PNP 2013-083, Response to Request for Additional Information - License Amendment Request to Adopt NFPA 805 Performance-Based Standard for Fire Protection for Light Water Reactors

Response to Request for Additional Information - License Amendment Request to Adopt NFPA 805 Performance-Based Standard for Fire Protection for Light Water Reactors

ML13336A649
Person / Time
Site:	Palisades
Issue date:	12/02/2013
From:	Vitale A Entergy Nuclear Operations
To:	Document Control Desk, Office of Nuclear Reactor Regulation
References
PNP 2013-083
Download: ML13336A649 (228)

Text

Entergy Nuclear Operations, Inc.

Palisades Nuclear Plant 27780 Blue Star Memorial Highway Covert, MI 49043-9530 Tel 269 764 2000 Anthony J Vitale Site Vice President PNP 2013-083 December 2, 2013 U. S. Nuclear Regulatory Commission ATTN: Document Control Desk Washington, DC 20555-0001

SUBJECT:

Response to Request for Additional Information - License Amendment Request to Adopt NFPA 805 Performance-Based Standard for Fire Protection for Light Water Reactors Palisades Nuclear Plant Docket 50-255 License No. DPR-20

References:

1. ENO letter, PNP 2012-106, License Amendment Request to Adopt NFPA 805 Performance-Based Standard for Fire Protection for Light Water Reactors, dated December 12, 2012 (ADAMS Accession Number ML12348A455)

2. ENO letter, PNP 2013-013, Response to Clarification Request License Amendment Request to Adopt NFPA 805 Performance-Based Standard for Fire Protection for Light Water Reactors, dated February 21, 2013 (ADAMS Accession Number ML13079A090)

3. NRC electronic mail of August 8, 2013, Palisades - Requests for Additional Information Regarding Transition to the Fire Protection Program to NFPA Standard 805 (TAC No. MF0382) (ADAMS Accession Number ML13220B131)

4. ENO letter, PNP 2013-075, Response to Request for Additional Information - License Amendment Request to Adopt NFPA 805 Performance-Based Standard for Fire Protection for Light Water Reactors, dated September 30, 2013

5. ENO letter, PNP 2013-079, Response to Request for Additional Information - License Amendment Request to Adopt NFPA 805 Performance-Based Standard for Fire Protection for Light Water Reactors, dated October 24, 2013

PNP 2013-083 Page 2 of 3

Dear Sir or Madam:

In Reference 1, Entergy Nuclear Operations, Inc. (ENO) submitted a license amendment request to adopt the NFPA 805 performance-based standard for fire protection for light water reactors. In Reference 2, ENO responded to a clarification request. In Reference 3, ENO received electronic Request for Additional Information (RAIs). In Reference 4, ENO submitted the 60-Day RAI responses. In Reference 5, ENO submitted the 90-Day RAI responses. provides the ENO responses to the 120-day RAIs, as follows:

Requests for Additional Information Response Time Response Date FM RAI 01, 02, 03, 04, 06 120 Days December 1, 2013 PRA RAI 01, 02, 03, 05, 07, 08, 09, 12, 13, 16, 17, 18, 19, 20, 23, 24, 25, 26, 27, 28 The PNP Fire PRA Model is currently being revised to incorporate RAI refinements. The aggregate numerical effect of the model refinements will be reflected in the base quantification of the RAI Response Fire PRA model. The RAI Response Fire PRA Model will represent the collection of changes identified in individual RAI responses that could impact the CDF, LERF and delta calculations that support the LAR submittal. The RAI Response Fire PRA Model will be quantified to provide updated Attachment W values as required for the LAR submittal once all issues that impact quantification have been identified and resolved via the RAI process. The final results are expected to provide reasonable assurance that the total risk remains within Region II of RG 1.174, Rev 2. A revised Attachment S will be provided when the numerical effect from the base case results of the RAI Response Fire PRA Model is provided.

At the end of Attachment 1, ENO is providing a revision to PRA RAI 06, which was provided in Reference 5, due to a reference revision.

A copy of this response has been provided to the designated representative of the State of Michigan.

This letter contains no new commitments and no revisions to existing commitments.

I declare under penalty of perjury that the foregoing is true and correct. Executed on December 2, 2013.

PNP 2013-083 Page 3 of 3 Sincerely, ajv/jpm Attachments:

1. Response to Request for Additional Information Regarding License Amendment Request to Adopt NFPA 805 Performance-Based Standard for Fire Protection for Light Water Reactors cc: Administrator, Region III, USNRC Project Manager, Palisades, USNRC Resident Inspector, Palisades, USNRC State of Michigan

ATTACHMENT 1 RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION REGARDING LICENSE AMENDMENT REQUEST TO ADOPT NFPA 805 PERFORMANCE-BASED STANDARD FOR FIRE PROTECTION FOR LIGHT WATER REACTORS Electronic RAIs were received from the Nuclear Regulatory Commission (NRC) on August 8, 2013. The Entergy Nuclear Operations, Inc. (ENO) responses to the 120-Day RAIs are provided below.

The PNP Fire PRA Model is currently being revised to incorporate RAI refinements. The aggregate numerical effect of the model refinements will be reflected in the base quantification of the RAI Response Fire PRA model. The RAI Response Fire PRA Model will represent the collection of changes identified in individual RAI responses that could impact the CDF, LERF and delta calculations that support the LAR submittal. The RAI Response Fire PRA Model will be quantified to provide updated Attachment W values as required for the LAR submittal once all issues that impact quantification have been identified and resolved via the RAI process. The final results are expected to provide reasonable assurance that the total risk remains within Region II of RG 1.174, Rev 2.

NRC Request Fire Modeling (FM) RAI 01 NFPA 805 Section 2.4.3.3, states The PSA [probabilistic safety assessment] approach, methods, and data shall be acceptable to the AHJ [authority having jurisdiction] ..." The NRC staff noted that fire modeling comprised the following:

- The Consolidated Fire Growth and Smoke Transport (CFAST) model was used to calculate control room abandonment times.

- The Generic Fire Modeling Treatments (GFMT) approach was used to determine the zone of influence (ZOI) in all fire areas throughout plant, with the exception of Switchgear Rooms 1-C and 1-D.

