ML18151A899

Summary Rept of :Grand Gulf Low Power & Shutdown Abridged Risk Analysis, Draft Ltr Rept

ML18151A899
Person / Time
Site:	Grand Gulf, Surry
Issue date:	05/31/1992
From:	Tony Brown, Kmetyk L, Miller L SANDIA NATIONAL LABORATORIES
To:	NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
Shared Package
ML18151A900	List: ML18151A899 ML18152A050
References
NUDOCS 9312020088
Download: ML18151A899 (46)

Text

SUMtvfARY REPORT OF:

GRAND GULF LOW POWER AND SHUTDOWN ABRIDGED RISK ANALYSIS POS 6: Early Refueling DRAFT LEITER REPORT Thomas D. Brown1 LeAnn A. Miller1 Lubomyra N. Kme7k1 Lanny N. Smith Donnie W. Whitehead1 John Darby3 John Forester2 May 31, 1992 1

Sandia National Laboratories 2

Sci.ence Application International Corporation 3

Sci.ence & Engineering Associates, Inc .

~-

,. Enclosure I Sandia Report of the Study of the Grand Gulf Plant

\..-

CONTENTS

• 1.0 INfR.ODUCTION . . . . . . . . . . . . . . . . . . .

1.1 1.2 1.3 Study Objectives ..... *. . . . . . . . . .

Scope of the Study . . . . . . . . . . . . .

Methods . . . . . . . . . . . . . . . . . . . .

1 1

1 1

1.4 Limitations and Strengths of the Study . . . . . . . . . . . . . . . . . . . 4

2. ACCIDENT PROGRESSION ANALYSIS ...  ; . . . . . . . . . . . . . . . . . . 8 2.1 Approach . . . . . . ."'. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2 POS 6 Plant Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.3 Level I Sequence Description . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.3.1 Sequence Description . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.3.2 Plant Damage State Description . . . . . . . . . . . . . . . . . . . 11 2.4 Event Tree Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3. SOURCE TERM ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.1 Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.2 Description of Parametric Model . . . . . . . . . . . . . . . . . . . . . . . 20 3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

4. CONSEQUENCE ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

• 5.

4.1 Onsite Consequences . . .

4.1.1 Building Doses . .

4.1.2 Parking Lot Doses 4.2 Offsite Consequences . . .

INTEGRATED RESULTS CONDIDONAL ON CORE DAMAGE . .. . . . . 33 24 24 28 30

6. INSIGHfS AND CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

7. REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~ . . . . 39

• • i

C FIGURES

• 1-1.

1-2.

2.4-1.

Scope of abridged study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Summary of abridged methodology . . . . . . . . . . . . . . . . . . . . . . . .

Grand Gulf POS 6 abridged accident progression event tree 6

7

. . . . . . . 19 4.1.1-1. Containment and auxiliary building dose rates for selected paths . . . . 28 4.1.2-1. Parking Lot dose rates for Ramsdell and Wilson/Regulatory Guide models for distances from 10 - 500 meters from the reactor: First Release Segment ....."'. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 4.1.2-2. Parking Lot dose rates for Ramsdell and Wilson/Regulatory Guide models for distances from 10 - 500 meters from the reactor: Second Release Segment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ._. 31 5-1. Grand Gulf POS 6 offsite consequences for LOSP and nonLOSP PDSs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 TABLES 2.3-1 Grand Gulf LP&S POS 6 Initiating Events . . . . . . . . . . . . . . . . . . . . 11

• 2.3-2 2.4-1 3.3-1 Grand Gulf LP&S POS 6 Plant Damage State Attributes . . . . . . . . . . 12 Accident Progression liming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Mean Source Terms for Accident Progression Paths (Total Release) . . . 23 4.1.1-1. Transit times through the containment and auxiliary building for Grand Gulf POS 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.1.1-2. Grand Gulf POS 6 mean containment doses and dose rates. . . . . . . . . 26 4.1.1-3. Grand Gulf POS 6 mean auxiliary building doses and dose rates. . . . . 27 4.1.2-1. Grand Gulf POS 6 mean doses and dose rates at 100 m based on the Ramsdell building wake effect model. . . . . . . . . . . . . . . . . . . . . . . . 29 4.1.2-2. Grand Gulf POS 6 mean doses and dose rates at 100 m based on the Wilson building wake effect model. . . . . .. . . . . . . . . . . . . . . . . . . . 30 4.2-1 Grand Gulf POS 6 Offsite Mean Consequences . . . . . . . . . . . . . . . . 32 ii

'<v ACKNOWLEDGMENfS

• We wish to thank the many people who work in various capacities to support this analysis: Fred T. Harper (SNL) who provided many helpful technical suggestions in the development and application of methods and who provided a much needed sanity check on the analysis and results; Jay Johnson (GRAM) who performed the MACCS calculations for the offsite consequences; David I. Chanin who provided helpful ideas for assessing onsite consequences; and Ann W. Shiver (SNL) who provided support in the data preparation for the uncertainty analysis.

We would also like to thank Chris Ryder (NRC) for his program and management support and for his words that form the introduction of this report.

Finally, we would like to thank the Source Term Advisory Group, which consisted of John E. Kelly (SNL), Hossein P. Nourbakhsh (BNL), Dana A. Powers (SNL), and Trevor Pratt (BNL), for their review of the source term analysis and their many helpful suggestions .

• iii

ACRONYMS AND INTI1ALISMS

• ADHR APET BNL BWR Alternate Decay Heat Removal System Accident Progression Event Tree Brookhaven National Laboratory Boiling Water Reactor CD Core Damage co Core Concrete Interactions CDS Condensate System .,

CF Containment Failure CNf Containment CRD Control Rod Drive System cs Containment Spray System ECCS Emergency Core Cooling System FW Firewater System ms Hydrogen Ignition System HPCS High Pressure Core Spray System I-IRA Human Reliability Analysis IE Initiating Event LHS Latin Hypercube Sample LOCA Loss of Coolant Accident LOSP Loss of Offsite Power

• LP&S NRC PDS POS PRA Low Power and Shutdown Nuclear Regulatory Commission Plant Damage State Plant Operating State Probabalistic Risk Assessment PSW Plant Service Water PWR Pressurized Water Reactor RFO Refueling Outage RHR Residual Heat Removal System RPV Reactor Pressure Vessel SBGT Standby Gas Treatment System SBO Station Blackout SDC Shutdown Cooling System SP Suppression Pool SPC Suppression Pool Cooling System SPMU Suppression Pool Makeup System SRV Safety Relief Valve SSW Standby Service Water TAF Top of Active Fuel TBCW Turbine Building Cooling Water VB Vessel Breach

• iv

1.0 INTRODUCTION

• 1.1 Study Objectives The Office of Nuclear Regulatory Research at the U. S. Nuclear Regulatory Com.mission established programs to investigate postulated accidents during low power and*

shutdown (LP&S) operations of a BWR (Grand Gulf) and a PWR (Surry). One such program is a risk study of accident progressions and consequences.

The objective of this study is to make a preliminary risk determination of the progressions (Level 2 analysis) al)d the consequences (Level 3 analysis) of accidents during low power and shutdown operations in the Grand Gulf plant. The study was designed to obtain results for regulatory decisions that are to be made in the early summer of 1992. This letter report documents the methods, findings, and implications of the study. A sister study of the Surry plant is reported separately by the staff at Brookhaven National Laboratory (BNL).

1.2 Scope of the Study The abbreviated risk analysis took place from January through April 1992. The study has been referred to as an abridged risk analysis. The term abridged means that simple event trees (about nine top event questions) were developed and used with assumptions and other approximate methods to compute rough estimates. The term risk means conditional consequences (probability of the various events during the accident progressions multiplied by the consequences), given that core damage has occurred. Traditional risk estimates, computed by multiplying the conditional consequences and the frequency of the sequences leading up to core damage, could not be made at this time because the frequencies have yet to be determined in companion Level 1 and HRA studies. Uncertainty has been taken into account in a manner consistent with the detail of the abridged study.

This study investigated the possible accident progressions and the associated consequences of a single plant operating state, POS 6, an early stage of refueling, where the reactor vessel head is removed, the steam dryers and separators are removed, the drywell is open, and the containment is open. The sister study at BNL investigated mid-loop operation. The scope of both studies is illustrated in Figure 1-1.

1.3 Methods The abridged process of computing conditional consequences is shown in Figure 1-2.

In general, both the study reported here and the study done at BNL follow this scheme. Some differences in the details of the procedure exist and are noted at the end

• 1

of Section 1.3. The process used here is an abbreviated form of the NUREG-1150 study

[1] .

Accident progressions The calculations begin with the assumption that core damage (CD) has occurred, making the consequences conditional. Given CD, the reasonable accident progressions are delineated with the accident progression event tree (APE1). Much of the delineation is based on deterministic calculations with a code used to compute source terms, such as ~LCOR [2]. The likelihood of the various accident progression is reflected vis-a-vis branch point probabilities.

Branch point probabilities were assigned to reflect the likelihood of various pathways thought to exist. In large scale risk studies, the assignment can be done by groups of people knowledgeable of the severe accident issues. Here, because of resource limitations, most of the assignments were done by the project staff. The probabilities are not as rigorous as they could be but this is one of many limitations of the study to be discussed. Some lack of rigor in determining the probabilities is taken into account by repeating the calculations with other possible probabilities; taken together, the repeated calculations as they were done constitute an uncertainty analysis.

Through the uncertainty analysis, distributions, instead of point values, were assigned to selected branch points. The distributions are subjective but account for many

• possible values of the branch points. Point values are selected from the distributions with a form of Monte Carlo sampling known as Latin Hypercube Sampling (LHS).

After making sets of inputs, each set, consisting of point values, is assigned to the branch points and multiplied though to the ends of the APET. The calculations are repeated using the sets of inputs, building a probability distribution at the end of each pathway.

