) l 9.19(-4) l

'Te l 1.6(-2)- 4.6(-2) 8.41(-2) 1.23(-3) l 2.16(-3) l l l Ba 1.0(-2) 1.7(-3) 9.94(-4) l 1.91(-5) 8.22(-5)

Ru 3.0(-3) -3.0(-3). l 4.95(-3) 7.39(-5) 1.39(-4)l l l 1 La 3.0(-4) 6.1(-4) , 9.89(-4) 1.46(-5) 2.61(-5) l l

DF for 12 100 '100. 100 l 100 l 100 I

DF for-Aerosols 100. 100 l 100 100 100 l l I Core Melt Start 1.3 1.75 1.5 1.5 l 1.5 I l Core Melt End 2.5 2.43 2.42 l 2.42 l 2.42 l l l l 1 i

'1st Vap. Release.

2.90 2.90 l 2.90 l l I 2nd :Vap. Release 3.40 .

3.40 I 3.40 I I I Vap. Release End . 4.90 4.90 4.90 l L ,

l I

' Containment Fail 6.5 5.23 5.17 5.17 . 5.17 4-13

-Table 4.3 ~ Fission product release fractions for Class II l- 1 l

-ASSESSMENT- LGS - PRA NUREG-3028 l DES CALCULATION l FAILURE MODE. CY-2 CY 2 II-T/WW II-T/WW l

0XIDATION RELEASE: -Yes Yes Yes l No l l Xe - Kr 1.0 1.0 9.8(-1) 9.8(-1)

! l

. Organic Iodine --- ---

'6.86(-3) l 6.86(-3) {

l l I2 6(-2) i 1.56(-1) .1 6.73(-1) j 6.16(-1) l l l 1 Cs .2.3(-2) , 2.58(-1) 3.36(-1) 3.36(-1)

! Te 4.0(-1) -4.21(-1) 2.31(-1) l 2.38(-1) i l Ba 6.3(-3) 2.7(-2) 4.1(-2) l 4.1(-2) i Ru 6.9(-2) 7.0(-2) i 4.0(-2) j. 2.2(-2) l 1

La 4.7(-3) 5.4(-3) 3.3(-3) 3.3(-3) .l I l

l l

1 l I i 0F for-12 10 10 1 l 1

. 0F for Aerosols 10 1 1 1 1 l l l l

' Core Melt Start 36.6-  : 35.5 24.92 24.92 .

l l .

t

' Core Melt End 39.0 38.3 26.83 l 26.83 l l I I I ist Vap. Release' l 26.83 l 26.83 l  !

.I l 2nd Vap.-Release I i 27.33 1 27.33 l

Vap. Release End 28.83 l 28.83 I -

4

' Containment Fail 30.0 29.2 19.6 l 19.6 l l 1 i l l l l 4-14

f Table 4.4 Fission product release fractions for Class III i i ASSESSMENT LGS .;A NUREG-3028 i DES CALCULATION -

\

l FAILURE MODE CY CY3 3 III-T/WW l  ;

0XIDATION RELEASE Yes Yes i No l Xe - Kr 1.0 1.0 9.99(-1)  ;

l [ .t Organic Iodine --- ---

6.99(-1) {

l 12 4.02(-2)I 1.22(-1) l 7.8(-2) l l

Cs 2.4(-2) 5.42(-2) 2.24(-1) l l

Te 7.3(-2) 1.85(-1) 5.74(-1) l Ba 2.7(-3) 3.61(-3) 1.95(-2)

Ru 8.6(-3) 1.7(-2) 3.65(-2) ( ,

I u

La 9.1(-4) 2.4(-3) 6.92(-3) l j

. I I

I DF for 12 10 10 1 l l

DF for Aerosols 10 1 1 1 I ,

Core-Melt Start .85 .76 .5 l l l

Core Melt End 2.5 2.22 l 1.8 l l l l

.1st Vap. Release

, 2.05 l l

C 2nd Vap. Release 2.55 Vap. Release End 4.05 l

Containment Fail 6.5 - 3.83 2.67 l 1  !

l I i

h 4-15

g

^ . .

Table 4.5 ' Fission product' release fractions for Class IV

-(failure location DW) l 1

. DES FES l ASSESSMENT LGS DRA NUREG-3028 CALCULATION CALCULATION

m. FAILURE MODE CY

-4 CY-4 IV-T/DW IV-T/DW l 1

0XIDATION RELEASE Yes Yes Yes Yes 1 i i Xe - Kr.. 1.0 l 1.0 9.99(-1) 9.99(-1) l

-Organic' Iodine ---

a 6.99(-3) 6.95(-3)

' l l l 12 2.61(-1) 1.54(-1) 9.39(-1) [ 4.74(-1) [

l Cs 2.02(-1) ,

7.49(-1) 8.61(-1) [ 4.86(-1) l Te 4.34 (-1) 7.47(-1) 8.62(-1) 5.09(-1) l

, Ba' ' 2.90(-2) ,

8.60(-2) 9.40(-2) 5.54 (-2) .I

'~ Ru 9.50(-2) 1.10(-1) 1.4 9(-1) ) 8.55(-2) l l La 5.20(-3) 1.03(-2) l 1.15(-2) l 6.82(-3) l l 1 ,

1 l DF for 12 10 10 1 1 DF for Aerosols 10 1 1 1 i

Core Melt Start 1.2 1.25 1.13 l 1.13-l l l Core Melt- End 2.2 2.7 2.20 1 2.20 1st Vap. Release 2.47 2.47 2nd Vap. Release 2.77 l 2.77 1 l Vap. , Release End -

4.47 4,.4 7' Containment Fail .67 .67 .67 .67 1

4-16

g .. .

[

Table 4.6 Fission product release fractions for Class IV (failure location WW) l l DES . FES ASSESSMENT LGS - PRA NUREG-3028 CALCULATION CALCULATION l FAILURE MODE C4Y' C4 Y' IV-T/WW IV-T/WW 1

-OXIDATION RELEASE l Yes Yes Yes Yes i I Xe - Kr 1.0 1.0 l 1.0 9.99(-1) l l l 1 l 1

-Organic Iodine l 6.99(-3) l 6.95(-3) l l l 12 1 7.0(-2).. 9.80(-2) 9.39(-1) l 4.61(-1) l

' l 1 Cs 9.0(-2) 7.49(-1) 7.72(-1) l 4.81(-1) j I

Te- l 2.0(-1) 7.47(-1) 6.88(-1) 4.45(-1)

Ba 1.6(-2) 8.60(-2) 9.0(-2) 5.60(-2) l l

Ru, 8.8(-2) .I 1.10(-1) 1. 1.19(-1) 7.81(-2) l l

La 6.0(-3) 1.03(-2) 9.40(-3) l 6.03(-3) 1 I l l 1 1 I DF_ for 12 10 10 1 1 l l

DF for Aerosols 10  ! 1 1 , 1 l 1 I l l l Core Melt Start 1.2 1.25 1.13 1.13 l l l 1 I

