ML21217A304
| ML21217A304 | |
| Person / Time | |
|---|---|
| Issue date: | 08/09/2021 |
| From: | Patrick Raynaud, David Rudland NRC/NRR/DNRL, NRC/RES/DE |
| To: | |
| Hiser A | |
| Shared Package | |
| ML21217A247 | List: |
| References | |
| Download: ML21217A304 (11) | |
Text
Industry / NRC Materials Programs Technical Information Exchange PFM Regulatory Guide Update David Rudland, Senior Technical Advisor, NRR/DNRL Patrick Raynaud, Senior Materials Engineer, RES/DE/REB August 10-11, 2021
Objectives and Disclaimers The goal of this presentation:
- 1. Provide an update on the draft PFM Regulatory Guide and Technical Basis Disclaimers:
- 1. All content shown in these slides is pre-decisional and does not represent an official position of the NRC
- 2. None of the ideas presented here are final
- 3. This content is not intended to be guidance on what constitutes an acceptable approach for PFM submittals to the NRC 08/10/2021 Industry / NRC Annual Technical Exchange 2
PFM Guidance Development Documents
- Technical Letter Report on NRCs preliminary thoughts on increasing confidence in PFM analyses (publicly available in ADAMS at ML18178A431)
- PFM NUREG technical basis
- NRR review complete
- Currently continuing internal review
- Draft Regulatory Guide
- NRR review complete
- Currently continuing internal review
- Report titled Application of Probabilistic Analysis Techniques in Probabilistic Fracture Mechanics
- Review Complete
- Will publish after public release of RG and NUREG
- FAVOR Example Study Report
- Review Complete
- Will publish after public release of RG and NUREG 08/10/2021 Industry / NRC Annual Technical Exchange 3
Analytical Steps in a Probabilistic Fracture Mechanics Demonstration Step Action
- Define the regulatory context
- 1. Translate regulatory *Define the QoI and how it relates to the model output and acceptance criteria Plan requirements into an
- Determine the suitability of the PFM code for the application analysis plan
- Identify key elements of the problem that impact analysis choices Analyze 2. Characterize model input uncertainty
- Identify uncertain model inputs
- Specify probability distributions on uncertain inputs
- Select a sampling scheme
- 3. Estimate QoIs and *Assess sampling uncertainty their associated Analysis uncertainty *Conduct sensitivity analysis results *Conduct output uncertainty analysis inconclusive, refinements 4. Conduct sensitivity studies to assess the *Determine a set of sensitivity studies needed.
credibility of modeling *Conduct sensitivity studies and present results assumptions
- 5. Draw conclusions *Interpret analysis results Synthesize From analysis results *Iterate on the analysis process to refine model results 08/10/2021 Industry / NRC Annual Technical Exchange 4
Detailed Thoughts on a Graded Approach for PFM
- PFM is complex
- Topics Covered
- Software QA and V&V
- The depth and breadth of a PFM analysis might vary - Models widely depending on several factors - Inputs
- Uncertainty Propagation
- It makes sense to take a graded approach - Convergence
- for PFM analyses themselves - Sensitivity Analyses
- for the level of detail to be presented as part of an - QoI Uncertainty Characterization evidence package - Sensitivity Studies
- General Principles
- Higher safety significance More analyses, more documentation
- Higher complexity Higher burden to create defensible and rigorous
- Higher level of novelty evidence 08/10/2021 Industry / NRC Annual Technical Exchange 5
PFM Graded Approach
- EPRI proposed minimum requirements for PFM applications and their documentation in EPRI BWRVIP 2019-016 white paper:
Suggested Content for PFM Submittals to the NRC, ML19241A545
- NRC reviewed the white paper and took EPRIs recommendations into consideration while developing a graded approach for PFM
- In many cases, NRC proposes to reduce the amount of documentation compared to EPRIs recommendations
- In some cases, NRC proposes additional documentation
- For each major topic in a PFM application, documentation categories are defined depending on the features of the specific application. Categories are independent from each other: can be in different categories for different topics
- SQA and V&V
- Models
- Inputs
- Uncertainty propagation
- Convergence
- Sensitivity analyses
- Output uncertainty characterization
- Sensitivity studies 08/10/2021 Industry / NRC Annual Technical Exchange 6
Example: Software QA and V&V DRAFT
- Safety demonstrations for the NRC usually require that a QA program be in place
- Pre-submittal meetings are very useful to help ensure that everyone agrees on the graded approach path Category Description Graded Approach QV-1 NRC-approved code QV-1A Exercised within validated range Demonstrate code applicability within the validated range.
