ML14191A672

Brunswick Steam Electric Plant, Unit Nos. 1 & 2 - Response to Request for Additional Information Regarding Voluntary Risk Initiative National Fire Protection Association Standard 805 (NRC TAC Nos. ME9623 and ME9624)

ML14191A672
Person / Time
Site:	Brunswick
Issue date:	06/26/2014
From:	Hamrick G T Duke Energy Carolinas
To:	Document Control Desk, Office of Nuclear Reactor Regulation
References
BSEP 14-0076, TAC ME9623, TAC ME9624
Download: ML14191A672 (10)

Text

DUKE George T. HamrickVice PresidentENERGY, Brunswick Nuclear PlantP.O. Box 10429Southport, NC 28461o: 910.457.3698June 26, 2014Serial: BSEP 14-0076U.S. Nuclear Regulatory CommissionATTN: Document Control DeskWashington, DC 20555-0001Subject: Brunswick Steam Electric Plant, Unit Nos. 1 and 2Renewed Facility Operating License Nos. DPR-71 and DPR-62Docket Nos. 50-325 and 50-324Response to Request for Additional Information Regarding Voluntary RiskInitiative National Fire Protection Association Standard 805 (NRC TACNos. ME9623 and ME9624)References: 1. Letter from Michael J. Annacone (Carolina Power & Light Company) to U.S.Nuclear Regulatory Commission (Serial: BSEP 12-0106), LicenseAmendment Request to Adopt NFPA 805 Performance-Based Standard forFire Protection for Light Water Reactor Electric Generating Plants (2001Edition), dated September 25, 2012, ADAMS Accession NumberML12285A4282. Letter from Michael J. Annacone (Carolina Power & Light Company) to U.S.Nuclear Regulatory Commission (Serial: BSEP 12-0140), AdditionalInformation Supporting License Amendment Request to Adopt NFPA 805Performance-Based Standard for Fire Protection for Light Water ReactorElectric Generating Plants (2001 Edition), dated December 17, 2012, ADAMSAccession Number ML12362A2843. Letter from Farideh Saba (USNRC) to George T. Hamrick (Duke EnergyProgress, Inc.), Second Request for Additional Information RegardingVoluntary Risk Initiative National Fire Protection Association Standard 805(TAC Nos. ME9623 and ME9624), dated February 12, 2014, ADAMSAccession Number ML14028A1784. Letter from George T. Hamrick (Duke Energy Progress, Inc.) to U.S. NuclearRegulatory Commission (Serial: BSEP 14-0029), Response to AdditionalInformation Regarding Voluntary Risk Initiative National Fire protectionAssociation Standard 805 (NRC TAC Nos. ME9623 and ME9624), datedMarch 14, 2014, ADAMS Accession Number ML14079A2335. Letter from George T. Hamrick (Duke Energy Progress, Inc.) to U.S. NuclearRegulatory Commission (Serial: BSEP 14-0035), Response to AdditionalInformation Regarding Voluntary Risk Initiative National Fire protectionAssociation Standard 805 (NRC TAC Nos. ME9623 and ME9624), datedApril 10, 2014, ADAMS Accession Number ML14118A105 U.S. Nuclear Regulatory CommissionPage 2 of 36. Electronic Mail from Andrew Hon (U.S. Nuclear Regulatory Commission) toWilliam R. Murray (Duke Energy Progress, Inc.), Brunswick Steam ElectricPlant, Units I and 2 -Request for Additional Information Regarding VoluntaryRisk Initiative National Fire Protection Association Standard 805 (TACNos. ME9623 and ME9624), dated June4, 2014, ADAMS Accession NumberML14155A209.Ladies and Gentlemen:By letter dated September 25, 2012 (i.e., Reference 1), as supplemented by letter datedDecember 17, 2012 (i.e., Reference 2), Duke Energy Progress, Inc., submitted a licenseamendment request (LAR) to adopt a new, risk-informed, performance-based (RI-PB) fireprotection licensing basis for the Brunswick Steam Electric Plant (BSEP), Unit Nos. 1 and 2. OnFebruary 12, 2014 (i.e., Reference 3), the NRC provided a request for additional information(RAI) regarding the fire probabilistic risk assessment. By letters dated March 14, 2014 (i.e.,Reference 4); and April 10, 2014 (i.e., Reference 5); Duke Energy responded to the RAI.Subsequently, on June 4, 2014 (i.e., Reference 6), the NRC provided an electronic, follow-upRAI. The response to the follow-up RAI is enclosed.This document contains no new regulatory commitments.Please refer any questions regarding this submittal to Mr. Lee Grzeck, Manager -RegulatoryAffairs, at (910) 457-2487.I declare, under penalty of perjury, that the foregoing is true and correct. Executed on June 26,2014.Sincerely,George T. HamrickEnclosure:Response to Request for Additional Information Regarding Voluntary Risk InitiativeNational Fire Protection Association Standard 805 U.S. Nuclear Regulatory CommissionPage 3 of 3cc (with enclosure):U. S. Nuclear Regulatory Commission, Region IIATTN: Mr. Victor M. McCree, Regional Administrator245 Peachtree Center Ave, NE, Suite 1200Atlanta, GA 30303-1257U. S. Nuclear Regulatory CommissionATTN: Mr. Andrew Hon (Mail Stop OWFN 8G9A) (Electronic Copy Only)11555 Rockville PikeRockville, MD 20852-2738U. S. Nuclear Regulatory CommissionATTN: Ms. Michelle P. Catts, NRC Senior Resident Inspector8470 River RoadSouthport, NC 28461-8869Chair -North Carolina Utilities CommissionP.O. Box 29510Raleigh, NC 27626-0510Mr. W. Lee Cox, Ill, Section Chief (Electronic Copy Only)Radiation Protection SectionNorth Carolina Department of Health and Human Services1645 Mail Service CenterRaleigh, NC 27699-1645lee.cox@dhhs.nc.gov

