ML23194A015

Level 3 PRA Project, Volume 4 Overview of Reactor, At-Power, Level 1, 2, and 3 PRAs for Internal Fires, Seismic Events, and High Winds; Draft Report for Comment

ML23194A015
Person / Time
Issue date:	08/31/2023
From:	Ball E, Keith Compton, Hathaway A, Alan Kuritzky, Selim Sancaktar NRC/RES/DRA/PRAB
To:
References
Download: ML23194A015 (1)

Text

NUREG-XXXX U.S. NRC Level 3 Probabilistic Risk Assessment Project Volume 4: Overview of Reactor, At-Power, Level 1, 2, and 3 PRAs for Internal Fires, Seismic Events, and High Winds This report, though formatted as a NUREG report, is currently being released as a draft (non-NUREG) technical report for comment.

Office of Nuclear Regulatory Research

U.S. NRC Level 3 Probabilistic Risk Assessment Project Volume 4: Overview of Reactor, At-Power, Level 1, 2, and 3 PRAs for Internal Fires, Seismic Events, and High Winds Manuscript Completed: August 2023 Date Published: xxxx 2023 Prepared by:

S. Sancaktar1 E. Ball2 A. Kuritzky2 A. Hathaway2 K. Compton2 1Formerly with the U.S. Nuclear Regulatory Commission 2U.S. Nuclear Regulatory Commission A. Kuritzky, NRC Level 3 PRA Project Program Manager Office of Nuclear Regulatory Research

ABSTRACT The U.S. Nuclear Regulatory Commission performed a full-scope site Level 3 probabilistic risk assessment (PRA) project (L3PRA project) for a two-unit pressurized-water reactor reference plant. The scope of the L3PRA project encompasses all major radiological sources on the site (i.e., reactors, spent fuel pools, and dry cask storage), all internal and external hazards, and all modes of plant operation. A full-scope site Level 3 PRA for a nuclear power plant site can provide valuable insights into the importance of various risk contributors by assessing accidents involving one or more reactor cores as well as other site radiological sources. This report, one of a series of reports documenting the models and analyses supporting the L3PRA project, provides an overview of the reactor, at-power, Level 1, 2, and 3 PRA models for internal fires, seismic events, and high winds. The analyses documented herein are based on information for the reference plant as it was designed and operated as of 2012 and do not reflect the plant as it is currently designed, licensed, operated, or maintained. To provide results and insights better aligned with the current design and operation of the reference plant, this report also provides the results of a parametric sensitivity analysis based on a set of new plant equipment and PRA model assumptions for all three PRA levels. The sensitivity analysis reflects the current reactor coolant pump shutdown seal design at the reference plant, as well as the potential impact of FLEX strategies, 1 both of which reduce the risk to the public.

CAUTION: While the L3PRA project is intended to be a state-of-practice study, due to limitations in time, resources, and plant information, some technical aspects of the study were subjected to simplifications or were not fully addressed. As such, inclusion of approaches in the L3PRA project documentation should not be viewed as an endorsement of these approaches for regulatory purposes.

1 FLEX refers to the U.S. nuclear power industry's proposed safety strategy, called Diverse and Flexible Coping Strategies. FLEX is intended to maintain long-term core and spent fuel cooling and containment integrity with installed plant equipment that is protected from natural hazards, as well as backup portable onsite equipment. If necessary, similar equipment can be brought from offsite.

iii

FOREWORD The U.S. Nuclear Regulatory Commission (NRC) performed a full-scope site Level 3 probabilistic risk assessment (PRA) project (L3PRA project) for a two-unit pressurized-water reactor reference plant. The staff undertook this project in response to Commission direction in the staff requirements memorandum dated September 21, 2011 (Agencywide Documents and Management System [ADAMS] Accession No. ML112640419) resulting from SECY-11-0089, Options for Proceeding with Future Level 3 Probabilistic Risk Assessment Activities, dated July 7, 2011 (ML11090A039).

Licensee information used in performing the Level 3 PRA project was voluntarily provided based on a licensed, operating nuclear power plant. The information provided reflects the plant as it was designed and operated as of 2012 and does not reflect the plant as it is currently designed, licensed, operated, or maintained. In addition, the information provided for the reference plant was changed based on additional information, assumptions, practices, methods, and conventions used by the NRC in the development of plant-specific PRA models used in its regulatory decision-making. As such, the L3PRA project reports will not be the sole basis for any regulatory decisions specific to the reference plant.

Each set of L3PRA project reports covering the Level 1, 2, and 3 PRAs for a specific site radiological source, plant operating state, and hazard group is accompanied by an overview report. The overview reports summarize the results and insights from all three PRA levels. This current document is the overview report for the reactor, at-power, Level 1, 2, and 3 PRAs for internal fires, seismic events, and high winds.

To provide results and insights better aligned with the current design and operation of the reference plant, this report also provides the results of a parametric sensitivity analysis based on a set of new plant equipment and PRA model assumptions for all three PRA levels. The sensitivity analysis reflects the current reactor coolant pump shutdown seal design at the reference plant, as well as the potential impact of FLEX strategies, 1 both of which reduce the risk to the public.

A full-scope site Level 3 PRA for a nuclear power plant site can provide valuable insights into the importance of various risk contributors by assessing accidents involving one or more reactor cores as well as other site radiological sources (i.e., spent fuel in pools and dry storage casks).

These insights may be used to further enhance the regulatory framework and decision-making and to help focus limited agency resources on issues most directly related to the agencys mission to protect public health and safety. More specifically, potential future uses of the L3PRA project can be categorized as follows (a more detailed list is provided in SECY-12-0123, Update on Staff Plans to Apply the Full-Scope Site Level 3 PRA Project Results to the NRCs Regulatory Framework, dated September 13, 2012 [ML12202B170]):

1 FLEX refers to the U.S. nuclear power industrys proposed safety strategy, called Diverse and Flexible Coping Strategies. FLEX is intended to maintain long-term core and spent fuel cooling and containment integrity with installed plant equipment that is protected from natural hazards, as well as backup portable onsite equipment. If necessary, similar equipment can be brought from off site.

v

• enhancing the technical basis for the use of risk information (e.g., obtaining updated and enhanced understanding of plant risk as compared to the Commissions safety goals)

• improving the PRA state of practice (e.g., demonstrating new methods for site risk assessments, which may be particularly advantageous in addressing the risk from advanced reactor designs, a multi-unit accident, or an accident involving spent fuel; and using PRA information to inform emergency planning)

• identifying safety and regulatory improvements (e.g., identifying potential safety improvements that may lead to either regulatory improvements or voluntary implementation by licensees)

• supporting knowledge management (e.g., developing or enhancing in-house PRA technical capabilities)

In addition, the overall L3PRA project model can be exercised to provide insights regarding other issues not explicitly included in the current project scope (e.g., security-related events or the use of accident tolerant fuel). Furthermore, some future advanced light-water reactor (ALWR) and advanced non-light-water reactor (NLWR) applicants may rely heavily on the results of analyses similar to those used in the L3PRA project to establish their licensing basis and design basis by using the Licensing Modernization Project (LMP) (NEI 18-04, Rev. 1) which was endorsed via Regulatory Guide 1.233 in June 2020. Licensees who use the LMP framework are required to perform Level 3 PRA analyses. Therefore, another potential use of the methodology and insights generated from this study is to inform regulatory, policy, and technical issues pertaining to ALWRs and NLWRs.

The results and perspectives from this report, as well as all other reports prepared in support of the L3PRA project, will be incorporated into a summary report to be published after all technical work for the L3PRA project has been completed.

vi

TABLE OF CONTENTS ABSTRACT ...............................................................................................................................iii FOREWORD ..............................................................................................................................v TABLE OF CONTENTS ...........................................................................................................vii LIST OF FIGURES ....................................................................................................................ix LIST OF TABLES ......................................................................................................................xi ACKNOWLEDGMENTS ..........................................................................................................xiii ABBREVIATIONS AND ACRONYMS ......................................................................................xv 1 INTRODUCTION................................................................................................................1-1 2 KEY MESSAGES ..............................................................................................................2-1 3

SUMMARY

OF RESULTS AND INSIGHTS .......................................................................3-1 3.1 Level 1 PRA .............................................................................................................3-1 3.1.1 Results of Circa-2012 and 2020-FLEX Cases ........................................3-1 3.1.2 Results of Alternative Analyses ...................................................................3-3 3.1.3 Initial Insights ..............................................................................................3-5 3.2 Level 2 PRA ...........................................................................................................3-14 3.2.1 Results of Circa-2012 and 2020-FLEX Cases ......................................3-14 3.2.2 Results of Alternative Analyses .................................................................3-18 3.2.3 Initial Insights ............................................................................................3-20 3.3 Level 3 PRA ...........................................................................................................3-33 3.3.1 Results of Circa-2012 and 2020-FLEX Cases ......................................3-33 3.3.2 Results of Alternative Analyses .................................................................3-37 3.3.3 Initial Insights ............................................................................................3-39 4 KEY ASSUMPTIONS, CONSIDERATIONS, AND UNCERTAINTIES FOR THE 2020-FLEX CASE ..............................................................................................................4-1 4.1 Level 1 PRA .............................................................................................................4-1 4.1.1 Key Assumptions ........................................................................................4-1 4.1.2 Key Uncertainties ........................................................................................4-3 4.2 Level 2 PRA .............................................................................................................4-3 4.2.1 Key Assumptions ........................................................................................4-4 4.2.2 Additional Considerations............................................................................4-5 4.2.3 Key Uncertainties ........................................................................................4-6 4.3 Level 3 PRA .............................................................................................................4-8 5 REFERENCES...................................................................................................................5-1 vii

LIST OF FIGURES Figure 3-1 CDF Percentages by Hazard Group for Circa-2012 Case ...............................3-10 Figure 3-2 CDF Percentages by Hazard Group for 2020-FLEX Case ..............................3-10 Figure 3-3 FLEX Impact on CDF by Hazard Category......................................................3-11 Figure 3-4 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Internal Events and Floods (Circa-2012) .........................................................3-44 Figure 3-5 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Internal Events and Floods (2020-FLEX) ........................................................3-44 Figure 3-6 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Internal Fires (Circa-2012 Case) .....................................................................3-45 Figure 3-7 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Internal Fires (2020-FLEX Case) ....................................................................3-45 Figure 3-8 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Seismic Events (Circa-2012 Case) .................................................................3-46 Figure 3-9 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Seismic Events (2020-FLEX Case) .................................................................3-46 Figure 3-10 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for High Winds (Circa-2012 Case) ................................................................................3-47 Figure 3-11 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for High Winds (2020-FLEX Case) ...............................................................................3-47 Figure 3-12 Individual Early Fatality Risk (0-1.8 Miles) ......................................................3-48 Figure 3-13 CCDF Curves for Population-Weighted Early Fatality Risk (0-1.8 Miles) for the 2020-FLEX Case (Internal Events and Floods) ....................................3-49 Figure 3-14 CCDF Curves for Population-Weighted Early Fatality Risk (0-1.8 Miles) for the 2020-FLEX Case (Internal Fires) .........................................................3-49 Figure 3-15 CCDF Curves for Population-Weighted Early Fatality Risk (0-1.8 Miles) for the 2020-FLEX Case (Seismic Events) ......................................................3-50 Figure 3-16 CCDF Curves for Population-Weighted Early Fatality Risk (0-1.8 Miles) for the 2020-FLEX Case (High Winds) ............................................................3-50 Figure 3-17 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Internal Events and Floods (Circa-2012) .........................................................3-51 ix

Figure 3-18 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Internal Events and Floods (2020-FLEX) ........................................................3-51 Figure 3-19 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Internal Fires (Circa-2012) ..............................................................................3-52 Figure 3-20 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Internal Fires (2020-FLEX)..............................................................................3-52 Figure 3-21 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Seismic Events (Circa-2012) ...........................................................................3-53 Figure 3-22 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Seismic Events (2020-FLEX) ..........................................................................3-53 Figure 3-23 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for High Winds (Circa-2012) .........................................................................................3-54 Figure 3-24 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for High Winds (2020-FLEX) ........................................................................................3-54 Figure 3-25 Individual Latent Cancer Fatality Risk (0-10 Miles) .........................................3-55 Figure 3-26 CCDF Curves for Population-Weighted Latent Cancer Fatality Risk (0-10 Miles) for the 2020-FLEX Case (Internal Events and Floods) ................3-56 Figure 3-27 CCDF Curves for Population-Weighted Latent Cancer Fatality Risk (0-10 Miles) for the 2020-FLEX Case (Internal Fires) .....................................3-56 Figure 3-28 CCDF Curves for Population-Weighted Latent Cancer Fatality Risk (0-10 Miles) for the 2020-FLEX Case (Seismic Events)..................................3-57 Figure 3-29 CCDF Curves for Population-Weighted Latent Cancer Fatality Risk (0-10 Miles) for the 2020-FLEX Case (High Winds)........................................3-57 Figure 3-30 Individual Latent Cancer Fatality Risk (0-10 Miles) for Internal Events and FloodsAlternative Analyses .........................................................................3-58 Figure 3-31 Individual Latent Cancer Fatality Risk (0-10 Miles) for Internal Fires Alternative Analyses .......................................................................................3-59 Figure 3-32 Individual Latent Cancer Fatality Risk (0-10 Miles) for Seismic Events Alternative Analyses .......................................................................................3-60 Figure 3-33 Individual Latent Cancer Fatality Risk (0-10 Miles) for High Winds Alternative Analyses .......................................................................................3-61 Figure 3-34 Individual Latent Cancer Fatality Risk (0-10 Miles) for All Hazards Alternative Analyses .......................................................................................3-62 x

LIST OF TABLES Table 2-1 Summary of Risk Metric Results for Circa-2012 Case .......................................2-3 Table 2-2 Summary of Risk Metric Results for 2020-FLEX Case ......................................2-3 Table 3-1 FLEX Failure Probabilities by Hazard Category ..............................................3-11 Table 3-2 CDF by Hazard Category ................................................................................3-12 Table 3-3 Summary of Hazard Category Parametric Uncertainty Analyses.....................3-12 Table 3-4 Summary of FLEX Cases for Internal Fires .....................................................3-13 Table 3-5 Summary of FLEX Cases for Seismic Events .................................................3-13 Table 3-6 Summary of FLEX Cases for High Winds........................................................3-13 Table 3-7 Comparison of Level 1 CDF and Level 2 Release Frequency for 2020-FLEX Case ............................................................................................3-21 Table 3-8 Description of Release Categories ..................................................................3-22 Table 3-9 Internal Fire Release Category Frequencies ...................................................3-23 Table 3-10 Seismic Event Release Category Frequencies ...............................................3-24 Table 3-11 High Wind Release Category Frequencies......................................................3-25 Table 3-12 Level 2 PRA Surrogate Risk Metrics ...............................................................3-26 Table 3-13 Level 2 PRA Representative Accident Scenario Timelines..............................3-27 Table 3-14 Internal Event and Flood Level 2 PRA Surrogate Risk Metric Results -

2020-FLEX Case ............................................................................................3-29 Table 3-15 Internal Fire Level 2 PRA Surrogate Risk Metric Results - 2020-FLEX Case ...............................................................................................................3-30 Table 3-16 Seismic Event Level 2 PRA Surrogate Risk Metric Results - 2020-FLEX Case ...............................................................................................................3-31 Table 3-17 High Wind Level 2 PRA Surrogate Risk Metric Results - 2020-FLEX Case ...............................................................................................................3-32 Table 3-18 Population-Weighted Early Fatality Risk, by Release Category, for the 0-1.8-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Internal Events and Floods ..........................................................................................3-63 Table 3-19 Population-Weighted Early Fatality Risk, by Release Category, for the 0-1.8-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Internal Fires................................................................................................................3-63 Table 3-20 Population-Weighted Early Fatality Risk, by Release Category, for the 0-1.8-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Seismic Events.............................................................................................................3-63 Table 3-21 Population-Weighted Early Fatality Risk, by Release Category, for the 0-1.8-Mile Interval for the Circa-2012 and 2020-FLEX Cases for High Winds..............................................................................................................3-64 xi

Table 3-22 Population-Weighted Early Fatality Risk, by Hazard Category, for the 0-1.8-Mile Interval for the Circa-2012 and 2020-FLEX Cases.........................3-64 Table 3-23 Statistical Analysis of the Population-Weighted Early Fatality Risk for the 0-1.8-Mile Interval for the 2020-FLEX Case ...................................................3-64 Table 3-24 Population-Weighted Latent Cancer Fatality Risk, by Release Category, for the 0-10-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Internal Events and Floods .............................................................................3-65 Table 3-25 Population-Weighted Latent Cancer Fatality Risk, by Release Category, for the 0-10-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Internal Fires ...................................................................................................3-65 Table 3-26 Population-Weighted Latent Cancer Fatality Risk, by Release Category, for the 0-10-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Seismic Events ...............................................................................................3-66 Table 3-27 Population-Weighted Latent Cancer Fatality Risk, by Release Category, for the 0-10-Mile Interval for the Circa-2012 and 2020-FLEX Cases for High Winds .....................................................................................................3-66 Table 3-28 Population-Weighted Latent Cancer Fatality Risk, by Hazard Category, for the 0-10-Mile Interval for the Circa-2012 and 2020-FLEX Cases...............3-66 Table 3-29 Statistical Analysis of the Population-Weighted Latent Cancer Fatality Risk for the 0-10-Mile Interval for the 2020-FLEX Case..................................3-67 xii

ACKNOWLEDGMENTS The authors of this report would like to thank Chris Hunter and Jeff Wood for their technical review of the work supporting the Level 1 and Level 2 PRA portions of this document, respectively, and Jonathan Evans and Kevin Coyne for their management review and guidance.

The authors would also like to thank the members of the Level 3 PRA project technical advisory group for providing helpful comments on the draft report.

xiii

ABBREVIATIONS AND ACRONYMS AC alternating current ADAMS Agencywide Documents and Management System AFW auxiliary feedwater ALWR advanced light-water reactor ARV atmospheric relief valve ATWS anticipated transient without scram CCDF complementary cumulative distribution function CCDP conditional core damage probability CCFP conditional containment failure probability CDF core damage frequency DC direct current EDG emergency diesel generator ELAP extended loss of all AC power EPA Environmental Protection Agency ET event tree FLEX Diverse and Flexible Coping Strategies FSG FLEX Strategy Guideline FV Fussell-Vesely (importance measure)

GE general emergency HFE human failure event HPS Health Physics Society HRA human reliability analysis ISLOCA interfacing systems loss-of-coolant accident L3PRA Level 3 Pobablisitic Risk Assessment (project)

LERF large early release frequency LMP Licensing Modernization Project LNT linear no-threshold LOCA loss-of-coolant accident LOOP loss of offsite power LRF large release frequency MAACS MELCOR Accident Consequence Code System NLWR non-light-water reactor NRC U.S. Nuclear Regulatory Commission NSCW nuclear service cooling water PAG protective action guideline PRA probabilistic risk assessment PRT pressurizer relief tank QHO quantitative health objective xv

RCF release category frequency RCP reactor coolant pump RCS reactor coolant system rcy reactor-critical-year SAMG severe accident management guideline SAT standby auxiliary transformer SBO station blackout SG steam generator SGTR steam generator tube rupture SOARCA State-of-the-Art Consequence Analyses SPAR Standardized Plant Analysis Risk (model)

SSC structure, system, and component TDAFW turbine-driven auxiliary feedwater UET unfavorable exposure time V volt xvi

1 INTRODUCTION The U.S. Nuclear Regulatory Commission (NRC) performed a full-scope site Level 3 probabilistic risk assessment (PRA) project (L3PRA project) for a two-unit pressurized-water reactor reference plant. The staff undertook this project in response to Commission direction in the staff requirements memorandum dated September 21, 2011 (Agencywide Documents and Management System [ADAMS] Accession No. ML112640419) resulting from SECY-11-0089, Options for Proceeding with Future Level 3 Probabilistic Risk Assessment Activities, dated July 7, 2011 (ML11090A039).

As described in SECY-11-0089, the objectives of the L3PRA project are the following:

• Develop a Level 3 PRA, generally based on current state-of-practice methods, tools, and data, 1 that (1) reflects technical advances since the last NRC-sponsored Level 3 PRAs (NRC, 1990), which were completed over 30 years ago, and (2) addresses scope considerations that were not previously considered (e.g., low-power and shutdown risk, multi-unit risk, other radiological sources).

• Extract new insights to enhance regulatory decision-making and to help focus limited NRC resources on issues most directly related to the agencys mission to protect public health and safety.

• Enhance PRA staff capability and expertise and improve documentation practices to make PRA information more accessible, retrievable, and understandable.

• Demonstrate technical feasibility and evaluate the realistic cost of developing new Level 3 PRAs.

Licensee information used in the L3PRA project was voluntarily provided based on a licensed, operating nuclear power plant. The information provided reflects the plant as it was designed and operated as of 2012 and does not reflect the plant as it is currently designed, licensed, operated, or maintained. In addition, the information provided for the reference plant was changed based on additional information, assumptions, practices, methods, and conventions used by the NRC in the development of plant-specific PRA models. As such, the L3PRA project reports will not be the sole basis for any regulatory decisions specific to the reference plant.

The series of reports for the L3PRA project are organized as follows:

Volume 1: Summary (to be published last)

Volume 2: Background, site and plant description, and technical approach 1 State-of-practice methods, tools, and data refer to those that are routinely used by the NRC and industry or have acceptance in the PRA technical community. While the L3PRA project is intended to be a state-of-practice study, note that there are several technical areas within the project scope that necessitated advancements in the state of practice (e.g., modeling of multi-unit site risk, modeling of spent fuel in pools or casks, and of human reliability analysis for other than internal events and internal fires).