" Core Melt End 2.2 2.7 2.2 2.2 l 1 l l

1st Vap. Release 2.47 ,

2.47 l

2nd Vap. Release 2.77 l 2.77

• 1 r Vap. Release End 4.47 4.47* ,l .

l l

Containment Fail .67 .67 l .67 .67 l I l l 4-17

b Table 4.7 Fission product release fractions for Class IV (failure location W below wetwell waterline) l DES FES ASSESSMENT. LGS - PRA- NUREG-3028 CALCULATION CALCULATION

. C y '.' C4 y" FAILURE MODE 4 IV-T/W l IV-T/W l 1

OXIDATION RELEASE Yes Yes l Yes Yes l l 1 -

i

.Xe - Kr 1.0 1.0 1.0 l 9.98(-1). l l l Organic lodine --- ---

I 6.99(-3) i 6.95(-3)~ l l

'12 7.30(-1) 7.08(-1) i 8.74(-1) 4.68(-1) l 1

Cs 7.0(-1) 7.49(-1) 1 8.04(-1) [ 5.18(-1) l

. I Te 5.50(-1) 7.47(-1) 5.82(-1) l 4.81(-1) l Ba. 9.0(-2) , 8.60(-2) 9.60(-2) [ 5.96(-2) l

-Ru 1.20(-1) 1.10(-1) 1.38(-1) l 8.31(-2) l La 7.0(-3) 1.03(-2) , 7.90(-3) [ 6.51(-3) l I

l DF for-12 10 10 1 l 1 l 1 l DF for Aerosols 10 1 1 I .1 l 1 l l

Core Melt Start 1.2 1.25 1.13 l 1.13 l .I Core Melt End 2.2 l 2.7 2.2 l 2.2 l I

\ , i l 1st Vap. Release 2.47 2.47 I .

2nd Vap. Release 2.77 2.77 l

-l Vap. Release End 4.47 4.47 l l

Containment Fail .67 .67 .67 .67 4-18

. Table 4.8 . Fission product release fractions for Class IS ASSESSMENT ' LGS-SARA DES CALCULATIONS l

FAILURE MODE- TSRB IS-f/DW IS-C/DW j i

-0XIDATION RELEASE - Yes Yes l I

-Xe-- Kr 1.0 9.99(-1) 9.99(-1) I l - I l

Organic Iodine 3.0(-4) l 6.99(-3) 6.99(-3) l

'1 .

t l l I2 -

l 5.0(-2) '

8.2(-2) 1 7.6(-2) {

l I i Cs 9.0(-2) 1.43(-1) 1.37(-1) l l

Te 9.0(-2) 6.06(-1) 5.68(-1) {

l Ba . 4.0(-3) 7.78(-3) 7.42(-3) l l

Ru 2.0(-2) 1.07(-1) 8.2(-2) i I

La 5.0(-3). 7.37(-3) l 7.05(-3) l 1 1 I I i

1

'0F for 12 l 100 100 l l

'0F for Aerosols 100 100 l 1

l

- Core Melt Start 1.47 .37 l.

I Core Melt End 2.32 l. 1.28 l 1

l 1st Vap.' Release 2.37 1.53 l 2nd Vap. Release 2.87 2.03

.Vap. Release End 4.37 3.53

. I -

Containment Fail 0.0 O.0 l 4-19

').

P

~

.5 Table 4.9 Fission product release fractions for Class S l

-ASSESSMENT LGS-SARA DES CALCULATIONS l- l i i S-H20/W l S-H20/W FAILORE MODE VRH2O VR I l

OXIDATION RELEASE I Yes l Yes t I

Xe - Kr 1.0 1.0 9.87(-1) 9.68(-1) l l l Organic Iodine 3.0(-4)~ 3.0(-4) l 6.99(-3), 6.98(-3) l l .

i 12 l 5.0(-1) i 1.0(-1) 1.09(-1) 2.55(-1) l l -

l -

l l Cs 7.3(-1) 3.3(-1) 1.62(-1) 2.74(-1) l

' l

_ Te 7.5(-1) 3.3(-1) '

2.90(-1) l 3.86(-1) l i l l Ba 3.5(-1) l 1.5(-1) l 1.20(-2) l 2.60(-2) l l Ru 7.0(-2) 4.0(-2) 4.90(-2) [ 6.20(-2) l l l La 5.0(-2) i 2.0(-2) 3.64(-3) 4.99(-3) l l 1

l i

DF for I2' i l 1 l I

l DF for Aerosols 1 'l 1 l l l I I Core Melt Start 2.67 l 2.83 l 1 l l Core Melt End 3.65 l 3.85 l l l l l l 1st Vap. Release 5.23 4.38 l l

2nd Vap. Release 5.73 1 4.88 l l

Vap. Release End 7.23 6.38 l l .

, Containment Fall 1 0.0 0'. 0 l' l

i 4-20

o . .

Table 4.10 A comparison of fission product release fractions for

. Class I sequences initiated by LOCAs and Transients l

ASSESSMENT DES CALCULATION l l

I 1

-FAILURE MODE I-S/DW l I-T/DW l l 1 0XIDATION RELEASE No l No (

l Xe- Kr 9.99(-1) 9.99(-1) [

l l Organic Iodine 6.99(-3)  !

6.99(-3) l I

12 3.31(-3) 1.78(-3) l l

Cs 4.89(-3) l 1.88(-2) {

l Te 2.80(-3) 8.41(-2) l I

Ba 6.01(-4) l 9. 94 (-4 ) l

'1 l Ru 2.87(-4) 4.95(-3) l l

La 4.01(-4) 9.89(-4 ) {

1

. _ l 1 1

, DF for 12 100 , 100 (

l i DF for Aerosols 100 l 100 l l l l l-Core Melt Start 1.35 l 1.5 i Core Melt End 2.44 2.42 i I, 1

l 1st Vap. Release , 2.83 i 2.90 1

2nd Vap. Release 3.33 3.40 l l I Vap. Release End l 4.83 l 4.90 l I

Containment Fail 5.11 , 5.17 I

I c .

4-21 L:

~

l  ;.w .y Y ,

4' ' g u  ; ,

' Table.4.11. A' comparison of fission product release fractions for (Class IV sequences initiated by LOCAs and Transients -

4 l

.l l

DES FES

' ASSESSMENT NUREG-3028' CALCULATIONS I CALCULATIONS- I.

FAILURE MODE CarLOCA IV-A/DW IV-T/0W IV A/DW Ne-Kr '1.0~ .9989 .999- ,

9.96(-1)

- Organic. Iodine _7,0(-3) l 6.99(-3) 6.99(-3) 6.94 (-3) l l 1

I2 8 23(-1). 9.68(-1) 9.39(-1) l 4.78(-1) l 7 1 Cs-- ,

.7.50(-1) 8.70(-1) 8.61(-1) l 5.06(-1) l 4

Te -7.5(-1) .8.74(-1) 8.62(-1) 5.18(-1)

Ba 8.6(-2)  ! 9.94(-2) 9.39(-2) 5.76(-2)

Ru 1.11(-1)" 1.5(-1) 1.4 9(-1) 8.86(-2) l

' I La. 11.0(-2) 1.17(-2) 1.15(-2) '6.95(-3) l l

-DF for 12 1 1 1

", DF;for. Aerosols -

1 1 1 b

Core Melt St' art . ,- 1.17 '

1.13' 1.13

~

'! eCore Melt End.' 1.58 2.20 2.20

t. l

, .lst. Vap. Release 2.20 - .