Source terms Having delineated accident progressions with the APET, the source terms of the progressions were calculated with a parametric code [3]. The parametric code is a mimic of the detailed source term codes; it is a collection of simple mass-balance equations, activated by characteristics of the progression.

The parametric code determines source terms given the characteristics of the accident progression and other inputs (typically various fractions, such as the inventory leaving the reactor vessel, involved in a core concrete interaction, entering the containment, etc.). Because these other variables are imprecisely known, many reasonable values can be assigned to the inputs. As in the APET calculations, distributions are assigned to the variables and sampled with LHS to form many sets of inputs values for repeated

• 2

calculations. The result is a distribution of source terms for each accident progression

• pathway.

An internal "Source Term Advisory Group" was formed to support this study. The results of the accident progression and source term analysis were presented to and discussed with the advisory group in two meetings during the course of this analysis.

The members of this group were W. T. Pratt and Hossein P. Nourbakhsh of BNL, and John E. Kelly and Dana A. Powers of SNL.

Consequences Three sets of consequences were determined, building dose, onsite dose (so called parking lot dose), and offsite consequences.

o Building dose was determined with source terms derived from the parametric source term code, then converting the source terms into dose with simple equations. Doses in the containment and auxiliary building were estimated.

o Parking lot dose was computed with the Ramsdell model, in which the release concentration is somewhat proportional to wind speed, and a combination of the Wilson model and model in Regulatory Guide 1.145, in which concentration is inversely proportional to wind speed .

• 0 Offsite dose was computed using the MACCS code[4].

In essence, the equations and code convert the source terms, accounting for other factors such as health effects, into doses and dose rates. Uncertainty was not propagated through the consequences as was through the APET and the source term calculations. While a sample size of 100 was used in the onsite analysis to propagate accident progression and source term uncertainties, a reduced sample size of 12 was used in the determination of offsite consequences.

Conditional consequences Conditional risk was computed by multiplying the consequences by their associated accident probability that was determined with the APET. This product of probability and consequences was computed for each accident progression pathway. The products of the pathways were summed. This process was repeated for each of the few samples of the source terms. Then, high, medium, and low results were reported.

Differences This study differs slightly from its sister program at BNL in three ways. (1) Here, one hundred samples of uncertainty distributions were taken in the accident progression

• . 3

and source term analyses whereas, in the BNL study, two hundred samples were*

taken. (2) Here, twelve samples were propagated through the APET to consequences whereas, in the BNL study, twenty samples from the source term distribution were used in consequence calculations and traced back through the APET for the probabilities needed to compute conditional risk. (3) Here, doses in the containment were calculated whereas, in the BNL study, these calculations were thought unnecessary due to the high doses predicted in the parking lot.

1.4 Limitations and Strengths of the Study The study had strengths and limitanons which are important to understand the context of the calculations.

Limitations o The subject of the study is one POS, early refueling. This POS was selected for study because it was identified in a preliminary Level 1 study, known as a coarse screening analysis [5], as potentially occurring at a relatively high frequency.

Also, the POS had characteristics (i.e., reactor vessel head removed) of interest to the staff in the Office of Nuclear Reactor Regulations at the NRC.

o The abridged study is based on the coarse screening analysis where accident sequences potentially having high frequencies, were identified. The

• consequences of these sequences were determined in the Level 2 and 3 abridged study reported here. The frequency is not incorporated into the merged with the Level 2 and 3 calculations to determine risk because the numerical value of the frequency estimate is believed to be too rough for such use.

0 The simple APET accounts for a limited number of factors. The APET consisted.

of nine top event questions, compared to about one hundred questions in a large scale PRA.

0 The onsite dose estimates stem from simple equations yielding rough estimates.

0 Variables were selected and assigned distributions for the uncertainty analysis by the project staff.

0 Because of gaps in knowledge of the plant configuration and operator actions, assumptions were necessary. The assumptions are documented in the sections to follow. Some of the gaps will be filled with more rigorous determinations with results from detailed Level 1 and HRA studies during a follow-up Level 2 and 3 study .

• 4

Strengths

• o Even with the limitations noted above, the abridged study is a systematic evaluation of severe accident progressions, with a limited treatment of uncertainty.

o The source term analysis was reviewed by an internal advisory group.

o The project staff and the NRC project staff believe that the APET represents the occurrence of key events during accident progressions.

o The relationship and timings of accident progression events and factors have been determined to at least a first approximation.

o Estimates of both onsite and offsite conditional consequences were made.

The sections to follow document the abridged study of the Grand Gulf plant. The discussion above is expanded, providing important details and results .

• 5

• Plant POS1 * *

• POSn F - - * (P)(C)

. (P)(C)

. (P)(C)

(P)(C)

. (P)(C)

.___ _ ... (P)(C)

• Level 1 Level 2 & 3 R

• risk R * (F] ~--IP] .IC] . :-.:*:. . *..** *. *.-*.. : --~ *.........:-.**.

F

• eeQuence freQuency P

• accident progression probability C

• conaeQuencee Figure 1-1. Scope of abridged study

• 6

Parametric Source Term Equations/

APET Code MACCS Condi tlonar Source Conse-

• Probablll ties Identifier Terms quences

~---,

ABCDEF <> CJ <> -z- c::J <> c,~ ...

BABBCC ct> D ¢> -z- w -t> c2~ '

L[P, uc, l CD

• • ADCCDE ABBDEF

¢>

D D

• • -z- [TI] <>

-z- CJ <>

c~ ,.,

c3?

4 Uncertainty In Condi tlonal Consequences Figure 1-2. Summary of abridged methodology 7

2. ACCIDENT PROGRESSION ANALYSIS

• 2.1 Approach The progression of accidents following core damage are analyzed in the Level II portion of the PRA. In this chapter the development and quantification of the accident progression scenarios will be presented. The input to the accident progression analysis are the core damage sequence definitio_ns developed in the Level I analysis [5]. The core damage sequences define the successes and failures of equipment and human actions that have resulted in the loss of core cooling and the onset of core damage. The sequence definitions provide information on the status of core cooling systems, containment cooling systems, and containment integrity at the time of core damage.

From this information the possible accident progression scenarios, which identify the response of the core and the containment following core damage, are determined.

These accident progressions are developed and displayed using an event tree approach.

In this abridged analysis only the most important events that affect the timing and the magnitude of the radionuclide release are addressed. The output from the accident progression analysis are the accident progression path definitions and the likelihood, conditional on core damage having occurred, of each path. In the source term analysis, the fission product release associated with each path is estimated. The estimation of the source term is addressed in Chapter 3 and the resulting consequences are presented in Chapter 4 .

• In the following subsections the configuration of the plant during POS 6 will be presented, the important characteristics of the Level I core damage sequences will be identified, and the development of the accident progression paths will be discussed.

2.2 POS 6 Plant Configuration The configuration of the plant at the onset of core damage is important because it will determine the framework within which the accident will unfold. That is, the plant configuration will define the boundary conditions for the analysis. For example, it will define the mitigative features of the plant that will be available during the accident (e.g., containment, suppression pool, containment sprays).

The abridged risk analysis was performed on the early portion of the refueling mode of operation. In the Level I coarse screening analysis this mode of operation is referred to as plant operating state 6 (POS 6). During a refueling outage the plant will enter POS 6 prior to loading fresh fuel (i.e., going down) and then following fuel transfer on the way back up to power conditions (i.e., going up). In the Level I analysis, the sequence definitions are based on the "going down" phase because 1) more systems are likely to be unavailable (i.e., on the way back up maintenance and repairs may already have been performed on many systems) and 2) the decay heat levels are higher and, therefore, there is less time to respond to events in the going down phase versus the

• 8

going up phase. Thus, in this study only the "going down" phase is analyzed. POS

• 6 begins when the vessel head is detached and ends when the upper reactor cavity has been filled with water. During this POS the following tasks are performed:

1. Steam dryers are removed,

2. Vessel water level is lowered to the bottom of the steam lines and the steam lines are plugged,

3. Water level is raised and the steam separators are removed, and

4. Vessel water level is raised to flood the upper reactor cavity.

Prior to this mode of operation, the containment equipment hatch and personnel locks have been opened, the drywell head has been removed, and the drywell equipment hatch and personnel locks have been opened. Thus, the suppression pool is effectively bypassed both from the vessel and from the drywell (i.e., steam lines are plugged and the drywell is open).

Urning information for the initiation of the accident in POS 6 is based on Grand Gulf refueling outage (RFO) data. Information was available for the first four RFOs.

However, because of the number of special tests that were conducted during the first refueling outage, it was felt that RF0-1 was not representative of a typical RFO and, therefore, data from this cycle was excluded from the analysis. Thus, only RFO 2,3, and 4 data were used in this study. Based on this data the fastest the plant will enter POS 6 from full power is approximately four days after shutdown and the longest the plant has been in POS 6 (in the "going down" phase) is approximately 12 days (i.e., 16 days from shutdown). In the Level I analysis the time window from the initiating event to core damage was based on the decay heat at four days. This assumption is carried through the Level II/ill analyses.

2.3 Level I Sequence Description 2.3.1 Sequence Description The initial conditions for the accident progression analysis are the core damage sequence descriptions from the Level I analysis [5]. That is, a list of attributes that describe the status of systems that can be used to mitigate the accident and *the configuration of the plant at the time of core damage. In the Level I analysis coarse screening analysis the sequences were placed into three groups: potentially high likelihood group, potentially medium likelihood group, and potentially low likelihood group. Only sequences from the high likelihood group were analyzed in this study.

Fourteen different initiating events (IE) are associa~ed with these sequences. A list of these 14 IEs is presented in Table 2.3-1. The initiating events can be divided into four major groups: Loss of Offsite Power (LOSP) Transients, Loss of Support System Transients, Loss of Coolant Accidents (LOCAs), and Decay Heat Removal Challenges.