1-1

Volume 3: Reactor, at-power, internal event and flood PRA (overview)

Volume 3a: Level 1 PRA for internal events Volume 3b: Level 1 PRA for internal floods Volume 3c: Level 2 PRA for internal events and floods Volume 3d: Level 3 PRA for internal events and floods Volume 4: Reactor, at-power, internal fire and external event PRA (overview)

Volume 4a: Level 1 PRA for internal fires Volume 4b: Level 1 PRA for seismic events Volume 4c: Level 1 PRA for high wind events and other hazards evaluation Volume 4d: Level 2 PRA for internal fires and seismic and wind-related events Volume 4e: Level 3 PRA for internal fires and seismic and wind-related events Volume 5: Reactor, low-power and shutdown, internal event PRA (overview)

Volume 5a: Level 1 PRA for internal events Volume 5b: Level 2 PRA for internal events Volume 5c: Level 3 PRA for internal events Volume 6: Spent fuel pool all hazards PRA (overview)

Volume 6a: Level 1 and Level 2 PRA Volume 6b: Level 3 PRA Volume 7: Dry cask storage, all hazards, Level 1, Level 2, and Level 3 PRA Volume 8: Integrated site risk, all hazards, Level 1, Level 2, and Level 3 PRA The original L3PRA project models are referred to as the Circa-2012 case and a description of the plant as modeled is given in Volume 2. Volumes 4a, 4b, and 4c were created to document the L3PRA project Level 1 PRA models and analyses for internal fires, seismic events, and high winds, respectively, during power operation for the Circa-2012 case. Volume 4c also documents the analyses used to screen out other hazards. Additionally, Volumes 4d and 4e were created to document the corresponding Level 2 and Level 3 PRA models and analyses. As indicated in the list above, other volumes address the risk contributions from internal events and internal floods, other plant operating states, and other site radiological sources (i.e., spent fuel pools and dry storage casks).

This report provides an overview of the reactor, at-power, Level 1, 2, and 3 PRA models and results for internal fires, seismic events, and high winds, as well as comparisons with the results for internal events and internal floods. The analyses documented herein are based on information for the reference plant as it was designed and operated as of 2012 and do not reflect the plant as it is currently designed, licensed, operated, or maintained. To provide results and insights better aligned with the current design and operation of the reference plant, this report also provides the results of a parametric sensitivity analysis based on a set of new plant equipment and PRA model assumptions for all three PRA levels. The sensitivity analysis, referred to as the 2020-FLEX case, reflects the current reactor coolant pump shutdown seal 1-2

design at the reference plant, as well as the potential impact of FLEX strategies, 2 both of which reduce the risk to the public.

Section 2 provides key messages from the reactor, at-power, Level 1, 2, and 3 PRAs for internal fires, seismic events, and high winds, while Section 3 provides a summary of the results and insights from these analyses, including comparisons between the Circa-2012 and 2020-FLEX cases. Section 4 documents key modeling assumptions, considerations, and uncertainties associated with the 2020-FLEX case.

Note, it is anticipated that the models and results of the L3PRA project are likely to evolve over time, as other parts of the project are developed, or as other technical issues are identified. As such, the final models and results of the project (which will be documented in the Volume 1 summary report after all technical work for the Level 3 PRA project has been completed) may differ in some ways from the models and results provided in the current report.

CAUTION: While the L3PRA project is intended to be a state-of-practice study, due to limitations in time, resources, and plant information, some technical aspects of the study were subjected to simplifications or were not fully addressed. As such, inclusion of approaches in the L3PRA project documentation should not be viewed as an endorsement of these approaches for regulatory purposes.

2 FLEX refers to the U.S. nuclear power industry's proposed safety strategy, called Diverse and Flexible Coping Strategies. FLEX is intended to maintain long-term core and spent fuel cooling and containment integrity with installed plant equipment that is protected from natural hazards, as well as backup portable onsite equipment. If necessary, similar equipment can be brought from offsite.

1-3

2 KEY MESSAGES This section provides some of the key messages resulting from the reactor, at-power, Level 1, 2, and 3 PRAs for internal fires, seismic events, and high winds for a single unit. Table 2-1 and Table 2-2 summarize some of the key risk metrics and surrogate risk metrics that were quantified as part of the analyses for the Circa-2012 and 2020-FLEX cases, respectively.

Overall, the estimated individual early fatality risk and latent cancer fatality risk results show that the combination of this plant design and site location has substantial margin to the two quantitative health objectives (QHOs) associated with the NRCs safety goal policy (NRC, 1986), though the margins are noticeably less for the surrogate risk metrics of core damage frequency (CDF) and large, early release frequency (LERF) that were endorsed by the Commission when it approved the issuance of Regulatory Guide 1.174.

Level 1 PRA Key Messages

• For the Circa-2012 case, the CDF from all hazards combined for the reactor, at-power, is 1.5x10-4 per reactor-critical-year (rcy), with major contributions from internal events (45 percent) and internal fires (43 percent).

• For the 2020-FLEX case, the CDF for all hazards combined for the reactor, at-power, is reduced by 38 percent to 9.3x10-5/rcy. This reduction is most pronounced for internal events and high winds because (1) the CDF for both these hazard categories is dominated by loss of offsite power (LOOP) or loss of nuclear service cooling water (NSCW) sequences, both of which benefit significantly from the types of measures incorporated into the 2020-FLEX case, 3 and (2) the relative likelihood of success for these measures is assumed to be greater than for internal fires and seismic events.

Level 2 PRA Key Messages

• A very small fraction of CDF leads to large early release 4approximately 1 percent for all hazard categories (for seismic events, it is approximately 2 percent because of the increased conditional probability of thermally induced steam generator tube rupture).

• A relatively large fraction of CDF results in later containment failure (approximately 60-90 percent, depending on hazard category).

o Late, large release does not result in any prompt fatalities, but can result in latent cancer fatalities and economic consequences.

• The frequency of late, large releases is highly dependent on the severe accident progression modeling time.

3 As mentioned in Section 3.1.1, besides the FLEX strategies and equipment, the 2020-FLEX case also incorporates the new reactor coolant pump (RCP) seals (shutdown seals) and, if FLEX is not successful, the potential for continued turbine-driven auxiliary feedwater (TDAFW) pump operation given a complete loss of all installed alternating current (AC) and direct current (DC) power. The impact of the new RCP seals alone is evaluated in a sensitivity analysis in Section 10.8 of NRC (2022d).

4 In Table 2-1 and Table 2-2, the large early release frequency is based on the early fatalities definition as discussed in Section 3.7.1 of NRC (2023d).

2-1

o The 2020-FLEX base case models severe accident progression for 7 days after accident initiation, though severe accident mitigation is generally only credited for the time up to, or shortly after, vessel breach (i.e., no credit is given for longer-term recovery actions, such as venting, steam-inerting, or implementing FLEX to restore electrical power).

o Reducing modeling time to approximately 2 days after accident initiation reduces large release frequency to between 18-34 percent of CDF, depending on hazard category. This demonstrates that significant reductions in risk can occur if credible mitigative actions can be successfully implemented in this timeframe.

Level 3 PRA Key Messages

• Early fatality risks to individuals are far below the QHO associated with the safety goals (due primarily to sufficient warning times for effective evacuation).

o For the 2020-FLEX case, there is only minimal change in the population-weighted early fatality risk within 1 mile of the site boundary, since this metric is dominated by interfacing system loss-of-coolant accidents (ISLOCAs), which do not generally benefit from the types of measures incorporated into the 2020-FLEX case.

• Latent fatality risks to individuals are well below the QHO associated with the safety goals (due to longer-term relocation of affected populations).

o Latent cancer fatalities occur from long-term reoccupation of land and use of the linear no-threshold (LNT) model.

o Radiogenic cancers are still not expected to be statistically detectable above norms.

o Economic impacts arise largely from these longer-term protective measures.

o For the 2020-FLEX case, individual latent cancer fatality risk within 10 miles of the plant is reduced by approximately 37 percent for all hazards combined. The reductions are greater for internal events and floods and for high winds because these hazard categories have a higher relative contribution from the types of sequences that most benefit from the types of measures incorporated into the 2020-FLEX case (i.e., station blackout and, to a lesser extent, losses of NSCW leading to RCP seal loss-of-coolant accidents [LOCAs]).

o For the Circa-2012 case, the margin to the QHO, when considering all hazards combined, is approximately a factor of 30. This margin increases to approximately a factor of 50 when accounting for the modeling changes associated with the 2020-FLEX case. Further, terminating the accident and radiological release analysis at approximately 2 days (as opposed to 7 days) after accident initiation increases the margin to the QHO to approximately a factor of 150. Finally, using an alternative dose truncation model (as opposed to the LNT model) suggests that the margin to the QHO may be significantly greater.

2-2

Table 2-1 Summary of Risk Metric Results for Circa-2012 Case Internal QHO or Risk Metric Seismic Events and Internal Fires High Winds Total Subsidiary (per reactor-critical-year) Events Floods 5 Risk Metric Core damage frequency 6.5E-05 6.1E-05 1.1E-05 1.4E-05 1.5E-04 1E-04 Large early release frequency 9.3E-07 5.5E-07 2.3E-07 1.6E-07 1.9E-06 1E-05 Large release frequency 6 4.4E-05 4.1E-05 1.2E-05 1.0E-05 1.1E-04 N/A Individual early fatality risk 7 ~0 ~0 ~0 ~0 ~0 5E-7 Individual latent cancer fatality risk 2.6E-08 2.5E-08 7.7E-09 6.3E-09 6.5E-08 2E-6 Table 2-2 Summary of Risk Metric Results for 2020-FLEX Case Internal QHO or Risk Metric Seismic Events and Internal Fires High Winds Total Subsidiary (per reactor-critical-year) Events Floods5 Risk Metric Core damage frequency 2.7E-05 5.3E-05 8.5E-06 4.8E-06 9.3E-05 1E-04 2-3 Large early release frequency 5.7E-07 4.9E-07 2.0E-07 7.0E-08 1.3E-06 1E-05 Large release frequency6 1.7E-05 3.6E-05 9.5E-06 4.1E-06 6.7E-05 N/A Individual early fatality risk 8 ~0 ~0 ~0 ~0 ~0 5E-7 Individual latent cancer fatality risk 9.7E-09 9 2.2E-08 6.2E-09 2.4E-09 4.0E-08 2E-6 5 Internal floods are a very minor contributor to this category (approximately 1 percent of internal event core damage frequency).

6 Large release frequency is obtained through addition of the frequencies for the relevant release categories. Due to Level 2 PRA release frequency inflation, these results can overstate the relative contribution of large release frequency when comparing to core damage frequency (as opposed to total release frequency, i.e., the sum of all release category frequencies). This frequency inflation is most pronounced for seismic events and internal fires due to the presence of very high failure probabilities for many components. For example, total Level 2 PRA release frequency for seismic events is 1.510-5/rcy (as compared to a CDF of 1.110-5/rcy), so the large release frequency represents approximately 80 percent of the total release frequency. For more information, see discussion of release category frequency inflation in Section 3.5.6 of NRC (2023d).

7 The actual calculated individual early fatality risk for the Circa-2012 case is 7.5x10-13/rcy for all hazards combined.

8 The actual calculated individual early fatality risk for the 2020-FLEX case is 6.6x10-13/rcy for all hazards combined.

9 Due to round-off differences, this value differs slightly from the value reported in NRC (2022a).

3

SUMMARY

OF RESULTS AND INSIGHTS This section provides a summary of the results and insights from the reactor, at-power, Level 1, 2, and 3 PRAs for internal fires, seismic events, and high winds for a single unit. To provide results and insights that are more reflective of the current design and operation of the reference plant, throughout this section, results of the Circa-2012 case are compared with the results of the 2020-FLEX case. These comparisons demonstrate how the plant risk profile associated with at-power, internal fires, seismic events, and high winds has been influenced by several key plant changes implemented at the reference plant since 2012. The Level 1, Level 2, and Level 3 PRAs are discussed in Sections 3.1, 3.2, and 3.3, respectively.

3.1 Level 1 PRA This section provides a summary of the results and insights from the reactor, at-power, Level 1 PRA for internal fires, seismic events, and high winds for a single unit. Section 3.1.1 provides the high-level results for both the Circa-2012 and 2020-FLEX cases. Section 3.1.2 discusses several alternative analyses that were performed to better assess the effect of introducing FLEX into the Level 1 PRA model. Section 3.1.3 discusses insights from the reactor, at-power, Level 1 PRA for internal fires, seismic events, and high winds, including a discussion of the dominant contributors to CDF for both the Circa-2012 and 2020-FLEX cases.

3.1.1 Results of Circa-2012 and 2020-FLEX Cases Detailed descriptions of the Circa-2012 Level 1 PRA models and results for internal fires, seismic events, and high winds during power operation are provided in their respective L3PRA project reports (NRC, 2023a; NRC, 2023b; NRC, 2023c). The total CDF from internal fires is reported as 6.1x10-5/rcy, the total CDF from seismic events is reported as 1.1x10-5/rcy, and the total CDF from high winds is reported as 1.4x10-5/rcy. The CDF for the different hazard categories is shown graphically in Figure 3-1. As is common for many of the tables and figures in this report, the results for internal events and floods are also included for comparison. 10 The 2020-FLEX case updates the Circa-2012 models to include the new reactor coolant pump (RCP) seals (passive shutdown seals) and FLEX strategies and equipment for responding to an extended loss of alternating current (AC) power (ELAP). In addition, if FLEX is not successful, the 2020-FLEX case credits the potential for continued turbine-driven auxiliary feedwater (TDAFW) pump operation given a complete loss of all installed AC and direct current (DC) power. 11 Continued TDAFW pump operation given a complete loss of all installed AC and DC power was not credited in the Circa-2012 Level 1 PRA models because, as discussed in Section 8.1.2 of NRC (2022d), there is a low likelihood of success for this action and, even if 10 Generally, a single combined value is used in reporting results for internal events and internal floods because internal flood CDF is only approximately 1 percent of internal event CDF.

11 In pre-FLEX PRA models, this was often referred to as blind feeding. For a post-FLEX PRA model, the current terminology is used, since for some FLEX failure modes (e.g., failure of the FLEX steam generator feed pump),

FLEX may still be able to provide control power for continued TDAFW pump operation. However, it is acknowledged that, in most instances, continued operation of TDAFW requires recovery of some form of installed AC power earlier than the time required to bring in offsite resources. The human error probabilities assigned to the basic events representing failure to successfully implement FLEX or continued TDAFW pump operation include the possibility of not recovering installed AC power in a timely manner.

3-1

successful, the plant would not be in a stable condition (without the FLEX equipment and strategies).

The installation of the shutdown seals affects (positively) all sequences where RCP seal leakage occurs. The FLEX strategies, as well as continued TDAFW pump operation given a complete loss of all installed AC and DC power, are only credited in the modeling of station blackout (SBO) accident sequences. General modeling assumptions and considerations associated with the 2020-FLEX case are addressed in Section 4.

Two new basic events are introduced into the model to allow credit for FLEX in the SBO sequences where ELAP is declared. These two basic events are (1) failure to implement FLEX and (2) failure to continue operation of the TDAFW pump given a complete loss of installed AC and DC power. These two basic events always appear together in the CDF cutsets (i.e., they must both fail for core damage to occur). A parameter p is defined as the joint failure probability of these two basic events and is used to parametrically examine FLEX effectiveness, as discussed further in Section 3.1.2. A summary of the FLEX failure probabilities for each hazard category is provided in Table 3-1.

The total CDF for the 2020-FLEX case during power operation is estimated to be 5.3x10-5/rcy for internal fires, 8.5x10-6/rcy for seismic events, and 4.8x10-6/rcy for high winds (see Figure 3-2). The impact of the FLEX-related changes (including the RCP shutdown seals and continued operation of TDAFW pumps) are summarized in Table 3-2.

From Table 3-2 it is seen that the reduction in the 2020-FLEX case is most pronounced for internal events and high winds (reduction in CDF on the order of 60 percent). This is because the CDF for both these hazard categories is dominated by loss of offsite power (LOOP) or loss of nuclear service cooling water (NSCW) sequences, both of which benefit significantly from the types of measures incorporated into the 2020-FLEX case, and because the relative likelihood of success for these measures is assumed to be greater than for internal fires and seismic events.

For all hazards combined, the FLEX-related changes reduce CDF from 1.5x10-4/rcy to 9.3x10-5/rcy (a reduction of 38 percent).

A parametric uncertainty analysis for the 2020-FLEX case was performed for each of the seismic, wind, and fire hazard categories, which addresses the uncertainties associated with all basic events in the models. A summary of the results is given in Table 3-3. The range of the output distribution (95th/5th) varies between approximately 4 and 14 for the different hazard categories. These are considered to be tight distributions. The relatively large number of basic events and cut sets used in the parametric uncertainty analysis is hypothesized to dilute (mask) the effect of those basic events with higher uncertainties. This hypothesis is supported by a test documented in the overview report for internal events and floods (NRC, 2022a).

The results of parametric uncertainty analysis only provide limited insights due to the reason stated above. However, greater insights can be obtained by focusing on modeling uncertainty; in particular, as related to the values of the three basic events introduced in 2020-FLEX case.

Such modeling uncertainty analyses were performed and are documented in the following section, where CDFs of various cases were quantified and compared.

3-2

3.1.2 Results of Alternative Analyses Several alternative analyses were performed to better assess the effect on plant CDF of introducing 2020-FLEX case changes into the model. The results of these analyses are reported separately below for internal fires, seismic events, and high winds.

As discussed for the 2020-FLEX case for internal events and floods (NRC, 2022a), the failure probabilities used for FLEX and manual TDAFW pump operations are parametric values chosen by expert judgment, based on PRA experience and specific experience with construction of NRCs 70 Standardized Plant Analysis Risk (SPAR) models. The cases studied with different parametric values are used to support the assertion that the selected base case values are reasonable, and they do not shift the results unduly in either direction.

Internal Fires FLEX effectiveness 12 for internal fire events is estimated by studying the different cases summarized in Table 3-4. It is expected that the conditional probability of FLEX/TDAFW failure (represented by the p-parameter in the model) for internal fires is higher than that for sunny-day events (i.e., internal events and internal flooding events). This is due to the following considerations:

• potential irreparable fire damage to some multiple support system trains (e.g., 480 volt [V] AC buses)

• effect of the fire on operator performance

• difficulty in identifying which structures, systems, and components (SSCs) are damaged in each fire scenario 13 In the estimation process, the p-parameter is applied to all SBO fire sequences with no consideration to basic events which may be intrinsically failed as a result of the specific fire scenario. To compensate for this, a p-parameter value of 0.50 is selected as the representative case for the fire events. This is considerably higher than the p-parameter value of 0.09 used for sunny-day events (NRC, 2022a) and is likely to be pessimistic for SBO sequences from internal fires. This conservative p-parameter value is used because potential fire damage to equipment that could prevent implementation of FLEX or continued operation of TDAFW is not considered in the evaluation, which is consistent with the limited level of rigor for this sensitivity analysis.

As shown in Table 3-4, internal fire CDF is reduced by 13 percent in the 2020-FLEX case. This table also shows that even with a perfect FLEX, the reduction in CDF from internal fires is limited to 18 percent. Based on these results, the p-parameter used for the 2020-FLEX case appears to be a reasonable choice to be further studied as part of the Level 2 and Level 3 PRA analyses.

12 The term FLEX effectiveness is defined as the reduction in CDF from the cumulative impact of the new RCP passive shutdown seals and potential implementation of the FLEX strategies in response to an ELAP and continued TDAFW pump operation given a complete loss of all installed AC and DC power.

13 This is an internal fire events modeling limitation that is not intrinsically related to the nature of the fire event.

3-3

Seismic Events FLEX effectiveness for seismic events is estimated by studying the different cases summarized in Table 3-5. The selected p-parameter value is higher for seismic events compared to internal events (0.5 vs. 0.09), indicating that the FLEX/TDAFW combination is more likely to fail during SBO events that result from a seismic event. For the selected p-parameter, the FLEX effectiveness was calculated to be 21 percent.

An additional case was evaluated with a slightly more optimistic p-parameter (0.3). This additional case results in a modest increase in FLEX effectiveness, from 21 percent to 31 percent. However, the p-parameter value of 0.5 was retained for the 2020-FLEX case so as not to imply a greater likelihood of FLEX success for seismic events as compared to internal fires.

Note that FLEX effectiveness is expected to be reduced as the seismic event intensity increases. However, a single p-parameter value was used across the spectrum of seismic events to avoid giving the impression of unwarranted accuracy. The chosen p-parameter value is deemed to represent (pessimistically) a mid-range seismic event (e.g., seismic bin 3 or 4).

Also note that in this sensitivity analysis, no credit is given for the FLEX/TDAFW combination for seismic bins 7 and 8, which are dominated by major structural failures and, therefore, SBO events do not contribute to the risk from these bins.

As shown in Table 3-5, seismic CDF is reduced by 21 percent in the 2020-FLEX case. This table also shows that even with a perfect FLEX, the reduction in CDF from seismic events is limited to 45 percent. Based on these results, the p-parameter used for the 2020-FLEX case appears to be a reasonable choice to be further studied as part of the Level 2 and Level 3 PRA analyses.

High Winds FLEX effectiveness for high winds is estimated by studying the different cases summarized in Table 3-6. It is expected that the conditional probability of FLEX/TDAFW failure (represented by the p-parameter in the model) for high winds is higher than for sunny day events (internal events and internal flooding events). This is due to potential damage to some structures that may occur during the high wind event and the effect that high winds may have on operator performance.

The wind-related events modeled in the base case almost always result in a LOOP. Therefore, the proportion of SBO sequences for the wind-related events is higher than for the other hazard categories. Thus, wind-related events have more sequences that may benefit from the FLEX strategies and equipment.

A p-parameter value of 0.25 is selected as the representative case for the wind-related 2020-FLEX case. This is considerably higher than the p-parameter value of 0.09 used for sunny-day events (NRC, 2022a). Note that the p-parameter value of 0.25 for wind-related events is smaller than the one for seismic events (0.5). It is expected that FLEX will be more effective for wind-related events than seismic events. Moreover, it is believed that there are more modeling uncertainties associated with FLEX use during seismic events.

The overall FLEX effectiveness for wind-related events is estimated to be 65 percent. This compares well with the weather-related LOOP (LOOPWR) FLEX effectiveness of 76 percent 3-4

previously calculated for internal events (NRC, 2022a). Wind-related FLEX effectiveness is expected to be somewhat lower than for weather-related LOOPs due to potential structural damage to SSCs and the effect of the event on operator performance.

As shown in Table 3-6, high wind CDF is reduced by 65 percent in the 2020-FLEX case. This table also shows that even with a perfect FLEX, the reduction in CDF from seismic events is limited to 88 percent. Based on these results, the p-parameter used for the 2020-FLEX case appears to be a reasonable choice to be further studied as part of the Level 2 and Level 3 PRA analyses.