2.47 l 2.47

' q .

T2 nd Vap. Release -.f-2.70 1 2.97 i 2.97 l Vap. Release. End A

, 4.20 ,

4.47 1 4.47 l

. !Contatnment Fail' .

, f.67 l .67 *

.67

~

t l>.

- nr 4-22 -

-. s

.---,_% r..m- ,. h . ..e-,- - ,-., + . , - + - , , , - - - .

a s . .

5.0 SOURCE TERM CHARACTERISTICS In this.section we generate representative source terms for the various fail-ure modes and release paths. This section therefore assembles the informa-tion . contained in Sections 2 through 4 of tnis report. The probabilities of the failure modes were calqu(ated tained 'fran the LGS-pRA,LlJ the in LGS-SARA,L2 Section J]andthe relyBNL on reviewsL3,4j inform 4tioq ob-of these reports. The timing of fission product release, energy of release, du-ration of release and warning time for the various failure modes were based on the MARCH analysis in Section 3. The quantities of the fission products re-leased were calculated in .Section 4. Source terms for 27 failure modes and release paths have been determined. Fourteen of these source terms were cal-culated as part of the present study and are described in detail in the body of this report. The remaining thirteen source terms are bayeq on the infor-mation in References [1-3] and on the Reactor Safety StudyL5J (with modifi-cations to reflect th,e present assessment),

i The information contained in this section is the data needed to perform a site consequence analysis. - The Accident Evaluation Branch (AEB) at NRC nas the responsibility of performing the site consequence analysis for the Limerick site as part of the Draft Environmental Statement (DES) and Final Environmen-tal Statement (FES). The information in Tables 5.1 through 5.7 was generated specifically as input to the DES. The information in Tables 5.6 througn 5.12 was used as input to the FES. In the, following sections we will briefly sum-marize the source terms.

5.1 LGS-DES Source Terms 5.1.1 Source Terms for Damage State I-T (Table 5.1)

This damage state is defined in Section 2 and basically consists of transients with loss-of-inventory make-up. Core melt is relatively fast and occurs into an intact containment. After vessel failure the majority of the core mate-rials are retained on the diaphragm floor. Containment failure occurs via gradual overpressurizatioa (except for SE, HB, LGT, and EGT releases) several b

hours after vessel failure due to core / concrete interactions. Each of tne

y. source terms in Table 6 are discussed below.

I-T/DW This release path assumes a failure in the drywell wall. The gap and melt re-leases are directed to the suppression pool and subjected to a DF of 100 (be-cause the water is subcooled) before reaching the wetwell airspace. The va-porization release is directed to the drywell without any pool scrubbing. All fission products in the drywell a4d wetwell are subjected to agglomeration and settling as predicted by CORRALL6] prior to vessel. failure several hours after the pressure vessel failure. -

I-T/WW This release path assumes a failure in the wetwell above the suppression pool.

The gap, melt, and vaporization releases are released to tne drywell and 5-1

wetwell as described above. The only difference is that at containment fail-ure fission products in the drywell must pass through the downcomers and sup-pression pool prior to release to atmosphere.

I-T/WW This release path assumes a failure in the wetwell below4the suppression pool, which drains the water. The gap, mel t, and vaporization releases are again released to the containment as described above. The only difference is tnat at containment failure the suppression pool is drained so that fission pro-ducts in the drywell no longer have to pass througn the suppression pool (as in the I-T/WW release path) prior to release to atmosphere.

I-T/SE This release path results from an in-vessel steam explosion generated missile.

We assume this occurs at core slump and opens a direct path from the primary system to atmosphere. In the LGS-PRA, this failure mode was similar to the RSS release category BW1. This release corresponds to an ATWS sequence analy-zed in Appendix V of the RSS, in which the steam explosion was assumed to oc-cur after only 13% of the core had melted. Consequently, most of the melt re-lease was released to containment without pool scrubbing. This is not consis-tent with our analysis of this sequence as we would subject all of the melt

release to pool scrubbing. We have therefore used a steam explosion release from the RSS that more appropriately reflects our analysis of the sequence.

I-T/HB This release path results from H2 burn fail.ures during tne time when the containment atmosphere is deinerted. We used the same release category as in the LGS-PRA but reduced the oxidation components associated with the Te and Ru releases.- (Note in the LGS-PRA, this release category was representative of ex-vessel steam explosions). ,

I-T/LGT and IUT These release paths results from' containment lehkage and assume that the SGTS operates (LGT) or that it does .not operate (L57). We use the LGS-PRA re-

. leases but changed the timing to c6rrespond to our analysis.

5.1.2 Source Terms for Damage State II-T (. Table 5.2)

' This damage state is defined in Section 2 and basice' ., cssumes loss of con-tainment heat removal . Eventually, the containmaa# fa 's, anicn causes the loss-of-inventory make-up'. As the ' containment ie ai prior to core melt and the suppression pool is ' saturated (DF of 1; cw ation of containment failure (DW, WW or 15i) is o'f rather less importance than ~for the I-T damage states. Each of the source terms in Table 7 are discussed below.

5-2

7 y- : s. . . .

'n,.

a 'II-T/WW

" This release path assumes a failure in the wetwell above the suppression pool.

, The melt ~ release is directed to- the suppression pool but is not subjected to

~

pool decontamination because the water is saturated. The vaporization release is directed to the drywell, then through tne downtomers- to 'the wetwell air-space _ and finally to the atmosphere. This one failure location was also used '

to represent Lfailures in _the drywell (DW) and wetwell below the suppression pool' (UE) . _This assumption 'is reasonable because the pool is saturated and hence the different flow paths do not result in significant differences in l calculated release 1 fractions (refer to the discussion on the IV-T damage  :

state)2-L II-T/SE' LThis' release path results from an in-vessei steam explosion generated missile..

~

The release path used in the LGS-PRA, which was taken from Appendix .V of the '

RSS, was considered appropriate and is used in Table 7. Differences related -

only to the timing, which now corresponds to the present analysis of a II-T

' damage state.

~

?

(5.1.3 Source Terms 'for Damage State III-T (Table 5.3) '

~ This damage state corresponds to a ~ transient event coupled with loss of scram j

-function (refer to Section 2). -Core melt is rapid and cccurs into an intact 1 containment. ' Containment. failure is predicted to occur afterivessel failure  ;

due to overpressurization. = However, the suppression pool is saturated so tnat '

'the gap, melt, and vaporization releases are not subjected to decontamination

by 'the pool.- Consequently, we again (as -for the II-T damage state)' used one failure location to represent the three potential' locations.

