CNL-17-050, Submission of Technical Reports to Support a Public Meeting Regarding the Transition to the Reactor Oversight Process for the Mitigating Systems Performance Index

From kanterella
Jump to navigation Jump to search
Submission of Technical Reports to Support a Public Meeting Regarding the Transition to the Reactor Oversight Process for the Mitigating Systems Performance Index
ML17151A181
Person / Time
Site: Watts Bar Tennessee Valley Authority icon.png
Issue date: 05/30/2017
From: Simmons P
Tennessee Valley Authority
To:
Document Control Desk, Office of Nuclear Reactor Regulation
References
CNL-17-050 006041-RPT-001, Rev. 0
Download: ML17151A181 (25)
v • d • e


Text

{{#Wiki_filter:Tennessee Valley Authority, 1101 Market Street, Chattanooga, Tennessee 37402 cNL-17-0s0 May 30, 2017 10 cFR 50.4 ATTN: Document Control Desk U.S. Nuclear Regulatory Commission Washington, D.C. 20555-0001 Watts Bar Nuclear Plant, Unit 2 Facility Operating License No. NPF-96 NRC Docket No. 50-391

Subject:

SUBMISSION OF TECHNICAL REPORTS TO SUPPORT A PUBLIC MEETING REGARDING THE TRANSITION TO THE REACTOR OVERSIGHT PROCESS FOR THE WATTS BAR UNIT 2 MITIGATING SYSTEMS PERFORMANCE INDEX

Reference:

NRC letter to TVA, "Watts Bar Nuclear Plant, Unit 2 - Reactor Oversight Process lmplementation and Partial Cornerstone Transition - Docket No. 50-0391" dated October 22,2015 (M1152954253). ln accordance with the provisions of Title 10 of the Code of Federal Regulations (10 CFR) 50.4, "Written communications," Tennessee Valley Authority (TVA) is submitting two technical reports to support a public meeting regarding the transition to the Reactor Oversight Process (ROP) for the Watts Bar Nuclear (WBN) Unit 2 Mitigating Systems Performance lndex (MSPI). ln the referenced letter, the NRC stated that the MSPI for WBN Unit 2 will remain grayed out for a minimum of four quarters after the cornerstone has been transitioned to the ROP. ln addition, the referenced letter stated that the NRC and TVA will reach a decision on the exact number of quarters for each MSPI indicator via the Frequently Asked Questions (FAO) process. This letter and proposed public meeting provide the first steps in achieving alignment on the number of quarters for each MSPI indicator. These two reports provide TVA's technical position on the appropriate number of quarters that the MSPI should remain grayed out. These technical reports do not contain proprietary information. The first technical report, Enclosure 1, provides the results of an evaluation performed to assess the sensitivity of the WBN Unit 2 MSPI of using shorter monitoring periods than the normal 12-quarter period. This paper provides background information, the key assumptions used, the results, and the conclusions from this evaluation. This evaluation used data from the current WBN Unit 2 Probability Risk Assessment (PRA). The second

U.S. Nuclear Regulatory Commission cNL-17-050 Page 2 May 30, 2017 report, Enclosure 2, provides the technical basis for the 3-year performance monitoring period in the MSPI. Enclosed is one copy of each technical report and Enclosure 3, one compact disk, containing a portable document format (PDF) version of this letter and the two technical reports. As noted above, the purpose of this letter is to provide the NRC Staff with TVA's technical basis regarding the transition to the MSPI indicator that would support a proposed public meeting. This letter, in addition to providing the above information, formally requests a public meeting following the NRC Staffs review of the provided information. There are no regulatory commitments associated with this submittal. Please address any questions regarding this request to Gordon Arent at 423-365-20M. Respectfully, .n Paul Simmons Vice President, Watts Bar Nuclear Plant

Enclosures:

1. Mitigating System Performance lndex Sensitivity Evaluation dated March 31,2017
2. Technical Basis For The 3-Year Monitoring Period ln The Mitigating Systems Performance I ndex, Report #: 00604 1 -RPT-00 1
3. Compact Disk containing a portable document format version of this letter and the two technical reports cc (Enclosures):

NRC Regional Administrator - Region ll NRC Senior Resident lnspector - Watts Bar Nuclear Plant NRR Project Manager - Watts Bar Nuclear Plant NRR Director Division of lnspection and Regional Support (DIRS) NRR Chief of Performance Assessment Branch, DIRS

Enclosure I Mitigating System Performance lndex Sensitivity Evaluation dated March 31,2017 cNL-17-050 E1-1 of 11

Watts Bar Unit 2 Mitigating System Performance !ndex Sensitivity Evaluation Date: 3131117 Prepared By: Clinton Woolson Reviewed By: Brad Dolan Program Manager Approval: Stuart Rymer Engineering Director Approval: Mike Casner cNL-17-050 E1-2 of 11

Purpose:

The purpose of this paper is to document an evaluation performed to assess the sensitivity of the Watts Bar (WBN) Unit 2 Mitigating System Performance lndex (MSPI) using shorter monitoring periods than the normal 12-quarter period. This paper provides background information, the key assumptions used, the results, and the conclusions from this evaluation. The paper will be provided to NRC to support a public meeting with the NRC to discuss TVA's position that a 12-quarter monitoring period for MSPI using the existing NRC endorsed NEI 99-02 guidance (Reference 1) as appropriate. Scope: The evaluation was performed for WBN Unit 2 by using an Excel spreadsheet to calculate the estimated MSPI values for MSPI periods of four, eight, and 12 quarters. The evaluation is considered a forecast because normally, values are determined based upon actual results and calculated by softurare in the INPO Consolidated Data Entry (CDE) application. The performance data used are documented in this evaluation.

