Text

PRA Realism Workshop- Industry Combined
	ML25272A201
Person / Time
Site:	Nuclear Energy Institute
Issue date:	09/29/2025
From:	True D; Nuclear Energy Institute
To:	; Office of Nuclear Regulatory Research
References
	Download: ML25272A201 (1)
	v • d • e

©2025 Nuclear Energy Institute 2 Risk-informed approaches have provided a better safety focus, improved safety and enabled efficiencies Uncertainties have been, and always will be, a challenge in risk-informed decision-making PRA realism is about finding the right balance to support good safety decision-making Risk-informed Truths

©2025 Nuclear Energy Institute 3 The use of PRA technology should be increased in all regulatory matters to the extent supported by the state-of-the-art in PRA methods and data and in a manner that complements the NRCs deterministic approach and supports the NRCs traditional defense-in-depth philosophy.

The use of PRA technology should be increased in all regulatory matters to the extent supported by the state-of-the-art in PRA methods and data and in a manner that complements the NRCs deterministic approach and supports the NRCs traditional defense-in-depth philosophy.

1995 PRA Policy Statement PRA is an expression of our state of knowledge

©2025 Nuclear Energy Institute 7 Treatment of Uncertainties 1 - In this context, Defense in Depth includes Redundancy, Diversity, and Barriers Contributor Deterministic Treatment Parametric Uncertainty (Failures Happen)

Redundancy Modeling Uncertainties (aka Known-Unknowns)

Safety Margin and/or Defense in Depth1 Completeness Uncertainties (aka Unknown-Unknowns)

Defense in Depth1 and/or Safety Margin

©2025 Nuclear Energy Institute 8 Treatment of Uncertainties Contributor Traditional Regulatory Treatment Parametric Uncertainty Redundancy 1 - In this context, Defense in Depth includes Redundancy, Diversity, and Barriers Contributor Traditional Regulatory Treatment Parametric Uncertainty Redundancy Modeling Uncertainties (aka Known-Unknowns)

Safety Margin and/or Defense in Depth1 Contributor Deterministic Treatment Risk-informed Treatment Parametric Uncertainty (Failures Happen)

Redundancy Statistically Propagated Modeling Uncertainties (aka Known-Unknowns)

Safety Margin and/or Defense in Depth1 Conservatism or Sensitivity Studies Completeness Uncertainties (aka Unknown-Unknowns)

Defense in Depth1 and/or Safety Margin Assessment of Defense in Depth

& Safety Margin

1. Uncertainties make assumptions unavoidable in PRA

2. The key issue is how those uncertainties impact our decision-making

3. Undue conservatisms can lead to poor decisions Conservative may not be safer

4. We must not let our technical hubris prevent us from understanding the true risks and uncertainties Risk-informed Decision-making

©2025 Nuclear Energy Institute 10 LLWR PRA Aspect of PRA Current Fleet Important Safety Systems Active Equipment Reliability Data Plentiful System Reliability Data Plentiful Human Actions for Accident Avoidance Many Nature of Important Human Actions Short-term Accident Progression Uncertainties Limited Modes of Operation At-Power Hazard Sources Core Predominant Risk Drivers Internal Hazards Accepted Methods Nearly All Margin to Safety Goals Adequate Useful Surrogate Risk Measures In Use Current Tools and Methods Reasonably Well Adapted for These

©2025 Nuclear Energy Institute 11 Aspect of PRA Current Fleet Advanced Reactors Important Safety Systems Active Passive Equipment Reliability Data Plentiful Limited System Reliability Data Plentiful None Human Actions for Accident Avoidance Many Few Nature of Important Human Actions Short-term Long-term Accident Progression Uncertainties Limited Plentiful Modes of Operation At-Power All Hazard Sources Core All Predominant Risk Drivers Internal Hazards External Hazards Accepted Methods Nearly All Some Margin to Safety Goals Adequate Significant Useful Surrogate Risk Measures In Use Not Defined LLWR PRA AR PRA