III-T/WW ,

Thisirelease path is-similar to' the I-T/WW sequence, however (because the pool is~ saturated) the melt release isLnot subjected to pool scrubbing in this dam-age, state, s

III-T/SE LThe steam explosion ' release category used in the LGS-PRA was considered appro- t priate and is used in Table 5.3.-

Differences relate only to timing, which was made Consistent with our MARCH analysis. -

t 5J.1.4 Source-Terms for Damage State I/-TE(Table 5.4)

, This damage state is defined'in Section 2 and essentially consists of ATWS se-

.quenc's in which continued coolantimake'-up results in overpressurization fail-e

', 'ure of containment' prior to core melt. The suppression pool is saturated for i c these - sequences and -hence the DF is unity. We analyzed the impact of tne j

5-3 w,m-- ~ ne ~

three potential failure locations (DW, WW and EU) and because of the saturated pool found similar release fractions (refer to Table 5.4). These calculations support _ our use of only one failure location for the II-T and III-T damage states. The release paths for the three locations have been discussed in de-tail above and will not be repeated here.

IV-T/SE The steam explosion release category used in the LGS-PRA for Class III (damage state III-T) was considered appropriate to this damage state. Consequently, this release category is used with the timing changed to be consistent with our MARCH analysis.

5.lo5 Source Terms for Damage States I-S and IV-A (Table 5.5)

These damage states are defined in Section 2 and correspond to LOCA initiated sequences.. They were calculated to have a low frequency but (because of dif-ferences in flow paths relative to transients) were analyzed separately. The I-S/DW flow path results in the release of the melt and vaporization releases to the drywell, thus bypassing pool scrubbing. However, as containment fails several hours after vessel failure, the release fractions are not significant-ly different fran the I-T/DW flow path (in wnich the gap and melt releases were subjected to suppression pool scrubbing).

5.1.6 Source Terms for Damage States IS-C and IS-C (Table 5.6)

These damage states are defined in Section 2 and are seismically induced. The RHR suction lines are severed resulting in partial loss of the suppression pool. The gap and melt releases are directed to the suppression pool and are subjected to decontamination (the water is subcooled and the DF=100) before re-lease via tne severed RHR suction lines. The vaporization release is directed to the drywell and then flows through the downcomers into the wetwell . How-ever, as the suppression pool has drained below the downcomer outlet, the vaporization rel ease is not subjected to pool scrubbing. The difference between IS-C and IS-C relates to the scram function and does not influence the flow paths, only the timing of the sequence is affected.

The in-vessel steam explosion failures (IS-C/SE and IS-C/SE) were assumed to be similar to the I-T/SE release. Only the timing was altered to reflect the MARCH analysis.

5.1.7 Source Terms for Damage States S-H2O and S-H2O (Table 5.7)

These damage states are defined in Section 2 and are also seismically induced.

The RHR suction lines are severed but also the vessel fails at the start of

. the accident. Thus the core melts into a failed. containment and none.of tne

. releases are subjected to pool scrubbing. The only differences between tne S-H2O and S-H7D sequences relates to the location of the failure in the ves-sel . For the S-H2O sequence, water remains in the vessel and is available for 5-4

A *' . .

Einteracting with the core debris as it slumps. This will affect movement of the fission products and also allows the potential for an in-vessel steam ex-pl os ion . - A. s the melt release is not subject to pool scrubbing, the steam ex-plosion release was considered similar to the release used for release paths III-T/SE and IV T/SE.

The S-H2O damage state involves a failure of the vessel, such the water is completely drained at :the start of the accident. Thus, there is no in-vessel debris / water interaction and no potential for an in-vessel steam explosion.

5.2 LGS-FES Source Terms The source terms .for damage states IS-C, IS-T, S-H20, and S-H2O (in Tables 5.6 and 5.7) .were not changed for use in the LGS-FES relative to the LGS-DES. The

frequencies lof the source terms for damage states I-T, II-T, and III-T were changed to reflect the revised probabilities of sequences initiated by loss-

~

of-offsite-power and fire (refer to Section 2). The revised source term prob -

abilities are given in Tables 5.2, 5.9, and 5.10. In addition, we recalcula-ted the source terms for damage states IV-T and IV-A for input to the LGS-FES (refer to Sections 3.5 and 4.4.3). The revised source terms are given in Ta-bles-5.11 and 5.12.

5.3 References to Section 5

1) Philadelphia Electric _ Company, " Limerick Generating Station, Probabilistic

-Risk Assessment," March '1981.

2) Philadelphia . Electric Company, " Limerick Generating Station, Severe Acci-dent Risk Assessment," April 1983.

3) I. A. Papazoglou, et al., " Review of the Limerick Generating Station Prob-abilistic Risk Assessment," NUREG/CR-3028,' February 1983.

-4) M. A._ Azarm, et al ., "A Preliminary Review of the Limerick Generating Sta-tion Severe Accident Risk Assessment, Volume 1: Core Melt Frequency,"

Draft BNL report dated August 15, 1983. ,

5) " Reactor Safety Study: An Assessment of Accident Risk in U. S. Commercial Nuclear. Power Plants," WASH-1400, NUREG-75/104,1975. '

6) R. 'd. Burian and P. Cybulskis, " CORRAL 2 User's Manual, BCL report, dated January 1977.

1

\

5-5

1 y . .

Table 5.11/ Summary. of . source terms for damage state -I-T for

~ input-tu-LGS-DES'

~

3

. Fa 1.l ure - _

Modes' and.

Release' ~ I-T/DW ~ _' l-T/ WW . I-T/W I-T/SE I-T/ HB I-T/LGT I-T/IGT

' Paths' Xe-Kr. 21.0 .l.0 1.0 - 1. 0 1. 0 - 0.73 0.73 01- 6.99(-3)*~ 5.99(-3) 6.99(-3) .- - - -

vl 1.78(-3) _1.48(-4) 2.09(-4) -9.6(-2) .2.0(-1) 2.7(-3) 1.9(-2)

Cs1 1.88(-2). '3.11(-4) 9.19(-4) 1.0(-1) 6.0(-2) 9.8(-5) 9.8(-2)

..Te -- '8.41(-2) 1.23(-3) 2.16(-3) 4.0(-1) 1.0(-1) 4.6(-4) 4.6(-2)

~Ba' ~9.94(-4). '1.91(-5). -8.22(-5) 1.0(-2) 7.0(-3) 1.6(-5) _ 1.6(-3)

' ~

Ru _4.95(-3) . 7.39(-5)- 1.39(-4)_ 4.0(-1) 8.0(-2) 3.2(-5) 3.2(-3)

-La- 9.89(-4) 1.46(-5) 2.61(-5) 2.0(-3) 1.0(-5) 5.8(-6) 5.8(-4)

~ eTime of' 5.17- 5.17 5.17 2.4 2.4 1.5 1.5

-Reiease:

(br)-

< Duration 0.5" 0.5 0.5 0.5 0.5 3.4 3.4 of.

t

~ Rel_ ease

- (h'r);

Warning .3.67- 3.67. 3.67 1.0 1.0 0 0 Time (hr)_

Energy of l 100_- 100 IS: 130 100 1.0 1.0

' Release

-(106 Btu /hr).