Background:

The WBN Unit 2 start-up testing program was completed late in September 2016. WBN Unit 2 reached commercialoperation on October 19, 20'16. The NRC letter to TVA (Reference 2) dated November 21, 2016, documented the full transition of WBN Unit 2 to the Reactor Oversight Process (ROP) on this date. The November 21,2016, letter referenced the NRC letter to TVA (Reference 3) dated October 22,2015, that outlined when the MSPI indicators would be in effect. After NRC clarification, the MSPI indicators will take effect for WBN Unit 2 after the fourth full monitored quarter. The NRC noted that the indicators would be officially "grayed out" until this time. The NRC informed TVA that the first WBN Unit 2 MSPI ROP submittalwas expected following the first quarter of data collection in 2017. This data was transmitted to the NRC in April 2017. The MSPI for WBN Unit 1 and other operating plants are based upon 12 quarters of performance data consistent with NEI 99-02. The use of 12 quarters of performance data included Browns Ferry Unit 1, which restarted in 2007 after an extended shutdown. ln a white paper presented at the February 21,2013, public meeting, the NRC and ROP task force documented that new plant MSPI results would be grayed out until four quarters of data are accumulated. This position was also referenced in FAQ 14-02 for the Fort Calhoun restart after its extended shutdown. Evaluation: An estimate of MSPI values and margins was calculated for time periods of four, eight, and 12 quarters. The WBN Unit 2 specific indicators for High Pressure Safety lnjection, Heat Removal, and Decay Heat Removalwere evaluated. The WBN Unit 2 Cooling Water MSPI calculation used a combination of \tr/BN Unit 2 specific equipment and common equipment that were monitored with WBN Unit 1. The Cooling Water indicator was evaluated below in the same manner as a specific system. The Emergency AC indicator was evaluated below with different assumptions than the unit specific indicators. cNL-17-050 E1-3 of 11

Assumotions:

1. Data used to estimate the effects of a shorter time span on the MSPI indicator are as follows.
a. Critical Hours - The Unit 2 critical hours were based on starting at the beginning of the first full quarter.
1) Critical Hours for evaluation The number of critical hours for Unit 2 was estimated based on a g5%

capacity factor (0.95 x 24 hours x 365 days). The number of critical hours were reduced for two assumed refueling outages of 35 days each, following the 4th quarter and 1Oth quarter of commercial operation. This reduction is based on an expected outage after one year (initialfuel load and startup) followed by a normal 18 month operating cycle.

b. Non Test and Test Run Hours - The \r/BN Basis Document (Reference 4) that includes Unit 2 assumes data collected are based on actual run time for Non Test Run Hours and estimates for Test Run Hours. The following values were used during this simulation:
1) Unit 2 Non Test Run Hour values were obtained by adjusting time periods of Unit 1 Non Test values. The Unit 1 Basis Document values were adjusted based on the time frame of analysis (e.9., four quarters represent a 12-month estimate).
2) Unit 2 Test Run Hour values were obtained by adjusting Unit 2 estimated values based upon the time period being evaluated (four quarters, eight quarters, and 12 quarters).
c. Non Test and Test Demands - The WBN Basis Document (Reference 4) includes Test Demands that are estimated values and Non Test Demands that are collected as actuals. The following values were used during this simulation:
1) Unit 2 Test Demand values were obtained by adjusting Unit 2 estimated values based upon the time period being evaluated (four quarters, eight quarters, and 12 quarters).
2) Non Test Demand values were obtained by adjusting time periods of Unit 1 Non Test values. The Unit 1 Basis Document values were adjusted based upon the time frame of analysis (four quarters represent a 12-month estimate).
d. Actual Engineering Safety Feature (ESF) Demand and Run Hours - These are entered monthly as actuals and are very low.
1) For WBN Unit 2, these are assumed to be zero.
e. Planned unavailability Baselines - NEI 99-02 states that these are based upon 2002 - 2004 values. Unit 1 uses these values or documents other periods and provides justification. Unit 2 does not have any documented performance data.