Summary The percentage CDF reduction (i.e., FLEX effectiveness) due to installation of passive RCP seals, implementation of FLEX strategies following declaration of ELAP, and continued TDAFW pump operation given the loss of all installed AC and DC power, is estimated for each hazard category in the previous sections. Figure 3-3 visually depicts FLEX effectiveness for two cases:

(1) the FLEX base case (i.e., the 2020-FLEX case), and (2) the perfect FLEX case. Note that a relatively flat slope in the CDF curve between the base case and the perfect FLEX case indicates that only limited additional CDF benefit can be obtained by demonstrating greater effectiveness for FLEX.

It should be understood that this model is a parametric study that uses p as the parameter of merit. In general, it is deemed that the p-parameter values quoted above may be pessimistic for all hazard categories with the possible exception of the internal events category. This pessimism is deemed to be appropriate since the three hazard categories in question (i.e.,

internal fire events, seismic events, and high wind events) may have larger modeling uncertainties due to the nature of the events and their impact on operator actions (both main control room actions and local FLEX actions). It should also be noted that the results of the case studies indicate that for the purposes of the L3PRA project, there would be very little value in performing a more rigorous and detailed assessment of the FLEX failure probability.

3.1.3 Initial Insights This section provides some initial insights from the Level 1 PRAs for internal fires, seismic events, and high winds, both for the Circa-2012 case and the 2020-FLEX case.

Internal Fires The following initial insights for the Level 1 PRA for internal fires for the Circa-2012 case are adapted from NRC (2023a). These insights have been augmented with information, when available, for the 2020-FLEX case.

Based on their Fussell-Vesely (FV) importance measure, 14 human failure events (HFEs) are the major contributors to CDF in the L3PRA project fire PRA. This is due to the limited mitigating systems available in many fire scenarios. The HFE with the highest percentage contribution to CDF (24 percent) is operator failure to initiate bleed and feed cooling in the absence of secondary-side heat removal. This action either occurs as an independent operator failure or in 14 The FV importance measure for a basic event represents the fraction of overall CDF contributed by the sum of all minimal cut sets containing that basic event.

3-5

cut sets with the HFE for failure to control auxiliary feedwater (AFW), for which there is a moderate dependency.

The HFE with the second highest percentage contribution to CDF (11 percent) is operator failure to trip the RCPs given loss of seal cooling. Per the WOG-2000 RCP seal loss-of-coolant accident (LOCA) model, failure to trip the RCPs within 13 minutes after the loss of RCP seal cooling results in an RCP seal LOCA. This can have a significant impact on CDF, since under most fire scenarios, long-term cooling is unavailable because the fire causes a direct failure of either the auxiliary component cooling water system or the nuclear service cooling water system.

The HFE with the third highest percentage contribution to CDF (approximately 10 percent) is failure to control AFW given a fire causes the spurious operation of the system (e.g., spurious starting of the AFW pumps). Failure to control AFW flow under these conditions will lead to a loss of secondary-side heat removal.

HFEs are also major contributors to CDF for the 2020-FLEX case. The three HFEs that contribute the most to fire CDF in the Circa-2012 case, discussed above, also contribute the most to CDF for the 2020-FLEX case, though their percent contributions are slightly higher (28 percent, 13 percent, and 11 percent, respectively).

Mitigating systems are affected by the different fire scenarios modeled in the Level 1 fire PRA, due to individual equipment or trains being directly failed. Since the fire scenarios are dominated by the direct effects of the fire and human errors, random hardware failures are typically not as important as they are for internal event scenarios.

For the Circa-2012 case, the hardware component failures with the highest percentage contribution to CDF (both approximately 9 percent) are (1) the RCP stage 2 seal failure given all seal cooling is lost and (2) spurious opening of the turbine-driven AFW pump steam inlet valve.

The spurious opening of this valve will cause the turbine-driven AFW pump to start, which can lead to overfilling the steam generators and an induced steam line break.

The two component failures with the next highest percentage contribution to CDF (both approximately 5 percent) are failure of the two emergency diesel generators (EDGs) to operate for the mission time. The EDGs have relatively high importance because multiple fire scenarios cause a LOOP, thereby requiring onsite emergency power to start and operate to provide essential AC power.

All other hardware component failures individually contribute less than 4 percent to total fire CDF.

For the 2020-FLEX case, hardware component failures also make only a limited contribution to internal fire CDF and the leading contributors are generally similar to those for the Circa-2012 case. However, unlike the Circa-2012 case, the RCP stage 2 seal failure is not a significant contributor in the 2020-FLEX case because the installation of the new RCP shutdown seals minimizes the importance of RCP seal LOCAs.

3-6

Seismic Events The following initial insights for the Level 1 PRA for seismic events for the Circa-2012 case are adapted from NRC (2023b). These insights have been augmented with information, when available, for the 2020-FLEX case.

As mentioned previously, the total plant CDF associated with seismic events for the Circa-2012 case is 1.110-5/rcy. Seventy-four percent of this CDF is due to seismic events in the range of 0.5g to 1.5g, represented by bins 3, 4, 5, and 6. Although higher seismic bins have conditional core damage probabilities (CCDPs) of (or nearly) 1.0, their CDF contribution is small (2 percent), due to their low seismic initiating event frequencies. The remaining CDF contribution (23 percent) comes from the lowest two bins, bins 1 and 2. It should be noted that some modeling assumptions may have increased the relative contribution from bins 1 and 2.

In the 2020-FLEX case, seismic bins 3, 4, 5, and 6 still make the most significant contribution to total seismic CDF. The contribution from these four seismic bins increases to approximately 82 percent in the 2020-FLEX case. The contribution from the lowest two bins (bins 1 and 2) is reduced to approximately 14 percent. However, as discussed in Section 3.1.2, a single p-parameter was selected to be used for all seismic bins. This simplification overstates the effectiveness of the FLEX strategies and continued TDAFW pump operation for the higher seismic bins and understates the effectiveness for the lower seismic bins. Accordingly, it is expected that a more rigorous analysis of FLEX effectiveness would lead to an even greater reduction in contribution to seismic CDF from the lower seismic bins (and a corresponding increase in the contribution from the higher seismic bins).

In the L3PRA project seismic PRA for the Circa-2012 case, seismically induced failures constitute the vast majority of the most important basic events, based on their FV importance measures. However, two of the most important basic events (in terms of FV importance) are independent random (non-seismic) failure of the two EDGs. The relative importance of EDG failures is due to the high likelihood of losing offsite power for the modeled seismic events. The other most important independent random failures involve maintenance on these EDGs, failure of the reserve auxiliary transformer supply breakers, and failure of the EDG load sequencers, again due to the high likelihood of losing offsite power.

For the Circa-2012 case, seismically induced LOOP basic events for seismic bins 1, 2, 3, and 4 represent four of the top five basic events (in terms of FV importance). These four basic events contribute to more than 40 percent of the total seismic CDF. However, this understates the importance of LOOP, since in the four higher seismic bins (5, 6, 7, and 8), LOOP is modeled to occur implicitly; thus, it is not assigned an FV importance.

Seismically induced uncorrelated failure of a 120V panel located in the auxiliary building also has a high collective FV importance (when considering all seismic bins), contributing nearly 14 percent to total seismic CDF in the Circa-2012 case. Failure of this panel leads to the loss of the train B load sequencer, which leads to core damage when combined with seismically induced anticipated transient without scram (ATWS) or random failures causing loss of critical train A equipment.

For seismically induced failures, the EDGs have a relatively high collective FV importance when considering all seismic bins and all EDG seismically induced failure modes. This includes seismically induced correlated failures of EDG supporting equipment, such as the lube oil heat 3-7

exchangers, the fuel oil transfer pumps, and the ventilation dampers, as well as seismically induced correlated failure of the EDG buildings.

For the 2020-FLEX case, seismically induced failures continue to constitute the vast majority of the most important basic events. However, in this case, the two most important basic events (in terms of FV importance) are the failure to implement FLEX strategies and the failure to continue TDAFW pump operation given the loss of all installed AC and DC power. Both basic events contribute approximately 30 percent to total seismic CDF since they always occur in cut sets together.

For the 2020-FLEX case, seismically induced LOOP basic events for seismic bins 1, 2, 3, and 4 represent four of the next five top basic events (in terms of FV importance). These four basic events contribute to approximately 34 percent of the total seismic CDF. However, as with the Circa-2012 case, this understates the importance of LOOP, since in the four higher seismic bins (5, 6, 7, and 8), LOOP is modeled to occur implicitly; thus, it is not assigned an FV importance.

High Winds The following initial insights for the Level 1 PRA for high winds for the Circa-2012 case are adapted from NRC (2023c). These insights have been augmented with information, when available, for the 2020-FLEX case.

As mentioned previously, the total plant CDF associated with high winds for the Circa-2012 case is 1.410-5/rcy. Approximately 75 percent of this CDF is due to straight-line winds, with the remaining 25 percent due to tornadoes. Over 77 percent of the straight-line wind CDF comes from the two lowest wind bins, that is, with wind speeds less than 129 mph.

The top five high wind accident sequences account for approximately 63 percent of the total high wind CDF. These accident sequences all involve a wind-generated LOOP, failure of the emergency AC power system, and failure to recover offsite power prior to core damage.

The basic events with the highest FV importance are the random failure to run of each EDG (contributing approximately 37 percent each to total high wind CDF). These are followed by the failure to recover offsite power in two hours, contributing approximately 15 percent. EDGs down for scheduled maintenance and other random or common-cause failures of emergency AC power system components are also significant risk contributors. The wind-related failures with the largest contribution to total wind CDF generally involve the transmission tower and structures and components associated with the standby auxiliary transformer.

The 2020-FLEX case is consistent with the Circa-2012 case in that approximately 75 percent of the CDF is due to straight-line winds, with the remaining 25 percent due to tornadoes. However, the absolute CDF in each of these cases is reduced by approximately 65 percent as compared to the Circa-2012 case. Again, consistent with the Circa-2012 case, over 77 percent of the straight-line wind CDF in the 2020-FLEX case comes from the two lowest wind bins, that is, with wind speeds less than 129 mph.

At the basic event level (i.e., hardware component failures and human errors), the major contributors in the 2020-FLEX case are fairly consistent with those in the Circa-2012 case. This is also true for the major wind-related failures.

3-8

Summary The changes associated with the 2020-FLEX case result in reduced CDF contributions from sequences involving SBO events or RCP seal failures. As stated previously and shown in Table 3-2, these changes are most effective at reducing CDF for internal events and high winds (reduction in CDF on the order of 60 percent). This is because the types of measures incorporated into the 2020-FLEX case have a higher relative likelihood of success for internal events and high winds than for internal fires and seismic events and because the CDF for internal events and high winds is dominated by LOOP and loss of NSCW sequences, both of which benefit significantly from these measures. For all hazards combined, the FLEX-related changes reduce CDF from 1.5x10-4/rcy to 9.3x10-5/rcy (a reduction of 38 percent).

The major contributors to CDF (in terms of hardware component failures and human errors) are similar for the Circa-2012 and 2020-FLEX cases, with two principal exceptions. The first exception is that unlike the Circa-2012 case, the RCP stage 2 seal failure is not a significant contributor in the 2020-FLEX case for internal fires because the installation of the new RCP shutdown seals minimizes the importance of RCP seal LOCAs. The second exception is that in the 2020-FLEX case, the two most significant contributors to seismic CDF are the failure to implement the specific strategies incorporated into the 2020-FLEX case (i.e., the FLEX strategies and continued operation of the TDAFW pump operation given the loss of all installed AC and DC power).

3-9

CDF % CIRCA-2012 CASE (1.51E-04/RCY)

High Winds (9.2%)

Seismic Events (7.2%)

Internal Events and Floods (42.9%)

Internal Fires (40.7%)

Figure 3-1 CDF Percentages by Hazard Group for Circa-2012 Case CDF% 2020-FLEX CASE (9.34E-05/RCY)

High Winds (5.2%)

Seismic Events (9.1%) Internal Events and Floods (28.6%)

Internal Fires (57.1%)

Figure 3-2 CDF Percentages by Hazard Group for 2020-FLEX Case 3-10

Figure 3-3 FLEX Impact on CDF by Hazard Category Table 3-1 FLEX Failure Probabilities by Hazard Category Basic Event Name Failure Probability Internal Seismic High Wind Internal Events Events Events Fires F 1-FLEX-FAILS 0.30 0.7 0.5 0.7 S 1-RCS-SDS-FC 0.01 0.01 0.01 0.01 T 1-AFW-SBO-NO-FLEX-FA 0.30 0.715 0.5 0.715 Combined FLEX failure 0.09 0.5 0.25 0.5 probability (p = F*T) 3-11

Table 3-2 CDF by Hazard Category Circa-2012 CDF 2020-FLEX CDF Hazard Category (/rcy) (/rcy) CDF Reduction Internal events and floods 6.47E-05 2.67E-05 59%

Internal fires 6.14E-05 5.34E-05 13%

Seismic events 1.08E-05 8.49E-06 21%

High winds 1.38E-05 4.85E-06 65%

Total 1.51E-04 9.34E-05 38%

Table 3-3 Summary of Hazard Category Parametric Uncertainty Analyses Quantity Fire CDF Seismic CDF Wind CDF Point Estimate* 5.11E-05/rcy 8.44E-06/rcy 4.54E-06/rcy 5th Percentile 2.36E-05/rcy 1.96E-06/rcy 8.06E-07/rcy Median 4.53E-05/rcy 5.87E-06/rcy 2.89E-06/rcy Mean 5.12E-05/rcy 8.27E-06/rcy 4.10E-06/rcy 95 Percentile th 9.62E-05/rcy 2.20E-05/rcy 1.10E-05/rcy 95th / 5th 4.08 11.3 13.7 95th / Mean 1.88 2.66 2.69 Standard Deviation 4.24E-06/rcy 8.50E-06/rcy 4.60E-06/rcy Skew 4.36 5.25 12.0 Kurtosis 36.9 58.6 252

• The point estimates reported in this table do not match the values in Table 3-2 because only a subset of the total cut sets was used for the parametric uncertainty analysis.

3-12

Table 3-4 Summary of FLEX Cases for Internal Fires Circa-2012 2020-FLEX Perfect FLEX Case Case Case F FLEX basic event failure probability TRUE 0.7 FALSE S SDS basic event failure probability TRUE 0.01 0.01 T TDAFW basic event failure probability TRUE 0.715 FALSE p=F*T 1 0.5 0 CDF 6.14E-05 5.34E-05 5.04E-05 FLEX effectiveness N/A 13% 18%

Table 3-5 Summary of FLEX Cases for Seismic Events Circa-2012 2020-FLEX Perfect FLEX FLEX-1 Case Case Case Case FLEX basic event failure F TRUE 0.7 0.5 FALSE probability SDS basic event failure S TRUE 0.01 0.01 0.01 probability TDAFW basic event T TRUE 0.715 0.6 FALSE failure probability p=F*T 1 0.5 0.3 0 CDF 1.08E-05 8.49E-06 7.46E-06 5.89E-06 FLEX effectiveness N/A 21% 31% 45%

Table 3-6 Summary of FLEX Cases for High Winds Circa-2012 2020-FLEX Perfect FLEX FLEX-1 Case Case Case Case FLEX basic event failure F TRUE 0.5 0.3 FALSE probability SDS basic event failure S TRUE 0.01 0.01 0.01 probability TDAFW basic event T TRUE 0.5 0.3 FALSE failure probability p=F*T 1 0.25 0.09 0 CDF 1.38E-05 4.85E-06 2.82E-06 1.70E-06 FLEX effectiveness N/A 65% 80% 88%

3-13

3.2 Level 2 PRA This section provides a summary of the results and insights from the reactor, at-power, Level 2 PRA for internal fires, seismic events, and high winds for a single unit. Section 3.2.1 provides the release frequency results for both the Circa-2012 and 2020-FLEX cases. Section 3.2.2 discusses alternative analyses to assess the effects of modeling assumptions on the Level 2 PRA results for the 2020-FLEX case. Section 3.2.3 discusses insights from the reactor, at-power, Level 2 PRA for internal fires, seismic events, and high winds, including a discussion of the dominant contributors to release category frequencies for both the Circa-2012 and 2020-FLEX cases.

3.2.1 Results of Circa-2012 and 2020-FLEX Cases The 2020-FLEX case updates the Circa-2012 models to include the new RCP shutdown seals and FLEX strategies and equipment for responding to an ELAP. The FLEX strategies are intended to provide coping capability to prevent core damage. Therefore, the primary effect of FLEX strategies on the PRA model is reduced CDF contributions from sequences involving SBO events or RCP seal failures in the Level 1 PRA model (as discussed in Section 3.1.1). The main impact on the Level 2 PRA model for FLEX strategies is carrying forward the modified Level 1 PRA sequences, which results in reduced frequencies for the applicable release categories.

This section provides a comparison of the 2020-FLEX case results to the Circa-2012 case. The description of the Circa-2012 Level 2 PRA model and results for internal fires, seismic events, and high winds during power operation are provided in the associated L3PRA project report (NRC, 2023d). The Circa-2012 case is based on the reference plant as it was designed and operated as of 2012 and does not reflect the FLEX strategies. However, the Circa-2012 case does include severe accident mitigating strategies that can delay or arrest core damage and subsequent releases. The Level 2 PRA for the Circa-2012 case considers extended manual operation of TDAFW pump for some SBO sequences. For the 2020-FLEX case, the Level 2 PRA model is revised to avoid applying conflicting credit for the same extended TDAFW operation that is represented in the 2020-FLEX case for the Level 1 PRA. The net effect on the model results is that the combined effects of the FLEX strategies and continued TDAFW pump operation in the 2020-FLEX case lead to significantly reduced frequency contribution from SBO sequences compared to the Circa-2012 case that included limited credit for extended TDAFW pump operation with a high probability of failure. 15 This section contains discussions on the following topics:

• Frequency inflation

• Release category frequencies

• Negative FLEX effectiveness 15 The Circa-2012 case Level 2 PRA uses a human error failure probability of 0.65 for extended TDAFW pump operation during certain slow-developing SBO sequences. For the FLEX parametric sensitivity cases documented in this report, the individual probabilities assigned for extended TDAFW pump operation (and for the FLEX strategies) have no particular significance by themselves and have been chosen independently of the failure probability of 0.65 previously used in the Circa-2012 Level 2 PRA model. Section 3.1.2 provides additional discussion about the failure probabilities selected for use in this sensitivity analysis. Given the lack of data and detailed modeling, the values selected for the FLEX parametric sensitivity analysis are only appropriate for the purposes of this analysis.

3-14

• FLEX impacts on representative accident scenarios and source terms

• Summary of Level 2 PRA surrogate risk metric results

• Assessment of parameter uncertainty Frequency Inflation The frequency results from Level 1 PRA CDF to Level 2 PRA release frequency can differ because of the approximation methods used for model quantification. The effect of non-minimal cut sets being carried forward to the Level 2 PRA sequences and the large failure probabilities inherent to the Level 2 PRA analysis can lead to inflation of the frequency results. Deflation of the frequency results can also occur due to truncation issues. Since release frequency includes additional success and failures (i.e., those occurring after the onset of core damage), more cut sets will fall below the truncation limit. Table 3-7 shows the inflation (or deflation) of frequency results going from Level 1 PRA CDF to Level 2 PRA release frequency for the 2020-FLEX case.

Fire and seismic sequences are especially prone to inflation due to the presence of very high failure probabilities for many components.

Release Category Frequencies The Level 2 PRA accident sequences are binned into release categories, as described in the Level 2 PRA report (NRC, 2023d). A description of each release category is provided in Table 3-8. The release category frequency results for the 2020-FLEX case and the Circa-2012 case are provided in Table 3-9, Table 3-10, and Table 3-11 for internal fires, seismic events, and high winds, respectively. The total release frequency is reduced in the 2020-FLEX cases because of the reduction in CDF.

Observations from the release category results:

• The overall impact on total release frequency for each hazard is a reduction of 16 percent for fire, 27 percent for seismic, and 72 percent for wind. Each of these is greater than the FLEX impact on CDF for the same hazard, due to numerical inaccuracies such as frequency inflation and cutset truncation.

• There is little to no impact on frequency of the pressure-induced steam generator tube rupture (SGTR) or ISLOCA release categories (1-REL-SGTR-O, 1-REL-SGTR-O-SC, and 1-REL-V-F). For fire, seismic, and wind, there are no SGTR or ISLOCA initiators.

ISLOCA occurs only as a result of earthquake in seismic bin 8, and core damage is assumed, so FLEX has no effect.

• As with internal events, some of the largest FLEX impacts are seen for the release categories that are dominated by SBO sequences: 1-REL-CIF, 1-REL-ICF-BURN, and 1-REL-LCF.

• Also as expected, release categories that are dominated by RCP seal failures are significantly reduced: 1-REL-ECF, 1-REL-LCF-SC, and 1-REL-NOCF

• For high winds, FLEX effectiveness in most release categories is similar to internal events. The exceptions are ICF-BURN-SC and LCF-SC, which in the wind model are mostly unaffected by FLEX.

3-15

• The seismic results do not follow the same trends. The 1-REL-NOCF release frequency is by far the most affected, with a reduction of over 50 percent. 1-REL-ICF-BURN and 1-REL-LCF see considerably smaller effects than they do for internal events or high winds (about 20 percent reduction). The effects on the early releases 1-REL-CIF and 1-REL-ISGTR are comparatively minor.

• Both fire and seismic show a substantial reduction in 1-REL-CIF-SC, unlike internal events where it is unaffected. This difference is likely not very meaningfulsince the overall frequency of 1-REL-CIF-SC is very low (especially for internal events, at 10-11), it can be more easily affected by cutset truncation issues and, in general, the influence of model changes is relatively noisy.

• The fire results include some negative FLEX effectiveness valuesthat is, for some release categories, the frequency in the 2020-FLEX case is very slightly higher than in the Circa-2012 case. These categories are 1-REL-SGTR-O and 1-REL-SGTR-O-SC, which, as noted above, would not be expected to show any effect from FLEX model changes. These are discussed below.

• The FLEX effects on fire results otherwise follow the same general trends as internal events, with the largest reductions in LCF, LCF-SC, NOCF, ICF-BURN, and ISGTR.