LHeight (ft) 82 82 82 82 82 82 -

82

. Probability 5.6(-7) 5.1(-7) 5.7(-8) 2.3(-10) 2.3(-8) 5.0(-7) 6.2(-7)

(Regional-

~

(Disasters).

. Probability < '2.41(-5) 2.18(-5) 2.44(-6) 9.77(-9) 9.77(-7) 2.17(-5) 2.67(-5)

(Non-Regional ,

Disa,sters) ' .

. Total' 2.47(-5) 2.23(-5) 2.5(-6) 1.0(-8) 1.0(-6) 2.22(-5) 2.73(-5)

Probability

• 6.99(-3)'='6.99 x 10-3 5-6

^

== - ,

xv  :.. .

E Table 5~

. 2 Summary of source terms for damage state II-T for : input to= LGS-DES Failure Modes-and Release- II-T/WW II-T/SE Paths Xe-Kr 9.8(-1)* 1.0 h

OI 6.86(-3) -

'I2 _6.73(-1) 9.6(-2)

Cs 3.36(-1) 1.0(-1)

Te 2.31(-1) 4.0(-1)

Ba - 4.1(-2) 1.0(-2)

.Ru 4.0(-2) 4.0(-1)

La 3.28(-3) 2.0(-3)

. Tine of' Release 24 . 9 2 27

_!(hr)

Duration of Release 3.91 0.5

_(hr)

Warning time (hr). 5.32 7

- _ Energy of Release 1.0- 130.0 6

(10 Stu/hr)

Height.(ft) 82 82 Probability 1 2.0(-8) 2.0(-12)

.(Regional Disasters)-

Probability. 2.04(-5) . 2.03(-10)

(Non-Regional Disasters)

, j Total P~robab,111ty. 2.06(-6) -

2.05(-10) ,

• 9.8(-1) = 9.8 x 10-1 5-7 L-

. Tabl e. 5.3 Sumary of source terms for damage state III-T for input to LGS-DES Failure Modes

'and Release III-T/WW III-T/SE ' III-T/ HB III-T/LG' III-T/l"5T

-Paths-

- Xe-Kr: 1.0 1.0 1.0 7.3(-1) 7.3(-1) 01 6.99(-3)*' - - - -

I2. :7.81(-2) 4.0(-1) 2.0(-1) 2.7(-3) 1.9(-2)

Cs' .2.24(-1). 4.0(-1) 6.0(-2) 9.8(-5) 9.8(-2)

Te 5.74(-1) 5.0(-1) 1.0(-1) .4.6(-4) 4.6(-2)

.Ba 1.95(-2)- 6.0(-2) 7.0(-3) 1.6(-5) 1.6(-3)

Ru 3.65(-2) 5.0(-1) 8.0(-2) 3.2(-5) 3.2(-3)

La 6.92(-3) 3.0(-3) 1.0(-5) 5.8(-6) -5.8(-4)

Time of 2.67 2. ') 2.0 0.5 0.5 Release

.(hr) 1

-Duration -1.38 0.5 0.5 3.5 3.5 of Release '

(hr)

Warning Time 2.17 1.0 1.0 0 0

_(hr)

Energy.of 100 130 100 1.0 1.0 Release

.(106 Btu /hr)

. Height (ft) 82 82 82 82 82 *

-Probability 3.7(-7) 7.4(-11) 7.4(-9) 1.6(-7) 2.0(-7)

(Regional

-Disasters)

Probability 1.66(-6) 3.4(-10) 3.4(-8) 7.5(-7) 9.2(-7)

(Non-Regional . -

Disasters)  ;

Total * ' 2.03(-6) 4.1(-10) 4.1(-8) 9.1(-7) '12(-6)

Probability 1

• 6.99(-3) = 6.99 x 10-3 5-8

.. .e ;. .

Table 5.4 Summary of source terms for damage state IV-T for input to LGS-DES

. Failure Modes

'and Release. . IV-T/DW IV-T/WW IV-T/W IV-T/SE Paths Xe-Kr 1.0 ' 1. 0 1.0 1.0

- 01 6.99(-3)* 6.99(-3) 6.99(-3) -

1'2 9.39(-1) 9.39(-1) 8. 74 (-1) 4.0(-1)

Cs 8.61(-1) 7.72(-1) 8.04(-1) 4.0(-1)

Te 6.88(-1) 5.82(-1) 8.62 (-1-) 5.0(-1)

Ba 9.39(-2) 9.0(-2) 9.55(-2) 5.0(-2)

Ru. 1.49(-1) 1.19(-1) 1.38(-1) 5.0(-1)

La 1.15(-2) 9.38(-3) 7.89(-3) 3.0(-3)

Time of -1.13- 1.13 1.13 2.0 Release

-(br)

Duration .

3 . 34 3.34 3.34 0.5 of Release (br)

Warning Time (hr) 0.5 0.5 0.5 1.5

. Energy of Reiease 1.0 1.0 1.0 130 (106 Stu/hr)

Height (ft) 82 82 82= 82 Probability 4.7(-8) 4.27(-8) 4.75(-9) 9.5(-12)

(Regional Disasters)

Probability 1.63(-7) 1.46(-7)- 1.63(-8) 3.25(-11)

_ (Non-Regional Disasters)

Total. , 2.1(-7) 1.89(-7) 2.1(-8) 4.2('11)

Probability

• 6.99(-3) = 6.99 x 10-3 5-9

o 9 ,

.g . .

ti) -

J

~

> Table -5.5. LSummary of source terms for damage _ states- 1-S and IV-A

..for-input'to LGS-DES-i Failure Modes: I-S/DW _ IV-A/DW f and Release' Paths

~< t -Xe-Kr 9.99(-1)* 9.99(-1) 01- 6.99(-3)- 6.99(-3)

M '

2 , 3.31(-3). 9.65(-1) o .Cs _ '4. 89( -3 ) _ _8.7(-1).

Te - 2.80(-3) 8.74(-1)

= -

Ba 6.01(-4) 9.9(-2) 4 Ru' 2. 87 ( -4 ) ~- ~1.51(-1)

La -4.01(-4) 1.2(-2)-

. - Time of Release 5.11 1.17'

.(hr)'

' Duration of Release 0.5 3.0

-(hr)

Warning Time'(hr) 3.76 0.5

' Energy ofiReiease 100 1.0

'(106-Stu/hr)

H eight'.(ft) 82 82 Probability (Regional - -

, . Di sa sters)

Probability.(Non-Regional 3.76(-8) 5.0(-9)

Disasters)

Total - Probability 3.76(-8) 5.0(-9)

• 9.~99(-1) = 9.99 x 10-1 5-10

t

. Table 5.6 Summary of source terms for. damage states IS-C and IS-f Failure Modes . IS-C/DW IS-C/SE IS-T/DW IS-f/SE and Release -Paths Xe-Kr 1.0 1.0 1.0 1.0 6.99(-3)*