As a result:

1) The WBN Basis Document used \AIBN Unit 1 values adjusted after three years of WBN Unit 2 operation as necessary and in accordance with NEI 99-02 based upon FAQ 16-02.
2) WBN Unit 2 values for this estimate were WBN Unit 1 values. The MSpl calculation adjusts these values based upon critical hours.
f. Unplanned Baselines - NEI 99-02 provides these values as inputs for all reactors.
1) Unit 2 estimates used these values. The MSPI calculation adjusts these values based upon critical hours.

cNL-17-050 E1-4 of 1 1

g. UnavailabilityHours
1) Actual Planned hours - Because these hours are unknown for WBN Unit 2, the calculation used adjusted values from wBN unit 1 data for three years ending on Fourth Quarter 2014. The totalvalue at the end of December 2014 was divided by 12 and multiplied by the number of quarters for the period being evaluated.
2) Actual Unplanned hours - Because these hours are unknown for \IVBN Unit 2, the calculation only assumed unplanned hours associated with the simulation of failures. lf a failure was postulated, then a representative number of hours were added as unplanned hours to repair the component.

Generally, a significant percentage of the Technical specifications Limiting condition for operations (TS LCo) time was used to return the equipment to operation.

h. Unreliability Events (Failures) - Any failures of common indicators used WBN Unit 1 actual values. \r/BN Unit 2 failures were inserted as part of the simulation to determine the effect of postulated events and the effect on MSPI values. The failures for Cooling Water can be applied to common equipment along with \rr/BN Unit 2 equipment.
i. The data used from WBN Unit 1 were through the end of December 2014.

Known inputs:

1. Probabilistic Risk Analysis (PRA) unavailability and unreliability data, and Core Damage Frequency (CDF) for WBN Unit 2.
2. Common Cause Failure (CCF) WBN Unit 2 values were approved by FAQ 16-02.
3. Excel spreadsheet to calculate the estimated MSPI values was verified to provide similar results as calculated for the industry by the lNpo cDE database.

Process:

1. Updated the calculation to represent WBN Unit 1 so that the spreadsheet data can be validated against INPO CDE data. The calculation applied the risk cap similar to the INPO CDE calculation. The calculation also included the component performance limits that are used in the margin results.
2. Copied the Unit 1 spreadsheet and updated the PRA parameters for WBN Unit 2.
3. Adjusted the parameters to represent four, eight, and 12 quarters of data.
4. Described the results using appropriate tables.

Analvsis: General lnformation: The following analysis was performed using an Excel spreadsheet. The different input values for each indicator were inserted into the spreadsheet based on the above assumptions, which determined MSPI and Margin values. The margin tables shown below are based on an 'gg' condition. The tables indicate how many failures or unavailability hours are possible while keeping the indicator green. A combination ot failures and unavailability hours reduce the margin of multiple components and can only be evaluated on a case-by-case basis. Failures are only possible as integer values (whole cNL-17-050 E1-5 of 11

numbers), thus the margin report lists these values rounded down to the smaller whole number. As an example, a margin value of 3 in the table could have been an actual value of anywhere between 3.00 and 3.99. A false positive condition (not a statistical definition) in this report was identified when the WBN Unit 2 indicator would have turned white for a given number of failures or unavailability hours using a reduced monitoring period, but otherwise would be green if a 12 quarter period was considered. All other things being equal, margin was lost when lower values were used for reactor critical hours and component demands/run hours. This is evident by the analysis shown below. Soecific Svstem Analvsis: H e at Re moval-A ux i I i a rv F eedwate r: The Unit 2 PRA information and the projected three-year critical hours were substituted in the copied version of the Unit 1 spreadsheet. The next step in the process was to develop Unit 2 spreadsheets (Base Cases) representing data for quarters 4, 8, and 12. These spreadsheets included critical hours for each time period and adjusted demands and run hours for each monitored component. The method to obtain the adjusted values is shown on each of the spreadsheets. A table of MSPI value and Margin Value was developed for each case. Data from the spreadsheet for the Base Case are shown below. Heat Removal - Auxiliary Feedwater Base Case Quarters I 41 81 12 Measured Element Marqin to Threshold Motor Driven Pump fails to start (FTS) 4 fai ures 4 failures 5 fai ures Motor Driven Pump fails to run (FTR) 4 fai ures 4 failures 5 fai ures Turb ne Dr ven Pump (FTS) 5 fai ure 6 failures 7 fai ures Turb ne Dr ven Pump (FTR) 5 fai ures 6 failures 7 fai ures Air Operated Valve 4 fai ures 4 fatures 4 fai ures Motor Operated Valve 4 fai ures 4 failures 4 fai ures Unplanned Unavailabilitv (UU) 2A train 2332 hours 4509 hours 6734 hours UU 28 train 2332 hours 4509 hours 6734 hours UU Turbine Driven train 973 hours 1882 hours 2810 hours MSPI Results The Heat Removal indicator contains false positive potential because of the loss of margin (two failures) for the Turbine Driven Pump when the monitoring period is reduced to four quarters from 12 quarters. cNL-17-050 E1-6 of 11