- CFAST, Fire Dynamics Simulator (FDS) and the Thermally Induced Electrical Failure (THIEF) model were used to calculate the damage time to raceways credited in the FPRA in Fire Area 3 (Switchgear Room 1-D) and 4 (Switchgear Room 1-C).

LAR Section 4.5.1.2, "Fire PRA" states that fire modeling was performed as part of the FPRA development (NFPA 805 Section 4.2.4.2). Reference is made to LAR Attachment J, "Fire Modeling V&V [verification and validation]," for a discussion of the acceptability of the fire models that were used.

Page 1 of 227

Specifically regarding the acceptability of CFAST for the CR abandonment time study:

a) Provide the basis for the assumption that the fire brigade is expected to arrive within 15 minutes. In addition, describe the uncertainty associated with this assumption, discuss possible adverse effects of not meeting this assumption on the results of the FPRA and explain how possible adverse effects will be mitigated.

b) LAR Table H-1, NEI 04-02 FAQs Utilized in LAR Submittal, credits FAQ 08-0052, Transient Fires Growth Rates and Control Room Non-Suppression, (ADAMS Accession No. ML092120501, closure memo). Provide justification for using transient fire growth rates that deviate from those specified in FAQ-08-0052, and discuss the effect of these deviations on the risk results (i.e., CDF, LERF, CDF and LERF).

c) During the audit, it was discussed how fires originating close to a wall or corner in the MCR were addressed. Explain the methodology for considering a fire (transient and fixed ignition source) located against a wall or in a corner and explain the effect on the CR abandonment times.

d) During the audit, it was discussed how some parameters (e.g. fire growth rate and fire base height) have a significant effect on MCR abandonment times. Explain how the results of the sensitivity analysis in Appendix B of the Evaluation of Control Room Abandonment Times calculation were used in the FPRA and provide a discussion of the impact on fire risk results (i.e., CDF, LERF, CDF and LERF).

Specifically regarding the acceptability of the GFMTs approach:

e) Explain how the modification to the critical heat flux for a target that is immersed in a thermal plume was used in the ZOI determination.

f) Provide technical justification to demonstrate that the GFMTs approach as used to determine the ZOI of fires that involve multiple burning items (e.g., an ignition source and an intervening combustible such as a cable tray) is conservative and bounding.

g) Describe how the flame spread and fire propagation in cable trays and the corresponding heat release rate (HRR) of cables was determined, and provide justification for deviations from NUREG/CR-6850, Fire PRA Methodology for Nuclear Power Facilities. Explain how the flame spread, fire propagation and HRR estimates affect the ZOI determination and hot gas layer (HGL) temperature calculations.

h) Describe how transient combustibles in an actual plant setting are characterized in terms of the three fuel package groupings in Supplement 3, Transient Ignition Source Strength of the GFMT. Identify areas, if any, where the NUREG/CR-6850 transient combustible HRR characterization (probability distribution and test data) may not encompass typical plant configurations. Finally, explain how any Page 2 of 227

administrative action will be used to control the type of transient combustibles in a fire area.

Detailed fire modeling was performed in Switchgear Room 1-C (Fire Area 4) and Switchgear Room 1-D (Fire Area 3). The following questions are regarding the acceptability of the use of CFAST, FDS and THIEF in Fire Areas 3 and 4:

i) Part of the approach described in the analysis is to treat the individual switchgear panel enclosures as a series of compartments, each capable of supporting a one-zone or well-stirred fire environment. CFAST is used to calculate the maximum steady fire size that can be supported within an enclosure given the vent flows under a one-zone assumption. Section 3.3 of the Technical Reference Guide for CFAST (National Institutes of Standards and Technology (NIST) Special Publication 1026, September 2010) describes the single zone approximation and states that this approximation is appropriate for smoke flow far from a fire source where the two-zone layer stratification is less pronounced than in compartments near the fire (e.g., elevator shafts, stairwells, etc.) Section 3.3 of the Technical Reference Guide for CFAST describes limitations of the zone model assumptions and provides a quantitative limit for when to consider the single-zone approximation. This limit is the ratio of a compartment height (H) to its length (L) which is 10 or more. Based on the figures in the detailed fire modeling reports, the upper cubicle of the individual panels in the switchgear cabinets, which were modeled in CFAST as worst-case have an H/L ratio of approximately 1.

Provide justification for the acceptability of using CFAST for this purpose and explain why the approach described in Supplement 1 to the GFMTs was not considered for the purpose of calculating the fire size specified in the FDS model.

j) During the audit, the licensee discussed how cable tray propagation and flame spread are considered in the analysis of the 1-C and 1-D switchgear rooms; however, it was not clear how the first cable tray is ignited and whether it is a function of time, temperature/heat flux, or both.

Clarify the mechanism for ignition of the first cable tray above an ignition source (upper cubicle of switchgear panel or other fixed cabinet source). In addition, provide justification for using a vertical flame spread rate of 0.0258 m/s as opposed to 0.258 m/s, which is the highest value in Table R-4 of NUREG/CR-6850.

Regarding the acceptability of the PSA approach, methods, and data in general:

k) Address how it was assured that non-cable intervening combustibles were not missed in areas of the plant. Provide information on how intervening combustibles were identified and accounted for in the fire modeling analyses and the FREs.

l) During the audit, the licensee stated that, Most Palisades cables are not IEEE-383 qualified. Therefore, vertical fire propagation of cable trays was postulated when the cable trays were separated by less than 2-3 feet and horizontal propagation was assumed when trays were in close proximity.

Page 3 of 227

Provide technical justification for the methodology used to determine when to consider horizontal and vertical cable tray propagation.

m) During the audit, the licensee stated that the fire location factor was used to account for this configuration for transient fire sources.

Explain the methodology used for considering a fire (transient and fixed ignition source) located against a wall or in a corner.

n) During the audit, the licensee stated that the activation of detection and suppression systems is credited in the FPRA.