The accident sequences that form the input to this study all progress to core damage

• . 9

in the following manner: the initiating event leads to the loss of the operating shutdown cooling system, subsequent random failures and unavailabilities complete the loss of core cooling and injection. Without a means to keep the core cool, the vessel inventory boils away and core damage ensues.

In the Level I screening analysis both the emergency core cooling system, ECCS, and Makeup (i.e., CRD and CDS) were assumed to be unavailable or unable, due to some postulated failure, to prevent core damage. Thus, only the firewater system, FW, and the standby service water (SSW) cross-tie were considered as potential injection .

systems.

In POS 6 the suppression pool (SP) can be either at its normal level, partially drained, or empty. Furthermore, the suppression pool makeup system (SPMU) is not available.

Because a supply of water to the SP is not available, ECCS systems that draw water from the SP could not be used in a continuous mode and, therefore, it was assumed in the Level I analysis that these system were not available to cool the core. Because the containment spray system (CS) is one mode of the residual heat removal system (i.e., part of ECCS) and draws water from the SP, it is also unavailable during these postulated accidents.

The CRD system has insufficient capacity to prevent the core inventory from boiling and, therefore, was not considered as a means to cool the core in the screening study.

(It should be noted, however, that if this system was used, the energy removed from

• the core via steaming would be sufficient to prevent core damage.) While CDS has more than enough capacity to cool the core, its unavailability due to random failures and maintenance precludes its use as a means to cool the core.

A general description of the core damage sequences for each class of initiators is presented below.

LOSP Transients The LOSP initiating event leads directly to the loss of the alternate decay heat removal system (ADI-IR). Subsequent random failures lead to the complete loss of shutdown cooling (SDC), makeup, the standby service water and the firewater system. With ECCS not available in this POS, as a result of support system failures, the accident proceeds to core damage because of the lack of core cooling.

Loss of Support System Transients In these sequences the IE leads directly to the loss of ADI-IR, makeup, and the firewater system. Subsequent random failures lead to the complete loss of SDC and the SSW system .

• • 10

Decay Heat Removal Challenges

• In these sequences the initiating event leads to the loss of the operating shutdown cooling system. In some of these sequences this system is recovered. However, subsequent random failures lead to the complete loss of SDC, the firewater system, and SSW.

LOCAs That Can Be Isolated In these sequences the isolatio!1 of the LOCA also isolates the SOC systems.

Subsequent random failures lead to the loss of both the firewater system and the standby service water cross-tie system.

Table 2.3-1 Grand Gulf LP&S POS 6 Initiating Events Initiating Initiating Event Event Group Nomenclature Description LOSP TI Loss of Offslte Power (LOSP) Transient Loss of TSB Loss of all TBCW Support System TSC Loss of all PSW (includes Radial Well)

Transient TIA Loss of all Instrument Air Decay ElB Isolation of SOC Loop B only Heat Removal E2B Loss of SOC Loop B only Challenge ElD Isolation of ADHRS E2D Loss of ADHRS only ElT Isolation of SOC Common Suction Line E2T Loss of SOC Common Suction Line ElV Isolation of Common Suction Line for ADHRS E2V Loss of Common Suction Line for ADHRS Isolated H1 Diversion to Suppression Pool via RHR LOCAs J2 LOCA In Connected System (RHR) 2.3.2 Plant Damage State Description The Level I sequences were divided into two plant damage state (PDS) groups: LOSP and nonLOSP. This distinction is made because of the effect that the LOSP has on injection recovery and containment closure. In the analysis of t~e nonLOSP PDS it is

assumed that if injection is not recovered prior to core damage, it will not be recovered

• during core damage. The reason for this assumption is that there is a considerable amount of time from the initiating event to core damage for the operators to align and use injection systems to cool the core. ff they have not done this by the time of core damage, there is no reason to believe that they will recover core cooling during core damage. Recovery of injection is considered in the LOSP PDS. In these sequences offsite power is not available and, therefc;>re, non-emergency systems are not available to provide injection to the core. Thus, for the LOSP PDS it is assumed that if offsite power is recovered, injection can be recovered. The availability of ac power also effects the likelihood that the containme!lt is closed prior to core damage. The crane that is used to position the equipment hatch is powered with offsite ac power. Thus, if the accident is initiated by a LOSP it is assumed that the containment can not be closed prior to core damage. The key attributes associated with these two PDSs are presented in Table 2.3-2.

Table 2.3-2 Grand GuH LP&:S POS 6 Plant Damage State Attributes PDS Attributes Plant Damage States (PDS)

LOSP Non-LOSP Offsite Power Not Available Available Vessel Head Off Off Containment Integrity Open Open Drywell lntegrity Open Open Suppression Pool Makeup Not Available Not Available Containment Sprays Not Available Not Available Containment Oosure Possible? No Yes Injection Recovery Possible? . Yes No From Table 2.3-2 it can be seen that the main differences between the LOSP and nonLOSP PDSs are 1) the containment can be closed only in the nonLOSP PDS and

2) injection can be recovered only in the LOSP PDS. Because sequence frequencies are not available from the Level I screening analysis, the relative likelihood of the two PDSs is not available. The remaining analysis that is presented in this report is conditional on the occurrence of these PDSs.

• 2.4 Event Tree Analysis A simplified accident progression event tree (APET) was used in this analysis to delineate and quantify the likelihood of the possible accident progression paths. A source term is estimated for each path. Thus, only the most important events that

• 12

characterize the accident progression and source term are included in this event tree.

The selection of these events is based on 1) insights gain from the NUREG-1150 full power PRAs [1], 2) results from :MELCOR calculations specifically performed for this analysis and 3) the plant configuration during POS 6. The magnitude of the source term depends on the amount of radionuclides released from the fuel and the integrity of the containment. The former depends on the extent of core damage (i.e., is the core damage process arrested in the vessel) and for accidents that involve vessel failure, the characteristics of the ex-vessel release (i.e., core concrete interactions).

The accident progression event tree addresses three general time regimes: prior to core damage, during core damage, and' following vessel failure. In the first time regime the issue of containment closure is addressed. Injection recovery, core damage arrest, in-vessel steam explosions and early containment failure are all addressed in the second time regime. The characteristics of the interaction between the core debris release from the vessel and the reactor pedestal are addressed in the last time regime. The times associated with these time regimes are based on results from a series of MELCOR calculations that were performed to support this analysis. The timing of key events in the accident progression analysis is presented in Table 2.4-1.

Table 2.4-1 Accident Progression Timlng Calculation Timlng of Key Events from Initiation of Accident {hr.s) llme to Core Vessel Aux. Bldg Contain.

TAF Damage Failure Failure failure PRA MODEL INPUT PRA Model: Containment Open 13.0 18.3 25.4 21.1 Cnt Open(2)

PRA Model: Containment Falls 13.0 19.4 28.6 No Fall. (3) 30.

MELCOR RESULTS Base Case (BC)- No Aux Bldg 12.7 18.3 25.4 (1) (2)

BC w/ Small Aux. Bldg 13.0 18.8 24.5 21.6 (2)

BC w/ Big Aux. Bldg 13.0 18.8 28.6 28.6 (2)

BC w/ Containment Oosed 13.6 19.4 28.6 (1) 22 - 80.

BC Initiated 15 days after SD 19.7 28.3 39.8 (1) (2)

Notes:

1. Auxlllary building model not tnduded

2. Containment Is open during the accident

3. Containment failure bypasses the auxlllary building

4. MELCOR POS 6 BC Calculation:

- Accident Initiated 4 days after shutdown

- Containment Is open (l.e, equipment hatch and both personnel locks)

- Injection, shutdown cooling, and containment sprays are all unavailable

5. Core damage Is defined as the first gap release

6. TAF - Collapsed water level at the top of the active fuel

• 13

In this table both the times estimated with 1viELCOR and the times assumed in this PRA are presented. From this table it is apparent that the timing of these accidents are quite different from accidents initiated at full power. For example, it takes approximately 18 hours2.083333e-4 days <br />0.005 hours <br />2.97619e-5 weeks <br />6.849e-6 months <br /> to progress from the initiation of the accident to the onset of core damage. In comparison, a fast station blackout initiated from full power progresses to a similar point in approximately 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br />. Another notable entry in this table is the predicted time of auxiliary building failure. The building is predicted to overpressurize and fail from the accumulation of steam and noncondensibles during core damage. The exact timing of building failure depends on the volume assumed for the auxiliary building (i.e., the eommunication between rooms and floors) and the building failure pressure. For this abridged study, the auxiliary building is estimated to fail approximately half way through the core damage process.

Nine events are used to characterize the accident progression. Each event is presented in the form of a question. A graphical depiction of the APET is presented in Figure 2.4-

1. The nine events or questions are presented at the top of the figure and the answers to these questions determine the path that is taken through the tree. The first nine paths are associated with the LOSP PDS and the remaining 7 paths (i.e., paths 10 through 16) are associated with the NonLOSP PDS. The mean probability for each path is also presented in this figure. The path probabilities for each PDS sum to 1.0.

The nine questions and a brief description of each event is presented below.

1. Is the containment closed prior to core damage?

The containment equipment hatch has been removed prior to entry into POS 6.

For the LOSP PDS the lack of offsite ac power precludes containment closure prior to core damage. However, for the nonLOSP PDS it is possible that the plant personnel will close the containment after the initiation of the accident but prior to core damage. The containment can be closed if the operators recognize that a problem exists early in the accident and decide that containment closure would be prudent. Because it takes between 8 to 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br /> to completely close the hatch, it is necessary that the operators begin the tasks required to close the containment within the first few hours of the accident. The equipment hatch is a pressure seating hatch which requires the personnel closing the hatch to be in the containment. Thus, the environment in the containment during the boiloff is an important parameter that will affect the personnel's ability to close the containment. :MELCOR calculations performed for this analysis indicate that the temperatures in the containment during this phase of the accident will be high (i.e., range from 100 to 140 degrees F) but not so high that it would preclude the personnel from carrying out their tasks. It was also assumed that the radiological environment in the containment will not preclude the closure tasks from being performed. These assumption will have to be verified in future analysis. In this analysis it was assumed that the containment was habitable up until the time of core uncovery.