Hiqh Pressurc Safetv lniection Svstems: The WBN Unit 2 PRA information and the projected three-year critical hours were substituted in the copied version of the Unit 1 spreadsheet. The next step was to develop Unit 2 spreadsheets (Base Cases) representing data for quarters 4, 8, and 12. These spreadsheets included critical hours for each time period and adjusted demands and run hours for each monitored component. The method to obtain the adjusted values is shown on each of the spreadsheets. A table of MSPI value and Margin Value was developed for each case. Data from the spreadsheet for the Base Case are shown below. High Pressure lnjection Systems Base Case Quarters 41 8 12 Measured Element Margin to Threshold Centrifugal Charging Pump FTS 4 fat ures 4 fai ures 5 fai ures Centrifugal Charging Pump FTR 4 fai ures 4 fat ures 5 fai ures Safety lnjection Pump FTS 4 fai ures 5 fai ures 5 fai ures Safety lnjection Pump FTR 4 fai u res 5 fai ure 5 fai ures Motor Operated Valve 4 fai ures 4 fai ures 4 fai ures UU Centrifugal Charging Pump 2A 437 hours 828 hours 1220 hours UU Centrifugal Charging Pump28 437 hours 828 hours 1220 hours UU Safety lnjection Pump 2A >25000 >25000 >25000 hours hours hours UU Safety lnjection Pump 28 >25000 >25000 >25000 hours hours hours MSPI Results The High Pressure Safety lnjection indicator contains false positive potential because of the loss of margin (one failure) for the Centrifugal Charging pumps when the monitoring period is reduced to four quarters from 12 quarters. Residual Heat Removal- RHR: The Unit 2 PRA information and the projected three-year critical hours were substituted in the copied version of the Unit 1 spreadsheet. The next step was to develop Unit 2 spreadsheets (Base Cases) representing data for quarters 4,8, and12. These spreadsheets included critical hours for each time period and adjusted demands and run hours for each monitored component. The method to obtain the adjusted values is shown on each of the spreadsheets. cNL-17-050 E1-7 of 11

A table of MSPI value and Margin Value was developed for each case. Data from the spreadsheet for the Base Case are shown below. Residual Heat Remova! Base Case Quarters 4 I 8 12 Measured Element Margin to Threshold Res dual Heat Removal Pump FTS 2 tailures 2fa lures 3 failures Res dual Heat Removal Pump FTR 3 failures 3fa lures 4 failures Motor Operated Valve 1 failures 2fa lures 2 failures UU Res dual Heat Removal Pump 2A 544 hours 1087 hours 1671 hours UU Res dual Heat Removal Pump 28 544 hours 1087 hours 1671 hours MSPI Results After reviewing the Base Case, it was determined that the four-quarter margin of 1 MOV failure is a low margin condition. Therefore, Case B was performed to evaluate the effects of 1 motor operated valve (MOV) failure and 60 hours of associated unplanned unavailability of the RHR. The results are shown below. Residual Heat Removal Case B Quarters 4 I 8 12 Measured Element Margin to Threshold Res dual Heat Removal Pump FTS 1fa Iure 1fa lures 1fa lure Res dual Heat Removal Pump FTR 1fa lure 2fa lures 2fa lures Motor Operated Valve 0fa lures 1fa lure 1fa lure UU Residual Heat Remova! Pump 2A 231 hours 547 hours 905 hours UU Residual Heat Removal Pump 28 231 hours 547 hours 905 hours MSPI Results Reviewing Case B and adding an additional MOV failure and 60 additional hours of unavailability was added for Case C. The results are shown below: Residual Heat Removal Case C Quarters I 4 8l 12 Measured Element Margin to Threshold Residual Heat Removal Pump FTS 0fa !ures 0fa Iures 0fa lure Residual Heat Removal Pump FTR 0fa lures 0fa lures 0fa Iures Motor Operated Valve 0fa lures 0fa lures 0fa lure UU Residual Heat Removal Pump 2A -153 hours -51 hours 140 hours UU Residual Heat Removal Pumo 28 -153 hours -51 hours 140 hours MSPI Results cNL-17-050 E1-8 of 11

At 8th quarter another MOV failure (2 total) and 60 (120 total) more associated unplanned hours results in a white indicator. The MSPI value at 12 quarters with two failures and 120 unplanned hours will result in a green indicator. Therefore, the RHR indicator contains significant false positive potential. Common Svstem Analvsis: Coolino Water: The WBN Unit 2 Cooling Water MSPI is a combination of Unit 2 specific trains and devices that were not monitored for Ul operation and common items that were monitored with Unit 1. The attached evaluation treated Cooling Water system as a specific system. The Unit 2 PRA information and the projected three year critical hours were substituted in the copied version of the Unit 1 spreadsheet. The next step was to develop Unit 2 spreadsheets (Base Cases) representing data for quarters 4, 8, and '12. These spreadsheets included critical hours for each time period and adjusted demands and run hours for each monitored component. The method to obtain the adjusted values is shown on each of the spreadsheets. A table of MSPI value and Margin Value was developed for each case. Data from the spreadsheet for the Base Case are shown below. Cooling Water System Base Case Quarters I 4 8I 12 Measured Element Margin to Threshold Component Cooling System Pump FTS 4fa lures 5fa lures 6 failures Component Cooling System Pump FTR 4fa lures 5fa lures 6 failures Emergency Raw Cooling Water Pump FTS 5fa lures 6fa lures 7 failures Emergency Raw Cooling Water Pump FTR 5fa Iures 6fa lures 7 failures UU Component Cooling System Pumo 2A 3393 hours 6490 hours 9619 hours UU Component Cooling System Pump 28 7026 hours 13440 hours 19920 hours UU Component Cooling System Pump CS > 25000 > 25000 > 25000 hours hours hours UU 28 Header 75 hours 143 hours 212 hours MSPI Results The Cooling Water indicator contains false positive potential as shown above as the number of pump failures increase from 5 to 6 between 4th and 12th quarters. Emeroencv AC The Emergency AC indicator is made up of 4 Diesel Generator sets that service Watts Bar Unit 1 and Unit 2. This equipment has been in service and monitored by Ul MSPI indicator. Therefore, the Unreliability portion of the indicator will be the same for both units. The Unavailability indicator is based upon Watts Bar U2 Critical hours since Commercial Operation. cNL-17-050 E1-9 of 11