Provide a technical justification for the methodology used to determine or calculate detector and sprinkler activation time (in any fire area). For instance, discuss the fire modeling tool used to perform these calculations and include the verification and validation justification for this tool in the response. If detector/sprinkler activation time was not calculated, provide a technical justification for not having to determine this time quantitatively.

o) During the audit, the non-abandonment fire scenarios considered in the CR envelope were discussed. It was stated that, consideration of many of these cables in common enclosures required a more detailed analysis not amendable to the source/ZOI/target methods used throughout the remainder of the plant. It was also stated that, the remaining control room ignition sources were more similar to standard electrical cabinet construction and were therefore analyzed in a manner consistent with the rest of the fire PRA.

Describe the methodology used to consider non-abandonment fire scenarios and whether the GFMTs were used to assess potential damage in the MCR envelope for ignition sources other than the main control board (MCB). In addition, state whether this analysis included fixed ignition sources as well as transient sources. If transient sources were not considered, provide a technical justification for not considering this type of ignition source.

p) During the audit, the licensee discussed the detailed calculations performed with MathCad, used to modify the non-suppression probability curves. The licensee stated that one of the underlying assumptions of the analysis involves converting damage times presented for thermoplastic cables in Appendix H of NUREG/CR-6850 to percent damage as a function of heat flux. For instance, the licensee stated that: It is assumed that these times can be converted to a percent of damage function as a function of heat flux.

Provide a technical justification for this assumption and explain the physical basis for a continuous damage function. In addition, describe how this assumption affects the CDF, LERF, CDF, and LERF.

q) During the audit, the licensee discussed the exposed structural steel analysis that was performed. Specifically, one of the fire scenarios postulated on the 590-ft Page 4 of 227

elevation of the turbine building was discussed. This scenario considered a feedwater pump lube oil fire and a CFAST analysis that was performed to evaluate the temperature rise of an exposed structural steel target from a defined lube oil pool fire source.

The fire scenario utilizing the CFAST modeling assumes a lube oil fire, which engulfs one structural steel column and exposes another structural steel column.

The licensee stated that this analysis assumes that the engulfed column fails and considers the temperature rise of the adjacent column. Provide a technical justification for not having to consider structural collapse of the compartment as a result of the failure of one structural steel column.

ENO Response a) The PNP fire brigade response time of 15 minutes is based on what is considered a reasonably conservative bounding response time and is consistent with response times assumed or noted at other plants within the Entergy fleet as well as at other utilities. This was validated by a review of fire brigade drills conducted between January 25, 2012 and December 4, 2012 for the fire brigade arrival to various plant areas. The results are summarized in Table FM01a-1 [1]. The plant areas listed in Table FM01a-1 include spaces near the control room as well as outlier areas such as the Construction Building. Table FM01a-1 provides an indication of the likely control room fire brigade response. The times shown in Table FM01a-1 represent the time interval between the fire brigade page and the arrival of the last required fire brigade member. The average response time is 8.2 minutes with a minimum time of six minutes and a maximum time of ten minutes.

The key aspect of the response with respect to the control room abandonment calculation is the potential for the ventilation conditions to change via an open door; as such, the times shown in Table FM01a-1 are conservative insofar as they are based on the arrival of the last team member.

Table FM01a Fire Brigade Arrival Times at Various Palisades Nuclear Plant Areas [1]

Fire Drill Date Location Time on Scene Brigade 01/25/12 T.B. Screenhouse 5 min Shift 1 01/25/12 T.B. Lube Oil Room 7 min Shift 5 01/31/12 MLO Room, T.B. 590 ft Elev Under 10 min Shift 4 02/07/12 L.O. Room, T.B. 590 ft Elev. 7 min Shift 3 Screenhouse, T.B. 590 ft 02/14/12 Under 10 min Shift 2 Elev.

09/27/12 Electrical Equipment Room Under 10 min Shift 2 10/12/12 FAB Shop 6 min Shift 1 11/12/12 Construction Building 9 min Shift 4 12/04/12 Construction Building 10 min Shift 5 Page 5 of 227

The fire brigade response time is incorporated into the control room abandonment calculation model via a change in the status of the boundary doors (closed to open) [2], though credit for manual suppression is independent of this assumption and the fire heat release rates in the CFAST models are not reduced at the brigade arrival time. The MCR boundary doors may open for reasons other than fire brigade arrival, such as operator actions or occupant egress, a value of fifteen minutes was selected as an intermediate value within the time interval considered in the calculation. The FPRA uses the natural ventilation configuration that produces the minimum abandonment time as a representative value to define the baseline abandonment scenarios. The updated fire scenarios report will clearly indicate that this is the process used to select the natural ventilation case from the control room abandonment calculation. Because the most adverse abandonment time is used for the range of natural ventilation conditions, the uncertainty in the door open time is bounded by the use of the data provided in the control room abandonment calculation.

For completeness, the control room abandonment calculation has been updated to include a sensitivity assessment of the model results to the time the boundary doors are assumed to open (see Section A2.2.8 of [3]). The sensitivity assessment considers the effect of opening the boundary door to the control room between ten and twenty minutes on both the calculated abandonment times and the total probability of control room abandonment. It is shown that opening the doors at ten minutes can reduce the total probability of abandonment time by up to twenty-seven percent in all but one scenario, confirming the conservative assumption of a fifteen minute abandonment time. The one scenario in which the total probability of abandonment is shown to increase when the door is opened at ten minutes causes a four percent increase, which is not considered significant. It is also shown that opening the door at twenty-minutes does not affect the total probability of abandonment relative to the assumed fifteen minute baseline time for opening a door. Note that given that the maximum time at which abandonment can affect the non-suppression probability is 20.9 minutes, the scenarios for which the door opens at twenty minutes are nearly the same scenarios as the closed door baseline scenarios.

Based on the actual response times of the fire brigade, the use of the abandonment times in the FPRA, and the sensitivity of the abandonment times to uncertainty in the fire brigade arrival time, there are no known adverse effects associated with not meeting this assumption.

References:

[1] EN-TQ-125 Attachments 9.1 and 9.3, Entergy, 2012.

[2] Report 1SPH02902.066, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, September, 2009.