01 -

6.99(-3) -

12 7.61(-2) 9.6(-2) 8.22(-2) 9.6(-2)

Cs 1.37(-1) 1.0(-1) ,1.43(-1) 1.0(-1) ,

Te 5.68(-1) 4.0(-1) -6.06(-1) 4.0(-1) -

Ba ' 7.42(-3) -1.0(-2) 7.78(-3) 1.0(-2)

Ru 8.17(-2) 4.0(-1) 1.07(-1) 4.0(-1)

La 7.05(-3) 2.0(-3) 7.37(-3) 2.0(-3)

. Time'of Release 0.37 1.3 1.47 2.3 -

r

.(hr)

. Duration of. '3.16 0.5 2.9 0.5

' Release (hr)

Warning Time-(br) 0.37 1.3 1.47 2.3  :

Energy of Release 1.0 130 1.0 130

(106 Btu /nr)

Height (ft) 82 .82 82 82 Probability 1.3(-7) '1.3(-11) 9.0(-7) 9.0(-11)

(Regional Disasters)

Probability 1.4(-8) -1.4(-12) 1.0( 7) 1.0(-11) - i (Non-Regional Disasters)

Total: 1.44(-7) 1.44(-11) 1. 0.( -6 ) 1.0(-10) -

~ Probability

~

• 6.99(-3) = 6.99 x 10-3 5-11

y3 .

,_ Table 5.7 Summary'of source terms for damage' states S-H2O and S-H2O

, Failure' Modes . ' S-H20/W S-H20/SE S-H20/W and Release Paths-Xe-Kr~ -9.87(-1)* 1.0 9.68(-1)

OI -: 6. 99 (-3) -

6.98(-3)

.I2- 1.09(-1) 4(-1) 2.56(-1)

Cs 1.62(-1). 4(-1) 2.74(-1)

. Te 2.89(-1) 5(-1) 3.86(-1)

.Ba 1.23(-2) 5(-2) 2.57(-2)

Ru 4.9(-2) 5(-1) 6.18(-2)

La 3.64(-3) 3.0(-3)' 4.99(-3)-

m.

. Time of Release ~ 2.67 3.5 2.83

.(nr)

Duration of Release 4.56 ~0.5 3.55 (hr)

Warning Time (br) -2.67 3.5 2.83 Ener,gy of Release _1.0 130.0 1.0 l(100 Btu /hr)

Height,(ft) -82 82 82

' Prob?bility 4.1(-8) 4.1(-12) 3.69(-7)

-(Regional Disasters)

' Probability (Non- 1.35(-8) 1.35(-12) 1.35(-8)

Regional Disasters)-

Total 5.45(-8) 5.45(-12) 3.83(-7)

Probability

• 9.87(-1) = 9.87 x 10-1 5-12

i a

. Tabie 5.8 Summary of source terms for damage state I-T for input to_ LGS-FES

- Failure

-Modes and' Release- - I-T/DW I-T/WW .I-T/EE I-T/SE I-T/HB I-T/LGT I-T/E67

Paths' Xe-Kr- 1.0 1.0 1.0 1.0 1.0 0.73 0.73

~

-01 6.99(-3)* 6.99(-3) .6.99(-3) - - - -

1 1.78(-3) 1.48(-4) 2.09(-4) 9.6(-2) 2.0(-1) 2.7(-3) 1.9(-2)

Cs. 1.88(-2) 3.11(-4) 9.19(-4) 1.0(-1)' 6.0(-2) 9.8(-5) 9.8(-2)

-Te' 8.41(-2) 1.23(-3)- 2.16(-3) 4.0(-1) 1.0(-1) 4.6(-4) 4.6(-2)

Ba 9.94(-4) 1.91(-5) 8.22(-5) 1.0(-2) 7.0(-3) 1.6(-5) l'. 6 ( -3)

-Ru 4.95(-3)' 7.39(-5) 1.39(-4) 4.0(-1) 8.0(-2) 3.2(-5) 3.2(-3)

La 9.89(-4) 1.46(-5) 2.61(-5) 2.0(-3) 1.0(-5) 5.8(-6) 5.8(-4)

. Time of- 5.17 5.17 5.17 2.4 2.4 1.5 1.5 Release

- (hr)

Duration - 0.5 0.5 0.5 0.5 0.5 3.4 3.4

- of Release

- (hr);

Warning 3.67 3.67- 3.67 1.0 1.0 0 0 Time (nc)-

Energy of 100 100 100 130 100 1.0 1.0

- Release

'(106 Btu /hr)

Height (ft) 82 82 82 82 82 82 82

.- - Probability 5.6(-7) 5.1(-7) 5.7(-8) 2.3(-10) 2.3(-8) 5.0(-7) 6.2(-7)

[

.g'

' (Regional-Disasters)

Probability 1.99(-5) 1.80(-5) 2.03(-6) 8.08(-9) 8.06(-7) 1.79(-5) 2.21(-5)

( Non-Rigional .

Disasters)

^ Total 2.05(-5) 1.85(-5) 2.03(-6) 8.31(-8) 8.31(-7) 1.84(-5) 2.27(-5)

Probability r

• 6.99(-3)=6.99x10-3 5-13

c. .-. _ _ _ _ _ _ _ _ _ _ _ - - _ - _ _ _ _

r #

Tabl e - 5.9 Summary of' source terms for damage state II-T for' input to LGS-FES Failure Modes-

'and Release II-T/WW II-T/SE Paths

-Xe-Kr 9.8(-1)* 1.0 O!- 6.86(-3) -

.12 6.73(-1) 9.6(-2)

Cs .3.36(-1) 1.0(-1)

.Te 2.31(-1) 4.0(-1)

. Ba 4.1(-2) 1.0(-2)

Ru 4.0(-2) 4.0(-1)

La. 3.28(-3) 2.0(-3)

Time of Release 24.92 27

-(hr)

I Duration of Release 3.91 0.5 (hr)

, Warning time (br) 5.32 7

,_ Energy of Release' 1.0 130.0 (106 Btu /hr)

Height (ft) 82 82 Probability 2.0(-8) 2.0(-12)

(Regional Disasters)

Probability 1.91(-6) 1.9(-10)

(Non-Regional Disasters)

W Total Probability 1.93(-6) 1.9(-10)

'g

• 9.8(-1) = 9.8 x 10-1

. 5-14 t

(. -..