Enclosure

1Page 1 of 7Response to Request for Additional Information Regarding Voluntary Risk InitiativeNational Fire Protection Association Standard 805By letter dated September 25, 2012, as supplemented by letter dated December 17, 2012, DukeEnergy Progress, Inc., submitted a license amendment request (LAR) to adopt a new, risk-informed, performance-based (RI-PB) fire protection licensing basis for the Brunswick SteamElectric Plant (BSEP), Unit Nos. 1 and 2. On February 12, 2014, the NRC provided a request foradditional information (RAI) regarding the fire probabilistic risk assessment. By letters datedMarch 14, 2014 and April 10, 2014; Duke Energy responded to the request for additionalinformation (RAI). Subsequently, on June 4, 2014, the NRC provided an electronic, follow-upRAI. The response to the follow-up RAI is provided below.Probabilistic Risk Assessment (PRA) Request for Additional Information (RAI) 1.d.02In a letter dated March 14, 2014, ADAMS Accession No. ML14079A233, the licenseeresponded to PRA RAI 1.d.01. The response to this RAI explains that, in addition to an in-depthreview of a sampling of plant records for performance of the "Transient Fire Load Evaluation"procedure, the licensee assessed the last three years of transient combustible violationsprovided in the Fire Protection Program System Health Reports. A number of violations appearto be dismissed based on the following rationale: "Based on their fire procedure, this wouldlikely not be a violation because Attachment 3 exempts this type of material from transientcombustible controls for both 'No Storage' locations and 'Non-Intervening Combustible Zones'."These violations appear to represent circumstances in which quantities of combustible sourcesexisted in the plant that could have contributed to a fire.a. Explain and justify how these violations were considered in the determination of reducedHeat Release Rates (HRRs) for transient fires. Include in this discussion explanation ofwhether these violations could have resulted in a transient fire exceeding the reducedHRR rates credited in the Fire PRA.ResponseIn evaluating the relevance of historical violations of the superseded administrative controlprogram, OFPP-01 3, Transient Fire Load Evaluation, with respect to the Fire PRA, the previousresponse discussed the hypothetical treatment under the current administrative control program,FIR-NGGC-0009, NFPA 805 Transient Combustibles and Ignition Source Controls Program.This was done to illustrate the difference between the reactive approach of the supersededprocedure and the pro-active approach in the current procedure. Because FIR-NGGC-0009includes both provisions for supporting the reduced HRRs credited in the Fire PRA and a list ofspecific combustible materials that have been pre-evaluated to be acceptable within thoseconstraints, none of the violations so identified could have resulted in a transient fire exceedingthe reduced HRRs credited in the Fire PRA.PRA RAI 01.f.ii.02In the same letter, the licensee responded to PRA RAI 1 .f.ii.01. The response to this RAIexplains that main control room (MCR) abandonment is only credited for loss of habitability (i.e.,not loss of control/function), and that MCR fire scenarios were separately evaluated for loss ofcontrol/function that leads to core damage without crediting alternate shutdown. This approachis asserted to be conservative. Given that MCR abandonment appears to be evaluated as asingle scenario using a single conditional core damage probability/conditional large early