14

2. If th_e containment is closed prior to core damage, does it fail prior to vessel

• failure?

The Grand Gulf plant utilizes a Mark ID containment to house its BWR-6 reactor.

Although the containment volume (1.6 million cubic feet) is comparable to a large dry PWR containment, its structural capacity is relatively low. The design pressure is only 15 psig and the mean estimated failure pressure is 56 psig. Thus, if the containment is closed prior to core damage, the pressure rise from the accumulation of steam and noncondensibles must be controlled. Containment venting was not considered in this analysis as a means to control pressure because venting would still result in an open containment. Actions must also be take to prevent the combustion of large quantities of hydrogen. The effectiveness of the Hydrogen Ignition System (HIS) to control the accumulation of hydrogen in this plant configuration will have to investigated in future studies. In POS 6 the suppression pool is bypassed and, therefore, the steam and noncondensibles are released directly to the containment atmosphere. Furthermore, the containment sprays are not available. Thus, the containment will pressurize during the core damage process. The peak pressure during this phase of the accident depends on the steam generation rate, the condensation rate in the containment, and the presence and magnitude of hydrogen burns. :MELCOR calculations indicate that the containment pressure can exceed the lower range of the containment failure pressure distribution if a burst of steam occurs at the time of vessel failure or if discrete hydrogen burns (not diffusion flames) occur during the core damage phase of the accident. Thus, it is possible that for some accident scenarios the containment will fail early in the accident. If the containment does not fail early, calculations indicate that it will take several days to reach the mean estimated failure pressure of 56 psig. Therefore, it was assumed that if the containment does not fail early, it will not fail in the time frame of this analysis.

3. If the containment fails, is the failure in the form of a leak or rupture?

The failure size will determine how fast the radionuclides are released from the containment and the amount of radionuclides deposited within the containment.

4. Is the auxiliary building bypassed?

This question distinguishes the accidents in which the releases pass through the auxiliary building from those accidents which result in a release from the containment directly into the environment. Accidents in which the containment equipment hatch is off will result in a release that passes through the auxiliary building; accidents in which the containment' fails bypass the auxiliary building.

Based on previous structural analysis of the Grand Gulf containment, it was concluded that the most likely location for failure is the region near the junction of the dome and the cylindrical wall. A failure in this location will result in a

• . 15

release to the enclosure building which surrounds the containment dome. The

• enclosure building has a virtually no pressure retaining capability and is essentially isolated from the auxiliary building. Therefore, it is assumed that following containment failure, the release goes directly from the containment to the environment. The release path is important because it will effect the amount of mitigation that the release experiences before entering the environment. The auxiliary building encompasses a very large volume and, therefore, acts as a hold up volume for the radionuclides which allows time for the radionuclides to deposit on surfaces within the building. The auxiliary building is predicted to overpressurize and fail froIIJ the accumulation of steam and noncondensibles during core damage. The exact timing of building failure depends on the volume assumed for the auxiliary building (i.e., the communication between rooms and floors) and the building failure pressure. For this abridged study, the auxiliary building is estimated to fail approximately half way through the core damage process.

5. Is injection recovered prior to vessel failure?

This question is used to identify those accidents in which injection is restored to the vessel during the core damage process. The recovery of injection allows for the possibility that the core damage process will be arrested in the vessel (i.e.,

prevent vessel failure). Injection can only be recovered for the LOSP PDS. The probability that injection is recovered is based on the probability that ac power is

• recovered during core damage.

6. If injection is recovered, when is it recovered?

The timing of injection recovery during core damage affects the likelihood that the core damage process will be arrested before the vessel fails. For this analysis, the in-vessel phase of the accident (i.e., core damage) has been divided into the following three time regimes: very early, early, and late. The "very early" time regime ranges from the initiation of core damage to the onset of autocatalytic oxidation. If injection is recovered during this phase of the accident the core damage process will be arrested in the vessel and the releases will be limited to the inventory in the gap. The "early" time regime ranges from onset of autocatalytic oxidation to 30 percent core damage. Based on extrapolation of analysis performed in NUREG-1150, if injection is recovered before 30% for the core has been damage it is very likely that the core damage process can be arrested. Because :MELCOR calculations indicate that core damage progresses rapidly from 30% to full core damage, the "late" time regime is defined as 30% core damage to vessel failure.

Recovery of injection during this phase of the accident will not prevent vessel failure. The time windows for each of these time regimes is based on results from

MELCOR calculations. The possibility of the reactor going critical following the restoration of injection was not address in this abridged analysis .

• 16

'\

7. Does an in-vessel steam explosion occur during core damage?

• In-vessel steam explosions are treated in a very limited fashion in this abridged analysis. A primary motivation for including this question in the APET is to highlight the fact that in-vessel steam explosions are possible. The effect of the steam explosion on the accident progression can be quite different from in-vessel steam explosions that occur at full power because the steam and radionuclides that are generated during this event are released directly to the containment atmosphere. In this analysis the treatment of in-vessel steam explosions was limited to the estimation of tJ1e source term that is associated with the debris that participates in the steam explosions. Neither the pressure loading from in-vessel steam explosions nor the relocation of intact fuel from the steam explosion was addressed in this study. Both issues were beyond the scope of this abridged study. Ex-vessel steam explosions were not considered in this analysis because the pedestal cavity below the vessel will be essentially dry at the time of vessel failure.

8. Is the core damage process arrested in the vessel?

This question addresses the coolability of the core debris following injection recovery. If the core damage process is arrested before the vessel fails, the core debris will remain in the vessel and core-concrete interactions (CO) will be prevented. Because only a portion of the core is damage and CO is prevented, the source term associated with recovered accidents is typically less than the source

• term associated with full core damage accidents. If injection is not restored during core damage the accident always progresses to vessel failure and the core debris relocates to the pedestal cavity below the vessel. The likelihood that the core damage process is arrested before vessel failure depends on when injection is restored during the core damage process (see question 6). If injection is restored during either the "very early" or "early" time regimes, analysis indicates that it is very likely the core damage process will be arrested. If, on the other hand, injection is not restore until the "late" time regime, it is very likely that the vessel will fail and the core debris will relocate to the pedestal cavity.

9. Do core-concrete interactions occur following vessel failure?

Core-concrete interactions consist of the thermal and chemical interactions between the core debris and the concrete pedestal. During this process the concrete is eroded and gases and radionuclides are released from the core/concrete mixture.

For the accidents analyzed in this study, the vessel will fail and the core debris will enter the cavity if 1) injection is not restore to vessel during core damage 2) injection is restore during the "late" time regime. The presence of water can effect CCI in two different ways. First, water can quench the debris and prevent CCI.

Second, if the debris is not quenched, the overlying pool of water will retain some of the radionuclides released during CO and thus, tend to mitigate the release.

17

Thus, for the accidents in which injection is restore but the vessel still fails, there

• is some probability that the core debris will be quenched and CO will be prevented. The probability of this occurring is based on information from the NUREG-1150 study [6]. If injection is not restored during core damage, CCI will always proceed in a dry cavity.

From inspection of Figure 2.4-1 it can be seen that there are several important clifferences between the LOSP PDS and the nonLOSP PDS. In the LOSP PDS injection can be recovered allowing for the possibility to arrest the core damage process in the vessel. If the vessel does fail, it is.,still possible to quench the core debris in the cavity (i.e., no CO). Thus, in many of the LOSP accidents the ex-vessel radionuclide release is prevented. The containment, however, cannot be closed in the LOSP PDS and, therefore, the releases always pass into the auxiliary building and then out into the environment. In the nonLOSP PDS, on the other hand, the containment can be closed, however, injection cannot be recovered. Thus, all of the nonLOSP accidents identified in the APET progress to full core damage, vessel failure, and involve CO. In some of the scenarios the containment is closed. However, because containment cooling is not available (i.e., containment sprays) even if the containment was closed it is possible that the it will fail early in the accident from pressure transients associated with events accompanying vessel failure and hydrogen combustion. Based on information from NUREG-1150, it is expected that the containment will fail above the auxiliary building roof. Thus, the releases from the containment will enter the environment without first going through the auxiliary building. Because so many of the mitigative features of the

• plant are bypassed in this POS (e.g., suppression pool, containment sprays, containment), the auxiliary building plays an important role in reducing the amount of radionuclide material that is released to the environment. Thus, for the nonLOSP accidents there are two extremes 1) if the containment was closed and remains intact, the releases to the environment are expected to be very small and 2) if the containment fails, the releases to the environment are expected to be quite large because all of the accidents involve full core damage and CO and the releases bypass the auxiliary building. The scenarios in which the containment was not closed are very similar to the LOSP accidents in which injection was not recovered .

• 18

• PDS Cnt Cloeed1 Don Cnt Fall LHk or Rupture?

Bldg Inject lo..,

llecover*d r Recovery Tl*lne lnVH StllExp?

CD Arrnted Node CCI No.

WTCOE Early? Bypened !

0.12 2 0.36 3 0.06 4 0.02 5 0.09 6 0.003 1 0.02 8 0.28 9 0.05 10 0.17 11 0.03 12 0.17 13 0.03 14 0.40 15 0.18 16 0.03 Figure 2.4-1. Grand Gulf POS 6 abridged accident progression event tree (APET) 19

3. SOURCE TERM ANALYSIS

• 3.1 Approach A source term is estimated for each accident progression path identified in the APET (see Figure 2.4-1). The simple parametric source term approach that was used in NUREG-1150 to estimate source terms is used in this study. The parametric source approach is used because 1) information from a wide variety of sources can be used in the model, 2) it is easily incorporated into uncertainty analysis, and 3) thousands of source terms can be estimated )Vith this model in a very efficient manner. The

• parametric source term code GGSOR that was developed in NUREG-1150 [6] was modified for this analysis. The modified parametric code is called GGLPSOR.