Page 6 of 227

[3] Report 0021-0019-000-001, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, October, 2013.

b) The heat release rate growth rates for transient fuel package fires were evaluated in the control room abandonment calculation [1] as Medium fires based on data provided in the SFPE Handbook of Fire Protection Engineering, Section 3-1

[2], which is different from the guidance provided in NUREG-6850, Supplement 1 (i.e., FAQ 08-0052). The heat release rate during the growth stage is defined by the following equation:

(FM1b-1) where is the heat release rate (kW [Btu/s]) at time (s), is the growth constant (0.0117 kW/s² [0.0111 Btu/s³]), and is the peak heat release rate for the fire scenario (kW [Btu/s]). The duration of the growth stage for transient fuel packages with heat release rates ranging from 22 - 578 kW (21.2 - 548 Btu/s) is 0.7 - 3.7 minutes. By contrast, recent guidance in NUREG/CR-6850, Supplement 1 [3] recommends that a constant growth time should be assumed: two minutes for loose or unconfined transient material and eight minutes for transient material located within containers or bins. Because the growth rate varies with the NUREG/CR-6850, Supplement 1 approach, the method assumed in the control room abandonment calculation [1] can be more or less conservative depending on the particular heat release rate bin considered and the type of transient postulated.

Specifically, the Medium growth rate is conservative and bounding for transient fires that are postulated to have an eight minute growth rate per NUREG/CR-6850, Supplement 1. The Medium growth rate is conservative and bounding for fires that are postulated to have a two minute growth rate per NUREG/CR-6850, Supplement 1 [3] when the peak heat release rate is less than 168 kW (159 Btu/s).

This heat release rate is greater than NUREG/CR-6850 transient heat release rate Bin 5 but less than NUREG/CR-6850 [4] transient heat release rate Bin 6.

Based on the results provided in Section B.3 of Appendix B to the control room abandonment calculation [1], the assumed growth rate for a transient fuel package fire can affect the abandonment times significantly, depending on the ventilation configuration and the fire location.

In lieu of quantifying the effect of the conservative/non-conservative bias for each transient fire scenario, the characterization of the transient fire scenarios has been revised in an updated control room abandonment calculation [5] such that they are consistent with the guidance provided in NUREG/CR-6850, Supplement 1 [3].

Further, the sensitivity analysis contained in Attachment 2 of [5] has been restructured in response to RAI FM-01d (MCR Abandonment Calculation Sensitivity Analysis) and is used to provide a basis for the baseline scenario assumptions, including the time to reach a peak heat release rate for the transient Page 7 of 227

fire scenarios. It is shown in Sections A2.2.3 and A2.2.4 of Attachment 2 that a two minute growth rate provides a conservative result for transient fire scenarios located in the control room proper and that these fire scenarios bound those that would occur in surrounding spaces [5].

Because the revised transient growth rates are consistent with current NUREG guidance, it is not necessary to compute the impact on fire risk results (i.e., CDF, LERF, CDF and LERF) for this RAI response.

References:

[1] Report 1SPH02902.066, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, September, 2009.

[2] Babrauskas, V., Heat Release Rates, Section 3-1, SFPE Handbook of Fire Protection Engineering, 4th Edition, Society of Fire Protection Engineers (SFPE), Bethesda, MD, 2008.

[3] NUREG/CR-6850, Supplement 1, Fire Probabilistic Risk Assessment Methods Enhancements, EPRI 1019259, Technical Report, NRC, Rockville, MD, September, 2010.

[4] NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, Electric Power Research Institute (EPRI) 1011989 Final Report, Nuclear Regulatory Commission (NRC),

Rockville, MD, September, 2005.

[5] Report 0021-0019-000-001, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, October, 2013.

c) The original control room abandonment calculation [1] did not postulate fixed or transient source fires in wall or corner locations. The original basis for fixed ignition sources (electrical panels) was that there are no locations for which a wall or corner location effect would be applicable since either the physical location of the panels is at least 0.6 m (2 ft) from a wall boundary or the vents for the panels are located at least 0.6 m (2 ft) from wall boundaries. The use of a 0.6 m (2 ft) offset for wall and corner locations is consistent with the guidance provided in the Significance Determination Process (SDP) [2]. In the case of transient source fires, it was assumed that the open location source fires would generate a more adverse hot gas layer than the wall and corner locations due to a faster hot gas layer descent time given that the layer descent time was the limiting constraint in nearly all of transient fires postulated.

In lieu of calculating the potential contribution on the risk for excluding wall and corner location transient fires in the control room, the control room abandonment calculation has been updated to include wall and corner location transient source Page 8 of 227

fires as baseline scenarios [3]. The original assumption for open location fixed ignition sources is considered valid in the updated control room abandonment calculation [3]. The Image Method [4] has been applied to simulate the effects of reduced air entrainment on wall and corner transient fire locations. Accordingly, the fire size is increased by a factor of two for wall locations and the enclosure volume, natural and forced ventilation, and wall area are increased by a factor of two to accommodate the virtual portion of the fire plume. Similarly, the fire size is increased by a factor of four for corner locations and the enclosure volume, natural and forced ventilation, and wall area are increased by a factor of four to accommodate the virtual portion of the fire plume. Because the abandonment criteria have been revised to correspond to the NUREG/CR-6850 [5] guidance, the layer descent time is not uniformly the limiting constraint in the updated control room abandonment calculation [3]. As a result, wall and corner transient source fire locations produce shorter abandonment times and have a significant effect (greater than fifteen percent) on the total probability of control room abandonment.

References:

[1] Report 1SPH02902.066, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, September, 2009.

[2] NRC, Fire Protection Significance Determination Process, Appendix F, NRC, Washington, D.C., February 28, 2005.

[3] Report 0021-0019-000-001, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, October, 2013.

[4] NIST-GCR-90-580, Development of an Instructional Program for Practicing Engineers Hazard I Users, Barnett, J. R. and Beyler, C. L., National Institute of Standards and Technology, Gaithersburg, MD, July, 1990.

[5] NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, Electric Power Research Institute (EPRI) 1011989 Final Report, Nuclear Regulatory Commission (NRC),

Rockville, MD, September, 2005.

d) The sensitivity analysis provided in Appendix B of the original control room abandonment calculation [1] was not used by the fire PRA because of the difficulties in propagating the parameter uncertainty into the fire PRA model. The sensitivity analysis in the control room abandonment report has been updated to provide a conservative basis for the baseline fire scenarios (see Attachment 2 of the updated control room abandonment calculation [2]). The conservative basis is assessed using both the absolute effect on the control room abandonment time and the cumulative effect on the total probability of abandonment as computed using the methods described in NUREG/CR-6850 [3].