Enclosure

1Page 2 of 7release probability (CCDP/CLERP) (though an event tree was used to calculate the singleCCDP/CLERP values), it is still not clear how the abandonment scenario addresses thepossibility of different fire induced impacts like spurious failures that can accompany a fire thatleads to abandonment.a. Explain how the single abandonment scenario addresses the various possible fire-induced failures. Specifically include discussion of how the following scenarios areaddressed:i. Scenarios where fire fails only a few functions aside from forcing MCRabandonment and successful alternate shutdown is straightforward.ii. Scenarios where fire could cause some recoverable functional failures orspurious operations that complicate the shutdown but successful alternateshutdown is likely.iii. Scenarios where the fire-induced failures cause great difficulty for shutdown byfailing multiple functions and/or complex spurious operations that makesuccessful shutdown unlikely.b. If fire-induced failures of MCR functions are not considered in abandonment scenarios,provide justification for their exclusion.i. Describe whether credited abandonment actions from the abandonmentprocedure is correct for loss of function or spurious actions that may occur as aresult of a fire leading to abandonment.ii. If abandonment actions do not account for these effects then describe how fire-induced failures are considered in modeling of abandonment scenarios andinclude those failures as part of the integrated analysis performed in response toPRA RAI 23.ResponseFor main control room fires, the Fire PRA broadly characterized the contributions to risk asattributable to either a loss of control or a loss of habitability. A loss of control leads directly tocore damage which is quantified by the Fire PRA. Core damage is assumed for a loss ofhabitability with some probability for operator recovery through the successful implementation ofthe alternate safe shutdown (ASSD) procedures. Although the contributions to risk wereseparately characterized, the fire events postulated to cause those contributions were not, inthat there was no assignment of a split fraction for the ignition frequency. To assess the risk dueto a loss of control, the applicable bin-specific ignition frequency was apportioned to eachignition source in the main control room. Likewise, to assess the risk due to a loss of habitability,the applicable bin-specific ignition frequency was again apportioned to each ignition source inthe main control room. These risk contributions are then summed to obtain the total risk. This"double-counting" of the ignition frequency simplifies the analysis and yields conservativeresults overall.The operator recovery of implementing the ASSD procedures as applied to the loss ofhabitability scenario is premised on two requirements. The first requirement concerns theoperator's ability to recognize when habitability makes it necessary to abandon the main controlroom. The second requirement is that the equipment needed to implement the ASSD capabilityis independent of and separated from that in the main control room. Consequently, fire-induced