Negative FLEX Effectiveness For two release categories in the fire results (Table 3-9), the frequency is higher for the 2020-FLEX case than for the Circa-2012 case. This outcome is unexpected, since the addition of FLEX mitigation strategies can only prevent core damage. As currently modeled, it should never affect containment status for scenarios where core damage does occur (and hence, it should never change the release category). The increase occurs because of sequence pruning for the fire model. Only the sequences contributing the top 99.9 percent of fire CDF are retained in the Level 2 PRA solution. Because some fire sequences have their CDF significantly reduced by the addition of FLEX modeling, they may fall below the pruning threshold (even though the total fire CDF is also reduced, this reduction is concentrated in certain sequences, primarily station blackouts). Other sequences that are not affected by FLEX, such as pressure-induced SGTR sequences, now constitute a larger fraction of the total, which may push some of them above the pruning threshold. Although this will affect a very small portion of the total release frequency (far less than 0.1 percent), it can be a noticeable fraction of a specific low-frequency release categoryin this case, over 1 percent of 1-REL-SGTR-O and 1-REL-SGTR-O-SC. This change can be safely disregarded.

FLEX Impacts on Representative Accident Scenarios and Source Terms The inclusion of FLEX strategies and passive RCP shutdown seals will generally result in delaying or preventing core damage events and subsequent releases. The primary impact on the model is the reduction of release category frequencies, but the model changes also impact the nature of the accident progression and related modeling assumptions. Each release category includes a mix of different accident sequences that have similar attributes but are not identical. The representative source term scenarios cannot exactly represent all the contributing elements. The representative source term should consider the balance of timing and magnitude to select a source term that conservatively bounds the range of outcomes for that release 3-16

category. The considerations related to within-release-category variability are discussed in Section 2.5.2 of the L3PRA project Level 2 PRA for internal events and floods (NRC, 2022b).

While the release category frequency results have changed in the 2020-FLEX case, the representative source terms are still consistent with the Level 2 PRA logic model assumptions that were used in defining the release categories. Therefore, the 2020-FLEX case uses the same representative source terms as the Circa-2012 case. The FLEX modeling changes can also impact the warning time available for evacuating the population and preventing early radiological consequences. The warning times are broadly informed by the timing of the representative accident scenarios. Therefore, the same warning times are also assumed to apply to the 2020-FLEX case. As such, no changes were made to the representative accident scenarios for the 2020-FLEX case.

Summary of Level 2 Surrogate Risk Metric Results It is often desirable to report the results of a Level 2 PRA in terms of one or more surrogate risk metrics. Section 2.6.1 of NRC (2022b) defines several such metrics, with more detail provided in Section D.13 of NRC (2022b). The three that are discussed in this document include large early release frequency (LERF) based on early fatalities, large release frequency (LRF), and conditional containment failure probability (CCFP). The definitions for these three metrics, adapted from NRC (2022b), are:

• LERF (early fatalities): release categories with a representative source term warning time (based on iodine release exceeding 1 percent) of less than 3.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> simultaneous with the cumulative iodine release fraction being greater than 4 percent

• LRF: the summation of the frequency of all release categories that include containment bypass or containment failure, excluding those where fission product scrubbing (or other mechanisms) result in a source term comparable to, or smaller than, the remainder of the (intact containment) source terms

• CCFP: the ratio of the release categories involving a failed or bypassed containment to the overall release frequency A comparison of the surrogate risk metric results for the Circa-2012 and 2020-FLEX cases is provided in Table 3-12. The contributing release categories for each surrogate risk metric are also provided in the notes to Table 3-12.

As can be seen from Table 3-12, LERF for the 2020-FLEX case is reduced by about 10 percent for fire, 14 percent for seismic, and 60 percent for wind, relative to the Circa-2012 case. These reductions are less than the overall reduction in total release frequency. Since LERF consists entirely of induced SGTR (ISGTR) (except for a tiny ISLOCA contribution to seismic LERF), this reflects largely the lack of importance of RCP seal LOCAs to ISGTR. Due to the very low failure probability of the shutdown seals, cutsets that contain that failure often fall below the truncation limit, so the exact contribution of shutdown seals to the reduction cannot be easily evaluated.

The FLEX events, on the other hand, appear in cutsets that contribute most of the wind ISGTR frequency, but less than half of seismic ISGTR and only about 10 percent of fire ISGTR.

LRF is reduced slightly more13 percent for fire, 20 percent for seismic, and 60 percent for wind. For internal events there is a large difference between FLEX effectiveness for LERF and 3-17

LRF (39 percent vs 60 percent), because ISLOCAs make a significant contribution to LERF; SBO sequences are a large contributor to LRF; and the new RCP shutdown seals, back-up power capabilities of FLEX, and the continued operation of TDAFW all help to mitigate SBO sequences, but not ISLOCAs. In addition, the types of measures incorporated into the 2020-FLEX case have a higher relative likelihood of success for internal events (and high winds) than for internal fires and seismic events.

Conditional containment failure probability is higher in all the 2020-FLEX cases when compared to the Circa-2012 cases. This is because, as can be seen from Table 3-9 to Table 3-11, the No containment failure release category (1-REL-NOCF) is greatly reduced in the FLEX cases (more so than most other release categories). Therefore, while the total frequency of severe accidents goes down significantly, the fraction of those accidents that lead to containment failure goes up. However, the most relevant fact is that the frequency of a severe accident that leads to containment failure goes down in the 2020-FLEX cases (though only modestly for fire and seismic).

Assessment of Parameter Uncertainty A parametric uncertainty analysis was performed on the integrated Level 1 and Level 2 PRA logic model for both the 2020-FLEX case and the Circa-2012 case to examine the influence of parameter uncertainty in the input parameters of the logic model on the release category frequencies. The parametric uncertainty can provide information about the overall uncertainty of the results. However, many of the significant sources of uncertainty arise from modeling choices and assumptions, rather than the input parameters. Still, the parameter uncertainty analysis is an important input to the Level 3 PRA consequence analysis. In the Level 3 PRA, the Level 2 PRA release category frequency uncertainty parameters are included with the statistical variation of offsite weather conditions to develop frequency-weighted complementary cumulative distribution function curves to illustrate the mean frequencies of exceeding different consequence levels for offsite consequence metrics (see Section 3.3.1).

3.2.2 Results of Alternative Analyses Alternative analyses were performed to assess the impact on the 2020-FLEX case results of alternate assumptions regarding the termination of radiological releases, a key source of uncertainty.

In both the Circa-2012 and 2020-FLEX cases the source term analyses assume a termination of radiological releases at 7 days (168 hours0.00194 days <br />0.0467 hours <br />2.777778e-4 weeks <br />6.3924e-5 months <br />) after the event initiation. However, it is possible that releases could be terminated earlier considering possible onsite and offsite resources and recovery actions that are beyond the scope of this model. Alternate termination times of 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br /> after severe accident management guideline (SAMG) entry and 60 hours6.944444e-4 days <br />0.0167 hours <br />9.920635e-5 weeks <br />2.283e-5 months <br /> after SAMG entry are considered to assess the impact of earlier release termination. Assumptions regarding these alternative release termination times are provided in Section D.21 of NRC (2022b).

Table 3-13 provides details of the key parameter timeline for each of the release category representative accident scenarios. This information assists in interpreting the alternate termination results. The key parameters are defined below.

• GE declaration - The timing of declaring a General Emergency (GE) is based on plant-specific Emergency Action Level determination guidance and the specific conditions of 3-18

the accident scenario. There are several different criteria and plant indications that can prompt the GE declaration. The assessments of each modeled accident scenario and the estimated times of GE declaration are discussed in Section 2.5.2 of NRC (2022b).

• SAMG entry - This marks the transition from the plant operators using the Emergency Operating Procedures to using the SAMGs to manage the accident response. The accident management staff will refer to the SAMGs when core damage is imminent or has occurred. The MELCOR simulated time to reach average temperature of coolant at core exit of 1200F is used as a surrogate for the timing of imminent core damage.

Navigation of the SAMGs is discussed further in Section D.14 of NRC (2022b).

• Cumulative I > 1 percent - This refers to the time when the cumulative environmental release of the iodine chemical class exceeds 1 percent of its initial core inventory mass.

This threshold is used as an indication of a release with potential to cause health effects.

The timing is an input into the calculation of warning time for the LERF definition.

• Warning time - The warning time is defined as the time when the cumulative environmental Iodine release fraction exceeds 1 percent minus the time that GE declaration occurs. Warning time is an input used in the LERF definition. The warning time gives an indication of the time available for evacuating populations, which can significantly influence the occurrence of early radiological health effects. Warning time is discussed further in the risk metric surrogate definitions in Section D.13 of NRC (2022b).

• Time to LERF threshold - The LERF definition includes criteria on the warning time and the cumulative environmental iodine release fraction exceeding 4 percent. In assessing the timing of reaching the threshold, the second criterion is the determining factor. The release categories that do not meet the LERF warning time criteria indicate N/A in this column.

• Time to LRF threshold - LRF is the summation of the frequency of all release categories that include containment bypass or containment failure, excluding those where fission product scrubbing (or other mechanisms) result in a source term comparable to, or smaller than, the remainder of the intact containment source terms. For the purposes of assessing timing of LRF, a threshold value is designated to determine when a release is significantly greater than the reference intact containment source term. The time when the cesium environmental release fraction exceeds 2.9x10-4 is used as the criterion for a large release that is not comparable to the intact containment source term. The LRF definition is discussed further in the risk metric surrogate definitions in Section D.13 of NRC (2022b).

• Time of Containment Failure - This refers to the timing of failure of the containment structure or timing of opening a containment bypass release pathway. A time of 0 hours0 days <br />0 hours <br />0 weeks <br />0 months <br /> indicates a containment bypass is open throughout the entire duration of the scenario.

Table 3-13 shows the times to reach the LERF, LRF, and containment failure criteria. If an earlier accident termination time is assumed, then some of the scenarios may not reach the thresholds for LERF, LRF, or containment failure.

Table 3-14 through Table 3-17 include results showing the impacts of alternate assumptions regarding the termination of radiological releases. As seen in the tables, the LERF result is 3-19

insensitive to the accident termination time assumptions. However, the LRF and CCFP results can be significantly reduced by the earlier accident termination alternatives. In the internal event and flood case (Table 3-14), accident termination at 36 or 60 hours6.944444e-4 days <br />0.0167 hours <br />9.920635e-5 weeks <br />2.283e-5 months <br /> after SAMG entry would reduce LRF by over 70 percent. This reduction is even more pronounced for high wind events (78 percent), but less so for seismic (62 percent) and internal fires (55 percent), because large releases for those hazards are not so heavily dominated by late containment failure due to overpressure. It should also be noted that besides reducing the LRF surrogate risk metric, earlier accident termination times also reduce the magnitude of the radiological releases (which is accounted for in the Level 3 PRA analyses). These reductions are largely independent of FLEX, and the reductions in LRF and CCDP for the Circa-2012 case (not shown) are similar.

This sensitivity analysis shows that selection of a shorter scenario modeling time results in reductions of LRF and CCFP. As discussed at the beginning of this section, if there are credible reasons to model an accident scenario termination time at 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br /> after SAMG entry (which, in many cases, is on the order of 2 days after accident initiation), then both LRF and CCDP would be reduced significantlyby a factor of approximately 2 to 4, depending on the hazard.

3.2.3 Initial Insights As discussed in Section 3.2.1, FLEX modeling reduces release frequency by 72 percent for high winds and tornados (similar to internal events), but just 16 percent for fire initiators and 27 percent for seismic initiators. The decrease is concentrated in release categories dominated by station blackout (late overpressure failure, intermediate combustion failure) and RCP seal LOCAs (intact containment, scrubbed overpressure failure). Containment bypass accidents are less amenable to these improvementsISLOCA and pressure-induced SGTR are not affected at all, and temperature-induced SGTR frequency is reduced to a lesser extent than overall CDF.

The dominant release categories for the 2020-FLEX case in terms of frequency contributions are 1-REL-LCF, 1-REL-NOCF, and 1-REL-ICF-BURN. The frequency results do not reflect the differences in the release magnitudes of the release categories and their overall contributions to risk. For example, the 1-REL-LCF and 1-REL-ICF-BURN release categories both contribute to LRF, while 1-REL-NOCF does not. The release categories 1-REL-CIF and 1-REL-ISGTR have a small contribution to the overall release frequency but have the highest contributions to the LERF risk metric. The release category contributions to the surrogate risk metrics are provided in Table 3-14 to Table 3-17.

The following points provide some other key results and insights:

• For the 2020-FLEX case, combined fire, seismic, and wind LERF is 7.6x10-7/rcy, exceeding the internal events and floods LERF of 5.7x10-7/rcy. In the Circa-2012 case, internal events contributed the majority of LERF, but the model changes for FLEX are more effective at reducing internal events LERF than they are for fire and seismic LERF.

This reduced FLEX effectiveness is partly inherent to the nature of the fire and seismic sequences that contribute to LERF, and partly created by the model assumptions about FLEX failure probabilities (see Table 3-1).

• While in the Circa-2012 case the combined LRF from fire, seismic, and wind was already greater than the internal events and floods LRF, that difference is magnified in the 2020-FLEX case, due in large part to the higher failure probabilities for FLEX and TDAFW extension. In the 2020-FLEX case, fire, seismic, and wind initiators together 3-20

contribute roughly 74 percent of total LRF. For some particularly consequential release categories, this difference is even greater. Fire, seismic, and wind initiators contribute over 81 percent of release frequency due to combustion (1-REL-ICF-BURN), 81 percent of unscrubbed pressure-induced SGTR (1-REL-SGTR-O), and 76 percent of thermally-induced SGTR (1-REL-ISGTR).

• The combined effects of the FLEX strategies and continued TDAFW pump operation in the 2020-FLEX cases lead to significantly reduced frequency contribution from SBO sequences compared to the Circa-2012 cases that included limited credit for extended TDAFW pump operation with a high probability of failure.

• The model changes for the 2020-FLEX cases are highly effective in reducing releases due to high wind events, but less effective for earthquakes and fires. High winds are more likely to have CDF caused by station blackout scenarios, and wind scenarios use a lower assumed failure probability for FLEX strategies.

• Because the effect of the model changes is to prevent core damage in scenarios where core damage is caused by station blackout or RCP seal LOCA, and many of these scenarios lead to intact containment, the 2020-FLEX cases show higher CCFP compared to the Circa-2012 cases (though the absolute frequency of severe accidents leading to containment failure is notably lower in the 2020-FLEX case).

• Given the relatively high probabilities of FLEX and TDAFW extension both failing (0.5 combined probability for fire and seismic, 0.25 for wind), some release category frequencies are reduced much more by the RCP shutdown seals than by the implementation of FLEX strategies. This is especially true for the intact containment release category (1-REL-NOCF).

Table 3-7 Comparison of Level 1 CDF and Level 2 Release Frequency for 2020-FLEX Case Release CDF - Ratio Frequency -

Hazard Category 2020-FLEX L2/L1 2020-FLEX Case (a) (b/a)

Case (b)

Internal Events & Floods 2.67E-05 2.63E-05 0.99 Internal Fires 5.34E-05 5.81E-05 1.09 Seismic 8.49E-06 1.08E-05 1.27 Wind 4.85E-06 4.64E-06 0.96 Total 9.34E-05 9.99E-05 1.07 3-21

Table 3-8 Description of Release Categories Name Description The containment eventually fails due to basemat ablation due to sustained core-concrete interaction. Only the airborne component of release to the 1-REL-BMT environment (which stems from normal containment leakage while the containment is pressurized) is modeled.

Release from the containment to the environment occurs via a containment 1-REL-CIF penetration that fails to be isolated by the containment isolation system, or a preexisting leakage path. The release is unmitigated.

Release from the containment to the environment occurs via a containment 1-REL-CIF-SC penetration that fails to be isolated by the containment isolation system, or a preexisting leakage path. The release is mitigated.

The containment fails before or around the time of vessel breach due to an 1-REL-ECF energetic event. This release may or may not benefit from any aerosol scrubbing.

The containment fails hours after vessel breach due to a global deflagration or 1-REL-ICF-BURN detonation. Releases to the environment are not mitigated significantly by sprays or water pools.

The containment fails hours after vessel breach due to a global deflagration or 1-REL-ICF-BURN-SC detonation. Releases to the environment benefit from scrubbing.

Release to the environment occurs via a thermally induced rupture of one or 1-REL-ISGTR more steam generator tubes after the time of core damage.

The containment fails tens of hours after the time of vessel breach due to long-1-REL-LCF term quasi-static overpressure. Releases to the environment are not mitigated significantly by sprays or water pools.

The containment fails tens of hours after the time of vessel breach due to long-1-REL-LCF-SC term quasi-static overpressure. Releases to the environment are mitigated by sprays and/or water pools.

Containment is not bypassed or failed, and radiological release to the 1-REL-NOCF environment occurs via design-basis containment leakage only. This release may or may not benefit from any aerosol scrubbing.

Release from the RCS to the environment occurs via ruptured steam generator 1-REL-SGTR-C tube(s), where the rupture occurred prior to core damage. ARVs and main steam relief valves remain predominantly closed.

Release from the RCS to the environment occurs via one or more ruptured SG tubes, where the rupture occurred prior to core damage. The release is not 1-REL-SGTR-O mitigated by water above the break point on the secondary side of the affected SG. One or more secondary-side relief valves are kept open during release as a deliberate action or fail in the open position.

Release from the RCS to the environment occurs via one or more ruptured SG tubes, where the rupture occurred prior to core damage. The release is 1-REL-SGTR-O-SC mitigated by water above the break point on the secondary side of the affected SG. One or more secondary-side relief valves are kept open during release as a deliberate action or fail in the open position.

Release occurs from the RCS to the auxiliary building via interfacing systems 1-REL-V LOCA. The break point may or may be not submerged. The auxiliary building remains intact.

Release occurs from the RCS to the auxiliary building via interfacing systems 1-REL-V-F LOCA. The break point is not submerged. The auxiliary building fails.

Release occurs from the RCS to the auxiliary building via interfacing systems 1-REL-V-F-SC LOCA. The break point is submerged. The auxiliary building fails.

3-22

Table 3-9 Internal Fire Release Category Frequencies Circa-2012 2020-FLEX FLEX RCF - RCF -

Release Category Case % of Case % of Impact Circa-2012 2020-FLEX Name Total Total (a-b)/a Case (a) Case (b)

Release Release  %

Total 6.88E-05 5.81E-05 15.51%

1-REL-BMT 3.70E-06 5.4% 3.53E-06 6.08% 4.62%

1-REL-CIF 4.05E-07 0.6% 3.71E-07 0.64% 8.25%

1-REL-CIF-SC 7.65E-08 0.1% 5.56E-08 0.10% 27.39%

1-REL-ECF 7.54E-09 0.0% 6.56E-09 0.01% 13.00%

1-REL-ICF-BURN 1.64E-05 23.8% 1.49E-05 25.59% 9.16%

1-REL-ICF-BURN-SC 2.71E-06 3.9% 2.58E-06 4.44% 4.83%

1-REL-ISGTR 5.51E-07 0.8% 4.94E-07 0.85% 10.31%

1-REL-LCF 2.13E-05 31.0% 1.79E-05 30.80% 16.00%

1-REL-LCF-SC 2.13E-06 3.1% 1.78E-06 3.06% 16.46%

1-REL-NOCF 2.13E-05 31.0% 1.63E-05 28.05% 23.47%

1-REL-SGTR-C 0.00E+00 0.0% 0.00E+00 0.00%

1-REL-SGTR-O 3.87E-08 0.1% 3.94E-08 0.07% -1.71%

1-REL-SGTR-O-SC 1.83E-07 0.3% 1.85E-07 0.32% -1.04%

1-REL-V 0.00E+00 0.0% 0.00E+00 0.00%

1-REL-V-F 0.00E+00 0.0% 0.00E+00 0.00%

1-REL-V-F-SC 0.00E+00 0.0% 0.00E+00 0.00%

3-23

Table 3-10 Seismic Event Release Category Frequencies Circa-2012 2020-FLEX FLEX RCF - RCF -

Release Category Case % of Case % of Impact Circa-2012 2020-FLEX Name Total Total (a-b)/a Case (a) Case (b)

Release Release  %

Total 1.48E-05 1.08E-05 26.89%

1-REL-BMT 1.31E-07 0.89% 1.24E-07 1.14% 5.73%

1-REL-CIF 1.15E-06 7.77% 1.00E-06 9.26% 12.87%

1-REL-CIF-SC 3.10E-08 0.21% 1.59E-08 0.15% 48.84%

1-REL-ECF 2.21E-09 0.01% 1.62E-09 0.01% 26.76%

1-REL-ICF-BURN 2.71E-06 18.33% 2.21E-06 20.41% 18.58%

1-REL-ICF-BURN-SC 2.71E-08 0.18% 2.07E-08 0.19% 23.63%

1-REL-ISGTR 2.27E-07 1.53% 1.95E-07 1.80% 14.27%

1-REL-LCF 7.46E-06 50.39% 5.80E-06 53.64% 22.18%

1-REL-LCF-SC 7.52E-08 0.51% 5.75E-08 0.53% 23.62%

1-REL-NOCF 2.75E-06 18.56% 1.15E-06 10.66% 58.01%

1-REL-SGTR-C 0.00E+00 0.00% 0.00E+00 0.00% 0.00%

1-REL-SGTR-O 6.38E-08 0.43% 6.38E-08 0.59% 0.00%

1-REL-SGTR-O-SC 1.72E-07 1.16% 1.72E-07 1.59% 0.00%

1-REL-V 0.00E+00 0.00% 0.00E+00 0.00% 0.00%

1-REL-V-F 2.32E-09 0.02% 2.32E-09 0.02% 0.00%

1-REL-V-F-SC 0.00E+00 0.00% 0.00E+00 0.00% 0.00%

3-24

Table 3-11 High Wind Release Category Frequencies Circa-2012 2020-FLEX FLEX RCF - RCF -

Release Category Case % of Case % of Impact Circa-2012 2020-FLEX Name Total Total (a-b)/a Case (a) Case (b)

Release Release  %

Total 1.67E-05 4.64E-06 72.2%

1-REL-BMT 8.44E-08 0.51% 8.44E-08 1.82% 0.0%

1-REL-CIF 1.46E-08 0.09% 3.29E-09 0.07% 77.5%

1-REL-CIF-SC 0 0.00% 0 0.00%

1-REL-ECF 9.58E-10 0.01% 2.25E-10 0.00% 76.5%

1-REL-ICF-BURN 1.97E-06 11.8% 7.63E-07 16.4% 61.3%

1-REL-ICF-BURN-SC 2.88E-07 1.72% 2.86E-07 6.17% 0.7%

1-REL-ISGTR 1.64E-07 0.98% 7.05E-08 1.52% 57.0%

1-REL-LCF 7.78E-06 46.5% 2.90E-06 62.5% 62.7%

1-REL-LCF-SC 6.96E-08 0.42% 6.58E-08 1.42% 5.5%

1-REL-NOCF 6.34E-06 38.0% 4.65E-07 10.0% 92.7%

1-REL-SGTR-C 0 0.00% 0 0.00%

1-REL-SGTR-O 4.17E-11 0.00% 4.17E-11 0.00% 0.0%

1-REL-SGTR-O-SC 3.71E-10 0.00% 3.71E-10 0.01% 0.0%

1-REL-V 0 0.00% 0 0.00%

1-REL-V-F 0 0.00% 0 0.00%

1-REL-V-F-SC 0 0.00% 0 0.00%

3-25

Table 3-12 Level 2 PRA Surrogate Risk Metrics LERF1 LRF2 CCFP3 Circa-2012 2020-FLEX Risk Metric Circa-2012 2020-FLEX Risk Metric Circa-2012 2020-FLEX Case Case Reduction Case Case Reduction Case Case Internal Events 9.31E-07/ry 5.70E-07/ry 38.8% 4.36E-05/ry 1.75E-05/ry 59.9% 0.656 0.785 Internal Fires 5.51E-07/ry 4.94E-07/ry 10.3% 4.11E-05/ry 3.57E-05/ry 13.1% 0.690 0.720 Seismic Events 2.29E-07/ry 1.97E-07/ry 14.1% 1.19E-05/ry 9.52E-06/ry 20.0% 0.814 0.893 High Winds 1.64E-07/ry 7.05E-08/ry 57.0% 1.03E-05/ry 3.80E-06/ry 60.2% 0.620 0.900

1. The release categories contributing to LERF (large early release frequency), which have the potential for early fatalities, are: 1-REL-ISGTR, 1-REL-V-F, and 1-REL-V-F-SC.