The Unit 2 PRA information and the projected three year critical hours were substituted in the Unit 1 spreadsheet. The next step was to develop Unit 2 spreadsheets (Base Cases) representing data for quarters 4,8, and 12. The following assumption were used for the estimate of U2 Emergency AC indicator: o Used U1 results through December 2016 

    . Used U2 PRA values.

o Reduced Critical Hours based upon 4, 8, 12 quarters. Same as other indicators. lncluded full demand and run hours 12 Quarters - (Since D/G have been monitored for Reliability during this time.) o Unplanned hours to zero due to no failures after tJ2 critical. 

    . Reduced planned unavailability hours to 4 Q or 8Q.

o Removed two FTL for 8 quarter case since they will roll off 3 year window o Removed the one each FTS, FTL, and FTR failures for 12 quarter case since they will roll off the 3 year window. A table of MSPI value and Margin Value was developed for each case. Data from the spreadsheet for the Base Case are shown below. Emergency AC Base Case Quarters 4J 8 12 Measured Element Margin to Threshold D ese Generator fa ls to start (FTS) 7fa lures 9fa lures 12 fa lures Diese Generator fa ls to load (FTL) 7fa lures 9fa lures 12 fa lures Diese Generator fa ls to run (FTR) 7fa lure 9fa Iures 11 fa lures Diesel Generator 1A Unplanned 3235 hours 6258 hours 10371 hours Unavailability Diesel Generator 1A Unplanned 3235 hours 6258 hours 10371 hours Unavailability Diesel Generator 1A Unplanned 3235 hours 6258 hours 10371 hours Unavailability Diesel Generator 1A Unplanned 3235 hours 6258 hours 10371 hours Unavailability MSPI Results The Emergency AC indicator contains false positive potential in the area of failures as the number of failures increase from 4th to 12 quarter.

== Conclusion:==

The above evaluation shows that all of the WBN Unit 2 specific indicators have significant loss of margin or false positive potential if implemented using a shorter monitoring period than 12 quarters. The WBN Unit 2 specific Residual Heat Removal System continue to have high potential for false positives even after eight quarters. cNL-17-050 E1-10 of 1 1

The Cooling Water system that contains of common and Unit 2 specific items can be treated as a specific system and has reduction in margin. The Emergency AC indicator contains false positive potential in the area of unreliability.

References:

1. NEI 99-02, Regulatory Assessment Performance !ndicator Guideline
2. NRC letter to TVA, "Watts Bar Nuclear Plant, Unit 2 - Full transition to the Reactor Oversight Process and Assessment letter-Docket No. 50-039'1' dated November 21, 2016 (M116326A210).
3. NRC letter to TVA, "Watts Bar Nuclear Plant, Unit 2- Reactor Oversight Process lmplementation and Partial Cornerstone Transition- Docket No. 50-0391" dated October 22, 201 5 (ML15295A253).
4. Watts Bar MSPI Basis Document Revision 1'1 plus recent update PRA values.

cNL-17-050 E1-11 of 11

Enclosure 2 Technical Basis For The S-Year Monitoring Period ln The Mitigating Systems Performance lndex, Report #: 006041-RPT-001 cNL-17-050 E2-1 of 12

JENSEN HUGHES s 158 W, Gay $tneet, Suite 40C rtuest C.hester, F'A 19-Et; L$A 1e r"i se nl"r ug hes cot"I'r C +1 61CI-4ii1-826* TECHNICAL BnSIS FoR THE 3-YEAR MONIToRING PTRIoD rN THE MllcATtNG SYSTEMS PTRFORMANCE IruDEX Prepared For The Tennessee Valley Authority (TVA) PO Box 2000 Spring City, TN 37381 Revision: 0 Project #: 1 DAD06041 .000.000 Project Name: Review of WBN U2 MSPI Report #: 006041 -RPT-001 Name and Date Preparer: Donald A. Dube PhD 2110117 Reviewer: Pat W. Baranowsky 2110117 Review Design Review X Method Alternate Calculation tr Approver: Jeff R. Gabor 210117 E2-2 of 12

006041-RPT-001 Revision Record Summary REVISION RECORD

SUMMARY

Revision Revision Summary 0 lnitial lssue Revision 0 Page ii E2-3 of 12

006041-RPT-001 Table cf Contents TABLE OF GONTENTS 1.0 Review of Historica! Documents............ ................i 1.1 NUREG-1753.............. .................1 1 .2 Overview of the Development of the MSPI ......... .... .. .. .. .. .. 1 1.3 NUREG-1816.............. .................1 

