ML21130A373
| ML21130A373 | |
| Person / Time | |
|---|---|
| Issue date: | 05/10/2021 |
| From: | Christopher Hunter NRC/RES/DRA/PRB |
| To: | |
| Hunter, Christopher 301 415 1394 | |
| References | |
| Download: ML21130A373 (13) | |
Text
Industry Feedback on Public Comments on Draft CCF NUREG (2012)
NEI The NUREG does not acknowledge that the CCF probabilities are estimated based on causes that did not affect the component that failed. For example, if the cause is a deficient maintenance program, common cause due to environment and design do not apply.
EXELON The draft NUREG attempts to provide a supporting position to justify that the potential shared failure is at a causal level (e.g., a deficiency in a maintenance process) as opposed to a failure mechanism. While there is merit in this argument, the draft NUREG does not acknowledge that the other causes included in the CCF probabilities did not affect the component that failed. For example, if the cause is a deficient maintenance program, common cause due to environment and design do not apply.
However, the CCF probabilities include all these causes.
1998 NUREG/CR-5485 The potential of evaluating common-cause failures (CCFs) on causal level is mentioned in Appendix E. However, little to no detail is provided.
When the cause for the CCF event is known, the CCF quantification can be tailored for that specific cause.
2004 CAFM Development As part of an NRC grant, University of Maryland MD (A.
Mosleh and A.
OConnor) developed the CAFM as documented in An Extension of the Alpha Factor Model for Cause-Based Treatment of Common-Cause Failure Events in PRA and Event Assessment.
2014 INL Evaluation INL complete review of alternative CCF methods for event assessment:
After the review, the CAFM appears to be the best choice as the alternate CCF model to replace the AFM currently used in the SPAR models. The CFM contains more detail than the AFM, which will make event assessments more straightforward. The causal alpha factors relate to causes that are expected to align well with those most often seen in event assessment.
2017 CAFM Priors INL developed priors for all five existing CCF cause groups:
Design Human Environment Internal Other
NRR requested that RES complete CAFM work and evaluate the suitability of the method for use in SDP risk evaluations.
Complete CAF calculations for all SSCs and failure modes included in the SPAR models.
Identify if any data gaps exist.
Develop guidance for using the CAFM in event and condition assessment (ECA).
RES developed a report documenting its analyses and conclusions (ADAMS Accession No. ML21055A027).
The CAFM best characterizes the potential CCF risk impact in SDP risk assessments by focusing on a specific failure cause as result of the licensee performance deficiency (PD).
The existing alpha factors consider all failure causes and, therefore, likely result in greater uncertainties when used in ECA.
There does not appear to be any significant data gaps for the various causes, components, failure modes, or common cause component groups (CCCGs).
CAFM Impacts - Using the CAFs can increase or decrease the CCF impact depending on the evaluation.
Sparse Data - Existing CCF data concerns may be exasperated using the CAFM.
Failure Cause Selection - The current CCF cause grouping scheme is not optimal for all the likely causes of licensee PDs.
The CAFs are not always smaller than the existing alpha factors.
For example, if a specific SSC and failure mode has a total of 4 CCF events, and 2 of these are due to design issues, the design CAF is not half of the existing alpha factor.
NUREG/CR-5485 states that the CCF impact will likely decrease when only a single cause is evaluated.
However, both the numerator (# of CCF events) and denominator (total number of failures) due to specific cause in the alpha factor ratio will change.
In addition, the Bayesian process (i.e., the prior) also affects the calculations.
=
=1
m = the # of redundant components in the CCCG k = the # of failed components nk = the # of failure events, which resulted in k components failing within a CCCG of size m
Preliminary calculations indicate that the CAFs for the environment, design, and human failure causes are typically greater than the existing alpha factors.
This seems to indicate that these failure causes result in stronger CCF potential.
However, revised calculations using only the last 10 years of data still need to be performed.
The same data concerns exist that are present in the existing alpha factors.
Slicing the data will likely result in in more CAFs for particular SSCs/failure modes not having CCFs.
- This is not believed to be significant concern; the CAF will decrease from the prior based on the number of total failures for the SSC/failure mode for that cause.
What about SSCs/failure modes with little total failure data for specific causes?
- This will result in CAFs nearly equal to the failure cause prior.
- While likely more frequent, this issue exists for the current alpha factors.
Cause Group Failure Cause PD Category Design Construction installation error Design and Engineering Design error or inadequacy Manufacturing error Human Accidental human action Corrective Action Program Maintenance Management Oversight Operations Procedures Inadequate maintenance Human action procedure Inadequate procedure Environment Ambient environmental stress Extreme environmental stress Internal environment Component Internal to component; piece-part Setpoint drift Age or wear Other/Unknown State of other component Other
Cause Group Failure Cause PD Category Design Construction installation error Design and Engineering Design error or inadequacy Human Action Accidental human action Management Oversight Operations Human action procedure Maintenance Inadequate maintenance CAP and Maintenance Procedures Inadequate procedure Procedures Environment Ambient environmental stress Environment Extreme environmental stress Internal environment Other Internal to component; piece-part Setpoint drift Age or wear State of other component Other
If issues regarding failure cause selection are identified as CAFM is implemented, staff will consider developing further guidance on appropriate failure cause selection.
Performed sensitivity calculations on priors and existing CCF parameters to determine if they represent current industry performance.
Work is largely rendered moot due to shift to 10-year data window for most of parameters, including CCF.
- However, new update will use 1998-2015 priors instead of the 1991-2005 prior used during the 2015 data update.
- Smaller data window will likely result in a greater number of CCF parameter estimates closer to prior.
- We will explore different priors for consideration of future data updates (e.g., 10-year prior date range, component-specific priors).