• Modifications were made to the code to incorporate the unique plant configuration associated with accidents initiated in POS 6. Wherever possible, data from NUREG-1150 was used to quantify the model. Results from :MELCOR were compared with both the input distributions and the final source terms to verify that distributions developed for full power accidents could be applied to shutdown accidents.

Because the estimation of the source terms is a critical component of this study, a panel of experts knowledgeable in source term phenomena and the treatment of this phenomena in PRAs was formed to support this study's source term effort. The members of the panel included: John E. Kelly (SNL), Hossein P. Nourbakhsh (BNL),

Dana A. Powers (SNL), and Trevor Pratt (BNL). The role of the Source Term Advisory

• Group was to 1) provide guidance on the identification of phenomena that may be important to the formation of the source term during these modes of operation, 2) assess the adequacy, relative to the study's objectives and scope, of the assumptions, methods and data used in this study.

A limited uncertainty analysis was performed in this section of the analysis. For each accident progression path, the model was repeatedly exercised with different combinations of selected input variables. The distributions for these input variables were obtained, when applicable, from NUREG-1150.

In the next section, key attributes of the parametric source term model are discussed.

3.2 Description of Parametric Model As was mentioned in the previous section, the parametric source term model GGSOR that was developed in NUREG-1150 was modified for this analysis. GGSOR had to be modified because of the unique features of POS 6 that have a strong impact on the source term. First, in POS 6 both the drywell head and the vessel head have been removed and the steam lines have been plugged. Thus, during the core damage process radionuclides released from the core debris will bypass the suppression pool and directly enter the containment. Furthermore, because most of the internal

• 20

structures above the core (e.g., steam dryers and separators) have been removed and

• the steam lines are plugged, there is very little deposition of radionuclides in the vessel.

Thus, the mitigative features of both the vessel and the suppression pool, which are present in many full power accident scenarios, are absent in this POS. For scenarios in which the containment hatch is open the residence time of the radionuclides in the containment atmosphere is fairly short and, thus, there will be limited deposition (i.e.,

from gravitational settling) of radionuclides in the containment. In this POS the drywell is open to the containment (i.e., the drywell hatch is open) and, therefore, an ex-vessel releases will also bypass the suppression pool. For these accidents, the containment sprays are not available and cannot be used to scrub the releases. Thus, the mitigative features of the vessel, suppression pool, containment sprays, and possibly the containment are bypassed or unavailable during this POS.

For scenarios with the containment open the only major mitigative feature of the plant is the auxiliary building. The auxiliary building encompasses a very large volume and, therefore, acts as a hold up volume for the radionuclides which allows time for the radionuclides to deposit on surfaces within the building. The auxiliary building can play an important role in POS 6 because so many of the other mitigative features of the plant are absent and the characteristics of the radionuclide transport to the auxiliary building are different from the transport associated with full power accidents. In full power accidents the containment pressurizes to the ultimate failure pressure and then blows down into the auxiliary/reactor building (i.e., Peach Bottom and LaSalle).

Following containment failure the auxiliary building rapidly pressurizes and fails (the failure pressure of the auxiliary building is only a few psi). Thus, the releases are swept through the auxiliary building fairly rapidly. In the POS 6 accident scenarios the steam and radionuclides are released to the auxiliary building much more slowly allowing more time for condensation and deposition. The scenarios that involve containment closure followed by containment failure will not benefit from the auxiliary building because the containment failure location is assumed to be above the roof the auxiliary building. Thus, the releases will bypass the auxiliary building resulting in essentially an unmitigated release.

Neither the normal ventilation system nor the standby gas treatment system (SBGT) were modeled in this analysis. The filters and charcoal beds in the SBGT system could act to mitigate the release or at least delay the release of radionuclides. Before credit can be given to this system, the capacity of the system and the performance of the filters under severe accident conditions will have to be addressed. The analysis of this system was beyond the scope of this study.

3.3 Results A source term is estimated for each path through the APET. In addition, because an uncertainty analysis was performed, a distribution of sources is available for each path.

For the sake of brevity, only the mean source terms, expressed as fractions of the core

• . 21

inventory, that enter the environment are presented in Table 3.3-1. When reviewing this table, it must be remembered that the initial inventory of radionuclides four days after shutdown is different from the inventory typical of full power accidents.

Inspection of Table 3.3-1 confirms that many of the releases are essentially unmitigated and, therefore, are quite large. Table 3.3-1 also highlights some of the differences between the various accident scenarios (i.e., paths). Paths 1 through 3 correspond to accidents in which injection is recovered early in the accident and the core damage process is arrested in the vessel. Thus, because only a portion of the core is damaged and there are no ex-vessel releases (i.e., no CO), the source terms associated with these accidents are relatively small compared to the other source terms presented in this table. The notable exception is Path 14 which corresponds to the scenario in which the containment is closed prior to core damage and remains intact throughout the accident.

Because the containment remains intact, only nominal leakage occurs and the resulting source term is quite small. Paths 4 through 9, on the other hand, correspond to full core damage accidents that have the containment open to the auxiliary building. The source terms associated with Paths 4 and 6 tend to be lower than the other full core damage source terms because the core debris is quenched in the pedestal cavity and, therefore, there are no releases associated with CO. This difference is fairly minor, however, and the fact still remains that these are large source terms. Paths 10 through 13 are nonLOSP accidents in which the containment fails around the time of vessel failure. All of t~ese accidents progress to full core damage and CCI. The containment fails via a leak in Paths 10 and 11; the containment ruptures in Paths 12 and 13. In all four of these scenarios the containment fails directly to the environment (i.e., the auxiliary building is bypassed). The source terms associated with the leak failure mode are similar to the source terms when the release passes through the auxiliary building.

In the leakage cases the radionuclides are held up in the containment for a period of time th~s allowing a fraction of the radionuclides to settle out of the containment atmosphere. For the rupture cases, however, the containment quickly depressurizes following containment failure and considerably less deposition occurs. Thus, the source terms associated with the rupture cases are quite large. Paths 15 and 16 correspond to the nonLOSP cases where the containment is not closed prior to core damage and the radionuclides pass through the auxiliary building. These source terms are essentially the same as the LOSP full core damage source terms .

• 22

Table 3.3-1 Mean Source Terms for Acddent Progression Paths (Total Release)

[:] NG I Cs Radionuclide Release Oasses Te Sr Ru la Ce Ba 1W Timlng of Release (hr.s) 11 D11 DT2 LOSPPDS 1 0.015 0.0C~ 5.9E-3 1.2E-5 0.0 0.0 0.0 0.0 1.2E-7 16.3 21.1 24.0 0.0 2 0.072 0.012 0.011 6.3E-3 2.lE-3 3.3E-4 1.4E-4 6.7E-4 2.2E-3 16.3 21.1 4.3 0.0 3 0.072 0.012 0.011 6.3E-3 2.tE-3 3.3E-4 1.4E-4 6.7E-4 2.2E-3 16.3 21.1 4.3 0.0 4 0.79 0.17 0.15 0.085 0.027 0.012 3.0E-3 8.7E-3 0.033 16.3 21.1 4.3 10.0 5 1.0 0.25 0.19 0.11 0.042 0.012 4.0E-3 0.011 0.047 16.3 21.1 4.3 10.0 6 0.74 0,15 0.13 0.075 0.022 4.9E-3 1.5E-3 7.lE-3 0.026 16.3 21.1 4.3 10.0 7 1.0 0.25 0.18 0.11 0.041 4.9E-3 2.7E-3 9.6E-3 0.042 16.3 ' 21.1 4;3 10.0 8 1.0 0.25 0.25 0.16 0.08 0.012 6.9E-3 0.012 0.084 16.3 21.1 4.3 10.0 9 1.0 0.25 0.25 0.17 0.088 5.4E-3 6.3E-3 0.011 0.089 16.3 21.1 4.3 10.0 nonLOSPPDS 10 1.0 0.27 0.27 0.18 0.094 0.013 6.8E-3 0.011 0.083 17.4 30.0 2.0 10.0 11 1.0 0.28 0.28 0.19 0.10 6.lE-3 6.2E-3 0.011 0.086 17.4 30.0 2.0 10.0 12 1.0 0.62 0.63 0.40 0.22 0.029 0.016 0.027 0.19 17.4 30.0 0.05 10.0 13 1.0 0.62 0.63 0.43 0.24 0.013 0.014 0.025 0.20 17.4 30.0 0.05 10.0 14 5.0E-3 4.lE-7 4.lE-7 2.9E-7 1.4E-7 9.4E-9 9.7E-9 1.9E-8 UE-7 16.3 21.1 4.3 10.0 15 1.0 0.25 0.25 0.16 0.08 0.012 6.9E-3 0.012 8.4E-2 16.3 21.1 4.3 10.0 16 1.0 0.25 0.25 0.17 0.088 5.4E-3 6.3E-3 0.011 0.089 16.3 21.1 4.3 10.0 Notes:

1. 1W - Warning Time

2. 11 - llmlng of first release

3. 011- Duration of first release

4. DT2- Duration of second release (start Immediately after first release ends) 23

4. CONSEQUENCE ANALYSIS

• The consequences of a severe accident during POS 6 were calculated as part of the abridged study. As is typically done, the offsite consequences were estimated. The onsite doses were also estimated, which is not typically done.

There are other differences between this analysis and those previously performed for full power accidents. The first and most obvious is that the radionuclides in the fuel have had at least four days to decay resulting in a different inventory than that present at shutdown. ORIGEN-PC was used to calculate the inventory in three different fuel assemblies, one which had been irradiated for three fuel cycles, one which had been irradiated for two fuel cycles, and one which had been irradiated for one fuel cycle.