Page 9 of 227

The revised sensitivity analysis provided in the updated control room abandonment calculation [2] indicates that the assumptions on parameter input values can be separated into one of three categories:

The parameter does not significantly affect the analysis results over the potential range of values that could be assigned to the parameter (Sensitivity Group 1);

The parameter does affect the analysis results, but the value selected for the baseline case is conservative (Sensitivity Group 2); and The parameter does affect the analysis results and the value selected for the baseline case is not conservative (Sensitivity Group 3).

A significant effect is defined as a fifteen percent variation in the probability of control room abandonment as summed over three heat release rate bins. The baseline parameter values are iteratively determined using the sensitivity analysis in such a way that the baseline fire scenarios are conservative or clear limitations on the applicability are established [2].

The final baseline fire scenarios contain parameters that fall into each of the three sensitivity groups identified [2]. Parameters that fall into the first group (Sensitivity Group 1) include the assumed boundary leakage fraction, the assumed fuel heat of combustion, the number of boundary doors opened, and several other parameters that are specific to some ignition sources or fire locations. Parameters that fall into the second group (Sensitivity Group 2) include the assumption that the MCR viewing window remains intact, the assumed burning regime for transient fires, the assumed radiant fraction for fires in the MCR, the assumed fire base height for transient fires, and the assumed time the boundary doors are open (at 10 minutes) for fires in the main control boards (i.e., the CFAST model sub-enclosures).

Two parameters fall into the third group (Sensitivity Group 3): the initial ambient temperature; and the fire base height for closed, multiple-bundle electrical cabinets that spread to adjacent cabinets in the MCR. In the case of the initial ambient temperatures, it is shown that the baseline scenarios are conservative or the effect is not significant for initial ambient temperatures up to about 34°C (93°F). The maximum design basis temperature for the control room is 32.2°C (90°F), so an initial temperature greater than 34°C (93°F) would represent an off-normal condition. The results of control room abandonment calculation are therefore limited to scenarios postulated under an initial temperature condition less than 34°C (93°F). Because this is greater than the maximum design temperature, this condition is expected to be met for the scenarios considered in the fire PRA. In the case of the fire base height, half-height Main Control Boards (MCBs) in the control room will be treated in the fire PRA with an increased probability of abandonment of twenty-five percent greater than the baseline [2] where they are characterized as multiple bundle panels that propagate to adjacent panels. The remainder of the Page 10 of 227

fixed ignition sources in the control room are adequately treated using the elevated fire base as prescribed in NUREG/CR-6850, Supplement 1 [4].

Based on the revised sensitivity analysis, the baseline results are considered conservative over the range of parameter uncertainty because the probability of abandonment is maximized, which in turn maximizes the CDF and LERF. As such, explicit consideration of the parameter uncertainty as contained in Attachment 2 to the control room abandonment calculation [2] is not required in the fire PRA and it is not necessary to compute the impact on fire risk results (i.e., CDF, LERF, CDF and LERF) for this RAI response.

References:

[1] Report 1SPH02902.066, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, September, 2009.

[2] Report 0021-0019-000-001, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, October, 2013.

[3] NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, Electric Power Research Institute (EPRI) 1011989 Final Report, Nuclear Regulatory Commission (NRC),

Rockville, MD, September, 2005.

[4] NUREG/CR-6850, Supplement 1, Fire Probabilistic Risk Assessment Methods Enhancements, EPRI 1019259, Technical Report, NRC, Rockville, MD, September, 2010.

e) The modified critical heat flux was implemented using a two point treatment for the updated fire scenarios in the fire PRA model. The first point corresponds to temperature conditions between ambient and 80°C (176°F) and represents the temperature interval in which the Zones of Influence (ZOIs) such as those documented in the Generic Fire Modeling Treatments report [1] are applicable.

The second point corresponds to temperature conditions greater than 80°C (176°F) and will be conservatively characterized in the fire PRA as a full-room burnout. This applies to both targets located in the thermal plume region and to targets that are located outside the thermal plume region. The updated fire scenarios will be in the base case results of the RAI Response Fire PRA model.

References:

[1] 1SPH02902.030, Rev. 0, Generic Fire Modeling Treatments, Hughes Associates, Baltimore, MD, January, 2008.

f) The Generic Fire Modeling Treatments (GFMT) report [1] provides tabulated ZOI information for ignition sources that do not involve secondary combustible Page 11 of 227

materials such as cable trays or multiple ignition sources. Although the original GFMT did not explicitly consider secondary combustible materials, the ZOI developed for electrical panel ignition sources includes a conservative lower ZOI dimension that would typically exceed the ZOI dimension during the early fire stages for scenarios that involved a relatively small number of cable trays. In addition, the vertical ZOI dimension is extended to the ceiling when secondary combustibles are located in the ignition source ZOI. However, there are situations where the lower ZOI dimension would not bound the ZOI dimension if secondary combustibles were explicitly included. In addition, the ZOI dimensions for transient ignition sources would typically be non-conservative if secondary combustibles are involved. In lieu of demonstrating which scenarios are conservative and which require further analysis, new ZOI tables have been developed that are applicable to ignition source - cable tray configurations at Palisades. These ZOI tables are documented in Report 0021-0019-000-002, Rev. 0 [2]. Because the method for determining the heat release rate development documented in Report 0021-0019-000-002, Rev. 0 [2] is consistent with applicable NUREG guidance (NUREG/CR-6850 [3] and NUREG/CR-7010, Volume 1 [4], the approach is considered conservative and bounding.

Supplemental plant walkdowns have been conducted to incorporate the new ZOI dimensions to identify ignition sources that involve secondary combustibles and to document the additional target sets that should be included in the RAI Response Fire PRA model for these ignition sources using the ZOI tables provided in Report 0021-0019-000-002, Rev. 0 [2].

References:

[1] 1SPH02902.030, Rev. 0, Generic Fire Modeling Treatments, Hughes Associates, Baltimore, MD, January, 2008.