Enclosure

1Page 3 of 7failures in the main control room do not affect the equipment needed to implement the ASSDprocedures.a. The operator recovery of implementing the ASSD procedures is only applied to the lossof habitability scenario. It does not address fire-induced failures because the equipmentrequired for implementation of the ASSD strategy is not affected by fires in the maincontrol room. For those scenarios where fire-induced failures result in a loss of controlleading to core damage, the analysis does not mitigate the associated risk by creditingmain control room abandonment to implement the ASSD procedures. Effectively, theassumption is that the main control room abandonment strategy always fails for loss ofcontrol scenarios regardless of whether the fire fails only a few functions, or multiplefunctions, or some recoverable functions, or causes spurious operations.b. The operator recovery of implementing the ASSD procedures is only applied to the lossof habitability scenario. Fire-induced failures of main control room functions are notconsidered in the abandonment scenario because the equipment required forimplementation of the ASSD strategy is not affected by fire-induced equipment failures inthe main control room. The credited abandonment actions include the transfer of controlfor the required equipment to a location outside the main control room, and theequipment controlled from there is sufficient to achieve and maintain safe and stableconditions. The abandonment risk includes random failures and unavailability of therequired equipment and related human errors including the failure to transfer control to alocation outside the main control room.PRA RAI O1.f.iii.02In the same letter, the licensee responded to PRA RAI 1 .f.iii.01. The disposition to this RAIstates that the large early release frequency (LERF) contribution from MCR abandonment dueto habitability was estimated to be 10% of the core damage frequency (CDF) for MCRabandonment, based on the Internal Events PRA where LERF is 8% of CDF. Containmentbypass scenarios, such as interfacing system loss-of-coolant-accident (ISLOCAs), are oftenmajor contributors to LERF.a. Justify that the relative likelihood/frequency of containment bypass scenarios for the FirePRA, as compared to that for core damage scenarios, is not higher than for the InternalEvents PRA.ResponseThe relative likelihood/frequency of containment bypass scenarios, as compared to that for coredamage scenarios, is justified as not being higher than that for the Internal Events PRA becausethe estimation of the LERF contribution as 10% of CDF is limited to main control roomabandonment due to loss of habitability.For main control room fires, the Fire PRA broadly characterized the contributions to risk asattributable to either a loss of control or a loss of habitability, with a loss of control leading tocore damage and a loss of habitability providing some probability for operator recovery throughthe implementation of the alternate safe shutdown procedures. Although the contributions to riskwere separately characterized, the fire events postulated to cause those contributions were not,in that there was no assignment of a split fraction for the ignition frequency. To assess the riskdue to a loss of control, the applicable bin-specific ignition frequency was apportioned to each

Enclosure

1Page 4 of 7ignition source in the main control room. Likewise, to assess the risk due to a loss of habitability,the applicable bin-specific ignition frequency was again apportioned to each ignition source inthe main control room. These risk contributions are then summed to obtain the total risk. This"double-counting" of the ignition frequency simplifies the analysis and yields conservativeresults overall. Effectively, there is no risk mitigation credit for abandoning the main control roomto implement the alternate safe shutdown procedures; there is only the additional risk of notimplementing the procedures successfully.The LERF contribution for containment bypass scenarios, such as Interfacing System Loss ofCoolant Accidents (ISLOCAs), would be attributed to fire-induced equipment failures resulting ina loss of control leading to core damage. As such, that contribution to LERF is quantifiedseparate from and in addition to the estimate of LERF for failing to implement the alternate safeshutdown procedures successfully following main control room abandonment due to habitability.When necessary, the credited abandonment actions transfer control of the required equipmentto a location outside the main control room, and the equipment controlled from there is sufficientto achieve and maintain safe and stable conditions. The quantification of the risk ofabandonment includes random failures and unavailability of the required equipment and relatedhuman errors including the failure to transfer control to a location outside the main control roomand the failure to close the main steam isolation valves (MSIVs) and remove power from theReactor Protection System panels. These later two steps assure Group 1 containment isolationoccurs and prevent/mitigate spurious opening due to fire damage. The failure to perform thesetwo actions is conservatively counted toward CDF, and therefore LERF, without regard towhether the fire damage set requires the actions.Since containment bypass scenarios are part of the Internal Events PRA but not part of themain control room abandonment, the 10% estimation of the LERF contribution from main controlroom abandonment compared to the 8% for the Internal Events PRA is a very conservativetreatment.PRA RAI 06.02The disposition to PRA RAI 06.01 presents results of a sensitivity in which Main Control Board(MCB) scenarios are multiplied by the whole MCB ignition frequency rather than a fraction of thefrequency. Given that this sensitivity study appears to only impact sequence frequencies, it isnot clear why there is asymmetry between CDF and LERF sensitivity results. Similarly, theresults of the integrated analysis provided in response to PRA 23 shows that, whereas changein (A) CDF, A LERF, and CDF increased as a result of the integrated analysis, the LERFdecreased in a number of cases. It is not clear why LERF would trend in an opposite directionfrom CDF.a. Please explain and justify these seeming anomalies.ResponseThe baseline method for calculation of risk associated with the Main Control Board used a firemodeling approach and credited in-cabinet incipient detection. The sensitivity analysis used thestatistical modeling approach in Appendix L of NUREG/CR-6850 and did not credit the incipientdetection. The "seeming anomaly" is that the calculated CDF for the sensitivity is higher thanthe baseline CDF while the calculated LERF is lower. The reason is the two methodologiesquantify fires which have different target sets. The baseline methodology credited incipient