All fuel assemblies were then allowed to decay for four days. Based on information from plant personnel, a fuel cycle consisted of 540 days of irradiation and 55 days of decay. The inventory for the whole core four days after shutdown was then summed.

This inventory, which was reduced to include only the sixty radionuclides currently available in the MACCS code, was then used as the basis for both the onsite and offsite consequence calculations. This inventory, which does not include short-lived radionuclides, is appropriate for both the onsite and offsite analyses since the reactor has been in shutdown for at least four days at the beginning of the accident thus allowing decay of the short-lived radionuclides.

The following sections detail the methodology and results for the onsite consequences, both in the buildings and in the parking lot, and the offsite consequences.

4.1 Onsite Consequences Onsite consequences have seldom been considered in the analysis of severe accidents at nuclear power plants. During shutdown there will be hundreds of onsite personnel and, thus, onsite consequences could be very important. For this reason a methodology for estimating the potential doses to onsite personnel had to be developed as part of this study. The primary simplifying assumption of the analysis was that radioactive decay was neglected during the exposure time. This assumption is justified by the fact that the accident under analysis typically occurs no earlier than four days after shutdown by which time the decay heat curve is fairly flat. Other assumptions were employed in the two aspects of the onsite consequences: (1) in building doses and (2) parking lot doses. The methodology, assumptions, and results of the analyses are discussed in the following two sections.

4.1.1 Building Doses The onsite consequences for POS 6 were estimated based on the source terms to both the containment and the auxiliary building that were determined in the parametric source term code, GGLPSOR. However, since GGLPSOR calculates integral releases, 24

the time dependence of the two release segments of the source terms was determined

• from MELCOR calculations. Three different sets of transit times were used based on the status of the containment. The fust set of transit times was used if the containment was open to the auxiliary building at the time of the accident. The transit times through both buildings were based on a MELCOR calculation modeling this scenario.

The residence time of the radioactive material in each building was directly proportional to the volume of that building. The second set of transit times was used if the containment ruptured directly to the environment. For this case, the same transit times were used as in the previous scenario, however, the residence time in the auxiliary building was set to zero. In other,, words, the amount of time the material spent in the containment was the same for both of these scenarios but in the latter scenario, the material did not pass through the auxiliary building. The third set of transit times was used if the containment leaked directly to the environment. In this case, the transit time for the first release was increased by two hours, and again the residence time in the auxiliary building was set to zero. The transit times used under the various conditions are summarized in Table 4.1.1-1.

Table 4.1.1-1. Transit times through the containment and auxiliary building for Grand Gull POS 6.

Accident ContaininentTransit Containment Transit Awclllary Building Awclllary Building Progression Path Time: Fust Segment Time: Second Transit Time: First TransltTime:Second Number (s) Segment (s) Segment (s) Segment (s)

LOSP PDS

• Paths 1-9 Path 10 Path 11 12141.0 1469'7.0 14697.0 2811.6 2811.6 2811.6 nonLOSPPDS 22059.0 0.0 0.0 5108.4 0.0 0.0 Path 12 12141.0 2811.6 0.0 0.0 Path 13 12141.0 2811.6 0.0 0.0 Path 14 0.0 0.0 0.0 0.0 Path 15 12141.0 2811.6 22059.0 5108.4 Path 16 12141.0 2811.6 22059.0 5108.4 To estimate the doses in the buildings, the average release fraction of each chemical group was determined for each building. The integrated concentration in the buildings of each radionuclide was then based on the average release fraction of its chemical group and the amount of time spent in that' building. Using the integrated concentration for each radionuclide, the immersion and 50 year committed inhalation dose were calculated over the entire exposure time. In addition, the immersion and 50 year committed inhalation dose were calculated for the first 30 minutes of exposure .

• 25

These doses should be viewed with caution since the integrated concentration in the

• building was based on an average concentration in the building and therefore the time dependence of the dose is not well represented. The final result estimated in the buildings was a dose rate. These results should also be viewed with caution since they are also based on average concentrations in the building. In addition, the dose rates are calculated by dividing the total dose during a release segment by the transit time through the building. This results in a very conservative estimate of the inhalation dose rate. The mean dose due to the entire release, the first 30 minutes of exposure, and the mean dose rates during the first and second release segments in the containment are shown in Table.. 4.1.1-2 for each of the paths through the APET.

Similar estimates are shown in Table 4.1.1-3 for the auxiliary building.

Table 4.1.1-2. Grand Gull POS 6 mean containment doses and dose rates.

Accident Path Consequence Measure Progression Path Conditional Number Probability Total Dose 30 minute Dose Dose Rate First Dose Rate Second (rem) (rem) Segment (rem/hr) Segment (rem/hr)

LOSPPDS Path 1 0.10 1.81E+6 2.69E+5 5.38E+5 0.0 Path 2 0.48 4.27E+7 6.35E+6 1.27E+7 0.0 Path 3 0.08 4.27E+7 6.35E+6 1.27E+7 0.0 Path 4 0.02 5.38E+6 7.95E+7 1.59E+8 0.0 Path 5 0.09 5.69E+8 7.95E+7 1.59E+8 4.04E+7 Path 6 0.003 4.27E+8 6.35E+7 1.27E+8 0.0 Path 7 0.01 4.66E+8 6.35E+7 1.27E+8 5.05E+7 Path 8 0.18 6.16E+8 7.95E+7 1.59E+8 1.01E+8 Path 9 0.03 5.25E+8 6.35E+7 1.27E+8 1.26E+8 nonLOSPPDS Path 10 0.16 5.78E+8 6.35E+7 1.27E+8 7.74E+7 Path 11 0.02 4.88E+8 5.05E+7 1.01E+8 9.68E+7 Path 12 0.16 6.16E+8 7.95E+7 1.59E+8 1.01E+8 Path 13 0.02 5.25E+8 6.35E+7 1.27E+8 1.26E+8 Path 14 1 0.37 0.0 0.0 0.0 0.0 Path 15 0.23 6.16E+8 7.95E+7 1.59E+8 1.01E+8 Path 16 0.04 5.25E+8 6.35E+7 1.27E+8 1.26E+8 1 Building doses were not calculated since the containment ls not open .

• 26

Table 4.1.1-3. Grand Gulf POS 6 mean auxiliary building doses and dose rates .

• Accident Progression Path Number Path Conditional Probability Total Dose (rem)

Consequence Measure 30 minute Dose (rem)

Dose Rate First Segment (rem/hr)

LOSPPDS Dose Rate Second Segment (rem/hr)

Path 1 0.10 9.41E+5 7.70E+4 1.54E+5 0.0 Path 2 0.48 2.13E+7 1.74E+6 3.47E+6 0.0 Path 3 0.08 2.J3E+7 1.74E+6 3_.47E+6 0.0 Path 4 0.02 2.80E+8 2.28E+7 4.56E+7 0.0 Path 5 0.09 2.98E+8 2.28E+7 4.56E+7 1.31E+7 Path 6 0.003 2.20E+8 1.79E+7 3.59E+7 0.0 Path 7 0.01 2.43E+8 l.79E+7 3.59E+7 l.64E+7 Path 8 0.18 3.23E+8 2.28E+7 4.56E+7 3.06E+7 Path 9 0.03 2.74E+8 l.79E+7 3.59E+7 3.83E+7 nonLOSP PDS Path 10 0.16 0.0 0.0 0.0 0.0 Path 11 0.02 0.0 0.0 0.0 0.0 Path 12 0.16 0.0 0.0 0.0 0.0 Path 13 0.02 0.0 0.0 0.0 0.0 Path 14 1 0.37 . 0.0 0.0 0.0 0.0 Path 15 0.23 3.23E+8 2.28E+7 4.56E+7 3.06E+7 Path 16 0.04 2.74E+8 1.79E+7 3.59E+7 3.83E+7 1 Building doses were not calculated since the containment Is not open.

To illustrate the uncertainty in the dose rate in the containment and the auxiliary building due to the source term uncertainty, the 5th, 50th, and 95th percentile dose rates as well as the mean dose rate for two pathways through the APET are shown in Figure 4.1.1-1. The first of these paths represents a scenario in which injection is recovered very early in the accident thus arresting core damage. Note that in the recovered accident, CCI does not occur therefore the source term consists of only one segment and only one dose rate was calculated. The second of the bins represents a scenario in which full core damage occurs .

• . 27

• 1010 Building Dose Rates for Recovered Accident (Path 1) and Full Core Damage Accident (Path 8)

Full Core Damage Accidenl Cnlmenl Aux.Bide 10i ht Relea.., 2nd Releue 1st Relea,oe 2nd Releue i

115th 108 Mean

'i:" r.otb 1

.c:

'c::e I)

JQ7 I) a, Recovered Accident Cntment Awe.Bide 5th c:: 10&

I I I) 0 Q

10~

104 103

• Figure 4.1.1-1. Containment and auxiliary building dose rates for selected paths

4.1.2 Parking

Lot Doses The dose due to immersion and inhalation was also estimated for several distances from the reactor. The source terms were obtained from the parametric source term code, GGLPSOR. In contrast to the building doses, the timing of the source terms was taken directly from GGLPSOR. For comparative purposes, three different wake effect models were used to estimate the relative concentrations downwind of the reactor.

