ML17128A452
| ML17128A452 | |
| Person / Time | |
|---|---|
| Site: | Palo Verde  |
| Issue date: | 05/08/2017 |
| From: | Michael Montecalvo NRC/NRR/DRA |
| To: | Sunil Weerakkody NRC/NRR/DRA |
| Mike Montecalvo, NRR/DRA 301-415-1678 | |
| References | |
| Download: ML17128A452 (5) | |
Text
UNITED STATES NUCLEAR REGULATORY COMMISSION WASHINGTON, D.C. 20555-0001 MEMORANDUM TO:
Sunil D. Weerakkody, Deputy Director (Acting) /RA/
Division of Risk Assessment Office of Nuclear Reactor Regulation FROM:
Michael T. Montecalvo PRA Operations and Human Factors Branch Division of Risk Assessment Office of Nuclear Reactor Regulation
SUBJECT:
LESSONS LEARNED FROM MODELING THE FLEX STEAM GENERATOR FEED PUMP FOR PALO VERDE NUCLEAR GENERATING STATION The purpose of this memo is to capture considerations and lessons learned from modeling the FLEX steam generator (SG) makeup pump for the Palo Verde Nuclear Generating Station (PVNGS) using SAPHIRE and the associated plant specific SPAR model. This memo does not constitute an exhaustive list of the considerations when modeling FLEX equipment in a PRA in a manner consistent with RG 1.200. Rather, it captures insights from a limited scope project completed to fulfill the requirements for the Learning Project as a capstone for the Grow Your Own PRA analyst program. To that extent, during the implementation of the project, the SPAR model for Palo Verde was modified and quantified in a manner that fulfills fundamental learning objectives of the GYO Learning Project. The key lessons learned from this project, documented below, provides insights and challenges associated with modifying SPAR models to incorporate FLEX strategies.
The major steps accomplished while completing the project were:
Review of data provided by the licensee to establish system success criteria.
Development of system fault trees based on design documents and procedures (e.g.,
electrical schematics, one-line diagrams, flow diagrams, process and instrumentation diagrams, operating procedures).
Event tree review for PVNGS SPAR model with particular focus on appropriate timing for use of the FLEX SG makeup pump.
Review of fault tree basic event probabilities with discussions on how to deal with lack of available data and steps to perform subsequent updates (e.g. Bayesian updates) to the data when it is obtained.
Developed appropriate human error probabilities (HEP) by reviewing licensee documentation (e.g. training materials, procedures).
Analysis of common cause failure issues.
Analysis of recovery events for additional trains of FLEX equipment.
CONTACT: Michael T. Montecalvo, NRR\\DRA 301-415-1678 May 8, 2017
Once the model was built with the above considerations, multiple runs were made using the SAPHIRE software while refining the model. Multiple sensitivities were performed varying equipment reliability parameters and human error probabilities to gain insights into some of the potential considerations when modeling FLEX equipment. Some of the considerations revealed were:
Feasibility Analysis o One of the first steps when considering providing credit for FLEX is to perform a feasibility analysis to make an initial determination if the equipment could be used to support the identified scenarios of interest.
o The feasibility analysis identifies the potential scenarios, evaluates the adequacy of the use of the equipment to support the success criteria for those scenarios, determines if the operators would be directed to implement the equipment after diagnosis, and evaluates the capability of using the equipment given its capacity and the conditions of the systems in which it needs to interact.
o The feasibility analysis allows the analyst to evaluate necessary considerations at a high level to establish whether or not a more detailed assessment is warranted.
Equipment Reliability o The lack of available data to establish appropriate failure probabilities is a continuing issue with crediting FLEX equipment. Some relevant information can be obtained using manufacturer testing and reliability information or generic industry information and operating experience. This should be balanced with plant specific operating experience and/or the testing and maintenance program currently implemented by the licensee.
o NEI 12-06 requires FLEX equipment to be subject to maintenance and testing guidance provided in INPO AP-913, Equipment Reliability Process, and EPRI 3002000623, Nuclear Maintenance Applications Center: Preventive Maintenance Basis for FLEX Equipment - Project Overview Report. The use of this equipment in applications that are not covered by NEI 12-06 should be considered in the analysis. The maintenance programs established to comply with NEI 12-06 may be adequate, but this should be a consideration for evaluation by the analyst.
o The location and storage of equipment must be considered, including deployment capabilities. Support equipment (e.g. for hauling or debris removal) should be available after the event, if required. Pre-deployment or pre-staging may be credited to ensure equipment is at the proper location to meet the required times established by analysis. This will obviously be dependent on the type of analysis that is being completed. When evaluating a performance deficiency in the significance determination process, pre-deployment would not be possible, however for a notice of enforcement discretion it may be appropriate. This will be discussed further below.
