ML17340B100
| ML17340B100 | |
| Person / Time | |
|---|---|
| Site: | Nuclear Energy Institute |
| Issue date: | 12/13/2017 |
| From: | Lindeman A Electric Power Research Institute, Nuclear Energy Institute |
| To: | Document Control Desk, Office of Nuclear Reactor Regulation |
| Shared Package | |
| ML17340B096 | List: |
| References | |
| Download: ML17340B100 (8) | |
Text
Fire PRA Research Plans for 2017-2019 Ashley Lindeman Senior Technical Leader Fire PRA FAQ Meeting December 13, 2017
© 2017 Electric Power Research Institute, Inc. All rights reserved.
Introduction Fire PRA methodology still in need of realism / refinement
- Recent research (electrical cabinet HRR / obstructed plume) an advancement, but not the holy grail
- Additional refinements needed to support risk-informed applications (50.69, 4b, 5b, etc.)
Obvious deltas still apparent from analysis of current FPRA results
- Clear overestimation of number of significant events (CCDPs > 1E-4) calculated from FPRAs as compared to OPEX
- For electrical cabinets: FPRAs calculate the rate of severe fires ~25 per decade (~70%), while OPEX shows that 75% of fires are limited to the ignition source only In the current dynamic, there is limited testing possible and limited OE available, and the bridging inputs of reasonable judgement and feedback are not working 2
© 2017 Electric Power Research Institute, Inc. All rights reserved.
Near-Term Fire PRA Research Plan Structured approach followed to set plan priorities Plan attributes:
- Need to accomplish meaningful improvement to fire PRA realism and provide results within 2 years (i.e. technical work by 2019) to support risk-informed applications (50.69, 4b, 5b, etc.)
- Data or technical bases can be obtained in the near term from established sources without the need for large scale experimental tests which may or may not produce conclusive results
- Priority was given to tasks which can be completed in accordance with the above principals and that have the highest probability of providing statistically significant improvement in realism
- Stakeholder input welcomed at all points 3
© 2017 Electric Power Research Institute, Inc. All rights reserved.
Key Contributors to Fire PRA Results Skyline 80%
Relative Contribution to Total Fire CDF 70%
60%
Elec.Cabinets 50%
Transients HEAF 40%
Other MCB 30%
Transformers Pumps 20%
Battery Chargers Junction Boxes 10%
Self Ign Cables Diesel Generators 0%
4
© 2017 Electric Power Research Institute, Inc. All rights reserved.
Comparison with Operating Experience Reviewed data in SECY 14-0107 Accident Sequence Precursor Data (2004-2013)
- 7 events with CCDPs > 1E-4/ year 2 involving fire
- Robinson (2010), estimated CCDP = 4E-4
- Fort Calhoun (2011), estimated CCDP => 1E-4 Extrapolating the FPRA results for the US industry over the same time period would have estimated:
- ~15 events with CCDPs > 1E-4
- ~5 events with CCDPs > 1E-3 5
© 2017 Electric Power Research Institute, Inc. All rights reserved.
Summary of FPRA Research Tasks Task 1a/1b: Technical basis and fire progression event tree Task 2: Plant personnel suppression Task 3: Fire resulting in plant trip Task 4: High energy arc fault (frequency and consequence)
Task 5: Transient fire scenarios Task 6: Main control board 6
© 2017 Electric Power Research Institute, Inc. All rights reserved.
Key Takeaways Skyline exercise in early 2017 provided a reset used to re-calibrate the path forward for fire PRA research The results from Fire PRAs are not reflective of the operating experience
- Screening level analysis and methodology assumptions may be inappropriately biasing our understanding of fire risk contributors EPRI research plan focused on most impactful research in near term (now through 2019)
- Top research tasks to improve the realism in the fire PRA models
- Not intended to be exhaustive list of low hanging fruit or listing of every task or data set to be improved 7
© 2017 Electric Power Research Institute, Inc. All rights reserved.
TogetherShaping the Future of Electricity 8
© 2017 Electric Power Research Institute, Inc. All rights reserved.