These models were developed by Ramsdell [7], Wilson [8], and the NRC (9]. For simplicity, the directional dependence of the weather was ignored and doses were calculated for several distances from the reactor. The weather used in each of the wake effect models was chosen to represent conservative values for the model. In the case of the Ramsdell model the relative concentration is somewhat proportional to the wind speed and the stability class. For this reason* the highest wind speed and the corresponding stability class in a year of weather data at Grand Gulf was chosen as input to this model.- In addition, the relative concentration is predicted to be somewhat inversely proportional to the area of the building, therefore, the minimum area was

• 28

utilized. In the case of the Wilson and NRC models the relative concentration is

• predicted to be inversely proportional to the wind speed. Therefore, a wind speed of 1 mis and a stability class of F were used in these models. Using the integrated air concentrations for each building wake effect model, the dose and dose rate due to immersion and inhalation for the entire source term was determined for each of the unique accident progression paths. As with the building dose rates, the dose rates in the parking lot are very conservative since the inhalation dose rate was determined by dividing the 50 year committed dose by the exposure time. The dose due to 30 minutes of exposure was also estimated. Table 4.1.2-1 contains the mean total dose, 30 minute dose, and dose rates for each segment of the release based on the Ramsdell building wake effect model at 100 m from the reactor. Similar estimates of the mean doses and dose rates at 100 m based on the Wilson model is shown in Table 4.1.2-2.

Table 4.1.2-1. Grand Gulf POS 6 mean doses and dose rates at 100 m based on the Ramsdell building wake effect model.

Accident Path Consequence Measure Progression Path Conditional Number Probability Total Dose 30 minute Dose Dose Rate First Dose Rate Second (rem) (rem) Segment (rem/hr) Segment (rem/hr)

LOSPPDS Path 1 0.10 4.23E+2 8.80 17.6 0.0 Path 2 0.48 9.42E+3 1.09E+3 2.19E+3 0.0 Path 3 0.08 9.42E+3 1.09E+3 2.19E+3 0.0 Path 4 0.02 1.32E+5 1.53E+4 3.06E+4 0.0 Path 5 0.09 1.73E+5 1.53E+4 3.06E+4 4.08E+3 Path 6 0.003 1.04E+5 1.21E+4 2.43E+4 0.0 Path 7 0.01 1.SSE+S 1.21E+4 2.43E+4 5.10E+3 Path 8 0.18 2.16E+5 1.53E+4 3.06E+4 8.44E+3 Path 9 0.03 2.10E+5 1.21E+4 2.43E+4 1.05E+4 nontOSPPDS Path 10 0.16 2.26E+5 3.41E+4 6.83E+4 8.94E+3 Path 11 0.02 2.19E+5 2.68E+4 5.37E+4 L12E+4 Path 12 0.16 5.24E+5 3.20E+5 6.22E+6 2.13E+4 Path 13 0.02 5.lOE+S 2.56E+5 4.88E+6 2.66E+4 Path 14 0.37 0.899 5.25E-2 0.105 4.SOE-2 Path 15 0.23 2.16E+5 1.53E+4 3.06E+4 8.44E+3 Path 16 0.04 2.10E+5 1.21E+4 2.43E+4 1.ffiE+4

• 29

Table 4.1.2-2. Grand Gulf POS 6 mean doses and dose rates at 100 m based on the Wilson building wake effect model.

Accident Path Consequence Measure Progression Path Conditional Number Probability Total Dose 30 minute Dose Dose Rate Arst Dose Rate Second (rem) (rem) Segment (remlhr) Segment (remlhr)

LOSPPDS Path 1 0.10 9.48E+3 1.97E+2 3.95E+2 0.0 Path 2 0.48 2.1,lE+S 2.45E+4 4.91E+4 0.0 Path 3 0.08 2.llE+S 2.45E+4 4.91E+4 0.0 Path4 0.02 2.95E+6 3.43E+5 6.86E+5 0.0 Path 5 0.09 3.56E+6 3.43E+5 6.86E+5 9.14E+4 Path 6 0.003 2.34E+6 2.72E+5 5.44E+5 0.0 Path 7 0.01 3.48E+6 2.72E+5 5.44E+5 1.14E+5 Path 8 0.18 4.84E+6 3.43E+5 6.86E+5 1.89E+5 Path 9 0.03 4.70E+6 2.72E+5 5.44E+5 2.36E+5 nonLOSPPDS Path 10 0.16 5.06E+6 7.65E+5 1.53E+6 2.00E+S Path 11 0.02 4.91E+6 6.00E+S 1.20E+6 2.50E+5 Path 12 0.16 1.17E+7 7.16E+6 1.39E+8 4.77E+5 Path 13 0.02 1.14E+7 5.72E+6 1.09E+8 5.96E+5 Path 14 0.37 20.1 1.17 2.34 1.01 Path 15 0.23 4.84E+6 3.43E+5 6.86E+5 1.89E+5 Path 16 0.04 4.70E+6 2.72E+5 5.44E+5 2.36E+5 Figure 4.1.2-1 contains the 5th, 50th, 95th percentile as well as the mean parking lot dose rates for the first release for both the Ramsdell and Wilson/Regulatory Guide models for distances of 10 - 500 m from the reactor. A similar plot for the second release is shown in Figure 4.1.2-2. The uncertainty in both the building wake effect models and the source term is shown by the wide range of dose rates at each distance.

4.2 Offsite Consequences The MACCS code was used to estimate the consequences to the general public.

MACCS models the transport and dispersion of plumes of radioactive material released from the plant. As the plumes travel through the atmosphere, material is deposited on the ground. Several of the pathways through which the general population can be exposed are considered. Emergency response and protective action guides are also considered as means to mitigate the extent of the public exposure .

30

• 10*

,oa FintSecment(Path 8) rlli Mean

• A Ram*dell Wilson 50th 107

~

.c

'e.

lOI 1*lli ,I a:

C

.0 106 104 l<>'

I I

,oi lOII 6011 10011 250M 500M Di*tance from Reactor Figure 4.1.2-1. Parking Lot dose rates for Ramsdell and Wilson/Regulatory Guide models for distances from 10 - 500 meters from the reactor: First Release Segment

• (()II 107 Second Secmeot(Path I!)

95th Mean

!,()th

• " Ram*dell lfil*oD IOI

.c 5th

'a:.e 106

-a:.,*. 104 0 l<>'

C

,oi 101 lOO lOll 50M lOOM -Z50M 500M Di*tance from Reactor Figure 4.1.2-2. Parking Lot dose rates for Ramsdell and Walson/Regulatory Guide models for distances from 10 - 500 meters from the reactor: Second Release Segment

• 31

The input used in this study is identical to that used for Grand Gulf in the NUREG-1150 study with the exception of the core inventory for which the inventory four days after shutdown was used and the source terms which resulted from GGLPSOR. The emergency response assumptions were not changed for this analysis.

Table 4.2-1 contains the estimated mean number of early fatalities, latent cancers, 50 mile population dose, and 1000 mile population dose for the sixteen paths through the APET along with the conditional probability of that path. The mean number of early fatalities ranged from O to 3. 9 x 10-2 while the mean number of latent cancers ranged from O to 1940. ,,

Table 4.2-1 Grand Gulf POS 6 Offslte Mean Consequences Accident Path Consequence Measure Progression Conditional Path Number Proba bill ty Early Total Latent 50 mile lOOJ Mile Fatalities Cancers Population Dose 1 Pop. Dose 1 LOSPPDS Path 1 2 0.10 0.00 0.00 0.00 0.00 Path 2 0.48 1.3E-5 102 77,000 591,00J Path 3 0.08 1.3E-5 102 77,000 591,000

• Path 4 Path 5 Path 6 Path 7 0.02 0.09 0.003 0.01 4.SE-3 4.SE-3 4.0E-3 4.0E-3 684 984 588 94{)

330,000 496,000 293,000 479,000 4,010,000 5,800,000 3,450,000 5,560,000 Path 8 0.18 5.2E-3 1270 652,000 7,480,000 Path 9 0.03 4.7E-3 1260 662,000 7,460,000 nonLOSP PDS Path 10 0.16 8.9E-3 1190 624,000 7,090,000 Path 11 0.02 9.3E-3 1200 640,000 7,130,000 Path 12 0.16 3.7E-2 1920 939,000 11,300,000 Path 13 0.02 3.9E-2 194{) 966,000 11,500,000 Path 14 2 0.37 0.00 0.00 0.00 0.00 Path 15 0.23 5.2E-3 1270 652,000 7,480,000 Path 16 0.04 4.7E-3 12o0 662,000 7,4o0,000 Table Notes:

1. Dose ls In Person Rem

2. Offsite consequences were not evaluated for these paths because the offsite consequences associated with these paths were assessed to be negligible .

• 32

• 5. INTEGRATED RESULTS CONDillONAL ON CORE DAMAGE In the previous section the consequences associated with individual accident progression paths were presented. In this section the offsite consequences conditional on the occurrence of the LOSP PDS and the nonLOSP PDS are presented and are compared to full power PRA results extracted from the Grand Gulf analysis presented in NUREG-1150. Onsite consequences were not evaluated in NUREG-1150 and, therefore, and analogous comparison is not provided for onsite consequences.

The consequences for a given PDS are calculated by taking a weighted average of the consequences for the individual paths. The weighted average is based on the conditional probability of each path. That is, the consequence for each path is multiplied by the probability of the path. The PDS consequence is the sum of all of the "weighted" path consequences for the given PDS.

The offsite consequence distributions associated with the LOSP and nonLOSP PDSs are presented in Figure 5.1. Because a relatively small LHS sample was used in the evaluation of offsite consequences, the presentation of exact quantiles (i.e., 95th) is not appropriate. Instead of quantiles, the high, low, median, and mean values are presented in this figure. From this figure it can be seen that the consequences associated with the nonLOSP PDS tend to be higher than the consequences associated with the LOSP PDS. This stems from the assumption that injection cannot be

• recovered in the nonLOSP PDS and, therefore, all of these accidents proceed to full core damage and CCI. Although the probability that the containment is closed during this PDS is significant, the lack of a means to control the containment pressure results in a significant probability of early containment failure. Containment failure bypasses the auxiliary building and results in essentially an unmitigated release.