Safety Margins =

Risk-Informed PRA Insights DID &

Safety Margins

=

Risk-Informed Current Fleet Licensing Modernization Project DID &

Safety Margins PRA Results

Better safety focus & improved efficiency We cant pretend that uncertainties dont exist Risk-informed decisions must acknowledge the strengths & limitations of the models and uncertainties Closing Thoughts:

The Goal Must Be Good Decisions

w w w. e p r i. c o m Ashley Lindeman Principal Project Manager NRC/Industry Realism Workshop September 30, 2025 Fire PRA Realism

2 Fire PRA Realism Efforts (Published 2019-2020)

EPRI / NRC Publication Title Notes NUREG-1921 Supplement 2 3002013023 EPRI/NRC-RES Fire Human Reliability Analysis GuidelinesQuantification Guidance for Main Control Room Abandonment Scenarios Quantitative guidance for quantifying HFEs related to main control room abandonment 3002016004 Alternative Method for Quantification of Decision Making for Main Control Room Abandonment 3002015992 Nuclear Station Electrical Distribution Systems and High-Energy Arcing Fault Events NUREG-2232 3002015997 Heat Release Rate and Fire Characteristics of Fuels Representative of Typical Transient Fire Events in Nuclear Power Plants Heat release rate testing for common NPP transient combustibles NUREG-2230 3002016051 Methodology for Modeling Fire Growth and Suppression Response for Electrical Cabinet Fires in Nuclear Power Plants Interruptible/growing fire profiles and split fractions, timing profiles for interruptible and growing fires, additional opportunities to credit plant personnel suppression NUREG-2178 Volume 2 3002016052 Refining and Characterizing Heat Release Rates from Electrical Enclosures During Fire, Volume 2: Fire Modeling Guidance for Electrical Cabinets, Electric Motors, Indoor Dry Transformers, and the Main Control Board Obstructed radiation, cabinet to cabinet propagation, fire location factor, heat release rates for motors and dry transformers, new main control board model, NSP floor for the main control room 3002016053 Methodology for Modeling Plant Trip Probabilities in Nuclear Power Plant Fire Probabilistic Risk Assessment NUREG-2233 3002016054 Methodology for Modeling Transient Fires in Nuclear Power Plant Fire Probabilistic Risk Assessment Will be jointly published as NUREG-2233

3 Recent Fire PRA Methods & Data (2021-2025)

EPRI / NRC Publication Title Notes NUREG-2262 &

3002025942 High Energy Arcing Fault Frequency and Consequence Modeling

Published April 2023

PRA methodology for HEAF RIL 2022-09 &

3002025123 Determining the Zone of Influence for High Energy Arcing Faults Using Fire Dynamics Simulator

Published November 2022

ZOIs used in NUREG-2262 RIL 2022-01 &

3002023400 Target Fragilities for Equipment Vulnerable to High Energy Arcing Faults

Published May 2022

Target fragilities used in NUREG-2262 3002026391 HEAF Prevention, Mitigation, and Lessons Learned

Published June 2023

White paper summarizing lessons learned from HEAF OPEX review and HEAF project 3002020746 Heat Release Rate Testing for Small Electrical Enclosures

Published September 2021

Testing results for small electrical panels 3002020747 Modeling of Oil Fires in Fire Probabilistic Risk Assessment

Published October 2021

Leak categories: pressurized spray and unpressurized (leaks)

NUREG-2180 Supplement 1 &

3002028821 Determining the Effectiveness, Limitations, and Operator Response for Very Early Warning Fire Detection Systems in Nuclear Facilities-Update to Event Tree Parameters (Alpha and Pi) and Integration of NUREG-2230 Methods:

Supplement 1

Joint EPRI/NRC report

Published April 2024

Updates Alpha and Pi parameters (derived from industry opex) 3002032020 Component-Based Fire Ignition Frequency Methodology

Published August 2025

Provides technical basis for calculating component-based fire ignition frequency, conceptual examples, and histograms of component counts

4 Component Based Fire Ignition Frequencies Recent Fire Event Data Collected and Analyzed Methodology Component-Based Frequencies In Progress:

Component-Based Fire Ignition Frequencies EPRI 3002029303 (July 2024) 2015-2019 Fire Event Data EPRI 3002032020 (August 2025)