                     '1.3.1 Formulation of the Risk Cap                    ...............1 1.3.2 Formulation of the Performance Limit.......        ............9 2.0    lssues with Other than 3-Year Monitoring Period          ...................s 3.0    Conclusions..............                                         ...........6 4.0    References                                                          .........7 LIST OF FIGURES Figure 1 Variable Backstop............                                ........3 Figure 2 1Q/2015 Performance lndicators - Fort   Calhoun                ....... s Revision 0                                                                                         Page iii E2-4 of 12

00604 r1 -Fr F I- -uti1 E xecr.lt ive S u ni rria ri/ EXECUTIVE

SUMMARY

This paper provides the technical basis for the 3-year performance monitoring period in the Mitigating Systems Performance lndex (MSPI). Historicaldocuments including NUREG-1753 and NUREG-1816 were reviewed and are discussed below. NUREG-1753 clearly established the 3-year monitoring period for unreliability based on statistical analyses. Those analyses assessed a range of 1 to 5 year periods and recommended the 3-year monitoring period based on the balancing of potential false positives from too short of a monitoring period, and non-responsiveness of the indicator from too long of a period. Furthermore, the appropriateness of the 3-year monitoring period was demonstrated in the 20-reactor pilot program, and subsequent analyses as described in NUREG-1816. Additionally, this paper discusses how both the risk cap and the performance limit in the current MSPI formulation were established based on a 3-year monitoring period. lmplementation of a shorter monitoring period such as 12 or 24 months using only unit-specific performance data is inconsistent with the original technical basis for the MSPI and could be problematic. Revision 0 Page iv E2-5 of 12 

         *"t*

1.0 REVIEW OF HISTORICAL DOCUMENTS 

         '1.1  NUREG-I753 ln NUREG-1753, "Risk-Based Performance lndicators: Results of Phase 1 Development" [1],

the technical feasibility of providing improved performance indicators for the Reactor Oversight Process (ROP) was examined by the U.S. Nuclear Regulatory Commission (NRC) staff and its contractors. lndicators related to the initiating events cornerstone, the mitigating systems cornerstone, and the containment portion of the barrier integrity cornerstone were considered. Unreliability indicators were developed at the componenVtrain/system level. As follow-up to NUREG-1753, the NRC and industry identified unreliability and unavailability indicators for six systems to be piloted starting in 2002. The indicators would come to be known as the MSPI. One to five year monitoring periods were considered in NUREG-1753. On page 3-2, it is noted that "the associated monitoring period must be long enough to reduce the probabilities of false negatives and false positives to acceptable levels, but no longer." Based on statistical analyses as documented in Appendix F, the authors recommended 1 year for the unavailability and 3 years for the unreliability indices as shown in Table F-8. However, this recommendation did not have the benefit of piloting. The unavailability index would have no prior distribution, while the unreliability index was to have the constrained noninformative prior (CNIP). The CNIP is a prior distribution using industry-averaged performance in a Bayesian process. lts technical basis is described in Appendix J of NUREG-1816. For practical data collection and reporting purposes, and since the unavailability index was found to be less sensitive to the monitoring period than unreliability, the MSPI pilot was implemented in 2002 with a 3-year rolling monitoring period for both unreliability and unavailability. 1.2 Overview of the Development of the MSPI The MSPI evolved following the conceptual development of risk-based performance indicators in NUREG-1753 and after several public workshops and meetings. Reference 2 provided an overview of the MSPI formulation, its characteristics, benefits and limitations, and key issues to be addressed. Table 2 with regard to the treatment of operating experience states: "Three year observation period is used to determine estimate of current performance; chosen duration balances the need for good statistics against need to detect performance changes within a reasonable time." ln Section 7 .2 on False Assignment Probabilities, the paper cautions that "As long as the MSPIs make use of the statistics of infrequent events occurring within short time windows, substantial false indication probability will remain for high-B (i.e., Birnbaum) elements." 1.3 NUREG-1816 ln NUREG-1816 [3], the NRC staff extensively tested and improved the MSPI methodology during the 12-month pilot plant application phase, and evaluated technical issues related to the new indicator's sensitivity to probabilistic risk assessment (PRA) modeling detail. Section 2.2 describes the Unreliability lndex (URl) formulation in great detail. Twelve quarters of data and the CNIP are key elements of the URl. ln addition, the extensive verification of the MSPI is described in Appendix A of NUREG-1816, while the numerical simulation is discussed in Appendix L. A rolling 3-year data collection period for the MSPI is central to the validity of the verification effort and the numerical simulation. 1.3.1 Formulation of the Risk Cap NUREG-1816 found some system indicators associated with the MSPI had significant "false positive" issues. That is, for statistical reasons, there was in the early formulation of the MSPI a Revision 0 Page 1 E2-6 of 12

significant probability that a plant system at baseline performance would cross over the GreenMhite threshold. As discussed in detail in Appendix D of that report, random failures that occur at a rate consistent with the industry performance are not indicative of a performance issue. That is, one failure over a 3-year performance monitoring period, or one failure above the normal expectation, can be argued not to constitute a significant trend. To address the concern with "false positives," the concept of a frontstop (or what has come to be known as the risk cap) was introduced. Any such risk cap should: 