= ..-

~

Table 5.10 Summary of source terms for damage state III-T for input to LGS-FES

' Failure Modes and Release III-T/WW - III-T/SE III-T/HB III-T/LGT III-T/13T Paths Xe-Kr- .1.0 1.0 1.0 7.3(-1) 7.3(-1) 01 6.99(-3)* - - - -

12 7.81(-2) 4.0(-1) 2.0(-1) 2.7(-3) 1.9(-2)

Cs. ~ 2.24(-1) 4.0(-1) 6.0(-2) 9.8(-5) 9.8(-2)

Te 5.74(-1)- 5.0(-1) 1.0(-1) 4.6(-4) 4.6(-2)

Ba 1.95(-2) 5.0(-2) 7.0(-3) 1.6(-5) 1.6(-3)

Ru 3.65(-2) 5.0(-1) 8.0(-2) 3.2(-5) 3.2(-3)

La 6.92(-3) 3.0(-3) 1.0(-5) 5.8(-6) 5.8(-4)

Time of 2.67 2. 0 2.0 0.5 0.5 Release (hr)

Duration 1.38 0.5 0.5 3.5 3.5 of Release (nr)

Warnin9 Time 2.17 1.0 1.0 0 0 (br)

Energy of 100 130 100 1.0 1.0 Reiease (106 8tu/hr)

Height (ft) 82 82 82 82 82 Probability 3.7(-7) 7.4(-11) 7.4(-9) 1.6(-7) 2.0(-7)

(Regional Disasters)

Probability 1.58(-6) 3.24(-10) 3.24(-8) 7.14(-7) 8.76(-7)

(ten-Regional ,

' Disasters) .

Total 1.95(-6) 3.98(-10) 3.98(-8) 8.74(-7)

Pr'obability 1.08(-6)

T

• 6.99(-3) = 6.99 x 10-3 5-15 k..

(( .

Table 5.11 - Summary of source terms for damage state IV-T for input to LGS-FES Failure Modes and. Release IV-T/DW IV-T/WW IV-T/W IV-T/ SE

. Paths Xe-Kr 9.99(-1) 9.99(-1) 9.98(-1) 1.0

'0I 6.95(-3)* 6.95(-3) 6.95(-3) -

'!2- 4.74(-1) 4.61(-1) 4.68(-1) 4.0(-1)

Cs- 4.86(-1) 4.81(-1) 5.18(-1) 4.0(-1)

Te 5.09(-1) 4.45(-1) 4.81(-1) 5.0(-1)

Ba' 5.54(-2)

. 5.60(-2) 5.96(-2) 5.0(-2)

.Ru' 8.85(-1) 7.81(-2) 8.31(-2) 5.0(-1)

La- 6.82(-2) 6.03(-3) 6.51(-3) 3.0(-3)

Time of 1.13 1.13 1.13 2.0

'u-Release

.(hr)

Duration 3.34 3.34 -3.34 0.5 of Release (hr)

Warning Time (br) 0.5 0.5 0.5 1.5 Energy of Release 1.0 1. 0 - 1.0 130 (106 8tu/hr)

Height'(ft) 82 82 82 82 Probability 4.7(-8) 4.27(-8) 4.75(-9) 9.5(-12)

(Regional Disasters)

Probability . 1.63(-7) 1.46(-7) 1.63(-8) 3.25(-11)

(Non-Regional Otsasters)

Total . .

2.1(-7') .

1.89(-7) 2.1(-8)

. 4.2(-11)

Probability -

• 6.99(-3) = 6.99 x 10-3 5-16 L - _ - - - _ - - - _

Table 5.12 Summary of source terms for damage states I-S and IV-A for input to LGS FES i P

' Failure Modes I-S/DW IV-A/DW and Release Paths i

Xe-Kr 9.99(-1)* 9.96(-1)

O! 6.99(-3) 6.94(-3) 12 3.31(-3) 4.78(-1)

Cs 4.89(-3) 5.06(-1)

Te 2.80(-3) 5.18(-1)

Ba 6.01(-4) 5.76(-2)

Ru 2.87(-4) 8.86(-2)

La 4.01(-4) 6.95(-3)

Time of Release 5.11 1.17 (nr)

Duration of Release 0.5 3.0

.(hr)

Warning Time (hr) - 3.76 0.5

- Energy of Release 100 1.0 (106 Btu /hr)

. Height (ft) 82 82 Probability (Regional - -

Disasters)

' Probability (Non-Regional 3.76(-8) 5.0(-9)

. Disasters)

Total Probability 3.76(-8) 5.0(-9) t

• e

• 9.99(-1) = 9.99 x 10-1 5-17

p~ , ~

h DISTRIBUTION

{

a Dr. Sarbeswar Acharya (1)

Accident Evaluation Branch

~

Division of Systems Integration U.S. Nuclear Regulatory Commission Mail. 'Stop P-802 L -Washington, D. C. 20555 Dr. Walt Butler, Chief (1)

Containment Systems- Branch Division of Systems Integration U.S.: Nuclear .Regulatcry Commission Mail Stop P-1000 Wasnington, D. C. 20555 Mr. James L.' Carter (6)

Reactor Systems Branch

~ Division of Systems Integration U.S. Nuclear Regulatory Commission Mail Stop P-1132 Washington, D. C. 20555 Mr.'Erulappa Chellian (1)

Reliability & Risk Assessment Branch Division of Safety Technology U.S. Nuclear Regulatory Commission

. Mail Stop P-216 -

i

. Washington, D. C. 20555 Mr. Frankl.in Coffman, Jr., (1)

Section Leader-Systems Interactions Section U.S. Nuclear Regulatory Commission Mail Stop P-216 Washington, D. C. 20555 Dr. Philip Hammond (1)

R&D Associates 4640 Admiralty Way 4

Marina del Rey,-California 90292 Mr. Brad Hardin ,- (1) ,

. Reactor Systems Branch Division of Systems Integration U.S. -Nuclear Regulatory Commission Mail Stop P-1132 Washington, D. C. 20555 b

~

Dr. R. Wayne Houston (1)

- Assistant Director for Reactor Safety Division of Systems Integration

U.S. Nuclear Regulatory Commission Msil Stop P-1102 Washington, D. C. 20555 Mr. Lewis J. Hulman, Chief (1)

Accident Evaluation Branch Division of Systems Integration

-U.S. Nuclear Regulatory Commission Mail Stop P-802 Washington, 0. C. 20555 Dr. Roger J.-Mattson, Director (1)

Division of Systems Integration

~U.S. Nuclear Regulatory Commission Mail Stop P-1102 Washington, D. C. 20555 Dr. Jocelyn A. Mitchell (1)

Reactor Systems Branch Division of Systems Integration U.S. Nuclear Regulatory Commission Mail Stop P-1132 Washington, D. C. 20555 Mr. Daniel R. Muller (1)

Assistant Director for Radiation Protection Division of Systems Integration U.S. Nuclear Regulatory Commission Mail Stop P-822 Washington, D. C. 20555 Dr. Walter Pasedag (1)

Accident Source Term Project Office U.S. Nuclear Regulatory Commission Mail Stop 1130SS Washington, D. C. 20555 Mr. Jack E. Rosenthal, Section Leader (1)

Reactor Systems Branch '

Division of Systems Integration U.S. Nuclear Regulatory Commission Mail Stop P-1132 Washington, D. C. 20555

- c;; .

