ML20241A063
| ML20241A063 | |
| Person / Time | |
|---|---|
| Issue date: | 08/17/2020 |
| From: | Matthew Homiack NRC/RES/DE |
| To: | Raj Iyengar NRC/RES/DE |
| M. Homiack | |
| Shared Package | |
| ML20241A062 | List: |
| References | |
| Download: ML20241A063 (4) | |
Text
.
August 17, 2020 MEMORANDUM TO: Raj M. Iyengar, Chief Component Integrity Branch Division of Engineering Office of Nuclear Regulatory Research FROM: Matthew J. Homiack, Materials Engineer /RA/
Component Integrity Branch Division of Engineering Office of Nuclear Regulatory Research
SUBJECT:
SUMMARY
OF THE AUGUST 5, 2020, RES CATEGORY 3 PUBLIC MEETING WITH EPRI AND STAKEHOLDERS TO DISCUSS ADVANCED METHODS FOR THE XLPR PROBABILISTIC FRACTURE MECHANICS CODE The U.S. Nuclear Regulatory Commission (NRC) staff from the Office of Nuclear Regulatory Research (RES) held a meeting on August 5, 2020, with representatives of the Electric Power Research Institute (EPRI) to provide information and guidance to new users on advanced methods for the Extremely Low Probability of Rupture (xLPR) probabilistic fracture mechanics code.
The agenda for the meeting is available in the NRCs Agencywide Documents Access and Management System under Accession Number ML20211L830. Enclosed is a list of the meeting participants and the slide presentation. Additionally, a video recording of the meeting will be posted on www.youtube.com. A summary of the meeting discussions follows by agenda topic.
- 1. Introduction and Opening Remarks The NRC staff welcomed participants and covered administrative items for the meeting.
It explained that the meeting was the final in a four-part technical seminar series on xLPR Version 2.1 (V2.1). Participants were reminded that summaries and video recordings of all the seminars were being made available to the public.
- 2. Advanced Methods for Inputs An EPRI contractor began the presentations by covering advanced methods in the area of setting up inputs for xLPR V2.1. This included proper definition of log-normal distributions, correlating inputs, and determining when inputs are outside the range of applicability. Resources for comparing input sets were also covered.
CONTACT: Matthew J. Homiack, RES/DE 301-415-2427
R. Iyengar 3. Advanced Methods for Sampling Advanced methods in the area of sampling strategies were covered next. There are four main ways to sample inputs in xLPR V2.1: (1) simple random sampling, (2) Latin hypercube sampling, (3) importance sampling, and (4) discrete probability distribution sampling. Some of these approaches can be combined. The advantages and disadvantages of each approach were explained, and it was shown how to implement them using the settings in xLPR V2.1. It was also shown how to set-up deterministic simulations and single-loop simulations, the latter of which would support a more classical Monte Carlo analysis.
- 4. Advanced Methods for Results and Outputs Then, advanced methods in the area of results and outputs were presented. A review was given on how to export results to external files. It was then explained how users can customize xLPR V2.1 by creating additional outputs using the features in GoldSim.
Because it may be difficult to view and extract results of interest from large sample sizes, it was shown how the results could be screened. To illustrate, a demonstration was given during which results were screened to display only the realizations that had cracks. Examples of post-processing were also discussed, including calculating outputs outside of xLPR V2.1 and performing sensitivity analyses.
- 5. Ways to Improve Efficiency The presentations were concluded by covering ways in which xLPR V2.1 simulations can be made more efficient. Efficiencies can be gained in terms of the amount of data saved and the speed. xLPR V2.1 outputs many results by default, so one way to reduce the amount of data retained from a simulation is to save only the results of interest. It was shown how to do this by disabling certain results or by saving only the final values instead of the time histories. It was also shown how the user can estimate the final file size of the results and adjust the output saving frequency. The default time step can also be adjusted. xLPR V2.1 runtimes can be dramatically reduced by running batches of realizations in parallel. This is possible using GoldSim Pro, where up to four slave processes can be run in parallel. With the GoldSim Distributed Processing Plus module, then there is no limit to the number of slave processes. A demonstration was given on how to set-up xLPR V2.1 for parallel processing.
- 6. Questions and Answers As an NRC Category 3 public meeting, stakeholders had the opportunity to participate by providing comments and asking questions throughout the meeting. This participation was facilitated through the electronic submission of questions and comments using the question and answer feature of the virtual platform used to host the meeting.
Clarifications were provided in response to a few attendee questions regarding the use of xLPR V2.1 to estimate probabilities in the range of one in a million. The use of different sampling strategies, as well as optimizing the time-step saving frequency and number of outputs saved, were presented as solutions to achieve such a goal. All questions were answered over the course of the meeting.
R. Iyengar 7. Closing Remarks The NRC staff concluded the meeting by providing a recap of the presentations. It also stated that RES and EPRI are planning to create an xLPR user group in the future. The user group will provide access to additional information and services pertaining to the code. Finally, participants were thanked for their time and encouraged to submit any future questions or comments to the NRC staff at xlpr@nrc.gov and to EPRI at xlpr@epri.com.
The NRC staff provided EPRI with an opportunity to review a draft of this meeting summary.
EPRI comments were incorporated as appropriate.
Enclosures:
As stated
R. Iyengar
SUBJECT:
SUMMARY
OF THE AUGUST 5, 2020, RES CATEGORY 3 PUBLIC MEETING WITH EPRI AND STAKEHOLDERS TO DISCUSS ADVANCED METHODS FOR THE XLPR PROBABILISTIC FRACTURE MECHANICS CODE DATED AUGUST 17, 2020 DISTRIBUTION:
DE/RES CIB/DE/RES CMB/DE/RES NPHP/DNRL/NRR NVIB/DNRL/NRR DRudland, NRR RKalikian, NRR RTregoning, RES JBowen, RES GFacco, RES ADAMS Accession Numbers: ML20241A063 *via email OFFICE RES/DE/CIB NAME MHomiack*
DATE 8/17/2020