2. The release categories contributing to LRF (large release frequency) are: 1-REL-CIF, 1-REL-CIF-SC, 1-REL-ECF, 1-REL-ICF-BURN, 1-REL-ISGTR, 1-REL-LCF, 1-REL-LCF-SC, 1-REL-SGTR-C, 1-REL-SGTR-O, 1-REL-SGTR-O-SC, 1-REL-V, 1-REL-V-F, and 1-REL-V-F-SC.

3. The release categories contributing to CCFP (conditional containment failure probability) include all release categories resulting in containment failure or bypass: 1-REL-BMT, 1-REL-CIF, 1-REL-CIF-SC, 1-REL-ECF, 1-REL-ICF-BURN, 1-REL-ICF-BURN-SC, 1-REL-ISGTR, 1-REL-LCF, 1-REL-LCF-SC, 1-REL-SGTR-C, 1-REL-SGTR-O, 1-REL-SGTR-O-SC, 1-REL-V, 1-REL-V-F, and 1-REL-V-F-SC.

3-26

Table 3-13 Level 2 PRA Representative Accident Scenario Timelines Time of Time to Time to 36 hr 60 hr MELCOR Warning End of Release GE1 SAMG Cumul. LERF LRF Time of Cont. after after Rep. Time3 Calc.

Category (hr) Entry2 I > 1% Thresh. Thresh. Failure (hr) SAMG SAMG Case (hr) (hr)

(hr) 6 7 Entry (hr) Entry (hr) 1-REL-BMT 6 13 14.7 Never >155 N/A Never 129 50.7 74.7 168 0

1-REL-CIF 7 3 15.6 18 15 4 ~30 ~17 51.6 75.6 168 (cont. bypass) 1-REL-CIF- 0 7A 3 15.6 18 15 4 Never ~17 51.6 75.6 168 SC (cont. bypass) 1-REL-ECF 2A 8 13.5 22 14 4 ~23 ~22 21.6 49.5 73.5 140 8 1-REL-ICF-1A2 3 15.5 33 30 N/A ~28 28 51.5 75.5 140 8 BURN 1-REL-ICF-1A2 3 15.5 33 30 N/A Never 28 51.5 75.5 28.0 9 BURN-SC 10.1 - SGTR 1-REL-3A2 8 10.1 11 35 ~12 10.1 87.8 - OP 46.1 70.1 168 ISGTR 3-27 failure 1-REL-LCF 1B 3 3.5 146 143 N/A ~68 47.9 39.5 63.5 168 1-REL-LCF-2R2 8 13.5 Never > 160 N/A ~140 120 49.5 73.5 168 SC 1-REL-NOCF 2R1 8 13.5 Never > 160 N/A Never Never 49.5 73.5 168 1-REL- 0 8 47 49.1 52 54 Never ~50 85.1 109.1 168 SGTR-C (cont. bypass) 1-REL- 0 8B 47 49.1 51 44 ~50 ~50 85.1 109.1 168 SGTR-O (cont. bypass) 1-REL- 0 8BR1 47 49.1 Never > 160 N/A ~50 85.1 109.1 58.8 10 SGTR-O-SC (cont. bypass) 0 1-REL-V 5 7.5 9.5 Never > 64 N/A ~11 45.5 69.5 72 (cont. bypass) 0 1-REL-V-F 5D 1.25 2.9 3.2 1.95 5 ~4 ~3 38.9 62.9 72 (cont. bypass) 1-REL-V-F- 0 5B 1.25 2.9 3.2 1.95 5 ~4 ~3 38.9 62.9 72 SC (cont. bypass)

Table 3-13 Level 2 PRA Representative Accident Scenario Timelines (cont.)

Notes 1 GE is declared according to plant-specific Emergency Action Level determination guidance. The GE declaration times are estimated for the representative accident scenarios in Section 2.5.2 and Table 2-20 of the Level 2 PRA (NRC, 2022b).

2 SAMG entry is indicated by average temperature of coolant at core exit exceeding 1,200F.

3 Warning time is defined as the time at which cumulative environmental iodine release fraction exceeds 1 percent minus the time that GE conditions are met.

4 The warning time meets the criteria for LERF resulting in early injuries, i.e., warning time < 20 hours2.314815e-4 days <br />0.00556 hours <br />3.306878e-5 weeks <br />7.61e-6 months <br />.

5 The warning time meets the criteria for LERF resulting in early fatalities, i.e., warning time < 3.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />.

6 The LERF criteria are met when the warning time is less than the designated time (3.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> for early fatalities and 20 hours2.314815e-4 days <br />0.00556 hours <br />3.306878e-5 weeks <br />7.61e-6 months <br /> for early injuries) and the cumulative environmental iodine release fraction exceeds 4 percent.

7 The time when cumulative environmental release fraction of cesium exceeds 2.9x10-4 is used to indicate a large release, which is significantly larger than releases from an intact containment.

8 The calculation was ended at 140 hours0.00162 days <br />0.0389 hours <br />2.314815e-4 weeks <br />5.327e-5 months <br /> because releases have stabilized at this time.

9 The source term for case 1A2 truncated at the time of containment failure is used as a surrogate for this release category.

10 The calculation terminated during in-vessel recovery due to numerical problems; no significant changes in the results after this time are expected.

3-28

Table 3-14 Internal Event and Flood Level 2 PRA Surrogate Risk Metric Results - 2020-FLEX Case Release Time at which Airborne Radiological Releases are Terminated Release Category Category SAMG Entry + 36 Hours SAMG Entry + 60 Hours 7 Days after Event Initiation Name Frequency LERF LRF LERF LRF LERF LRF

(/rcy) CCFP CCFP CCFP

(/rcy) (/rcy) (/rcy) (/rcy) (/rcy) (/rcy) 1-REL-BMT 7.86E-07 (Note 1) 3.0%

1-REL-CIF 2.36E-08 2.36E-08 0.1% 2.36E-08 0.1% 2.36E-08 0.1%

1-REL-CIF-SC 1.11E-11 1.11E-11 0.0% 1.11E-11 0.0% 1.11E-11 0.0%

1-REL-ECF 3.14E-09 3.14E-09 0.0% 3.14E-09 0.0% 3.14E-09 0.0%

1-REL-ICF-BURN 4.14E-06 4.14E-06 15.8% 4.14E-06 15.8% 4.14E-06 15.8%

1-REL-ICF-BURN-SC 2.33E-06 8.9% 8.9% 8.9%

1-REL-ISGTR 2.39E-07 2.39E-07 2.39E-07 0.9% 2.39E-07 2.39E-07 0.9% 2.39E-07 2.39E-07 0.9%

1-REL-LCF 1.04E-05 39.6% 1.04E-05 39.6%

1-REL-LCF-SC 2.05E-06 2.05E-06 7.8%

3-29 1-REL-NOCF 5.64E-06 1-REL-SGTR-C 4.17E-08 4.17E-08 0.2% 4.17E-08 0.2% 4.17E-08 0.2%

1-REL-SGTR-O 2.42E-08 2.42E-08 0.1% 2.42E-08 0.1% 2.42E-08 0.1%

1-REL-SGTR-O-SC 2.31E-07 2.31E-07 0.9% 2.31E-07 0.9% 2.31E-07 0.9%

1-REL-V 1.27E-08 1.27E-08 0.0% 1.27E-08 0.0% 1.27E-08 0.0%

1-REL-V-F 1.01E-07 1.01E-07 1.01E-07 0.4% 1.01E-07 1.01E-07 0.4% 1.01E-07 1.01E-07 0.4%

1-REL-V-F-SC 2.30E-07 2.30E-07 2.30E-07 0.9% 2.30E-07 2.30E-07 0.9% 2.30E-07 2.30E-07 0.9%

Total 2.63E-05 5.70E-07 5.04E-06 28.1% 5.70E-07 5.04E-06 67.7% 5.70E-07 1.75E-05 78.5%

Notes

1. A similar set of results is shown in Table 2-23 of the Level 2 PRA (NRC, 2022b) for the Circa-2012 case. However, Table 2-23 includes an error. The CCFP contribution for the 1-REL-BMT release category was incorrectly included for the SAMG entry + 60 hours6.944444e-4 days <br />0.0167 hours <br />9.920635e-5 weeks <br />2.283e-5 months <br /> alternate termination time. The basemat melt-through scenario sees gradual erosion of the reactor vessel cavity. Containment is considered failed when radial erosion exceeds the thickness of the cavity wall. This occurs at 129 hours0.00149 days <br />0.0358 hours <br />2.132936e-4 weeks <br />4.90845e-5 months <br /> after the initiating event. The SAMG entry + 60 hours6.944444e-4 days <br />0.0167 hours <br />9.920635e-5 weeks <br />2.283e-5 months <br /> accident termination occurs at approximately 75 hours8.680556e-4 days <br />0.0208 hours <br />1.240079e-4 weeks <br />2.85375e-5 months <br /> after the initiating event.

Table 3-15 Internal Fire Level 2 PRA Surrogate Risk Metric Results - 2020-FLEX Case Release Time at which Airborne Radiological Releases are Terminated Release Category Category SAMG Entry + 36 Hours SAMG Entry + 60 Hours 7 Days after Event Initiation Name Frequency LERF LRF LERF LRF LERF LRF

(/rcy) CCFP CCFP CCFP

(/rcy) (/rcy) (/rcy) (/rcy) (/rcy) (/rcy) 1-REL-BMT 3.53E-06 6.1%

1-REL-CIF 3.71E-07 3.71E-07 0.6% 3.71E-07 0.6% 3.71E-07 0.6%

1-REL-CIF-SC 5.56E-08 5.56E-08 0.1% 5.56E-08 0.1% 5.56E-08 0.1%

1-REL-ECF 6.56E-09 6.56E-09 0.0% 6.56E-09 0.0% 6.56E-09 0.0%

1-REL-ICF-BURN 1.49E-05 1.49E-05 25.6% 1.49E-05 25.6% 1.49E-05 25.6%

1-REL-ICF-BURN-SC 2.58E-06 4.4% 4.4% 4.4%

1-REL-ISGTR 4.94E-07 4.94E-07 4.94E-07 0.9% 4.94E-07 4.94E-07 0.9% 4.94E-07 4.94E-07 0.9%

1-REL-LCF 1.79E-05 30.8% 1.79E-05 30.8%

1-REL-LCF-SC 1.78E-06 1.78E-06 3.1%

3-30 1-REL-NOCF 1.63E-05 1-REL-SGTR-C 0.00E+00 0.00E+00 0.0% 0.00E+00 0.0% 0.00E+00 0.0%

1-REL-SGTR-O 3.94E-08 3.94E-08 0.1% 3.94E-08 0.1% 3.94E-08 0.1%

1-REL-SGTR-O-SC 1.85E-07 1.85E-07 0.3% 1.85E-07 0.3% 1.85E-07 0.3%

1-REL-V 0.00E+00 0.00E+00 0.0% 0.00E+00 0.0% 0.00E+00 0.0%

1-REL-V-F 0.00E+00 0.00E+00 0.00E+00 0.0% 0.00E+00 0.00E+00 0.0% 0.00E+00 0.00E+00 0.0%

1-REL-V-F-SC 0.00E+00 0.00E+00 0.00E+00 0.0% 0.00E+00 0.00E+00 0.0% 0.00E+00 0.00E+00 0.0%

Total 5.81E-05 4.94E-07 1.60E-05 32.0% 4.94E-07 1.60E-05 62.8% 4.94E-07 3.57E-05 72.0%

Table 3-16 Seismic Event Level 2 PRA Surrogate Risk Metric Results - 2020-FLEX Case Release Time at which Airborne Radiological Releases are Terminated Release Category Category SAMG Entry + 36 Hours SAMG Entry + 60 Hours 7 Days after Event Initiation Name Frequency LERF LRF LERF LRF LERF LRF

(/rcy) CCFP CCFP CCFP

(/rcy) (/rcy) (/rcy) (/rcy) (/rcy) (/rcy) 1-REL-BMT 1.24E-07 1.1%

1-REL-CIF 1.00E-06 1.00E-06 9.3% 1.00E-06 9.3% 1.00E-06 9.3%

1-REL-CIF-SC 1.59E-08 1.59E-08 0.1% 1.59E-08 0.1% 1.59E-08 0.1%

1-REL-ECF 1.62E-09 1.62E-09 0.0% 1.62E-09 0.0% 1.62E-09 0.0%

1-REL-ICF-BURN 2.21E-06 2.21E-06 20.4% 2.21E-06 20.4% 2.21E-06 20.4%

1-REL-ICF-BURN-SC 2.07E-08 0.2% 0.2% 0.2%

1-REL-ISGTR 1.95E-07 1.95E-07 1.95E-07 1.8% 1.95E-07 1.95E-07 1.8% 1.95E-07 1.95E-07 1.8%

1-REL-LCF 5.80E-06 53.6% 5.80E-06 53.6%

1-REL-LCF-SC 5.75E-08 5.75E-08 0.5%

3-31 1-REL-NOCF 1.15E-06 1-REL-SGTR-C 0.00E+00 0.00E+00 0.0% 0.00E+00 0.0% 0.00E+00 0.0%

1-REL-SGTR-O 6.38E-08 6.38E-08 0.6% 6.38E-08 0.6% 6.38E-08 0.6%

1-REL-SGTR-O-SC 1.72E-07 1.72E-07 1.6% 1.72E-07 1.6% 1.72E-07 1.6%

1-REL-V 0.00E+00 0.00E+00 0.0% 0.00E+00 0.0% 0.00E+00 0.0%

1-REL-V-F 2.32E-09 2.32E-09 2.32E-09 0.0% 2.32E-09 2.32E-09 0.0% 2.32E-09 2.32E-09 0.0%

1-REL-V-F-SC 0.00E+00 0.00E+00 0.00E+00 0.0% 0.00E+00 0.00E+00 0.0% 0.00E+00 0.00E+00 0.0%

Total 1.08E-05 1.97E-07 3.66E-06 34.0% 1.97E-07 3.66E-06 87.7% 1.97E-07 9.52E-06 89.3%

Table 3-17 High Wind Level 2 PRA Surrogate Risk Metric Results - 2020-FLEX Case Release Time at which Airborne Radiological Releases are Terminated Release Category Category SAMG Entry + 36 Hours SAMG Entry + 60 Hours 7 Days after Event Initiation Name Frequency LERF LRF LERF LRF LERF LRF

(/rcy) CCFP CCFP CCFP

(/rcy) (/rcy) (/rcy) (/rcy) (/rcy) (/rcy) 1-REL-BMT 8.44E-08 1.8%

1-REL-CIF 3.29E-09 3.29E-09 0.1% 3.29E-09 0.1% 3.29E-09 0.1%

1-REL-CIF-SC 0.00E+00 0.00E+00 0.0% 0.00E+00 0.0% 0.00E+00 0.0%

1-REL-ECF 2.25E-10 2.25E-10 0.0% 2.25E-10 0.0% 2.25E-10 0.0%

1-REL-ICF-BURN 7.63E-07 7.63E-07 16.4% 7.63E-07 16.4% 7.63E-07 16.4%

1-REL-ICF-BURN-SC 2.86E-07 6.2% 6.2% 6.2%

1-REL-ISGTR 7.05E-08 7.05E-08 7.05E-08 1.5% 7.05E-08 7.05E-08 1.5% 7.05E-08 7.05E-08 1.5%

1-REL-LCF 2.90E-06 62.5% 2.90E-06 62.5%

1-REL-LCF-SC 6.58E-08 6.58E-08 1.4%

3-32 1-REL-NOCF 4.65E-07 1-REL-SGTR-C 0.00E+00 0.00E+00 0.0% 0.00E+00 0.0% 0.00E+00 0.0%

1-REL-SGTR-O 4.17E-11 4.17E-11 0.0% 4.17E-11 0.0% 4.17E-11 0.0%

1-REL-SGTR-O-SC 3.71E-10 3.71E-10 0.0% 3.71E-10 0.0% 3.71E-10 0.0%

1-REL-V 0.00E+00 0.00E+00 0.0% 0.00E+00 0.0% 0.00E+00 0.0%

1-REL-V-F 0.00E+00 0.00E+00 0.00E+00 0.0% 0.00E+00 0.00E+00 0.0% 0.00E+00 0.00E+00 0.0%

1-REL-V-F-SC 0.00E+00 0.00E+00 0.00E+00 0.0% 0.00E+00 0.00E+00 0.0% 0.00E+00 0.00E+00 0.0%

Total 4.64E-06 7.05E-08 8.37E-07 24.2% 7.05E-08 8.37E-07 86.7% 7.05E-08 3.80E-06 90.0%

3.3 Level 3 PRA This section provides a summary of the results and insights from the reactor, at-power, Level 3 PRA for internal fires, seismic events, and high winds for a single unit. Results are provided for the following two risk metrics:

• Population-weighted early fatality risk (0-1.8 miles) measures the average annual risk to individuals within 1 mile of the site boundary of incurring a fatality within 1 year from acute exposures to radiation due to modeled accidental releases of radiological materials from the reference nuclear power plant site. Results for this metric can be compared to the average individual early fatality risk quantitative health objective (QHO) to obtain insights related to the NRCs safety goal policy (NRC, 1986).

• Population-weighted latent cancer fatality risk (0-10 miles) measures the average annual risk to individuals within 10 miles of the site of incurring a fatality from cancers caused by doses arising from modeled accidental releases of radiological materials from the reference nuclear power plant site. This result, by weighting health effects cases across the entire 10-mile population, reflects the occurrence of exposures relative to the distribution of population around the site. Results for this metric can be compared to the average individual latent cancer fatality risk QHO to obtain insights related to the NRCs safety goal policy (NRC, 1986).

Note that while this report focuses only on the two risk metrics associated with the QHOs, Volume 4e (NRC, 2023e) addresses a more complete set of risk metrics, including those associated with land contamination, population relocation, and economic costs.

Section 3.3.1 provides the results for these two risk metrics for the 2020-FLEX case and a comparison to the results for the Circa-2012 case. Section 3.3.2 discusses alternative analyses to assess the effects of modeling assumptions on the Level 3 PRA results. Section 3.3.3 discusses insights from the Level 3 PRA portion of the 2020-FLEX case, including a discussion of the significant risk contributors.

3.3.1 Results of Circa-2012 and 2020-FLEX Cases The 2020-FLEX case updates the Circa-2012 models to include the new RCP shutdown seals, FLEX strategies and equipment for responding to an ELAP, and continued TDAFW pump operation given a complete loss of all installed AC and DC power. The FLEX strategies are intended to provide coping capability to prevent core damage. Therefore, the primary effect of FLEX strategies on the PRA model is a reduction of the CDF in the Level 1 PRA model (as discussed in Section 3.1.1). The 2020-FLEX case model changes result in reduced CDF contributions from the sequences involving SBO events or RCP seal failures. The main impact on the Level 2 PRA model for FLEX strategies is carrying forward the modified Level 1 PRA sequences, which results in reduced frequencies for the applicable release categories. The 2020-FLEX case does not consider the impact of FLEX strategies on severe accident timing; therefore, the conditional consequences do not change and the only impact on the Level 3 PRA model derives from the change in the Level 2 PRA release category frequencies.

This report provides a comparison of the 2020-FLEX case results to the Circa-2012 case. The description of the Circa-2012 Level 3 PRA model and results for internal fires, seismic events, 3-33

and high winds during power operation are provided in the associated L3PRA project report (NRC, 2023e). Separate discussions are provided below for early and latent fatality risk.

Individual Early Fatality Risk Table 3-18 to Table 3-21 compare mean annual population-weighted early fatality risk within 1 mile of the site boundary for the Circa-2012 and 2020-FLEX cases, for internal events and floods, internal fires, seismic events, and high winds, respectively. 16 Only the release categories that appreciably contribute to early fatality risk for each hazard category are included in these tables. This information is displayed graphically in Figure 3-4 to Figure 3-11. 17 As seen in Table 3-18, for internal events and floods in the Circa-2012 case, nearly 90 percent of the mean annual population-weighted individual early fatality risk within 1 mile of the site boundary comes from interfacing systems loss-of-coolant accidents (ISLOCAs) that involve auxiliary building failure (1-REL-V-F and 1-REL-V-F-SC). However, all ISLOCA pathways were screened out of the Level 1 fire PRA because they met at least one of the following conditions (NRC, 2023a):

• The path includes flow restrictions that would restrict leakage to a rate below the capacity of normal charging.

• The path is a closed loop inside containment.