             . Reduce false positives
             . Be compatible with, but not ignore, the Unavailability index contribution
             . Maintain sensitivity (without adversely impacting false negatives)

Mathematically, the risk cap of 5x10-7 was placed on the single most risk-significant failure within a system in the 3-year monitoring period. This addressed the concern with false positives without adversely impacting false negatives by virtue of the following properties: 

             . Two significant failures (each with a risk contribution greater than 5x10-7) would very likely result in a White indication
             . One significant failure with other less-significant failures could exceed the GreenMhite threshold
             . One significant failure with a significant UAI contribution could exceed the GreenA//hite threshold The risk cap of 5x10-7 was chosen with great deliberation. Given baseline system performance (MSPI - 0) just prior to the failure, the risk cap of 5x10-7 from a single failure combined with incremental unavailability (i.e., UAI) due to corrective maintenance should normally result in an MSPI under the GreenMhite threshold of 1x10{. ln effect, the risk cap allowed up to an incremental 5x10-7 for UAl. As discussed in NUREG-1816 Appendix d, the incremental UAI of 5x10-7 was itself consistent with Regulatory Guide 1.177 atthat time [4J. Moreover, data from the 20-reactor pilot program were used to provide reasonable confidence that most, although not all, outage times from corrective maintenance in combination with the component failure would place the MSPljust below the GreenMhite threshold.

The 3-year monitoring pgriod is a key element of the risk cap formulation As discussed above, given^a risk cap of 5x10-7 allows up to an incremental 5x10-7 for the UAI contribution before 1x10-' is reached. The incremental UAI comes about from the product of the Birnbaum for unavailability of the train or segment and incremental unavailability from the corrective (unplanned) maintenance. The unplanned unavailability is the unplanned maintenance hours divided by the critical hours of the 3-year monitoring period. lncremental UAI = B;* [incremental unplanned maintenance/critical hrs in 12 qtrs.] What if a monitoring period other than 3 years was used, for example, l year? The incremental train or segment unplanned unavailability would be about 3 times higher because of 3 times fewer critical hours in the 1-year period compared to a 3-year period, and the incremental UAI would also be about 3 times higher. ln effect, substantial margin to the GreenMhite threshold would be lost, negating the full benefit of the risk cap as currently formulated, and increasing the probability of false positives. ln effect, a monitoring period other than 3-years is inconsrsfenf with the basic formulation of the risk cap. Revision 0 Page 2 E2-7 of 12

00S04 i -RFT-tr0'1 Review of t-{tstoricat Documents 1.3.2 Formulation of the Performance Limit lf the risk cap was formulated to address false positives in the MSPI, the backsfop (or what has come to be called the component performance limit) was instituted to address, to some degree, false negatives. The concern was that during the MSPI pilot, large numbers of component failures were observed to be needed for some components in some systems in certain pilot plants before the MSPI turned White at 1x106. Conceptually, the performance limit is a limit on the total number of failures, of all failure modes and of all components of one type in one system of a single nuclear power plant unit. lf the number of failures seen in the 3-year performance period exceeds the performance limit, the system MSPI is denoted as White. The criterion is based on statistical significance of the observed number of failures, relative to prior expectations. As described in Appendix E of NUREG-1816, the performance limit was chosen to be the smallest number indicative of degraded performance such that: 

             . The probability of false positive < 0.01
             . The fraction of positives that are false s 5%.

Thus, the performance limit is defined to ensure that false positives are very rare, and if a positive occurs, it is most likely a true positive. The performance limit given in Section F.4 of NEI 99-02 [5] is derived from a linear regression as shown in Figure E.3 of NUREG-1816 and is reproduced below as Figure '1-1. Gorrelation of Backstop with Expected Number of Failures 20 18 y=4.65x+4.2 16 R2 = 0.9294 14 EL Uo 12 ga 10 (, m (E I 6 4 2 0 0.0 0.5 1.0 1.5 2,0 2.5 3.0 Expected number of failures (in 3 years) Figure 1 Variable Backstop The correlation in Figure 1-1 shows the backstop, or number of allowable component failures of a given component type within a system, as a function of the expected number of failures. The expected number of failures, in turn, depends on the number of demands (and run hours if Revision 0 Page 3 E2-8 of 12