[2] Report 0021-0019-000-002, Rev. 0, Combined Ignition Source - Cable Tray Fire Scenario ZOIs for Palisades Nuclear Power Plant Applications, Hughes Associates, Inc., Baltimore, MD, November, 2013.

[3] NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, Electric Power Research Institute (EPRI) 1011989 Final Report, Nuclear Regulatory Commission (NRC),

Rockville, MD, September, 2005.

[4] NUREG/CR-7010, Volume 1, Cable Heat Release Rate, Ignition, and Spread in Tray Installations During Fire (CHRISTIFIRE) Phase 1: Horizontal Trays, NUREG/CR-7010, Vol. 1., McGrattan, K., Lock., A., Marsh, N., Nyden, Bareham, S., M., Price, M., Morgan, A. B., Galaska, M., and Schenck, K.,

Office of Nuclear Regulatory Research, NRC, Washington, DC., July, 2012.

g) Fire spread and fire propagation in cable trays is considered at PNP for fire scenarios in which the ignition source fire propagates into one or more cable trays.

Page 12 of 227

Supplement 2 to the Generic Fire Modeling Treatments (GFMT) report [4] provides tables that document the time to reach threshold hot gas layer (HGL) temperatures for electrical panel ignition sources that ignite two adjacent 0.61 m (24 in) wide cable trays. The cable trays ignite at plane section through the cable trays five minutes after the ignition source ignites and propagate fire laterally in two directions in a manner consistent with the NUREG/CR-6850, Appendix R guidance for thermoplastic cables. The configuration is considered to be an average representation of an ignition source - cable tray configuration and consequently may be over conservative in some situations and under-conservative in other situations when compared against the FLASH-CAT methodology provided in NUREG/CR-7010, Volume 1 [3]. The GFMT report [1] provides tabulated Zone of Influence (ZOI) and HGL information for ignition sources that do not involve secondary combustible materials such as cable trays or multiple ignition sources.

No consideration is provided for the additional heat release rate on the ZOI and HGL temperatures in this report; however, the general practice is to increase the vertical ZOI dimension to the ceiling when secondary combustibles are in the ZOI of the ignition source. This practice, coupled with the conservative method for which the ZOIs are developed, is considered conservative in some situations and non-conservative in other situations.

In lieu of demonstrating which scenarios are conservative relative to NUREG guidance and which require further analysis, new ZOI and HGL have been developed that are applicable to ignition source - cable tray configurations at PNP.

The new ZOI and HGL tables are used to characterize the target sets where secondary combustibles (cable trays) are ignited by an ignition source and replace those provided in Supplement 2 to the GFMT Report [4]. The new ZOI and HGL tables are documented in Report 0021-0019-000-002, Rev. 0 [5] and Report 0021-0019-000-003 [6]. Because the method for determining the heat release rate development documented in Report 0021-0019-000-002, Rev. 0 [5] and Report 0021-0019-000-003, Rev. 0 [6] is consistent with applicable NUREG guidance, the approach is considered conservative and bounding.

Supplemental plant walkdowns have been conducted to incorporate the new ZOI dimensions as described in the response to RAI FM 01(f).

References:

[1] 1SPH02902.030, Rev. 0, Generic Fire Modeling Treatments, Hughes Associates, Baltimore, MD, January, 2008.

[2] NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, Electric Power Research Institute (EPRI) 1011989 Final Report, Nuclear Regulatory Commission (NRC),

Rockville, MD, September, 2005.

[3] NUREG/CR-7010, Volume 1, Cable Heat Release Rate, Ignition, and Spread in Tray Installations During Fire (CHRISTIFIRE) Phase 1: Horizontal Trays, Page 13 of 227

NUREG/CR-7010, Vol. 1., McGrattan, K., Lock., A., Marsh, N., Nyden, Bareham, S., M., Price, M., Morgan, A. B., Galaska, M., and Scheneck, K.,

Office of Nuclear Regulatory Research, NRC, Washington, DC., July, 2012.

[4] Hughes Associates, Supplemental Generic Fire model Treatments: Hot Gas Layer Tables, Revision F, Hughes Associates, Inc., Baltimore, MD, March 1, 2011.

[5] Report 0021-0019-000-002, Rev. 0, Combined Ignition Source - Cable Tray Fire Scenario ZOIs for Palisades Nuclear Power Plant Applications, Hughes Associates, Inc., Baltimore, MD, November, 2013.

[6] Report 0021-0019-000-003, Rev. 0, Evaluation of the Development and Timing of Hot Gas Layer Conditions in Generic Palisades Fire Compartments with Secondary Combustibles, Hughes Associates, Inc., Baltimore, MD, November, 2013.

h) The transient fuel packages are assumed to be miscellaneous materials (trash configurations) that do not contain acetone or other combustible liquids. This corresponds to the Group 3 and Group 4 transient fuel packages described in Supplement 3 of the Generic Fire Modeling Treatments report [1]. The 98th percentile transient fuel packages are considered a special case of the Group 3 and Group 4 transient fuel packages with a specific heat release rate per unit area as described in the Supplement 3 of the Generic Fire Modeling Treatments report [1].

The transient fire heat release rate conditional probability distribution specified in NUREG/CR-6850 [2], with a 317 kW (300 Btu/s) 98th percentile peak heat release rate fire, is considered to be generically applicable to nuclear power plants. PNP does not differ in any significant manner with respect to its transient combustible controls to warrant a significant increase or decrease in the applicable heat release rate profile or heat release rate conditional probability distribution as specified in NUREG/CR-6850 [2].

The control of combustibles will be ensured under procedure EN-DC-161, which limits the accumulation and composition of materials using a graded approach (Level 1 [highest risk) through Level 4 [lowest risk]). The procedure provides the framework for the introduction of combustibles into each hazard level area and the required conditions that apply when combustibles are introduced. Combustibles that do not meet the specified requirements for each hazard level require Transient Combustible Analysis (TCA) to be performed [3]. High hazard areas (Level 1) require a TCA when any transient combustible material is introduced whereas lower hazard areas require a TCA only when the exempt quantity is exceeded. The use of the combustible control procedure will limit the combustible configurations in high hazard areas to configurations that are bound by the analysis provided in Supplement 3 of the Generic Fire Modeling Treatments report [1] or where impractical provide for the necessary compensatory measures Page 14 of 227

via the TCA.