Methodology EPRI 3002005302 (May 2016) 2010-2014 Fire Event Data Tech Transfer Future:

Tech Transfer Webcast

5 Fire Ignition Frequency, Non-Suppression Probability, and Split Fraction Update Work is ramping up on the next revision of fire ignition frequencies, split fractions, and non-suppression probabilities Fire ignition frequencies

- Plant-based

- Component-based

- Bin 15 subdivision Split fractions

- Contained among various reports

- Update and consolidate Non-suppression probabilities

- Update consistent with NSP categories (NUREG-2230, NUREG-2178 V.2)

6 2022 Skyline - Core Damage Frequency Per Source 1.0E-11 1.0E-10 1.0E-09 1.0E-08 1.0E-07 1.0E-06 1.0E-05 1.0E-04 1.0E-03 1.0E-02 1.0E-01 1.0E+00 Electrical cabinets Main control board High voltage HEAF Trans. fires - welding/cutting Transients Pumps Transformers (Oil/dry)

Ventilation Subsystems Bus Duct Battery Chargers Transients T/G Oil T/G Oil Cable run - self-ignited Transformer - Catastrophic Transformer - Non Catastrophic Electric motors Transients Trans. fires - welding/cutting Diesel generators Low voltage HEAF Transients T/G Hydrogen T/G Hydrogen Air Compressors Yard transformers (Others)

Misc. Hydrogen Fires Junction box Cable fires - welding/cutting Trans. fires - welding/cutting Main feedwater pumps Reactor Coolant Pump Off-gas/H2 Recomb. (BWR)

RPS MG sets Batteries Dryers Hydrogen Tanks Boiler Iso phase ducts Turbine-Generator Excitor CDF Chart Average Maximum Minimum

7 Cable Tray Ignition Testing complete

- Test data reviewed Working through how to bin data in a sensible manner (fitting, conduit, seals)

Simulation and model validation

- Restarting now that testing is complete

- To address additional configurations and parameters Fire PRA incorporation

- Initial model developed

- Documentation

8 Fire PRA - Overview Schedule of major activities through 2027 Ongoing activities

- Fire PRA training & wiki updates 2025 2026 2027 Task 1st half 2025 2nd half 2025 1st half 2026 2nd half 2026 1st half 2027 2nd half 2027 Fire Event Data Review & classify 2020-2025 fire event data Fire Ignition Frequencies Methodology Fire Ignition Frequency, Non-suppression, and split fraction update (plant & component based)

Hydrogen Co-located Hydrogen Update to Appendix N (Hydrogen Fires)

Lithium-Ion Batteries Li-Battery Whitepaper Limited testing Fire PRA Guidance for Li-Batteries Bulk Cable Tray Ignition Testing Review test data and perform simulations and validation Incorporate test insights into Fire PRAs (method development & implementation guidance)

Cable Coatings NRC to perform accelerating aging Update to Appendix Q.2.1 (Cable Coatings)

Fire PRA Maintenance Fire PRA Maintenance Fire PRA Workshop Workshop Option 1 Workshop Option 2

9 Fire PRA Potential Future Topics Better guidance for modeling large yard transformer fires Updating automatic detection and suppression probabilities Improved multi-compartment analysis guidance

- How to be more efficient based on lessons learned from implementing Use of advanced computing

- Screening for multi-compartment, hot gas layer, MCR abandonment Fire PRA model maintenance

- How to efficiently and systematically assess new information

10

w w w. e p r i. c o m TOGETHERSHAPING THE FUTURE OF ENERGY

Copyright 2024, BWR Owners Group, All Rights Reserved BWR Expertise Proven Solutions Modeling Conservatisms in Utility PRAs Suzanne Loyd, Constellation BWROG IRIR Chair NRC PRA Realism Public Workshop September /October 2025

Utility model development has progressed significantly over the past several years in support of risk-informed program license applications