Mr_. Frank Rowsome (1)

Assistant Director for Technology Division of Safety Technology U.S.~ Nuclear Regulatory Commission Mail Stop P-266 Washington,.D. C. 20555-Mr. Brian Sheron', Chief (1)

Reactor Systems Branch Division of Systems Integration U.S. Nuclear Regulatory Commission Mail Stop P-1132 Wasnington, D. C. 20555 Dr. Tnemis P. Speis Director (1)

. Division of Safety Technology U.S. Nuclear Regulatory Commission Mail Stop 278 Washington, D. C. 20555 Dr. Ashok C. Thadani (1)

Reliability and Risk Assessment Branch Division of Safety Technology

- U.S. Nuclear Regulatory Commission Mail Stop 216 Washington, O. C. 20555 Prof. Theo. Theofanous (1)

Purdue University-132 Pathway Lane W. Lafayette Indiana 47905 DNE Chairman (1)

DNE Deputy Chairman (1)

ONE Associate Chiirman (4)

ERA Division Heads and Group Leaders (6)

DNE Nuclear Safety Library .

.(2) e

BROOKHAVEN NATIONAL LABORATORY ,

MEMORANDUM

.DATE: dune 11, 1984 To: W. T. Pratt FROM: d.W. Yang 9W susJECT: BNL-NUREG-33835 INTRODUCTION An inconsistency has been found between the text and the numerical re-sults in the subject BNL informal report. Specifically, the Class S accident sequence is described as being equivalent to a large break LOCA (refer to pages 3-6, 4-9, and 5-4) but the timing of events (Table 3.9) release frac-tions (Table 4.9) and source-terms (Table 5.7) are not consistent with a large break LOCA calculation. A number of calculations were perfonned for accident Class S to determine the sensitivity of the results to various primary system i assumptions. The large break LOCA calculation resulted in slightly higher source terms and lower warning times, so that it was selected as the represen-tative sequence for this class. The text reflects this decision; unfor-

-tunately, the numerical results do not. Tables 3.9, 4.9 and 5.7 have there-fore been reproduced from BNL-NUREG-33835 and modified to reflect the large .

break LOCA assumptions.

I The original source terms in BNL-NUREG-33835)were used by the NRC staff to perform a site consequence analysis in support of the Final Environmental Statement

• related to the operation of the LGS. The modified Class S source terms attached to this memorandum should have been used rather than the origi-nal source terms in BNL-NUREG-33835. However, it can be demonstrated that if the modified source terms were used in place of the original source terms the overall risk, as calculated by the NRC staff in the LGS-FES, would not change significantly. The Class S sequences contribute to only 2% of the long-tenn damage indices (e.g., latent fatalities). Changes in the fission product re-  ;

lease fractions indicated in Table 5.7 would not therefore significantly in-fluence these damage indices at Limerick, in addition, differences in the warning time do not affect long-term damage indices. Class S sequences con-tribute to 20% of the early fatalities at Limerick. However, most of this

' contribution comes from seismically initiated events. The evacuation model used for regional disasters by the NRC staff assumes a 20-hour delay so that

• Final Environmental Statement related to the operation of Limerick Generating Station, Units I and 2 NUREG-0974.

0 -

l W. T. Pratt June 11, 1984 Page 2 differences in the warning times (which could influence early damage indices) are not important for these seismically initiated events. In addition, most of the contribution to early fatalities for Class S sequences is due to the S-IIRT sequences and differences between the modified and original release fractions are minimal for this sequence. In summary, the modified release fractions and warning times for the Class S sequences do not significantly change overall risk at the Limerick facility.

DISCUSSION OF RESULTS For the S-H20$ case (in which the water is assumed to drain from the vessel prior to core melt and the containment is assumed to fail in the wet-well below the water level) a comparison between the BNL releases and those reported in the LGS-SARA for the equivalent release (VR) shows reasonable agreement, except for 01, Ba, and La. These differences are all indications of dlfferences in source term methodology. Fission product releases are based in part on NUREG-0772 in the LGS-SARA and on RSS methods in BNL-NUREG-33835 A decreased release of organic iodine and an increased release of the barium group are characteristic of NUREG-0772 releases relative to RSS releases.

However, for the S-H20S case (in which water is assumed to remain in the vessel during core degradation and the containment is assumed to fall in the wetwell below the water line) the BNL release fractions are lower than the equivalent LGS-SARA releases (VRH2O). The largest discrepancy occurs for the 01, Te, Ba, and I.a groups. Discrepancies in the first three fission product groups are partially explained by differences in methodology.

The LGS-SARA release fractions for the case with water (VRH2O) are pre-dicted to be significantly higher than without water (VR). This is not the case for the equivalent BNL calculations. Only for the iodine group is there an increase of approximately a factor of four in the RNL calculations for the case with water in the bottom head compared to the case with no water. Fur-thermore (for the aerosol groups), it is seen that the species released during the melt release phase bottom head (S-H20/GO. (Cs andspecies However, Ba) arereleased higherprimarily for the case duringwith corewater

/ con- in the crete interactions (Te, Ru, and La) show an increased release for the case withnowaterpresent(S-H20/W). These effects are shown graphically on Fig-ures 1 and 2. Figure 1 shows the variation with time of the 1 2 release which shows a significant increase at the time of core slump. This same character-istic is also true of Cs and Ba, Figure 2 shows the variation with time of the Te release, which shows a rapid release during core / concrete interactions for the case with no water. The' case with water shows an initial increase -

during core slumping 'followed by a reduced release rate during core / concrete interactions. The release characteristics for Ru and t.a are similar to the Te release.

JWY:jr/tr

(_.

l a -

4 l

Table 3.9 BNL analyses of Class S (T RPVRB)

S i h

Key Events (hours) S-H2O S-Ti26 Containment fails 0 0 j Core melt begins 0.5 0.3 Core melt ends 1.2 1.3 Vessel head fails 2.5 1.3 70-cm penetration of floor 6.0 3.8 l

4 e

B 3-12

w. . _ .

6_.__-__-------------------------------------------------------------------.---------------------

l . .

l' >

h Table 4.9 ' Fission product release fractions for Class S

?

MODIFIED

' ASSESSMENT LGS-SARA CALCULATIONS

< FAILURE MODE VRH2O VR S-H20/W S-IPl6/W I OXIDATION RELEASE Yes Yes Xe --Kr 1.0 l'. 0 9.99(-1) 9.99(-1)

Organic Iodine 3.0(-4) 3.0(-4) 6.99(-3) 6.99(-3) 1_

2 5.0(-1) 1.0(-1) 2.2(-1) 5.4(-2)

Cs 7.3(-1) 3.3(-1) 3.7(-1) 3.2(-1).

. Te 7.5(-1) 3.3(-1) 3.0(-1) 4.1(-1) I L

Ba 3.5(-1) 1.5(-1) 3.8(-2) 3.4(-2) ,

Ru 7.0(-2) 4.0(-2) 5.3(-2) 6.6(-2)

La 5.0(-2) 2.0(-2) '4.1(-3) 5.5(-3)

DF for 12 1 1 0F for Aerosols 1 1 Core Melt Start 0.34 0.25 0.5 0.3 Core Melt End 0.34 0.25 1.2 1.3 1st Vap. Release 1.0 3.75 2.5 1.3 2nd Vap. Release 3.0 1.8 Vap. Release .End 4.5 3.3 Containment Fail -

0.0 0.0 4-20

,,