It is noted that there is one case considered in the control room abandonment calculation [4] in which a transient fuel package fire scenario is characterized using a heat release rate profile that is more adverse than the standard NUREG/CR-6850 Appendix E Case 8 transient fuel package fire scenario. Specifically, an office type fuel arrangement is postulated and characterized using a heat release rate profile applicable to such fuel packages. This configuration is unique to the control room area among risk significant plant areas.

References:

[1] 0021-0019-000-004, Rev. 0, Supplemental Generic Fire Modeling Treatments: Transient Fuel Package Ignition Source Characteristics, Hughes Associates, Inc., Baltimore, MD, 2013.

[2] NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, Electric Power Research Institute (EPRI) 1011989 Final Report, Nuclear Regulatory Commission (NRC),

Rockville, MD, September, 2005.

[3] EN-DC-161, Rev. 9, Control of Combustibles, Entergy, 10/21/2013.

[4] Report 0021-0019-000-001, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, October, 2013.

i) The CFAST analysis of the upper switchgear cubicle is a preliminary assessment of the types of fires that should be input into the FDS model. In particular, the maximum internal heat release rate is estimated and this is used as a means of estimating the maximum exterior heat release rate given the known upper limit on the equivalence ratio within an enclosure. The decision to treat the upper switchgear cubicle as a one-zone enclosure is based on the dimensions of the space as compared to a typical room enclosure for which the CFAST model was developed and the objective of the CFAST calculation.

The recommended guidelines for using the one-zone assumption in CFAST per NIST SP 1026 [1] consist of the following:

Spaces with large length to height ratios (i.e., a shaft-like enclosure). The position of the smoke layer within such spaces would stabilize near the source fire, and can be expected to be at a position that is less than twenty percent of the total enclosure height.

Spaces that are relatively far from the source fire for which the smoke products cool and may no longer be stratified.

Page 15 of 227

A third class of fire scenarios that are not identified in the CFAST technical manual consist of fully-developed enclosure fires. These types of fires are normally described as post-flashover or ventilation controlled and are characterized by well-mixed conditions and a burning rate /heat release rate that is a function of the ventilation factor (see [2], for example). CFAST is typically used to calculate the enclosure conditions during initial stages of a fire development prior to flashover using a two-zone approximation. When a fire transitions to a fully-developed condition, the two-zone assumptions break down and the use of CFAST would correctly be considered to be outside its limitations. Because the objective of the upper cubicle analysis is to identify the maximum fully-developed fire size that can be supported by the actual vent configuration, the two zone assumption was not considered to be applicable within the enclosure. In addition, dimensions of the upper switchgear cubicle are not typical of a room enclosure:

Length: 0.66 m (2.16 ft)

Width: 2.02 m (6.61 ft)

Average height: 0.6 m (1.98 ft)

The momentum driven flows from a fire plume within such a small enclosure would create significant mixing between the combustion products and the inflow and further contribute to a well-stirred internal condition.

Utiskul [3] investigated the applicability of the one-zone assumption to small enclosures (0.4 0.4 1.2 m [1.3 1.3 4 ft) and identified three types of configurations for which a one-zone assumption is appropriate:

A post-flashover environment (i.e., a fully-developed enclosure fire)

A fuel area that is large compared to the compartment floor area A hot gas layer height that is lower than twenty percent of the ceiling height The objective of the CFAST analysis of the upper switchgear cubicle is to identify the maximum fire size that can be supported by the actual vent configuration and this is by definition a fully-developed enclosure fire. In addition, the hot gas layer height within the upper switchgear cubicle is expected to be at or near the cubicle base indicating at least two of the three configurations or conditions for which a one-zone model is appropriate to apply are present.

Based on these considerations, a one-zone enclosure model is considered appropriate for the upper switchgear cubicle. The use of the one-zone model in CFAST was selected because of the ability to position the vents at the actual height and thus approximate the mass-flows into and out of the enclosure correctly. In addition, the CFAST one-zone model allows for the computation of the Page 16 of 227

heat flows through the steel enclosure and provides a better approximation of the internal temperature as compared to correlations developed for concrete or gypsum enclosures (see [2], for example). The selection of the limiting oxygen index (ten percent) was based on experimental data applicable to the calculated internal temperature (see [4] and [5]).

The one-zone CFAST model in Reports 1JMW20053.000-1, Rev. 0 and 1JMW20053.000-2, Rev. 0 was used in parallel with an empirical model for calculating the maximum fire size that can be supported by the vent configuration (Kawagoe equation). The Kawagoe model has been shown to be applicable to fully-developed fires within small enclosures in both the Utiskul [3] dataset and in the CIB [6] dataset, which consisted of about four-hundred tests in enclosures having a length scale between 0.5 - 1.5 m (0.6 - 5 ft) [6]. The two methods (CFAST and Kawagoe) predict a maximum fire size ranging from 34 - 37 kW (32 -

35 Btu/s) and are within ten percent of one another. This provides both the validation and confirmation of the CFAST model analysis of the upper switchgear cubicle fires.

Supplement 1 to the Generic Fire Modeling Treatments consists of a one-zone enclosure model treatment of panel fires using the same basic method as CFAST.

It involves a mass and energy balance across a control volume that is coincident with an electrical panel boundary. The predictions in Supplement 1 are validated against the available full scale panel test data and are found to provide a conservative estimate of the maximum fire size a panel vent configuration can support. The key difference between CFAST and Supplement 1 to the Generic Fire Modeling Treatments involves the assumption of the vent location. Because the analysis in Supplement 1 to the Generic Fire Modeling Treatments is generic, the outflow vent is placed at the ceiling of the enclosure and the inflow vent is placed at the floor of the enclosure in order to maximize the potential fire size. The one-zone CFAST analysis places the vents at the actual location, which in the case of the upper switchgear cubicle is slightly below the ceiling for the outflow vent and at the floor for the inflow vent.

References

[1] NIST SP 1026, CFAST - Consolidated Model of Fire Growth and Smoke Transport (Version 6) Technical Reference Guide, Jones, W. W., Peacock, R.