Fire PRA modeling has been a major focus area for the past 10+ years

Development of fire PRAs

Refinement of models to reduce conservatism

Fire risk typically is the largest contributor to overall site risk

Conservatisms in the fire modeling identified during development

New method development to reduce conservatisms is still ongoing

Utilities must balance resources while developing PRA models that are acceptable to support risk-informed programs

Many utility models have incorporated new methods that are available

Use a targeted approach to refine risk-significant areas

Most methods are not fully implemented for all scenarios

Some refinements require significant resources to apply new methods

New method use requires focused-scope peer review

Depending on the method, refinement can be very labor intensive

Walkdowns

Operator interviews/site interface

Example - HEAF methodology updates require significant resources that can take months of person-hours to incorporate

Schedule also limits full implementation of refinements

Required updates every 2 cycles

Scope of full update often limits extra refinements that can be done

Utilities continue to refine as much as possible or an as-needed basis

Often driven by emergent issues like SDPs

Time pressure for emergent issues limit scope of refinements to very narrow focus

Utility models continue to improve, but more refinement to reduce conservatisms is possible

Focus on risk-significant refinements give confidence that the models are sufficiently realistic to support risk-informed applications

Resources for model refinements are limited with large scope of work that must be done to do routine model updates

Copyright 2024, BWR Owners Group, All Rights Reserved BWR Expertise Proven Solutions Modeling Conservatisms in Utility PRAs Suzanne Loyd, Constellation BWROG IRIR Chair NRC PRA Realism Public Workshop September /October 2025

Utility model development has progressed significantly over the past several years in support of risk-informed program license applications

Fire PRA modeling has been a major focus area for the past 10+ years

Development of fire PRAs

Refinement of models to reduce conservatism

Fire risk typically is the largest contributor to overall site risk

Conservatisms in the fire modeling identified during development

New method development to reduce conservatisms is still ongoing

Utilities must balance resources while developing PRA models that are acceptable to support risk-informed programs

Many utility models have incorporated new methods that are available

Use a targeted approach to refine risk-significant areas

Most methods are not fully implemented for all scenarios

Some refinements require significant resources to apply new methods

New method use requires focused-scope peer review

Depending on the method, refinement can be very labor intensive

Walkdowns

Operator interviews/site interface

Example - HEAF methodology updates require significant resources that can take months of person-hours to incorporate

Schedule also limits full implementation of refinements

Required updates every 2 cycles

Scope of full update often limits extra refinements that can be done

Utilities continue to refine as much as possible or an as-needed basis

Often driven by emergent issues like SDPs

Time pressure for emergent issues limit scope of refinements to very narrow focus

Utility models continue to improve, but more refinement to reduce conservatisms is possible

Focus on risk-significant refinements give confidence that the models are sufficiently realistic to support risk-informed applications

Resources for model refinements are limited with large scope of work that must be done to do routine model updates

IRIR = Integrated Risk Informed Regulation Committee

Provides a forum to allow its member utilities to

Maintain and improve plant safety

Achieve higher plant reliability

Minimize and share costs

Facilitate regulatory interaction.

IRIR Leadership

Suzanne Loyd, Chair

Gary DeMoss, Vice Chair

Tony Mangan, Vice Chair

IRIR Technical and Administrative Support - GE Vernova

IRIR works with the PWROG RMC closely for generic PRA topics - complementary work

NRC-funded equipment reliability data analysis has been performed at INL for many years

Predominant data source is the INPO ISIS data set, with consideration of LER data

INL data is used by most US nuclear power plants

As a prior for a Bayesian calculation when plant specific data is used

As a generic data set for other PRA equipment

INL data is used to quantify the NRC SPAR models

Many international plants also use this US based data

Turbine Driven Pump (HPCI & RCIC) failure to run rates seemed unrealistically high

Anecdotally, members knew of specific instances where INL categorizations seemed to be incorrect

The simplified example is the reported EDG A failed its 24-hr surveillance test due to...

Often, the EDG could have successfully responded to a real plant event.

Initial data report - TP20-3-143, Rev. 1, Review of BWR Component Reliability Data Issues

Found a number of failure to run events that were not PRA failures

Adjusted priors are being used by industry

Expanded data report - TP20-3-143, Rev. TBD, Review of Component Reliability Data Issues

Updated to include EDG, HPCI/RCIC, TDAFW and CCF failure events reviewed by both BWROG/IRIR and PWROG/RMC members.