006041 -RFT-[i0r Rev'lew crf Hlstorical Docurnents applicable) as well as industry-averaged component failure rates. The component performance limit clearly relied on 3 years'worth of performance data from the pilot plants. What if a monitoring period other than 3 years was used, for example, 1 year? The correlation per Figure 1-1 would remain valid since the expected number of failures would clearly lie within the domain of the graph given about one-third of the demand data (demands and run-hours). However, for some components within a system, the licensee could be disadvantaged. With only about one-third of the demands and run-hours, the expected number of failures (based on industry-averaged failure rates) is about one-third the value for a 3-year monitoring period. Thus, the backstop or performance limit would be somewhat less for a 1-year monitoring period, all other things being equal. By way of a simplified example, presume that the expected number of failures for a component type in a given system is 1.5 for a 3-year monitoring period based on actual demand and test data. This might be typicalfor a case with four emergency dieselgenerators (EDGs), for example. The component performance limit would be exceeded at 12 failures according to the correlation. lf the monitoring period were just 1 year, there would be about 3 times fewer demands, with the expected number of failures equaling about 0.5, all other things being equal. The component performance limit would then be exceeded at 7 failures. ln practice, this difference may or may not be significant depending on the risk worths of the EDGs, unplanned maintenance hours, and remaining margin to White. This author is not aware of any instance where exceedance of the performance limit caused a system indicator to turn White. Nevertheless, the linear regression assumed 3 years of performance data in deriving the statistical correlation, not 1 or 2 years. Revision 0 Page 4 E2-9 ot 12

006041-RpT-001 issues with Other Thank 3-Year Monitoring Period 2.0 ISSUES WITH OTHER THAN 3.YEAR MONITORING PERIOD A monitoring period other than 3 years would be inconsistent with the basic formulation of the MSPI. While it would be theoretically possible to use a shorter monitoring period, the MSPI values would be volatile and subject to large fluctuations as component failures and uncharacteristic high train/segment unavailabilities roll in and out of the monitored period. For a newly commissioned plant, volatility could be expected for the first few years of monitoring. Whether the MSPI gave "reasonable" results based on indicated values for each system and margins to White would be based mostly on fortune rather than the deliberately crafted and statistically tested MSPI formulation. One need only look at the one indicator for Fort Calhoun when it returned to service after an extended outage to realize that a volatile indicator is not in any of the stakeholders' best interest (see Figure 2-1). The volatility in the Emergency AC Power indicator arose because the denominators in the UAI and URI formulations were not sufficiently large to buffer small deviations in diesel generator maintenance, and a failure. Mitlgatlng Systems Performance lnder, Emergency AC Power System 

            <1.0E               l.m-6 1,0E-5 1.0E-4.

l.0E-3. ThesholE llfftc > l.ltr-6 Yelorf, > l"mE-6 Red > tJ0E-4 Figure 2 1Q12015 Performance lndicatorc - Fort Calhoun Revision 0 Page 5 E2-10 of 12

006041 -ffipT-riil 1 C,:nc{us ions

3.0 CONCLUSION

S The key features of the MSPI are that it: 

             . ls Risk-informed and performance-based.
             . Allows trade-offs between Unreliability and Unavailability to optimize system performance.

Has no penalty for on-line preventive maintenance hours up to the pre-planned baseline. 

             ' Reflects plant-specific design and operation.
             . Has features to address false positives and false negatives.
             ' Uses 3 years of performance data to provide stability against large fluctuations in the indicators.

Historical documents such as NUREG-1753 and NUREG-I816 founded the MSpl based on a 3-year monitoring period. The 3 year observation period was found to balance the need for good statistics against the need to detect performance changes within a reasonable time. The verification of the MSPI and numerical simulation that led to confidence in the robustness of the MSPI were both based on a 3-year monitoring period. Both the risk cap and the performance limit are formulated on a 3-year rolling indicator. Any option that phases in the MSPI at a newly operating plant before 3 years of plant-specific performance data have been obtained compromises the original basis of the MSPI. Whether these options give "reasonable" results based on indicated values for each system and margins to White would be based mostly on fortune rather than the deliberately crafted and statisticaily tested MSPI formulation. The Emergency AC Power MSPI for Fort Calhoun after an extended outage is not a good precedent. Revision 0 Page 6 E2-11 of 12

Cf;BC41 -R PT-((i1 Referenc"es

4.0 REFERENCES

1. NUREG-1753, "Risk-Based Performance lndicators: Results of Phase 1 Development," U.S.

Nuclear Regulatory Commission, April 2002.

2. U.S. Nuclear Regulatory Commission, lnteroffice Memorandum from Scott F. Newberry (RES/DRAA) to John A. Zwolinksi (NRR), "Request for Review of Mitigating Systems Performance lndices White Paper," May 12,2003. (Adams Accession No. ML031350208 and M1031360121).
3. NUREG-18't6, "lndependent Verification of the Mitigating Systems Performance lndex (MSPI) Results for the Pilot Plants, Final Report," U.S. Nuclear Regulatory Commission, February 2005.
4. Regulatory Guide 1.177, "An Approach for Plant-Specific, Risk lnformed Decisionmaking:

Technical Specifications," U.S. Nuclear Regulatory Commission, August 1998.

5. NEI 99-02, "Regulatory Assessment Performance lndicator Guideline," Revision 7, Nuclear Energy lnstitute, August 31,2013.

Revision 0 Page 7 E2-12 of 12}}

Retrieved from "https://kanterella.com/w/index.php?title=CNL-17-050,_Submission_of_Technical_Reports_to_Support_a_Public_Meeting_Regarding_the_Transition_to_the_Reactor_Oversight_Process_for_the_Mitigating_Systems_Performance_Index&oldid=2624499"