Also presented in Figure 5.1 are the conditional consequences from the Grand Gulf full power PRA. The full power results are for internal events and are "averaged" over all of the accidents analyzed in the study. In addiqon to the global consequences, the mean consequences associated with a selected full power accident are also presented (i.e., triangle on the full power distribution). This selected accident is a fast station blackout that progresses to full core damage. The containment is ruptured during core

• damage; the containment sprays are unavailable throughout the accident. Thus, this accident is similar to the accidents analyzed in this abridged study in that many of the mitigative features of the plant are unavailable (i.e., the containment fails and the sprays are unavailable). In this full power accident, however, the in-vessel releases are typically scrubbed by the suppression pool. From Figure 5.1 it can be seen that the number of early fatalities associated with POS 6 are very similar to the number of early fatalities associated with full power accidents. This may seem somewhat surprising at first because the inventory of radionuclides important to early fatalities during POS 6 is less than the inventory at full power. However, this difference is compensated by the lack of mitigative features in POS 6. In full power accidents a considerable fraction

• 33

of these radionuclides are retained in the suppression pool. Thus, in POS 6 the

• inventory has been reduced but because of the lack of mitigative features, a significant amount of the radionuclides are released to the environment. In the full power accidents, on the other hand, there is a large inventory, however, mitigative features of the plant limited the size of the release. The net effect is that the number of early*

fatalities is roughly the same. *1ne number of latent cancers associated with POS 6 accidents is greater than the number of latent cancers associated with full power accidents. The differences between a full power inventory and the inventory 4 days after shutdown will not significantly affect the number of latent cancers because the radionuclides that are important to latent cancers will still be present in the fuel. Thus, the magnitude of the release is the driving factor for latent cancer fatalities. Because in POS 6 the releases tend to be higher than the full power accidents, the number of latent cancers associated with POS 6 are greater than the number of latent cancers associated with full power accidents. The factors that influence latent cancers also affect the population dose .

• 34

• 10-1 10-2 2

Earl Fatalities B

B Wean r.llb 103 I

II Total Lalenl Cancers H

lle,m

!'.*. 95lb Ill Mean 5

'2 Me&D '

10-3 ' L Mean

~lh rio1,11 I s M L 2

102 10-,

e s 5 lQ-5 2

3 6lh 2

'z lilb L 101 10-e LOSP nonLOSP NR-1150 LOSP nonLOSP NR-1150 Po ulation Dose 50 miles (Person Rem) Po ulation Dose 1000 miles (Person Rem) 1oe 107 B

7 II B

e H Kean II II Ill

' Kean Me&D Ill Ill L

z 2 llle&D Mean L

L liOlh 105 1oe II 'e 5 5 4

3

!>lb s

2 2 10' LOSP nonLOSP NR-1150 LOSP nonLOSP NR-1150 I* tstSBO-ICF-YB-aoCS I Figure 5-1. Grand Gulf POS 6 offsite consequences for LOSP and -nonLOSP PDSs

• 35

6. INSIGHTS AND CONCLUSIONS The results and insights presented in this study are conditional on the occurrence of core damage. Thus, this study gives no indication about the likelihood of these postulated accidents, but rather what could be expected given that core damage does occur. The input to this analysis is the core damage sequence definitions from the Level I coarse screening analysis. In this Level I scoping analysis some very conservative assumptions were made with regard to the availability of certain systems and the performance of the plant operators. These assumptions provided the necessary simplifications such that the dominant sequences could be identified and still keep the scope of the study manageable. While the calculated frequencies from the Level I study are used to rank the sequences, the absolute values of these frequencies were not reported due to the conservative nature of many of the necessary simplifications. Thus, frequencies were not propagated through to the Level II/ID analyses. It is within this framework that the abridged study was performed. Therefore, when interpreting these results it must be remembered that frequency information is not available to indicate the likelihood of accidents and simplifying assumptions were made in both the Level I and the Level II/ill studies.

• The following is a list of insights obtained from this study:

o During POS 6 the majority of the mitigative features of the plant are bypassed or are unavailable. The vessel and drywell are open to the containment and, thus, the suppression pool is effectively bypassed. Furthermore, the containment spray system is unavailable during these accidents. Thus, steam and radionuclides are released directly into the containment atmosphere without being scrubbed by either the suppression pool or the containment sprays. For the accidents in which

• the containment hatch is removed, the only significant plant mitigative feature is the deposition that occurs in the auxiliary building. li the containment is closed but then fails during core damage, the auxiliary building is also bypassed.

o Because of the lack of mitigative features associated with these accidents, the source terms tend to be quite large.

o The consequences associated with these accidents are also significant. Offsite consequences are comparable with consequences associated with full power accidents. Onsite consequences are large.

o The time from the accident initiation to the onset of core damage is significant (i.e.,

from 18 to 28 hours3.240741e-4 days <br />0.00778 hours <br />4.62963e-5 weeks <br />1.0654e-5 months <br />). Thus, there is a considerable amount of time to restore core cooling and to close the containment. li ac power is available, it is likely that the operators will close the containment prior to core damage.

0 Because the Grand Gulf containment is relatively weak (i.e., design pressure of 15

• 36

psig), even if the containment is closed prior to core damage, steps must be take

• 0 to control containment pressurization (i.e., from steam, noncondensibles, and hydrogen burns). Failure to control the pressure could result in early containment failure.

Because of the large recovery potential associated with these accidents (i.e., which were not fully accounted for in both the Level I analysis and this abridged analysis because of simplifying assumptions), POS 6 offsite risk could be significantly lower than the risk associated with full power accidents .

o Recovered accidents can still pose a significant threat to onsite personnel.

o Because of the lack of mitigation features associated with accidents initiated in this POS, the auxiliary building and the SBGTs could play a significant role in the mitigation of the release, especially for recovered accidents.

There were many issues that were identified in this study that could affect the possible accident progressions and consequences. The resolution of many of these issues was beyond the scope of this abridged analysis and will have to be addressed in any more detailed analysis that is performed in the future. The following is a list of potentially significant issues:

Containment Oosure. The effects that the temperature, humidity, and radiological 0

condition of the containment atmosphere have on the plant personnel's ability to close the containment needs to be addressed in more detail. Containment closure is a critical issue that will affect the consequences associated with these accidents.

0 Containment* Loading. Hydrogen combustion phenomena associated with this plant configuration need to be investigated. In this plant configuration steam and hot hydrogen are released directly into the containment atmosphere. The amount of steam blanketing and air ingression and the availability of ignition sources will all affect the likelihood and magnitude of hydrogen burns. As part of this assessment the effectiveness of the hydrogen ignition system in this plant configuration also needs to be investigated. The loading from in-vessel steam explosions is another issue that needs to be addressed. With the vessel head off in this POS and the relatively low failure pressure of this containment, in-vessel steam explosions could be a significant mechanism for early containment failure.

0 Source Term. There are several events that can enhance the source term that were not included in the PRA model. First, the role that air ingression plays during core damage needs to be investigated. If significant air ingression does occur, the in-vessel phase of the core damage process could be significantly altered and the release of certain radionuclides enhanced. Second, the relocation of intact fuel from an in-vessel steam explosion could also result in the enhancement of an early

• 37

source term. This issue was not addressed in this analysis. Finally, for recovered accidents the embrittlement and failure of the clad could lead to a release earlier than what is currently modeled. 1his could be particularly important for onsite consequences.

0 Auxiliary Building. For accidents in which the containment is open during core damage, the auxiliary building could play a major role in mitigating the release.

The radionuclide retention capabilities of this building need to be assessed in more detail than what was done ii) this abridged analysis. Furthermore, effectiveness of the SBGT system to mitigate the release, especially for recovered accidents, also needs to be assessed.

0 Onsite Consequences. Only a scoping type analysis of onsite consequences was performed in this study. In the calculation*of doses in the building, the integrated concentrations used were based on average concentrations from GGLPSOR and on crude transit times in the buildings. More detailed information on the concentration as a function of time and on the transit time would produce more realistic dose estimates .

• 38

~*>*

• 7. REFERENCES

1. U.S. NuclearRegulatoryCommission, "Severe Accident Risks: AnAssessment for Five U.S. Nuclear Power Plants," NUREG-1150, Vols. 1-3, December 1990-January 1991.

2. R. M. Summers, et. al., "1.fELCOR 1.8.0: A Computer Code for Nuclear Reactor Severe Accident Source Term and Risk Assessment Analyses,"

NUREG/CR-5531, Sandi~ National Laboratories, SAND90-0364, January 1991.

3. P. Cybulskis, "Assessment of the XSOR Codes,* NUREG/CR-5346, Battelle Columbus Division, BMl-2171, November 1989.

4. D. I. Chanin, et. al., "1.fELCOR Accident Analysis Consequence Code System," Sandia National Laboratories, NUREG/CR-4691, Sandia National Laboratories, SAND86-1562, Vols. 1 - 3, February 1990.

5. D. W. Whitehead, et. al., "BWR Low Power and Shutdown Accident Frequencies Program: Phase 1 - Coarse Screening Analysis," Vols. 1-3, Draft Letter Report, October 1991 - November 1991. Available in the NRC Public Document Room, 2120 L Street, NW .

• 6. T. D. Brown, et. al., "Evaluation of Severe Accident Risks: Grand Gulf Unit 1," NUREG/CR-4551, SAND86-1309, Vol. 6, Rev. 1, Sandia National Laboratories, December 1990.

7. J. V. Ramsdell Jr., "Diffusion in Building Wakes for Ground-Level Releases,"

Atmospheric Environment, Vol. 24B, No. 3, pp 377-388, 1990.

8. Wilson in "Atmospheric Science and Power Production," Ed. Randerson, D.,

DOE/TIC-27601, 1984, pg. 299.

9. NRC-Reg. Guide 1.145, Revision 1, November 1982 .

• 39

'I

\..,/

Enclosure 2 Brookhaven Report of the Study of the Surry Plant