Document will be delivered to INL & NRC later this fall for consideration in scheduled 2026 INL data update

Regular industry review is valuable to ensure that the INL data is a realistic as possible.

Industry review needs to be provided in a timely manor that INL can consider the input when publishing data reports

INL/NRC should continue this valuable data collection

Scope may need to be expanded further. Suggestions include

Relief valves

Air Compressors

Initiating Events

Industry needs to improve the clarity of ISIS reports

Industry and NRC/INL need to consider eliminating the impact vectors for non-CCF events that had some of the characteristics of actual PRA CCF events

Global Expertise

One Voice FLEX Data Matthew Degonish - Vistra

In May of 2017 [ML17031A269] the US NRC assessed guidance related to crediting FLEX equipment in PRA models.

The NRC staff found certain elements of NEI 16-06 lacked sufficient technical justification for crediting FLEX.

Conclusions 4 through 10 pertain to PRA Data Analysis.

It was noted that The staff recommends that organizations such as INPO, EPRI, and NRC collect and share operating experience data of portable equipment to improve the technical basis for relevant reliability data.

In early 2020, the PWROG led an industry effort to collect and analyze operating experience for FLEX equipment.

History

The overall goal of the PWROG was to develop failure rate parameters for FLEX equipment. Additionally, the following goals were established:

Use data from comprehensive and consistently collected and interpreted sources that are maintained and updated.

Characterized the current industry performance

Structure the characterization of industry-average performance such that results can be updated periodically

Gain insights on the quality and effectiveness of FLEX equipment and PM tasks in ensuring long-term reliability of equipment.

Provide recommendations to ease the process for future data updates History

The PWROG developed a project team to assist in defining the needs of utilities. This team, along with industry input, defined the scope of equipment to be considered in the data analysis:

Portable Diesel Generators

Portable Combustion Turbine Generators

Portable Diesel-Driven Pumps

Portable Motor-Driven Positive Displacement Pumps

Portable Air Compressors History

Original efforts collected random failure data from the implementation of FLEX at each site through roughly 2019.

Subsequent efforts have collected data through December 2021 with future efforts which are planned to collect data through December 2025.

Data Analysis Included

Collection of Failure Data and Success Data

Classification of Failures

Pooling Analysis of Data

Bayesian Update of Industry Data

Outlier Evaluation

Original data analysis was subjected to an audit by the US NRC and INL

[ML20155K827]. Observations were provided back to the PWROG which were then incorporated into the analysis.

History

History Component Failure Mode Dist.(1)

Mean(2)

Method(3)

Portable Diesel Generator FTR G

1.03E-02 0.856 82.9 EB FTS B

4.35E-02 67.5 1482.5 JNI Portable Combustion Turbine Generator FTR G

1.86E-02 2.5 134.2 JNI FTS B

3.30E-02 7.5 219.5 JNI Portable Diesel-Driven Pump FTR G

1.55E-02 16.5 1065.2 JNI FTS B

3.38E-02 75.5 2161.5 JNI Portable Motor-Driven Positive Displacement Pump FTR G

1.56E-02 3.5 223.7 JNI FTS B

7.35E-03 3.5 472.5 JNI Portable Air Compressor FTR G

2.82E-02 4.5 159.8 JNI FTS B

2.46E-02 6.5 257.5 JNI

1. Gamma Distribution Parameterized as =

() 1()

2. Failure Rates from Tabel 6-1 of PWROG-18042-P/NP Revision 1

3. JNI - Bayesian updated using Jeffreys Noninformative Prior, EB - Empirical Bayes Analysis

PWROG continues to pursue collection and evaluation of data.

Additional data has been collected and evaluated through December 2021.

No changes to approved data analysis

Future efforts will collect and analyze data through December 2025.

Have performed an initial evaluation of common cause failures based on industry operating experience.

Follows mapping-up/mapping down process as describe in various approved NUREGs.