• The path contains at least two isolation valves that cannot be impacted by fire.

Therefore, the mean annual population-weighted individual early fatality risk for internal fires in the Circa-2012 case comes entirely from just the following two release categories:

• 1-REL-ISGTR - a thermally-induced steam generator tube rupture (SGTR) after core damage (contributing 76 percent)

• 1-REL-SGTR-O - an SGTR occurring prior to core damage and the release is not mitigated by water above the break point on the secondary side of the affected steam generator (SG) and one or more secondary-side relief valves are either kept open during the release as a deliberate action or fail in the open position (contributing 24 percent)

As seen in Table 3.1-3 in NRC (2023e), the two release categories mentioned above drive early fatality risk because 1-REL-ISGTR has a short warning time combined with large radiological release fractions and 1-REL-SGTR-O has a relatively longer warning time, but particularly large release fractions.

As seen in Table 3-19, the changes in the 2020-FLEX case for internal fires only reduce population-weighted early fatality risk within 1 mile of the site boundary by 7 percent (the reduction comes entirely from 1-REL-ISGTR; 1-REL-SGTR-O actually increases slightly due to cutset truncation impacts and other minor quantification peculiarities). The reduction is limited 16 These values were obtained by weighting the mean (over all weather trials) consequence values for individual release categories by the point estimate of the individual release category frequencies.

17 Several figures in this section label the risk metric results in terms of per reactor year (/ry). In actuality, these risk metric results are in terms of per reactor-critical-year (/rcy).

3-34

since the top contributors to 1-REL-ISGTR for internal fires do not involve extended losses of offsite power or RCP seal LOCAs.

As seen in Table 3-20, early fatality risk from seismic events is mostly driven by the same two release categories as for internal fires (1-REL-SGTR-O and 1-REL-ISGTR), with ISLOCAs (1-REL-V-F) making an additional modest contribution. For seismic events, the relative contribution from release category SGTR-O is greater than for internal fires or high winds. This appears to be primarily due to the contribution from seismically induced ATWS events, in combination with inadequate primary pressure relief, leading to a pressure-induced SGTR. The ISLOCA contribution for seismic events derives from an assumption in the seismic Level 2 PRA (NRC, 2023d) to assign the core damage frequency associated with seismic bin 8 to the 1-REL-V-F release category, since it is considered to be the most conservative with respect to early fatalities.

The changes in the 2020-FLEX case for seismic events only reduce population-weighted early fatality risk within 1 mile of the site boundary by 6 percent (as for internal fires, the reduction comes entirely from 1-REL-ISGTR, which is reduced by 14 percent). The reduction in early fatality risk for 1-REL-ISGTR is limited since major structural failures, which are not impacted by the FLEX changes, are significant contributors to 1-REL-ISGTR.

As seen in Table 3-21, population-weighted early fatality risk within 1 mile of the site boundary for high winds comes entirely from release category 1-REL-ISGTR. As with internal fires, no high wind events are assumed to directly result in ISLOCAs. Nonetheless, the changes in the 2020-FLEX case for high winds reduce early fatality risk by 57 percent because 1-REL-ISGTR for high winds almost exclusively involves wind-related LOOP and a combination of wind-related and random failures of the onsite emergency AC system.

Table 3-22 shows the contribution to population-weighted early fatality risk within 1 mile of the site boundary by hazard category, for both the Circa-2012 case and the 2020-FLEX case. As can be seen from Table 3-22, for the Circa-2012 case, internal events and floods and seismic events collectively contribute over 80 percent to this risk metric. The contribution from these hazard categories is even greater (nearly 90 percent) in the 2020-FLEX case, since the changes in the 2020-FLEX case only significantly reduce early fatality risk for the high winds hazard category (for the reason stated in the previous paragraph).

While the changes in the 2020-FLEX case generally have a limited impact on population-weighted early fatality risk within 1 mile of the site boundary (except for high winds), Figure 3-12 shows that the margins to the associated QHO are already substantial.

As discussed in Section 3.2.1, a parameter uncertainty analysis considering the uncertainty in the Level 2 PRA release category frequencies was performed for the 2020-FLEX case for internal fires, seismic events, and high winds. Table 3-23 presents the mean, median, 5th-percentile, and 95th-percentile 0-1.8-mile population-weighted early fatality risk considering the uncertainty in the Level 2 PRA release category frequency estimates for these hazards (for comparison, Table 3-23 also includes the respective values for internal events and floods).

Figure 3-13 to Figure 3-16 present complementary cumulative distribution function (CCDF) curves that illustrate the frequencies of exceeding specified levels of population-weighted early fatality risk in the 0-1.8-mile range (i.e., within 1 mile of the site boundary) for internal events and floods, internal fires, seismic events, and high winds, respectively. These curves include the contributions from the full spectrum of accident scenarios modeled in the reactor, at-power PRA 3-35

for each hazard group. The red curve reflects the mean CCDF curve, the green curve reflects the median CCDF curve, and the blue curves represent the 5th-percentile and 95th-percentile CCDF curves.

Individual Latent Cancer Fatality Risk Table 3-24 to Table 3-27 compare mean population-weighted individual latent cancer fatality risk within 10 miles of the site for the Circa-2012 and 2020-FLEX cases, for internal events and floods, internal fires, seismic events, and high winds, respectively. 18 Only those release categories that contribute at least 1 percent to latent cancer fatality risk are included in these tables. This information is displayed graphically in Figure 3-17 to Figure 3-24.

As seen in Table 3-24, Table 3-25, and Table 3-27, in the Circa-2012 case for internal events and floods, internal fires, and high winds, respectively, two radiological release categories collectively contribute well over 90 percent of the mean population-weighted individual latent cancer fatality risk within 10 miles of the site: (1) a late containment failure release category in which the containment fails tens of hours after the time of vessel breach, due to long-term quasi-static overpressure, and releases to the environment are not mitigated significantly by sprays or water pools (1-REL-LCF); and (2) an intermediate containment failure release category in which the containment fails hours after vessel breach, due to a global deflagration or detonation, and releases to the environment are not mitigated significantly by sprays or water pools (1-REL-ICF-BURN). As can be seen from Table 3.1-3 in NRC (2023e), these release categories combine relatively high frequencies of occurrence with relatively high radiological release fractions.

As seen in Table 3-26, for seismic events, these same two radiological release categories collectively contribute over 80 percent of the mean population-weighted individual latent cancer fatality risk within 10 miles of the site, with an additional 12 percent contributed by the containment isolation failure release category (1-REL-CIF), in which an unmitigated release from the containment to the environment occurs either via a containment penetration that fails to be isolated by the containment isolation system or via a pre-existing leakage path. The greater contribution of 1-REL-CIF for seismic events arises from the elevated probability of direct seismic failure of the containment isolation system for the higher seismicity bins (i.e., bins 5-7).

As can also be seen from Table 3-24 and Table 3-27, the changes in the 2020-FLEX case reduce population-weighted latent cancer fatality risk within 10 miles of the site by over 60 percent for internal events and floods and for high winds. For internal fires and seismic events, Table 3-25 and Table 3-26 show more modest reductions (13 percent and 20 percent, respectively). This is because release categories 1-REL-LCF and 1-REL-ICF-BURN, for internal events and high winds, have a significant contribution from SBO sequences and, to a lesser extent, losses of nuclear service cooling water (NSCW) leading to RCP seal LOCAs, both of which significantly benefit from the types of measures incorporated into the 2020-FLEX case. In addition, as shown in Table 3-1, FLEX and TDAFW extension are assumed to have a greater likelihood of success for internal events and floods and high winds, as compared to internal fires and seismic events.

18 These values were obtained by weighting the mean (over all weather trials) consequence values for individual release categories by the point estimate of the individual release category frequencies.

3-36

Table 3-28 shows the contribution to population-weighted latent cancer fatality risk within 10 miles of the site by hazard category, for both the Circa-2012 case and the 2020-FLEX case.

As can be seen from Table 3-28, for the Circa-2012 case, internal events and floods and internal fires collectively contribute nearly 80 percent to this risk metric. The contribution from these hazard categories is nearly the same in the 2020-FLEX case; however, the contribution from internal fires rises from 38 percent to 55 percent and the contribution from internal events and floods drops from 40 percent to 24 percent. This reordering of hazard category contributions in the 2020-FLEX case is due to the reasons stated in the previous paragraph with regard to internal events and high winds (which also explains the widening gap in the contributions between seismic events and high winds).

As can be seen from Figure 3-25, for the reactor at-power for the Circa-2012 case, the margin to the latent cancer fatality QHO is around a factor of 30 when considering all hazards. For the 2020-FLEX case, the margin to the latent cancer fatality QHO increases to a factor of around 50.

As mentioned previously for early fatality risk, a parameter uncertainty analysis considering the uncertainty in the Level 2 PRA release category frequencies was performed for the 2020-FLEX case for internal fires, seismic events, and high winds. Table 3-29 presents the mean, median, 5th-percentile, and 95th-percentile 0-10-mile population-weighted latent cancer fatality risk considering the uncertainty in the Level 2 PRA release category frequency estimates for these hazards (for comparison, Table 3-29 also includes the respective values for internal events and floods). Figure 3-26 to Figure 3-29 present CCDF curves that illustrate the frequencies of exceeding specified levels of population-weighted latent cancer fatality risk in the 0-10-mile range for internal events and floods, internal fires, seismic events, and high winds, respectively.

These curves include the contributions from the full spectrum of accident scenarios modeled in the reactor, at-power PRA for each hazard group. The red curve reflects the mean CCDF curve, the green curve reflects the median CCDF curve, and the blue curves represent the 5th-percentile and 95th-percentile CCDF curves.

As shown in Figure 3-26 to Figure 3-29, the slopes of the CCDF curves for all modeled hazards appear to increase (i.e., become more negative) at about 5x10-4. This indicates the likelihood of exceeding population-weighted latent cancer fatality risk levels beyond this value becomes increasingly less likely as the risk level increases.

3.3.2 Results of Alternative Analyses Several alternative analyses were performed to assess the impacts of modeling assumptions and sources of uncertainty on the Level 3 PRA results. The two alternative analyses discussed here involve the accident termination time (as previously discussed in Section 3.2.2 for the Level 2 PRA) and the dose truncation model for evaluating radiological health effects.

As discussed in Section 3.2.2, the L3PRA project does not explicitly model the role of long-term onsite, or offsite, resources in terminating accidents after core damage has occurred. The issue of the timing of accident termination and termination of radiological release is treated as a global modeling uncertainty, as described in Appendix D of the Level 2 reactor at-power internal event and flood PRA report (NRC, 2022b). In both the Circa-2012 and 2020-FLEX cases, the accident and release termination time for many accident sequences is 7 days after event initiation. To gain insight into the range of consequence/risk results from different accident termination times, consequence calculations were performed terminating radiological releases from all the 3-37

representative accident sequences 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br /> after SAMG entry. Actions to terminate the accident in this timeframe could include those that were successfully or unsuccessfully carried out at the various Fukushima Daiini and Daiichi units in March 2011 (e.g., restoration of DC power, restoration of AC power, and containment venting).

For releases that can lead to early fatalities, most of the release occurs within 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br /> after SAMG entry; therefore, this alternative termination time has no appreciable impact on early fatality risk. However, the alternative termination time does have a significant impact on latent cancer risk. This can be seen in Figure 3-30 to Figure 3-34 for internal events and floods, internal fires, seismic events, high winds, and all hazards, respectively.

For the 1-REL-LCF release category, the release is prolonged and occurs over a period of several days, with a steady increase in released material. Terminating the releases 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br /> after SAMG entry therefore significantly reduces the total amount of radiological material released for the 1-REL-LCF release category. The same is true, albeit to a to a lesser extent, for the 1-REL-ICF-BURN release category. Because these two release categories are dominant contributors to the latent cancer fatality risk, terminating the analysis at 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br /> after SAMG entry reduces the latent cancer fatality risk for the 2020-FLEX case by approximately 70 percent for all hazards combined. The largest decrease is for high winds (approximately 80 percent); the smallest is for internal fires (approximately 60 percent).

As discussed in Section 3.6.4 of the Level 3 PRA report for internal fires, seismic events, and high winds (NRC, 2022c), it is unclear what health consequences, if any, are attributable to very low radiation exposure. The NRC currently relies on the hypothesis that a linear no-threshold (LNT) dose-response relationship is the appropriate approach to use in making its regulatory decisions. The LNT approach is based on scientific evidence supported by many in the technical community. However, there is also the belief by many in the technical community that estimating latent cancer fatalities based on very small doses to large populations is inappropriate, though there is no consensus on what dose threshold is appropriate.

Consistent with the State-of-the-Art Reactor Consequence Analyses (SOARCA) (NRC, 2012) and current NRC policy for regulatory applications, the LNT model is used as the base-case dose-response model for both the Circa-2012 and 2020-FLEX cases for evaluating radiological health effects. However, an alternative dose truncation model was also considered to allow examination of the cancer risks arising only from moderate (>10 rem) or high (>100 rem) lifetime doses, where the level of uncertainty in cancer risk estimation is less than in the low- and very-low-dose range. The alternate dose truncation model is based on the model documented in a Health Physics Society (HPS) position paper on radiation risk (HPS 2010), which estimates cancer risk based only on annual individual doses greater than 0.05 Sv (5 rem), or lifetime individual doses greater than 0.1 Sv (10 rem). 19 The impact of the alternative dose truncation model on the 2020-FLEX case (while retaining the 7-day accident and release termination time) is shown in Figure 3-30 to Figure 3-34 for internal events and floods, internal fires, seismic events, high winds, and all hazards, respectively. As 19 It is noted that the HPS position statement on radiation risk was updated in February 2019 to state simply that The Health Physics Society advises against estimating health risks to people from exposures to ionizing radiation that are near or less than natural background levels because statistical uncertainties at these low levels are great (HPS, 2019). However, the numerical values corresponding to the 2010 position statement were used in this analysis for consistency with recent NRC analyses using the MELCOR Accident Consequence Code System (MACCS) dose truncation model.

3-38

can be seen from these figures, use of the alternative dose truncation model reduces latent cancer fatality risk significantly. This is to be expected since the latent cancer fatality risk estimated in this study primarily results from long-term, low-dose exposure to individuals after they are allowed to return to their homes following decontamination. 20 3.3.3 Initial Insights This section provides initial insights from the Level 3 PRA portion of both the Circa-2012 and 2020-FLEX cases. The focus is on significant risk contributors, but some other insights are provided at the end of the section.

To gain insight into the relative risk significance of individual basic events to selected offsite public risk metrics, composite FV importance measures were calculated for each event. The composite FV importance measure for a particular basic event is used to approximate the relative contribution to the total mean annual risk for each selected offsite public risk metric from accident scenarios that include that basic event. In practice, this composite FV importance measure is calculated as a weighted sum of the standard FV importance measure for the basic event with respect to each radiological release category frequency, weighted by the relative contribution of each radiological release category to the mean annual risk for each selected offsite public risk metric. For more information on the composite FV importance measure, see Section 5.2.5.1 of the Level 3 PRA report for internal events and floods (NRC, 2022c).

For both the Circa-2012 and 2020-FLEX cases, composite FV importances were calculated for (1) mean annual population-weighted individual early fatality risk within 1 mile of the site boundary and (2) mean annual population-weighted individual latent cancer fatality risk within 10 miles.

Significant Risk Contributors for Individual Early Fatality Risk For internal fires, since the 1-REL-ISGTR release category is such a large contributor to mean annual population-weighted individual early fatality risk within 1 mile of the site boundary, it is not surprising that the largest individual contributor (with a composite FV of 0.76) is the Level 2 PRA basic event that represents occurrence of a thermally induced SGTR given high-dry-low conditions (i.e., high primary-side pressure, dry SGs, and low secondary-side pressure). The Level 1 PRA basic event for conditional probability of a pressure induced SGTR (typically following an ATWS event or secondary-side break) is also a significant contributor (composite FV of 0.13), due to its large contribution to the frequency of release category 1-REL-SGTR-O.

Many other significant contributors to mean annual population-weighted individual early fatality risk within 1 mile of the site boundary, for internal fires, include various operator errors from both the Level 1 and Level 2 PRA models. Examples from the Level 1 PRA model include failure to initiate feed and bleed cooling after loss of all steam generator cooling and failure to control TDAFW after the fire causes it to spuriously start. Examples from the Level 2 PRA model 20 The dose criterion for the required decontamination after a severe accident is uncertain. The current state of practice is to model decontamination to the level of meeting the habitability criteria as defined by the Environmental Protection Agency (EPA) intermediate-phase protective action guidelines (PAGs). The use of the EPA intermediate-phase PAGs (2 rem in the year of the accident and 500 mrem in subsequent years) is assumed as a surrogate for decisions on cleanup and reoccupation.

3-39

include failure of SAMG actions to depressurize the RCS via secondary-side cooling and to feed the steam generators before or after vessel breach.

The significant risk contributors for early fatality risk for the 2020-FLEX case for internal fires are very similar to those for the Circa-2012 case. This is expected since the same release categories (1-REL-ISGTR and 1-REL-SGTR-O) dominate risk in both cases. Since SBO sequences make a moderate contribution to 1-REL-ISGTR, and these sequences do significantly benefit from the types of measures incorporated into the 2020-FLEX case, there is a corresponding decrease in the relative importance of SBO-related basic events (i.e., failures of EDGs, reserve auxiliary transformer input breakers, and other emergency AC system basic events).

For seismic events, there are two groups of basic events that contribute the most to mean annual population-weighted individual early fatality risk within 1 mile of the site boundary. The first group involves various seismically induced failures leading to a station blackout that then progresses to a thermally induced SGTR given high-dry-low conditions. The second group involves various seismically induced failures leading to an ATWS, combined with insufficient primary pressure relief due to UET (Unfavorable Exposure Time) or seismically induced failure of power to the power-operated relief valves, causing a pressure-induced SGTR. This latter group also includes failure of the post-core-damage action to feed the ruptured SG before or after vessel breach, either due to operator error or seismically induced failures leading to main control room abandonment.

The significant risk contributors for early fatality risk for the 2020-FLEX case for seismic events are very similar to those for the Circa-2012 case. This is expected since, as pointed out earlier, the frequency of release category 1-REL-SGTR-O, as well as the substantial portion of 1-REL-ISGTR frequency that comes from major structural failures, are not impacted by the types of measures incorporated into the 2020-FLEX case. However, since SBO sequences, which do significantly benefit from the types of measures incorporated into the 2020-FLEX case, contribute somewhat to 1-REL-ISGTR, the two basic events representing failure of FLEX and continued TDAFW pump operation each have a composite FV value of approximately 0.14.

For high winds, the most significant contributors to mean annual population-weighted individual early fatality risk within 1 mile of the site boundary involve various wind-induced and random failures and operator errors leading to core damage from a station blackout that then progresses to a thermally induced SGTR given high-dry-low conditions. The Level 2 PRA basic events for induced SGTR given high-dry-low conditions and failure of manual extension of TDAFW in an SBO scenario have composite FV importances of 1.0 and 0.89, respectively. The random failures of the A and B EDGs to each run for a 24-hour mission time have a composite FV importance of approximately 0.4. The most significant wind-induced failures include failure of a transmission tower directly from the high winds (bins 2 and 3) and failure from a wind missile of the standby auxiliary transformer (SAT) itself, the SAT relay house, or SAT cable trays (bins 1 and 2).

The significant risk contributors for early fatality risk for the 2020-FLEX case for high winds are very similar to those for the Circa-2012 case. In both cases, virtually the entire early fatality risk comes from SBO sequences that lead to a thermally induced SGTR. While the absolute risk is significantly reduced in the 2020-FLEX case, the relative contributions of the different basic events are similar. The one significant difference is the importance in the 2020-FLEX case of the basic event representing failure of FLEX, with a composite FV value of 0.93. The failure of 3-40

manual extension of TDAFW in an SBO scenario has the same FV value, 21 since both these events always appear in cut sets together.

Significant Risk Contributors for Individual Latent Cancer Fatality Risk For internal fires, many of the most significant contributors to mean annual population-weighted individual latent cancer fatality risk within 10 miles are related to combustion (detonations or deflagrations) within containment. Note, some combustion events result in direct failure of the containment, while others occur early in the accident progression before there is sufficient combustible gas to result in containment failure. In these latter cases, the early combustible events can reduce the amount of combustible gas in containment, thereby significantly reducing the likelihood of a larger combustible event later in the accident progression.

Other significant contributors to latent cancer fatality risk for internal fires include (1) failure of in-vessel recovery when applicable, (2) various operator errors from both the Level 1 and Level 2 PRA models, and (3) various spurious operations due to fire (most significantly, the steam inlet motor-operated valve for the TDAFW pump). Examples of significant operator errors from the Level 1 PRA model include failure to initiate feed and bleed cooling after loss of all steam generator cooling or failure to control TDAFW after the fire causes it to spuriously start.

Examples from the Level 2 PRA model include failure of SAMG actions to depressurize the RCS via secondary-side cooling and to spray containment with firewater to scrub the release.

The significant risk contributors for latent cancer fatality risk for the 2020-FLEX case for internal fires are very similar to those for the Circa-2012 case. This is expected since the same release categories (1-REL-LCF and 1-REL-ICF-BURN) dominate risk in both cases. Since SBO and RCP seal LOCA sequences make a moderate contribution to these release categories, and these sequences do significantly benefit from the types of measures incorporated into the 2020-FLEX case, there is a corresponding decrease in the relative importance of basic events associated with these types of sequences (e.g., failures of EDGs, reserve auxiliary transformer input breakers, and RCP seal integrity).

For seismic events, many of the most significant contributors to mean annual population-weighted individual latent cancer fatality risk within 10 miles are related to combustion within containment, as described above for internal fires. Examples of other significant contributors for seismic events include:

• various failures of emergency on-site AC system components leading to the occurrence of an SBO

• seismically induced failure of major structures and systems (e.g., containment isolation) for the higher seismicity bins

• induced hot leg failure after core damage 21 In the Level 2 PRA of the Circa-2012 case, some modest credit was given to continued TDAFW pump operation to preclude containment failure for a subset of SBO sequences. The failure of this action has a composite FV value of 0.89 in the Circa-2012 case. For the 2020-FLEX case, this modeling is superseded by the changes described in Section 3.2.1.