Enclosure

1Page 5 of 7detection for early suppression but postulated scenarios where suppression and detectionfailed, permitting the fire to grow large enough to damage not only the entire panel but alsoexternal cables in trays/conduits above the panel. In these cases, the target sets sometimescontained sufficient components to cause both core damage and a large early release such thatthe calculated LERF value equaled or nearly equaled the calculated CDF value. In contrast, theAppendix L methodology typically quantified target sets for smaller fires which were containedwithin a panel, or multiple panels where propagation was appropriate, but with no damage tocables outside of the Main Control Board. The calculated LERF value for these fires was usuallyless than and often significantly less than the calculated CDF value. So while the Appendix Lmethodology had a higher CDF value, the postulated scenarios were less likely to cause a largeearly release and hence a lower LERF value.A similar "seeming anomaly" exists in response to PRA RAI 23 where the calculated change inUnit 1 LERF for the control power transformer (CPT) and state of knowledge correlations(SOKC) sensitivity was negative while the change in Unit 1 CDF was positive. In this case, thevalues represent small changes, less than 1% from the baseline, and the small negative valueresulted from the UNCERT's use of Monte Carlo simulation combined with removal of the CPTcredit having created no change in Unit 1 LERF. The change was close enough to zero that aMonte Carlo simulation caused a slight negative value.PRA RAI 22.c.01In the same letter dated March 14, 2014, the licensee responded to PRA RAI 22.c. Regardingthe final five bullets of part (c) of this RAI, on improvements made to facilitate fire modeling, twoof the explanations are not clear.a. Explain more fully what the constant of 6 minutes was used for and the modeling withwhich it was replaced.b. Also, describe more fully how the hot gas layer (HGL) modeling basis changed, includinghow the "total energy released" is modeled in the updated HGL analysis.Responsea. The constant of 6 minutes was included in the calculation of the time-to-damage for theclosest external target in the zone of influence (ZOI) fire scenario. Although the technicalbasis was poorly documented, the effect of including the 6 minutes in the time-to-damage calculation was to decrease the probability of non-suppression for the scenario.As a replacement for the constant 6 minutes, each ZOI scenario was split into two time-dependent scenarios.* The first scenario included targets that would be damaged within 5 minutes of thetime-to-damage of the closest target. Those targets were modeled as failed withthe closest external target at the time-to-damage of the closest target." The second scenario included the remainder of the targets in the ZOI. Thosetargets were modeled as failed 5 minutes after the time-to-damage of the closesttarget.