Selects a Prior Distribution

Generates Alpha Factor parameter estimates up to a group size of 4 for various component types

Has not been reviewed by the NRC. Current NRC conclusion is that currently available CCF parameters (as published by INL) should be used.

History

What is the context of realism for various forms of risk informed regulation?

How can lack of realism impact risk informed regulatory decisions?

What role do SDPs and supporting documents play in the level of realism available for risk informed decisions?

How do the principles of risk informed decision making interact with realism?

Overview

©2025 Nuclear Energy Institute 4 Basic Premises of Risk Informed Regulation Improves safety by highlighting vulnerabilities and hidden dependencies Improves efficiency by allowing flexibility (e.g., in maintenance, testing, inspection) where risk is very low

©2025 Nuclear Energy Institute 5 Context of Realism in Various Risk Informed Applications Context-specific realism depends on how PRA is used in decision-making Generally, best-estimate modeling that preserves the correct relative importance of risk contributors Realism may refer to a PRA model, or scenario specific risk

Average risk

Facility modifications

Risk Informed Tech Spec Surveillance Frequencies

Scenario specific risk

Risk Informed Tech Spec Completion Times

Significance Determination Process

Instead of using the best-estimate failure probability of the EDGs (e.g., based on actual plant data), the PRA uses a conservative bounding value

This inflates the calculated contribution of EDG failures to plant risk Resulting PRA Output

EDGs appear to dominate risk

Risk-informed evaluation conclude that EDG allowed outage times cannot be extended to support maintenance flexibility Incorrect Decision

Resources are over-concentrated on EDG surveillance and maintenance.

Other contributors to actual plant risk are under-addressed because they appear less important in the PRA model

Any electrical cabinet fire grows to full-room involvement with no credit for suppression

Realistic data show many fires are detected and suppressed within a few minutes.

Effect on PRA

The calculated conditional core damage probability from cabinet fires is exaggerated

Fire risk dominates the total plant risk Incorrect Decision

Plant management invests heavily in cabinet fire barriers and redundant suppression systems.

More meaningful risk reductions (like early detection upgrades) are overlooked.

©2025 Nuclear Energy Institute 8 PRA conservatism shifts the apparent risk profile, which misleads prioritization in risk-informed applications Best-estimate values with uncertainty characterization give much more useful insights for decision-making Importance of Realism in Risk-Informed Decision Making

©2025 Nuclear Energy Institute 9 SDPs and guidance supporting their conduct (e.g. RASP Handbook) can lack realism in key areas Reliance on generic CCF values leads to over-prediction of CCF risk, especially in cases where actual evidence or design features might reduce CCF likelihood Limited opportunities for refining conservatisms when better data are available SPAR-H has known limitations Potentially overstates the risk of findings compared to what would be found with refined modeling Creates disconnect when utilities use realistic models for licensing applications and daily operational decisions Critical to move forward on improving realism in SDPs to support consistency in risk informed decision making Role of SDPs and supporting documents

©2025 Nuclear Energy Institute 10 Realism and the Principles of Risk Informed Decision Making PRA realism is central to all five principles Realism ensures that risk-informed regulation is credible, balanced, and effective

A realistic PRA demonstrates that proposed changes are consistent with the regulations

If overly conservative: Could wrongly suggest non-compliance, blocking justified changes Compliance with Current Regulations

PRA realism helps identify where layers truly provide protection

If conservative: Might suggest defense-in-depth is eroded when it isnt.

Defense-in-Depth is Maintained

Realistic modeling of uncertainties and plant performance ensures safety margins are neither understated nor overstated

If conservative: May imply safety margins are too thin and block changes with net benefit to safety Safety Margins are Maintained

Realistic PRA is critical to correctly quantify CDF and LERF.

If conservative: Inflates CDF/LERF, preventing risk-informed improvements.

Changes in Risk are Small and Consistent with Safety Goals

A realistic PRA baseline makes performance monitoring meaningful deviations stand out

If conservative: Creates false alarms performance looks worse than it really is Implementation is Monitored Using Performance Measurement Strategies

ML25272A201

Text

Navigation menu

Search