3-41

• failure of manual extension of TDAFW in an SBO scenario

• failure of the RCP stage 2 seal The significant risk contributors for latent cancer fatality risk for the 2020-FLEX case for seismic events are very similar to those for the Circa-2012 case. This is expected since the same release categories (1-REL-LCF, 1-REL-ICF-BURN, and 1-REL-CIF) dominate risk in both cases. The major difference for the 2020-FLEX case is the importance of the two new basic events representing implementation of the FLEX strategies and continued operation of TDAFW under SBO conditions (since both events always appear in cutsets together, they identically contribute 29 percent). Other significant differences include the elimination of the basic event for manual extension of TDAFW used in the Circa-2012 case (it is replaced by the new event in the 2020-FLEX model) and a much lesser contribution form the RCP seal failures because of the new RCP seal design.

For high winds, the most significant contributor to mean annual population-weighted individual latent cancer fatality risk within 10 miles is failure of manual extension of TDAFW in an SBO scenario, with a composite FV of approximately 0.77. Examples of other significant contributors include:

• events related to combustion within containment

• various failures of emergency on-site AC system components leading to the occurrence of an SBO

• induced hot leg failure after core damage

• operator failure to recover offsite power within 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />

• failure of the RCP stage 2 seal The most significant wind-induced failures include failure of a transmission tower directly from the high winds (bins 2 and 3) and failure from a wind missile of the SAT itself, the SAT relay house, or SAT cable trays (bins 1 and 2).

The significant risk contributors for latent cancer fatality risk for the 2020-FLEX case for high winds are very similar to those for the Circa-2012 case. This is expected since the same release categories (1-REL-LCF and 1-REL-ICF-BURN) dominate risk in both cases. The major difference for the 2020-FLEX case is the importance of the two new basic events representing implementation of the FLEX strategies and continued operation of TDAFW under SBO conditions (since both events always appear in cutsets together, they identically contribute approximately 87 percent). Other significant differences include the elimination of the basic event for manual extension of TDAFW used in the Circa-2012 case (it is replaced by the new event in the 2020-FLEX model) and a much lesser contribution form the RCP seal failures because of the new RCP seal design.

Additional Insights The changes in the 2020-FLEX case generally have a limited impact on population-weighted early fatality risk within 1 mile of the site boundary since these changes do not generally impact 3-42

the types of sequences that drive this risk (except in the case of high winds). However, the margins to the associated QHO are already substantial in the Circa-2012 case (i.e., several orders of magnitude even when combining the risk from all hazard categories).

In addition, for releases that can lead to early fatalities, most of the release occurs within 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br /> after SAMG entry. Therefore, use of a reduced accident analysis termination time (i.e., crediting additional operator actions to terminate the accident in this timeframe) has no appreciable impact on early fatality risk.

For mean annual population-weighted individual latent cancer fatality risk within 10 miles for all hazards, the Circa-2012 case is over a factor of 30 below the respective QHO associated with the Commissions safety goals. When accounting for the modeling changes associated with the 2020-FLEX case, this margin increases to a factor of over 50 for all hazards combined. Further, terminating the accident and radiological release analysis at 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br /> after SAMG entry, as opposed to 7 days after event initiation, further increases the margin to the QHO to approximately a factor of 150. Finally, using the alternative dose truncation model described previously (as opposed to the LNT model) suggests that there may be significantly more margin to the QHO.

3-43

Figure 3-4 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Internal Events and Floods (Circa-2012)

Figure 3-5 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Internal Events and Floods (2020-FLEX) 3-44

Figure 3-6 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Internal Fires (Circa-2012 Case)

Figure 3-7 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Internal Fires (2020-FLEX Case) 3-45

Figure 3-8 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Seismic Events (Circa-2012 Case)

Figure 3-9 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for Seismic Events (2020-FLEX Case) 3-46

Figure 3-10 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for High Winds (Circa-2012 Case)

Figure 3-11 Release Category Contribution to the 0-1.8-Mile Population-Weighted Early Fatality Risk Using Mean Release Category Frequencies for High Winds (2020-FLEX Case) 3-47

IEIF: internal events and floods F: internal fires S: seismic events W: high winds ALL: all hazards combined Figure 3-12 Individual Early Fatality Risk (0-1.8 Miles) 3-48

Figure 3-13 CCDF Curves for Population-Weighted Early Fatality Risk (0-1.8 Miles) for the 2020-FLEX Case (Internal Events and Floods)

Figure 3-14 CCDF Curves for Population-Weighted Early Fatality Risk (0-1.8 Miles) for the 2020-FLEX Case (Internal Fires) 3-49

Figure 3-15 CCDF Curves for Population-Weighted Early Fatality Risk (0-1.8 Miles) for the 2020-FLEX Case (Seismic Events)

Figure 3-16 CCDF Curves for Population-Weighted Early Fatality Risk (0-1.8 Miles) for the 2020-FLEX Case (High Winds) 3-50

Figure 3-17 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Internal Events and Floods (Circa-2012)

Figure 3-18 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Internal Events and Floods (2020-FLEX) 3-51

Figure 3-19 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Internal Fires (Circa-2012)

Figure 3-20 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Internal Fires (2020-FLEX) 3-52

Figure 3-21 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Seismic Events (Circa-2012)

Figure 3-22 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for Seismic Events (2020-FLEX) 3-53

Figure 3-23 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for High Winds (Circa-2012)

Figure 3-24 Release Category Contribution to the 10-Mile Population-Weighted Latent Cancer Fatality Risk Using Mean Release Category Frequencies for High Winds (2020-FLEX) 3-54

IEIF: internal events and floods F: internal fires S: seismic events W: high winds ALL: all hazards combined Figure 3-25 Individual Latent Cancer Fatality Risk (0-10 Miles) 3-55

Figure 3-26 CCDF Curves for Population-Weighted Latent Cancer Fatality Risk (0-10 Miles) for the 2020-FLEX Case (Internal Events and Floods)

Figure 3-27 CCDF Curves for Population-Weighted Latent Cancer Fatality Risk (0-10 Miles) for the 2020-FLEX Case (Internal Fires) 3-56

Figure 3-28 CCDF Curves for Population-Weighted Latent Cancer Fatality Risk (0-10 Miles) for the 2020-FLEX Case (Seismic Events)

Figure 3-29 CCDF Curves for Population-Weighted Latent Cancer Fatality Risk (0-10 Miles) for the 2020-FLEX Case (High Winds) 3-57

Figure 3-30 Individual Latent Cancer Fatality Risk (0-10 Miles) for Internal Events and FloodsAlternative Analyses 3-58

Figure 3-31 Individual Latent Cancer Fatality Risk (0-10 Miles) for Internal Fires Alternative Analyses 3-59

Figure 3-32 Individual Latent Cancer Fatality Risk (0-10 Miles) for Seismic Events Alternative Analyses 3-60

Figure 3-33 Individual Latent Cancer Fatality Risk (0-10 Miles) for High Winds Alternative Analyses 3-61

Figure 3-34 Individual Latent Cancer Fatality Risk (0-10 Miles) for All Hazards Alternative Analyses 3-62

Table 3-18 Population-Weighted Early Fatality Risk, by Release Category, for the 0-1.8-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Internal Events and Floods 2020-FLEX Circa-2012 Case 2020-FLEX Case Impact Release Category Early Fatality  % of Early Fatality  % of (a-b)/a %

Name Risk (/rcy) (a) Total Risk (/rcy) (b) Total Total 3.4E-13 100% 3.2E-13 100% 6%

1-REL-V-F 2.1E-13 62% 2.1E-13 66%

1-REL-V-F-SC 9.2E-14 27% 9.2E-14 28%

1-REL-ISGTR 3.4E-14 10% 1.3E-14 4% 60%

1-REL-SGTR-O 6.0E-15 2% 6.0E-15 2%

Table 3-19 Population-Weighted Early Fatality Risk, by Release Category, for the 0-1.8-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Internal Fires 2020-FLEX Circa-2012 Case 2020-FLEX Case Impact Release Category Early Fatality  % of Early Fatality  % of (a-b)/a %

Name Risk (/rcy) (a) Total Risk (/rcy) (b) Total Total 4.1E-14 100% 3.8E-14 100% 7%

1-REL-ISGTR 3.1E-14 76% 2.8E-14 74% 10%

1-REL-SGTR-O 9.7E-15 24% 9.8E-15 26%

Table 3-20 Population-Weighted Early Fatality Risk, by Release Category, for the 0-1.8-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Seismic Events 2020-FLEX Circa-2012 Case 2020-FLEX Case Impact Release Category Early Fatality  % of Early Fatality  % of (a-b)/a %

Name Risk (/rcy) (a) Total Risk (/rcy) (b) Total Total 2.8E-13 100% 2.6E-13 100% 6%

1-REL-SGTR-O 1.3E-13 47% 1.3E-13 51%

1-REL-ISGTR 1.3E-13 45% 1.1E-13 42% 14%

1-REL-V-F 2.1E-14 7% 2.1E-14 8%

3-63

Table 3-21 Population-Weighted Early Fatality Risk, by Release Category, for the 0-1.8-Mile Interval for the Circa-2012 and 2020-FLEX Cases for High Winds 2020-FLEX Circa-2012 Case 2020-FLEX Case Impact Release Category Early Fatality  % of Early Fatality  % of (a-b)/a %

Name Risk (/rcy) (a) Total Risk (/rcy) (b) Total Total 9.2E-14 100% 3.9E-14 100% 57%

1-REL-ISGTR 9.2E-14 100% 3.9E-14 100% 57%

Table 3-22 Population-Weighted Early Fatality Risk, by Hazard Category, for the 0-1.8-Mile Interval for the Circa-2012 and 2020-FLEX Cases Circa-2012 Case 2020-FLEX Case Early Fatality Early Fatality Hazard Category  % of Total  % of Total Risk (/rcy) Risk (/rcy)

Total 7.5E-13 100% 6.6E-13 100%

Internal Events and Floods 3.4E-13 45% 3.2E-13 49%

Internal Fires 4.1E-14 5% 3.8E-14 6%

Seismic Events 2.8E-13 37% 2.6E-13 40%

High Winds 9.2E-14 12% 3.9E-14 6%

Table 3-23 Statistical Analysis of the Population-Weighted Early Fatality Risk for the 0-1.8-Mile Interval for the 2020-FLEX Case Point 5th- 95th-Median Estimate* Mean (/rcy) Percentile Percentile

(/rcy)

(/rcy) (/rcy) (/rcy)

Internal events and floods 3.2E-13 3.9E-13 2.8E-13 8.0E-14 1.0E-12 Internal fires 3.8E-14 3.7E-14 2.3E-14 5.7E-15 1.1E-13 Seismic events 2.6E-13 2.4E-13 1.3E-13 3.3E-14 7.7E-13 High winds 4.0E-14 1.2E-13 4.8E-14 6.2E-15 4.2E-13

• The reported point estimate value is the frequency-weighted sum of all release categories using the release category frequency point estimate values convolved with the conditional consequence mean values over all weather trials for each release category.

3-64

Table 3-24 Population-Weighted Latent Cancer Fatality Risk, by Release Category, for the 0-10-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Internal Events and Floods 2020-FLEX Circa-2012 Case 2020-FLEX Case Impact Latent Cancer Latent Cancer Release Category  % of  % of Fatality Risk Fatality Risk (a-b)/a %

Name Total Total

(/rcy) (a) (/rcy) (b)

Total 2.6E-08 100% 9.7E-09 100% 62%

1-REL-LCF 1.8E-08 70% 6.3E-09 65% 65%

1-REL-ICF-BURN 6.1E-09 24% 2.8E-09 28% 55%

1-REL-ISGTR 5.0E-10 2% 2.0E-10 2% 60%

1-REL-LCF-SC 4.8E-10 2% 3.0E-10 3% 38%

1-REL-NOCF 3.5E-10 1% 8.1E-11 <1% 77%

Table 3-25 Population-Weighted Latent Cancer Fatality Risk, by Release Category, for the 0-10-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Internal Fires 2020-FLEX Circa-2012 Case 2020-FLEX Case Impact Latent Cancer Latent Cancer Release Category  % of  % of Fatality Risk Fatality Risk (a-b)/a %

Name Total Total

(/rcy) (a) (/rcy) (b)

Total 2.5E-08 100% 2.2E-08 100% 13%

1-REL-LCF 1.3E-08 51% 1.1E-08 49% 16%

1-REL-ICF-BURN 1.1E-08 43% 1.0E-08 45% 9%

1-REL-ISGTR 4.6E-10 2% 4.1E-10 2% 10%

1-REL-CIF 3.2E-10 1% 2.9E-10 1% 8%

1-REL-LCF-SC 3.1E-10 1% 2.6E-10 1% 16%

1-REL-NOCF 3.0E-10 1% 2.3E-10 1% 23%

3-65

Table 3-26 Population-Weighted Latent Cancer Fatality Risk, by Release Category, for the 0-10-Mile Interval for the Circa-2012 and 2020-FLEX Cases for Seismic Events 2020-FLEX Circa-2012 Case 2020-FLEX Case Impact Latent Cancer Latent Cancer Release Category  % of  % of Fatality Risk Fatality Risk (a-b)/a %

Name Total Total

(/rcy) (a) (/rcy) (b)

Total 7.7E-09 100% 6.2E-09 100% 20%

1-REL-LCF 4.5E-09 59% 3.5E-09 57% 22%

1-REL-ICF-BURN 1.8E-09 24% 1.5E-09 24% 19%

1-REL-CIF 9.1E-10 12% 8.0E-10 13% 13%

1-REL-ISGTR 2.6E-10 3% 2.2E-10 4% 14%

1-REL-SGTR-O 7.7E-11 1% 7.7E-11 1%

Table 3-27 Population-Weighted Latent Cancer Fatality Risk, by Release Category, for the 0-10-Mile Interval for the Circa-2012 and 2020-FLEX Cases for High Winds 2020-FLEX Circa-2012 Case 2020-FLEX Case Impact Latent Cancer Latent Cancer Release Category  % of  % of Fatality Risk Fatality Risk (a-b)/a %

Name Total Total

(/rcy) (a) (/rcy) (b)

Total 6.3E-09 100% 2.4E-09 100% 63%

1-REL-LCF 4.7E-09 75% 1.8E-09 74% 63%

1-REL-ICF-BURN 1.3E-09 21% 5.1E-10 21% 61%

1-REL-ISGTR 1.9E-10 3% 8.0E-11 3% 57%

1-REL-NOCF 9.0E-11 1% 6.6E-12 <1% 93%

Table 3-28 Population-Weighted Latent Cancer Fatality Risk, by Hazard Category, for the 0-10-Mile Interval for the Circa-2012 and 2020-FLEX Cases Circa-2012 Case 2020-FLEX Case Latent Cancer Latent Cancer Hazard Category Fatality Risk  % of Total Fatality Risk  % of Total

(/rcy) (/rcy)

Total 6.5E-08 100% 4.0E-08 100%

Internal Events and Floods 2.6E-08 40% 9.7E-09 24%

Internal Fires 2.5E-08 38% 2.2E-08 55%

Seismic Events 7.7E-09 12% 6.2E-09 15%

High Winds 6.3E-09 10% 2.4E-09 6%

3-66

Table 3-29 Statistical Analysis of the Population-Weighted Latent Cancer Fatality Risk for the 0-10-Mile Interval for the 2020-FLEX Case Point 5th- 95th-Median Estimate* Mean (/rcy) Percentile Percentile

(/rcy)

(/rcy) (/rcy) (/rcy)

Internal events and floods 9.8E-09 9.9E-09 8.1E-09 3.3E-09 2.2E-08 Internal fires 2.2E-08 2.1E-08 1.9E-08 8.8E-09 4.0E-08 Seismic events 6.2E-09 5.6E-09 4.6E-09 1.9E-09 1.2E-08 High winds 2.4E-09 7.5E-09 5.8E-09 1.9E-09 1.8E-08

• The reported point estimate value is the frequency-weighted sum of all release categories using the release category frequency point estimate values convolved with the conditional consequence mean values over all weather trials for each release category.

3-67

4 KEY ASSUMPTIONS, CONSIDERATIONS, AND UNCERTAINTIES FOR THE 2020-FLEX CASE This section documents key modeling assumptions, additional considerations (if any), and uncertainties associated with the 2020-FLEX case. This information is provided separately for the Level 1 PRA, Level 2 PRA, and Level 3 PRA in Sections 4.1, 4.2, and 4.3, respectively.

4.1 Level 1 PRA This section contains a summary of the 2020-FLEX case model. FLEX Support Guidelines (FSGs) are intended to provide preplanned FLEX strategies for performing specific tasks in support of emergency operating procedure and abnormal occurrence procedure functions to improve the capability to cope with beyond-design-basis external events. Section 4.1.1 describes the key modeling assumptions and Section 4.1.2 summarizes the key uncertainties and their impact on the results. A description of the changes incorporated into the Circa-2012 Level 1 PRA model to develop the 2020-FLEX model is provided in Appendix A of the overview report for internal events and floods (NRC, 2022a).

It should be noted that the PRAs for internal fires and seismic events for the Circa-2012 case (and, by extension, the 2020-FLEX case) were performed using PRA methods and data commonly available in the 2012 timeframe and, therefore, do not reflect more recent advancements in these areas. Some of the subsequent advancements in methods and data are identified in the Level 1 PRA reports for internal fires (NRC, 2023a) and seismic events (NRC, 2023b) as sources of uncertainty and either addressed as sensitivity analyses or identified as candidates for future study.

4.1.1 Key Assumptions Major assumptions for the Level 1 PRA portion of the 2020-FLEX case model are summarized below. Much of the information in this section is adapted from the overview report for internal events and floods (NRC, 2022a). As stated in that report, the following list includes the desired characteristics of a FLEX model as already used in the NRCs SPAR models; however, it should be noted that not all these modeling points are necessarily used explicitly in the FLEX model parametric sensitivity analyses performed for the L3PRA project. Some of these points are only included here to elaborate on the context and scope of the FLEX model basic events used in this sensitivity analysis.

• FLEX is considered for accident sequences modeled in the SBO event tree (ET). The SBO ET is entered when the Total Loss of All AC Power procedure is implemented.

The Circa-2012 case models were revised to account for FLEX if extended loss of all AC power (ELAP) is declared, since FSGs are only activated after ELAP is declared in Section B of the procedure.

• ELAP may be declared as early as within the first hour following a reactor trip, and as late as within the fourth hour following a reactor trip.

• Operators are directed to the FSGs from ECA-0.0 (Loss of All AC Power). Accident sequences while ECA-0.0 is invoked are modeled in the SBO ET. If ELAP is not invoked, or before ELAP is invoked, Section A of ECA-0.0 is followed.

4-1

• When ELAP is invoked, Section B of ECA-0.0 is used. The accident sequences in this case are modeled in the SBO ET.

• FLEX strategies and equipment are assumed to be unable to satisfy PRA success criteria for LOCA events.

• ELAP declaration can apply to all hazard categories modeled as long as the SBO ET is entered in the sequence. Different hazard categories can lead to additional failure modes that can negatively impact the ability of the FLEX strategies to successfully accomplish the core cooling and RCS makeup functions.

• The mission time is assumed to be 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />.

• It is assumed that if the TDAFW pump is available, operators will use it as long as possible to cope with the sequence, until AC power is restored. That is, operators will not voluntarily switch to low pressure SG injection if the TDAFW pump is still operational and can remove decay heat.

• Success of FLEX or TDAFW operation includes eventual installed plant AC power recovery to reach a safe and stable end state. Indefinite FLEX operation (e.g., beyond 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />) is not considered a success. However, this does not impact the 2020-FLEX model quantification since a simplified modeling approach is used.

• The SBO accident sequences in which the TDAFW pump is modeled to fail (to start or run) are not transferred to the SBO-FLEX ET. These sequences require AC power recovery within 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br />, which is not considered to be enough time to implement the FLEX actions, and failure of the TDAFW pump would preclude continued TDAFW operation.

• Highly accurate human error probabilities could not be assigned to individual FLEX actions due to the state of knowledge during this sensitivity analysis. Therefore, no attempt was made to separately model equipment and human failure events. Instead, the failure probabilities assigned to FLEX and continued TDAFW operation are meant to include equipment and operator failures as well as failure to recover AC power within 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />. This simplified approach is deemed appropriate due to the uncertainties in modeling deployment actions and valid when the new nodes are independent from the other ET nodes.

• As with the FLEX cases for internal events and floods (NRC, 2022a), no detailed analysis was performed to obtain failure probabilities for implementation of the FLEX strategies (F) or continuation of TDAFW without AC and DC power (T). Rather, a parametric sensitivity analysis was performed for the p-parameter, which is defined as the product of F and T.

• The failure probabilities used for FLEX and manual TDAFW pump operations are parametric values chosen by expert judgment, based on PRA experience, and experience with 70 SPAR models. The cases studied with different values of the parameter values are used to support the assertion that the selected base case values are reasonable and do not overly shift the results in either direction.

4-2

• Since the failure probabilities for FLEX strategies are assigned based on the hazard, not the hazard bin, they are treated as equivalent regardless of the severity of a seismic or wind event. This causes some distortion of the results since, realistically, the more severe initiating events should also have higher FLEX failure probabilities. However, based on event-specific constraints accounted for in the PRA model and the p-parameter values selected for this sensitivity analysis, the distortion is not expected to significantly affect the results.

• All FLEX model cases credit the installed hardware for the RCP shutdown seals, and a value of 0.01 is assigned as the combined failure probability of the shutdown seals to close and remain closed. It is also assumed that the reliability of these seals is not impacted by any of the hazards analyzed.

• For the purposes of propagating parametric uncertainty, the failure of the passive shutdown seals, FLEX strategies, and manual TDAFW pump operations are all represented by broad distributions. However, as mentioned previously, the relatively large number of basic events and cut sets used in the parametric uncertainty analysis appears to dilute (mask) the effect of basic events with higher uncertainties.

4.1.2 Key Uncertainties The most important modeling uncertainties introduced in the 2020-FLEX case are associated with the probabilities assigned to the three new ET nodes characterizing the new shutdown seals and the two FLEX-related failures, as discussed below.

The FLEX-related revisions to the CDF model consist of adding three new ET nodes to the existing SBO ET to expand the number of possible success paths. These nodes address (1) whether the RCP shutdown seals operate successfully, (2) whether ELAP is declared and FLEX strategies are successfully implemented, and (3) if FLEX is unsuccessful, whether extended TDAFW pump operation is successful under extended SBO conditions. 22 As discussed in Section 3.1.2, sensitivity analyses (referred to as alternative cases) were performed to assure that the failure probabilities assumed for these ET nodes are reasonable; namely, they do not significantly sway the results and insights in either a conservative or nonconservative direction.