Enclosure

1Page 6 of 7A probability of non-suppression was then calculated for each scenario based on theapplicable time of failure.b. The basis used to determine whether an HGL develops for a particular scenario waschanged because the comparison between the conditions required for an HGL in aparticular room and the conditions produced by a particular fire scenario in that roomwas using dissimilar parameters. In particular, the conditions required for an HGL in aparticular room were expressed as an average heat release rate over a 30 minute timeperiod, while the conditions produced by a particular fire scenario in that room wereexpressed as an instantaneous heat release rate. This comparison of dissimilarparameters was performed at 1 minute time intervals over a 60 minute time period and,consequently, sometimes produced non-conservative results. To more reliablydetermine whether a particular scenario resulted in the formation of an HGL, bothparameters were changed to be expressed as the total energy released. Currently, thetotal energy released by a scenario is accumulated at 1 minute time intervals andcompared over a 60 minute time period to the total energy required to be released intothe room to form an HGL.PRA RAI 23.01With respect to the response to RAI 01 .d.02, if the transient combustible violations cited cannotbe explained or justified to support use of reduced transient heat release rates (HRRs), then useanother value for the HRR toa. Justify its use and incorporate that value in the integrated analysis provided in responseto PRA RAI 23. In addition, based on the response to RAI 24.01 below, revise allestimates of the risk and delta-risk metrics to exclude the credit for the "panel methods"approach.ResponseNo modification of the response to PRA RAI 23 was proposed because the response toRAI 01 .d.02 explained that the transient combustible violations have no impact on the reducedtransient heat release rates.In addition, the estimates of the risk and delta-risk metrics were not revised because theresponse to RAI 24.01 clarified that the frequency multiplier used for closed 480V motor controlcenters (MCCs) was not a "panel methods" approach.PRA RAI 24.01In the same letter dated March 14, 2014, the licensee responded to PRA RAI 24. Thedisposition to this RAI presents a new implementation item (i.e., #13) that commits to replacingunacceptable methods with acceptable methods prior to self-approval in cases where impactwas shown to be minimal in the current submittal, with one exception. That exception is statedto be replacing the "panel methods" approach (assumption that 10% of the electrical panelsmeet the definition of an open panel).a. Either provide justification for this, including a phenomenological basis beyond thehistorical fire events database, or

Enclosure

1Page 7 of 7b. Confirm how you will modify the new implementation item #13 to include removal ofcredit for use of the "panel factors" method.ResponseThe 10% multiplier is neither a "panel methods" approach nor a "panel factors" method.Brunswick MCCs were evaluated as either a vented ignition source or an un-vented ignitionsource (i.e., closed) depending upon the individual MCC construction. About 2/3 of the 480Vand 250VDC MCCs were determined to be vented ignition sources. The remaining 1/3 of theMCCs were determined to be unvented ignition sources and consistent with the writtenguidance in NUREG/CR-6850 section 11.5.1.7.3. The 10% multiplier is not a "panel factors"method, which was described in the NRC's June 21, 2012, (i.e., "Gitter") letter (i.e., ADAMSAccession Number ML12171A583), as:... using a single "factor," blends together all the case-specific phenomena that contributeto growth of a fire within an electrical cabinet and then assigns an "average" or"aggregate" propagation probability on a generic basis for igniting the first cable tray.In marked contrast, unvented MCC scenarios postulated with the 10% multiplier were assessedon a case-by-case basis, using accepted fire modeling tools and practices, as applicable.Ignition sources were characterized using a two-point HRR model, and target sets were definedby the spatial relationships to the ignition source of interest. Fire growth and decay weremodeled for individual ignition sources in terms of fire intensity and duration. Target damagewas based on the exposure environment exceeding the applicable damage threshold with thenon-suppression probability based, in part, on the time to reach the damage threshold.Propagation to secondary combustibles reflected the spatial relationships to the ignition sourcewith consideration for the possible presence of credited barriers.As used in the BSEP Fire PRA, the 10% multiplier for otherwise non-propagating ignitionsources constitutes a reasonably realistic yet conservative treatment of modeling uncertainties,including arcing faults in 480V MCCs for which no specific guidance is provided inNUREG/CR-6850. The use of this 10% multiplier results in a risk increase relative to what isrequired by the guidance in NUREG/CR-6850.This 10% multiplier has been reviewed with the NRC on multiple occasions and does notrepresent a method that has been deemed unacceptable to the NRC. This approach waspresented on slide 17 (i.e., ADAMS Accession Number ML12152A145) at the May 31, 2012,BSEP Pre-LAR Application Meeting with the NRC. Section 3.4.7 of the Safety Evaluationaccompanying the June 28, 2010, NFPA 805 License Amendment for the Harris Plant (i.e.,ADAMS Accession Number ML1 01750602) described this as a reasonable basis for consideringthe Harris Plant MCCs as closed cabinets and concluded that the risk evaluations werereasonable and conservative.Duke Energy is an active participant in industry efforts to improve Fire PRA methods and isaware of both a short-term and a long-term approach currently being developed for this issue.The short-term approach is expected to be bounded by the conservative risk assessmentdescribed in the response to this RAI. The long-term approach is expected to be more realisticand would be addressed through the normal PRA maintenance process.Based on the above discussion, no modification of implementation item #13 was proposed.