Human Error Probabilities o Human reliability analysis is one of the more unique and important aspects of modeling the use of this equipment. The HRA methods frequently used (e.g.
SPAR-H, EPRI HRA Calculator) were developed using observed behavior in controlled scenarios, but this would not necessarily preclude their use for these situations. Other HRA methods exist that can directly handle these situations (e.g. ATHEANA, SLIM) but rely heavily on expert elicitation and engineering judgement to estimate HEPs. The use of these tools in time-sensitive applications may not be feasible.
o The existing performance shaping factors used in SPAR-H (time available, stress/stressors, complexity, experience/training, procedures, ergonomics/human-machine interface, fitness for duty, and work processes) seem fit to evaluate the use of FLEX equipment. Adjustment of the base human error probabilities and development of guidance on the selection of PSFs and their values for these situations should be developed to ensure consistency among analysts if it is desired to use SPAR-H for crediting FLEX.
o Sensitivity studies can be performed to evaluate the impact of specific HEPs on the PRA results. If particular human failure events are found to significantly contribute to the results, an HRA practitioner can be involved to provide a more comprehensive assessment. Many of these assessments will be generic in nature and the compilation of these assessments backed up with data from the industry can be used to provide more accurate results.
o The actions required for successfully implementing the FLEX equipment must include clear cues to enter previously developed procedures that are adequate for the scenarios of interest. Additionally, the analyst should consider staffing and available time to determine that a human action can be completed successfully.
o A detailed dependency analysis must be completed to assess the additional human actions modeled. This is especially important when FLEX HEPs are present in multiple cutsets.
o Portable equipment could be susceptible to unique human failure events not considered for installed equipment. An example of this is the reliance on diesel powered equipment on batteries for operation. The procedures associated with this equipment should be reviewed for the possibility of pre-initiator human failure events that leaves the equipment unavailable for use during an event.
The impact of these errors can be reduced significantly by using appropriate human performance error reduction tools or relying on additional operator responsibilities (e.g. operator rounds), but this should be evaluated by the analyst.
Passive components (e.g. equipment connections, manual valves) are not usually modeled in a PRA. This practice should be reconsidered if operating experience or data indicates issues with these components.
The use of change sets could be considered for different regulatory applications (e.g.
NOED vs. SDP). Analysis in the SDP is a backward looking assessment where the plant conditions were known and it is unlikely that FLEX equipment was prestaged or pretested beyond the normal maintenance requirements. Analysis for a NOED can most likely credit prestaging and pretesting equipment. This would provide increased confidence the equipment will work and would affect the equipment reliability component. Additionally, prestaging and pretesting could completely eliminate some HEPs or at least greatly improve confidence that the actions can be reliably completed.
The benefits of doing this should be weighed with acknowledgement of the unavailability of the equipment for its designated use for complying with NEI 12-06.
Additionally, the prestaging of the equipment may expose it to additional hazards (e.g.
staging in the turbine building would expose the equipment to a steam break initiating event). Some licensees have staging areas that are outside of plant buildings, exposure to external events should be considered as appropriate in these cases.
When building the FLEX equipment into the PRA model, it should be very clear to the analyst when credit is being enabled so as to ensure the appropriate analysis has been done for the specific situation being considered.
Common cause groupings for FLEX/portable equipment should be considered for equipment that is similar in construction or operation. Portable diesel equipment is generally started using batteries, so a failure of a specific class of batteries may expose many different pieces of equipment to a potential failure. If the same type of battery charger is used for multiple pieces of equipment, they may belong in the same common cause group even though they may be drastically different pieces of equipment (e.g. a FLEX diesel generator and a FLEX SG feed pump).
SUBJECT:
LESSONS LEARNED FROM MODELING THE FLEX STEAM GENERATOR FEED PUMP FOR PALO VERDE NUCLEAR GENERATING STATION DATED May 8, 2017 DISTRIBUTION:
SWeerakkody MMontecalvo RidsNrrDraAphb Resource ADAMS Accession No. ML17128A452 OFFICE NRR\\DRA NRR\\DRA NAME MMontecalvo SWeerakkody DATE 5/8/17 5/8/17