There are other modeling uncertainties inherited from the Circa-2012 models, upon which the FLEX scenarios are superimposed. These modeling uncertainties are already discussed for the Circa-2012 models in the corresponding Level 1 PRA reports for all hazards (NRC, 2022d; NRC, 2022e; NRC, 2023a; NRC, 2023b; NRC, 2023c).

4.2 Level 2 PRA This section contains a summary of the key model assumptions, additional considerations, and sources of uncertainty for the Level 2 PRA portion of the 2020-FLEX case. The Level 2 PRA model is influenced by the Level 1 PRA FLEX-related modeling changes that result in a reduction of the CDF. Section 4.2.1 describes the key modeling assumptions. Section 4.2.2 22 To simplify the ET structure, the extended TDAFW node is assumed to inherently include failures associated with the safe/stable node from the original (Circa-2012 case) SBO ET model (note, the alternate charging node from the original SBO ET does not apply to the FLEX case).

4-3

discusses other modeling considerations. Section 4.2.3 summarizes the key uncertainties and their impact on the results.

4.2.1 Key Assumptions The key assumptions for the Level 2 PRA for the 2020-FLEX case model are summarized below. Other assumptions, in common with the Circa-2012 case for either internal events and floods or internal fires, seismic events, and high winds, can be found in those respective Level 2 PRA reports (NRC, 2022b; NRC, 2023d).

• The Level 2 PRA for the Circa-2012 case considers extended manual operation of the TDAFW pump during certain slow-developing SBO sequences. For the 2020-FLEX case, the credit for extended TDAFW pump operation is removed because it duplicates model changes that are made to the Level 1 portion of the model. The continued TDAFW pump operation is addressed in the Level 1 PRA portion of the 2020-FLEX case.

• The Level 2 PRA considers a variety of post-core damage actions to mitigate radiological releases. The credit for such actions is based on review of the reference plants accident management guidelines and severe accident MELCOR analyses. This approach does not credit actions following core damage during station blackout, or actions in the long-term after vessel breach. It is expected that operators would continue to take actions under station blackout conditions and during the longer timeframes, including possibly making use of off-site resources (e.g., Phase 3 FLEX strategies).

Nevertheless, modeling the reliability of such actions is beyond the scope of the Level 2 PRA human reliability analysis (HRA) approach for the L3PRA project and generally beyond the state-of-practice in Level 2 PRA.

• The modeled SAMG-directed actions are developed based on versions of the guidelines provided by the reference plant circa 2012. Updated versions of the SAMGs for the reference plant were later obtained, but these were not provided in time to be incorporated into the current analysis. In the updated SAMGs, FLEX equipment is listed as an option along with other installed plant equipment as possible ways to implement the strategies. Since the SAMGs are not prescriptive, but focus on accomplishing a function (e.g., control containment pressure) and provide options to accomplish it, the Level 2 PRA modeling of SAMGs is primarily driven by HRA and plant conditions due to earlier failures, not by availability of equipment. Availability of FLEX equipment provides another layer of defense-in-depth but is not expected to have significant impact on the post-core damage modeling.

• The existing representative source terms are generally no less applicable to the fire, seismic, and wind 2020-FLEX cases than to the Circa-2012 cases. The Circa-2012 analysis determined that it would be appropriate to retain the internal events source terms for every release category, despite significant changes in the makeup of the sequences for some categories, since the existing analyses represent very similar containment failure modes to those expected in fire/seismic/wind scenarios. That decision is applicable to the 2020-FLEX case as well. The primary impact on the model with the inclusion of FLEX strategies and passive RCP shutdown seals is the reduction of release category frequencies, but the model changes also influence the accident progression sequences and related modeling assumptions. Each release category 4-4

includes a mix of different accident sequences that have similar attributes but are not identical. The representative source term scenarios cannot exactly represent all the contributing elements. The representative source term should consider the balance of timing and magnitude to select a source term that conservatively bounds the range of outcomes for that release category. The considerations related to within-release-category variability are discussed in Section 2.5.2 of NRC (2022b), and generally apply to the 2020-FLEX case for fire, seismic, and wind. The changes resulting from the modeling of FLEX strategies were judged insufficient to justify defining additional representative sequences.

• The reference plants containment design does not require cooling or venting during the Phase 1 or Phase 2 FLEX implementation. Therefore, the inclusion of FLEX has no impact on the ability to control containment pressure or containment heat removal. The Level 2 PRA does consider longer-term actions directed by SAMGs for containment pressure control and containment heat removal. These actions are considered to be unaffected by the FLEX model changes.

4.2.2 Additional Considerations The inclusion of FLEX strategies significantly reduces the release frequencies, but the representative release source terms are not changed. As discussed in the key assumptions above, the modeled release categories each include a range of accident sequences, which are collectively represented by a single representative source term. A significant contribution to the variation within release categories is the timing of equipment failures that can delay core damage and containment failure, which can reduce the resulting source terms. As part of the Level 2 PRA for internal events and floods, a range of accident scenario simulations and sensitivity studies have been performed to account for this variation. The boundary conditions of the cases that were simulated to generate potential source terms are described in Appendix B of the Level 2 PRA report for internal events and floods (NRC, 2022b). The following discussion is adapted from the overview report for internal events and floods (NRC, 2022a).

One aspect of the accident simulation cases that significantly influences the timing of core damage and containment failure is the continued operation of auxiliary feedwater (AFW) during the accident sequence. The following variations were evaluated and are described in Appendix B of the Level 2 PRA report for internal events and floods (NRC, 2022b).

• Case 1 - A SBO scenario with the TDAFW pump available throughout the accident progression. This situation results in significant delay of core damage relative to the other simulated scenarios, and gradual containment overpressure failure does not occur within the accident simulation time of 7 days.

• Case 1A and its variations - A SBO scenario with the TDAFW pump available for 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> after the time of the initiating event. The scenario results in much earlier loss of heat removal from the SGs and onset of core damage compared to Case 1. In this scenario, the gradual overpressure of containment results in containment failure at approximately 68 hours7.87037e-4 days <br />0.0189 hours <br />1.124339e-4 weeks <br />2.5874e-5 months <br /> after the initiating event. Other variations of this case consider different modeling assumptions that can influence the occurrence of combustion events in containment.

4-5

• Case 1B and its variations - A SBO scenario with TDAFW unavailable from the start of the accident. Like Case 1A, the accident progresses more rapidly without feedwater to the SGs. Containment overpressure failure is predicted at approximately 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> after the initiating event.

• Case MU-1.2 - A variation of Case 1A that extends the AFW availability from 4 to 13 hours1.50463e-4 days <br />0.00361 hours <br />2.149471e-5 weeks <br />4.9465e-6 months <br />. The extended availability of AFW delays accident progression and eventual containment failure. Containment overpressure failure is predicted at approximately 86 hours9.953704e-4 days <br />0.0239 hours <br />1.421958e-4 weeks <br />3.2723e-5 months <br /> after the initiating event. The delayed containment failure and delayed environmental release results in a smaller total cumulative radiological release compared with Cases 1A and 1B. (This sensitivity analysis was performed as part of the uncertainty analysis in Appendix C of the Level 2 PRA report for internal events and floods [NRC, 2022b]).

In addition, other variations on the availability of AFW are addressed in Cases 3, 6, 7, and their variations. The different times of AFW availability give insights into the impacts of delaying the accident progression. As discussed in the MU-1.2 sensitivity analysis, the timing of AFW availability has a somewhat linear impact on the timing of containment failure and environmental release.

With the implementation of FLEX strategies, it may be more likely that core cooling can be successfully extended. The reference plants FLEX implementation plan states that critical DC power can be extended to a minimum of 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br /> by shedding unnecessary loads. These conditions would be more aligned with the MU-1.2 sensitivity case than with Cases 1A and 1B.

The sensitivity case results show that by delaying core damage, the eventual environmental releases are lessened. However, the representative source terms for the 2020-FLEX case are based on the Case 1A and 1B variations. The reasoning for using those representative cases is that the Level 1 PRA SBO sequences that are passed to the Level 2 PRA include failures to implement FLEX and extend TDAFW operation. If those actions fail, then SG cooling is assumed to be lost early in the event progression, similar to the assumptions in Cases 1A and 1B. If those actions are successful, then core damage is avoided, and the sequences would not be addressed in the Level 2 PRA. Realistically, there could be a range of intermediate outcomes with different combinations of human and equipment failures that influence the timing of event progression, which would tend to reduce the environmental releases.

4.2.3 Key Uncertainties There are several important modeling uncertainties that can affect the Level 2 PRA results. The model uncertainties for the Circa-2012 case are documented in the Level 2 PRA reports for internal events and floods (NRC, 2022b) and internal fires, seismic events, and high winds (NRC, 2023d). In general, the same model uncertainties apply to the 2020-FLEX case for internal fires, seismic events, and high winds.

As described in Section 3.2.2, the assumptions regarding accident termination significantly impact the surrogate risk metric results for LRF and CCFP. If there are credible reasons to model an accident scenario termination time at 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br /> after SAMG entry (as opposed to 7 days after event initiation), then both LRF and CCDP would be reduced significantly.

As discussed in the Level 2 PRA report for internal fires, seismic events, and high winds (NRC, 2023d), the source terms, in general, are expected to be similar to those for internal 4-6

events, since the hazards are generally not present by the time core damage occurs. Also, the model uncertainties that impact release magnitudes are not affected by the FLEX model changes. Therefore, the key model uncertainties that can significantly impact the releases for the 2020-FLEX case for internal fires, seismic events, and high winds are essentially the same as those for the Circa-2012 case for internal events and floods. For convenience, the discussion of some of these uncertainties is provided here, adapted from the overview report for internal events and floods (NRC, 2022a).

• The impacts of assumptions regarding the timing of Level 1 PRA failures and system availabilities (e.g., extended battery life during SBO) are discussed in Section 4.2.2, above.

• The timing of primary-side relief valve failure and realistic modeling of pressurizer relief tank (PRT) behavior can be very important in terms of cumulative iodine release if it is proximate to the time of containment failure. The PRT drying out and increasing temperature cause the iodine and tellurium classes (in particular) to re-volatilize and increases the environmental release. Refer to sensitivity cases MU-4.2, MU-5.1A, and MU-5.1B in Appendix C of the Level 2 PRA report for internal events and floods (NRC, 2022b) for additional discussion.

• Modeling assumptions related to accumulator injection can significantly affect the in-vessel melt progression. Increased accumulator injection flow rate delays aspects of the accident progression and aids in limiting the release. Refer to sensitivity case MU-4.1 in Appendix C of the Level 2 PRA report for internal events and floods (NRC, 2022b) for additional discussion.

• The susceptibility of the containment fan coolers to combustion-induced failure would tend to shift the release category frequency from the 1-REL-BMT release category to the 1-REL-LCF release category. Refer to sensitivity case MU-2.1 in Appendix C of the Level 2 PRA report for internal events and floods (NRC, 2022b) for additional discussion.

• The assumptions about location and size of containment failure can have a significant impact on environmental release. The long-term containment overpressure failure pathway is modeled through the tendon gallery. From there, two pathways are open, one to the environment with two-thirds of the flow area and the other to the auxiliary building with the remaining one-third flow area. If the release pathway is assumed only to open to the auxiliary building rather than directly to the environment, then significant auxiliary building fission product retention is seen. Refer to sensitivity case MU-8.1 in Appendix C of the Level 2 PRA report for internal events and floods (NRC, 2022b) for additional discussion.

• Alternative treatments are explored for the uncertainty of the timing of SG tube and hot leg nozzle creep rupture for severe accident-induced SGTR, as well as uncertainty related to secondary-side retention (e.g., in the dryers and separators) of fission products in all SGTRs. The alternative treatments tend to reduce release to the environment. Refer to sensitivity cases MU-11.1 and MU-11.2 in Appendix C of the Level 2 PRA report for internal events and floods (NRC, 2022b) for additional discussion.

• Uncertainties in ISLOCA modeling can impact the release. This includes modeling choices regarding the initial break size, whether the break is covered, turbulent 4-7

deposition in the piping, and downstream effects on auxiliary building status. Refer to Section 4.10 of Appendix C of the Level 2 PRA report for internal events and floods (NRC, 2022b) for additional discussion.

• Uncertainty in the containment isolation failure can significantly influence the release.

The modeling approach considers many potential isolation failure pathways (see Section D.2 of the Level 2 PRA report for internal events and floods [NRC, 2022b]). A representative failure of a 2-inch equivalent diameter failure area leading to the environment is used for the containment isolation failure release category. If a larger equivalent diameter failure area is assumed, then the magnitude of the environmental release increases significantly. Refer to sensitivity case MU-12.1 in Appendix C of the Level 2 PRA report for internal events and floods (NRC, 2022b) for additional discussion.

• The likelihood of containment failure is influenced by the uncertainty in the approach for modeling energetic burning of combustible gases; in particular, whether the combustion is most appropriately modeled as a deterministic process or a stochastic process. For this study a stochastic process was used. Refer to sensitivity case MU-7.1 in Appendix C of the Level 2 PRA report for internal events and floods (NRC, 2022b) for discussion of the approach and related sensitivity analyses. The analyzed sensitivity cases demonstrate that a combustion event large enough to over-pressurize and fail containment is unlikely in the early phases of accident progression. However, during the phase after vessel breach, combustion is much more likely, and without prior burn events early in the progression, the pressure caused by the combustion event could challenge the containment. The likelihood of the prior burn events is a key uncertainty in the combustion-induced containment failure modeling. The uncertainty in these and other modeling parameters may not be fully represented in a deterministic approach.

However, the deterministic analysis could suggest a lower likelihood for combustion events challenging containment integrity. This would tend to shift severe accident sequences from the 1-REL-ICF-BURN release category to the 1-REL-LCF release category.

4.3 Level 3 PRA As discussed previously, FLEX strategies are intended to provide coping capability to prevent core damage. Therefore, the primary effect of FLEX strategies on the PRA model is a reduction of the CDF from sequences involving SBO events or RCP seal failures. The main impact on the Level 2 and Level 3 models for FLEX strategies is carrying forward the modified Level 1 sequences, which results in reduced frequencies for the applicable release categories.

As discussed in Section 4.2, the implementation of FLEX strategies may affect accident progression modeling impacting the timing and magnitude of releases. However, while the release category frequency results have changed in the 2020-FLEX case, the representative source terms are still consistent with the Level 2 PRA logic model assumptions that were used in defining the release categories. Therefore, the 2020-FLEX case for internal fires, seismic events, and high winds uses the same representative source terms as the Circa-2012 case for internal fires, seismic events, and high winds (which, in turn, uses the same representative source terms as the Circa-2012 case for internal events and floods).

To the extent that the implementation of FLEX strategies alters accident progression timelines, the warning time (i.e., the time between the declaration of a GE and the onset of a major 4-8

release) could either increase or decrease. An increase in the warning time would not have a significant effect on the results because the warning time is already sufficiently long for most release categories to significantly reduce early phase exposures. In principle, a decrease in the warning time could result in increased early phase exposures; the significance of this hypothetical situation would need to be balanced against the long warning times in the base case model, the reduction of CDF associated with the 2020-FLEX case, and likelihood of lower release magnitudes associated with the more delayed accident progression.

The potential influences of the FLEX modeling changes on the scenario assumptions and timing are discussed in more detail in Section 4.2, but ultimately no changes were made to the representative accident scenarios or source terms. It is also not expected that there would be changes to other aspects of the Level 3 modeling (e.g., meteorology, atmospheric transport and diffusion, emergency response, economic factors, dosimetry, or health effects).

As indicated above, the 2020-FLEX case does not include any changes specific to the consequence analysis portion of the analysis; therefore, no specific assumptions, considerations, or uncertainties are identified here. However, it should be noted that the key assumptions, considerations, and uncertainties from the Circa-2012 case, as discussed in the Level 3 PRA reports for internal events and floods (NRC, 2022c) and internal fires, seismic events, and high winds (NRC, 2023e), also apply to the 2020-FLEX case for internal fires, seismic events, and high winds.

4-9

5 REFERENCES HPS, 2010 Health Physics Society, Radiation Risk in Perspective: Position Statement of the Health Physics Society (PS010-2), adopted January 1996, revised July 2010, McLean, Va.

HPS, 2019 Health Physics Society, Radiation Risk in Perspective: Position Statement of the Health Physics Society (PS010-4), adopted January 1996, revised February 2019, McLean, Va.

NRC, 1986 U.S. Nuclear Regulatory Commission, Safety Goals for the Operations of Nuclear Power Plants, Policy Statement, Republication (51 FR 30028), Federal Register, 1986.

NRC, 1990 U.S. Nuclear Regulatory Commission, Severe Accident Risks: An Assessmente for Five U.S. Nuclear Power Plants, NUREG-1150, December 1990 (Agencywide Documents Access and Management System (ADAMS) Accession No. ML040140729).

NRC, 2012 U.S. Nuclear Regulatory Commission, State-of-the-Art Reactor Consequence Analyses (SOARCA) Report, NUREG-1935, November 2012 (ADAMS Accession No. ML12332A057).

NRC, 2022a U.S. Nuclear Regulatory Commission, U.S. NRC Level 3 Probabilistic Risk Assessment Project, Volume 3x: Overview of Reactor, At-Power, Level 1, 2, and 3 PRAs for Internal Events and Internal Floods, April 2022 [Draft for Comment] (ADAMS Accession No. ML22067A210).

NRC, 2022b U.S. Nuclear Regulatory Commission, U.S. NRC Level 3 Probabilistic Risk Assessment (PRA) Project, Volume 3c: Reactor, At-Power, Level 2 PRA for Internal Events and Floods, April 2022 [Draft] (ADAMS Accession No. ML22067A214).

NRC, 2022c U.S. Nuclear Regulatory Commission, U.S. NRC Level 3 Probabilistic Risk Assessment (PRA) Project, Volume 3d: Reactor, At-Power, Level 3 PRA for Internal Events and Floods, April 2022 [Draft] (ADAMS Accession No. ML22067A215).

NRC, 2022d U.S. Nuclear Regulatory Commission, U.S. NRC Level 3 Probabilistic Risk Assessment (PRA) Project, Volume 3a, Part 1: Reactor, At-Power, Level 1 PRA for Internal Events, Part 1 - Main Report, April 2022 [Draft] (ADAMS Accession No. ML22067A211).

NRC, 2022e U.S. Nuclear Regulatory Commission, U.S. NRC Level 3 Probabilistic Risk Assessment (PRA) Project, Volume 3b: Reactor, At-Power, Level 1 PRA for Internal Flooding, April 2022 [Draft] (ADAMS Accession No. ML22067A213).

NRC, 2023a U.S. Nuclear Regulatory Commission, U.S. NRC Level 3 Probabilistic Risk Assessment Project, Volume 4a: Reactor, At-Power, Level 1 PRA for Internal Fires, August 2023 [Draft] (ADAMS Accession No. ML23166A036).

NRC, 2023b U.S. Nuclear Regulatory Commission, U.S. NRC Level 3 Probabilistic Risk Assessment Project, Volume 4b: Reactor, At-Power, Level 1 PRA for Seismic Events, August 2023 [Draft] (ADAMS Accession No. ML23166A038).

5-1

NRC, 2023c U.S. Nuclear Regulatory Commission, U.S. NRC Level 3 Probabilistic Risk Assessment Project, Volume 4c: Reactor, At-Power, Level 1 PRA for High Winds and Other Hazards, August 2023 [Draft] (ADAMS Accession No. ML23166A041).

NRC, 2023d U.S. Nuclear Regulatory Commission, U.S. NRC Level 3 Probabilistic Risk Assessment Project, Volume 4d: Level 2 PRA for Internal Fires, Seismic Events, and High Winds, August 2023 [Draft] (ADAMS Accession No. ML23166A060).

NRC, 2023e U.S. Nuclear Regulatory Commission, U.S. NRC Level 3 Probabilistic Risk Assessment Project, Volume 4e: Level 3 PRA for Internal Fires, Seismic Events, and High Winds, August 2023 [Draft] (ADAMS Accession No. ML23166A061).

5-2

NUREG-XXXX U.S. NRC Level 3 Probabilistic Risk Assessment Project Volume 4: Overview of Reactor, At-Power, Level 1, 2, and 3 PRAs for Internal Fires, Seismic Events, and High Winds Selim Sancaktar, Erick Ball, Alan Kuritzky, Alfred (Trey) Hathaway, Technical Keith Compton Division of Risk Analysis Office of Nuclear Regulatory Research U.S. Nuclear Regulatory Commission Washington, D.C. 20555-0001 Same as above Alan Kuritzky, NRC Level 3 PRA Project Program Manager The U.S. Nuclear Regulatory Commission performed a full-scope site Level 3 probabilistic risk assessment (PRA) project (L3PRA project) for a two-unit pressurized-water reactor reference plant. The scope of the L3PRA project encompasses all major radiological sources on the site (i.e., reactors, spent fuel pools, and dry cask storage), all internal and external hazards, and all modes of plant operation. A full-scope site Level 3 PRA for a nuclear power plant site can provide valuable insights into the importance of various risk contributors by assessing accidents involving one or more reactor cores as well as other site radiological sources. This report, one of a series of reports documenting the models and analyses supporting the L3PRA project, provides an overview of the reactor, at-power, Level 1, 2, and 3 PRA models for internal fires, seismic events, and high winds. The analyses documented herein are based on information for the reference plant as it was designed and operated as of 2012 and do not reflect the plant as it is currently designed, licensed, operated, or maintained. To provide results and insights better aligned with the current design and operation of the reference plant, this report also provides the results of a parametric sensitivity analysis based on a set of new plant equipment and PRA model assumptions for all three PRA levels. The sensitivity analysis reflects the current reactor coolant pump shutdown seal design at the reference plant, as well as the potential impact of FLEX strategies, both of which reduce the risk to the public.

PRA Level 1 PRA Internal fires Level 2 PRA Seismic events Level 3 PRA High winds Plant damage states Core damage Release categories CDF Source terms Level 3 PRA project Consequence analysis L3PRA project LERF Risk

NUREG-XXXX U.S. NRC LEVEL 3 PROBABILISTIC RISK ASSESSMENT (PRA) PROJECT