Describe features of the specific application where the code is validated and applicable (i.e., areas of known code capability).
QV-1B Exercised outside of validated range Provide evidence for the applicability of the code to the specific application with respect to the areas of unknown code capability.
Describe features of the specific application where the code has not been previously validated and applied (i.e., areas of unknown code capability).
QV-1C Modified Give an SQA summary and V&V description for modified portions of the code.
Demonstrate that the code was not broken as a result of changes.
Make detailed documentation available for further review upon request.
QV-2 Commercial off-the-shelf software designed Demonstrate code applicability.
for the specific purpose of the application Describe the software and its pedigree.
Make software and documentation available for review upon request.
QV-3 Custom code Summarize the SQA program and its implementation.
Provide a basic description of the measures for quality assurance, including V&V of the PFM analysis code as applied in the subject report.
For very simple applications, possibly provide the source code instead of standardized SQA and V&V.
Include separate deterministic fracture mechanics analyses to support other validation results, as appropriate for a given application.
08/10/2021 Industry / NRC Annual Technical Exchange 7
Example: Inputs DRAFT Category Graded Approach I-1D List input value.
I-1R List input distribution type and parameters. Input Category Low Knowledge of Input High Knowledge of Input If applicable, list uncertainty classification (aleatory or epistemic). Characteristics Characteristics I-2D List input value. Deterministic Uncertain Deterministic Uncertain If there is a lack of data, justify the use of expert judgment. High Importance I-4D I-4R I-3D I-3R I-2R List input distribution type and parameters. Low Importance I-2D I-2R I-1D I-1R If applicable, list uncertainty classification (aleatory or epistemic).
If there is a lack of data, justify the use of expert judgment.
I-3D List input value.
State the rationale for setting the input to a deterministic value.
For each deterministic input, give the rationale (method and data) for the selection of its numerical value, along with any known conservatisms in that numerical value and the rationale for such conservatisms.
Reference documents that contain the foundation for input choices.
Explain the correlations between inputs and how they are modeled, and verify that correlated inputs remain consistent and physically valid.
Describe any sensitivity analyses/studies performed to show that the input or its classification does not have a significant effect on the QoI.
I-3R List input distribution type and parameters.
If applicable, list uncertainty classification (aleatory or epistemic).
If relevant, classify uncertain inputs as aleatory or epistemic and give the corresponding rationale.
For each uncertain input, describe both its distribution parameter values and its distributional form. Give the rationale (method and data) for selecting each distribution, including any known conservatisms in the specified input distributions and the rationale for the conservatism. Detail the distributional fitting method, including interpolation, extrapolation, distribution truncation, and curve fitting.
Reference documents that contain the foundation for input choices.
Explain the correlations between inputs and how they are modeled, and verify that correlated inputs remain consistent and physically valid.
Describe any sensitivity analyses/studies performed to show that the input or its classification does not have a significant effect on the QoI.
I-4D See I-3D.
If there is a lack of data, justify the use of expert judgment.
I-4R See I-3R.
If there is a lack of data, justify the use of expert judgment.
08/10/2021 Industry / NRC Annual Technical Exchange 8
Experience from Recent PFM Submittals
- Several PFM related alternatives have been reviewed. At the time of the recent reviews, details of graded approach were still under development (and are currently under internal review)
- PFM code audit is a useful tool to better understand code details and basis
- The (draft, pre-decisional) PFM guidance
- helps the industry prepare submittals that have a consistent level of quality
- helps the staff understand more complex analyses such as sensitivity analysis and sensitivity studies and their importance in the interpreting the PFM results
- helps put a framework around the various topics, thus organizing the review 08/10/2021 Industry / NRC Annual Technical Exchange 9
ACRS Interaction
- Joint Subcommittees on Metallurgy & Reactor Fuels and Structural & Seismic Analysis
- Along with the documents, staff discussed the need for PFM guidance and reviewers perspective of the guidance.
- Brief was well received with many questions from several members.
- Since this was an informational brief, no letter from ACRS is expected.
08/10/2021 Industry / NRC Annual Technical Exchange 10
Next Steps
- NRC will publish draft regulatory guidance and an accompanying draft technical basis NUREG, both for public comments (August 2021, hopefully)
- NRC will gather public comments on the draft guidance and draft NUREG
- NRC will address all comments received
- Public meetings may take place, if needed
- At least one more ACRS briefing on the topic of PFM and PFM regulatory guidance is expected in the future 08/10/2021 Industry / NRC Annual Technical Exchange 11