D., Forney, G. P., and Reneke, P. A., NIST, Gaithersburg, MD, April, 2009.

[2] Walton, W. D. and Thomas, P. H., Estimating Temperatures in Compartment Fires, Section 3-6, SFPE Handbook of Fire Protection Engineering, 4th Edition, Society of Fire Protection Engineers (SFPE), Bethesda, MD, 2008.

[3] Utiskul, Y., Theoretical and Experimental Study of Fully-Developed Compartment Fires, University of Maryland Ph. D. Dissertation, 2006.

Page 17 of 227

[4] Beyler, C. L., Flammability Limits of Premixed and Diffusion Flames, Section 2-7, SFPE Handbook of Fire Protection Engineering, 4th Edition, Society of Fire Protection Engineers (SFPE), Bethesda, MD, 2008.

[5] NIST SP 1018-5, Fire Dynamics Simulator (Version 50 Technical Reference Guide, Volume 1: Mathematical Model, McGrattan, K., Hostikka, S., Floyd, J.,

Baum, H., Rehm, R., Mell, W., and McDermott, R., NIST, Gaithersburg, MD, October, 2010.

[6] Thomas, P. H. and Heselden, A. J. M., CIB Report No. 20, Fire Research Note 923/197, 1972.

j) The mechanism for cable tray ignition is a timed delay after ignition of the panel below. The delay is equal to the time required for fire to propagate up a vertical airdrop, and for a flame spread rate of 0.0258 m/s (1 in/s) is about six seconds after the panel ignites.

The justification for using a vertical flame spread rate of 25.8 mm/s (1 in/s) in lieu of 258 mm/s (10 in/s) as listed in Table R-4 of NUREG/CR-6850 [1] was based on an assessment of the method used to derive the recommended values and a review of existing vertical flame spread data for cables and cable trays. The flame spread values provided in Table R-4 of NUREG/CR-6850 [1] are deduced from a very conservative model consisting of three basic components:

A flame height calculation for wall fires A heat flux estimate in the pre-heat region above the pyrolysis zone An ignition model The method postulates an initial pyrolysis zone equal to the flame height of an exposure fire, determines the flame height above the pyrolysis zone, calculates the ignition time for a constant heat flux exposure of 25 kW/m² (2.2 Btu/s-ft²), and divides the distance between the flame tip and pyrolysis front by the cable ignition time to obtain a vertical flame spread rate. The flame height calculation used in NUREG/CR-6850 is a linear approximation of a wall flame height model that embeds a twenty percent conservative bias for the maximum value provided in Table R-4 of [1] (see [2]). The wall flame height model itself is conservative relative to other similar models (see [3] and [4]). Additional conservatism in the NUREG/CR-6850 model is introduced through the assumption of a constant flame heat flux in the pre-heat zone. Data for burning flat surfaces suggest that the heat flux is constant and equal to about 25 kW/m² (2.2 Btu/s-ft²) up to about one-half the flame height and decreases to about 5 kW/m² (0.44 Btu/s-ft²) at the flame tip (see Figure 2-14.26 in [4]). A final conservatism involves the ignition model used, which is based on the physical properties of the cables rather than correlated ignition test data. The SFPE [5] provides documentation for an ignition model that is applicable to cables (Tewarson model) and indicates the ignition time would be Page 18 of 227

between 42 - 173 seconds for the same types of cables exposed to a constant 25 kW/m² (2.2 Btu/s-ft²) radiant heat flux as compared to 16 seconds for the cables listed in Table R-4 of NUREG/CR-6850 [1]. This introduces a conservative factor for the flame spread rate that is on the order of 3 - 10. Collectively, the conservative features of the NUREG/CR-6850 [1] model produce flame spread rates that are significantly higher than observed values and approach values associated with surface flame spread on flammable liquid pools [6]. As such, a review of test data was conducted to determine a vertical flame spread rate applicable to the vertical cable tray arrangements found in FA3 and FA4.

The existing vertical cable and cable tray segments in FA3 and FA4 range from 0.15 - 1.8 m (0.5 - 6 ft) in length and, except for the 0.15 m (0.5 ft) airdrops to the switchgear, are all postulated to ignite via propagation from adjacent horizontal cable trays. Tewarson and Kahn [7] conducted thirty-eight tests on single and multiple vertical cable configurations. The tests included both thermoset and thermoplastic cables with vertical segments ranging from 0.5 - 1.3 m (1.6 - 4.3 ft),

which are comparable in length to the vertical cable tray configurations found in FA3 and FA4. The data indicate the typical vertical propagation rate for thermoplastic cables was between 2 - 6 mm/s (0.08 - 0.24 in/s) and the maximum value was 14 mm/s (0.55 in/s). Likewise, the typical vertical propagation rate for thermoset cables was between 1 - 3 mm/s (0.04 - 0.12 in/s) and the maximum value was 7.8 mm/s (0.31 in/s), about one-half the thermoplastic values. In addition, there was no clear pattern observed between the 0.5m (1.6 ft) and 1.3 m (4.3 ft) segments; the results were similar when both lengths were tested for a given cable but the fastest spread rate was not uniformly associated with either the long or short segment. Given that the fires were ignited near the cable base, as would be expected in the FA3 and FA4 scenarios, and that the lengths are comparable to the vertical cable lengths considered in FA3 and FA4, the Tewarson and Kahn [7] data are considered applicable. A value of 25.8 mm/s (1 in/s) was selected as a bounding representation of this dataset and, based on the maximum vertical propagation value report (14 mm/s [0.55 in/s]), it introduces a conservative bias factor of about 1.8.

References:

[1] NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, Electric Power Research Institute (EPRI) 1011989 Final Report, Nuclear Regulatory Commission (NRC),

Rockville, MD, September, 2005.

[2] Quintiere, J. G., A Simulation Model for Fire Growth on Materials Subject to a Room-Corner Test, Fire Safety Journal, 1993.

[3] Beyler, C. L., Fire Plumes and Ceiling Jets, Fire Safety Journal, 1986.

Page 19 of 227

[4] Lattimer, B. Y., Heat Fluxes from Fires to Surfaces, Section 2-14, SFPE Handbook of Fire Protection Engineering, 4th Edition, Society of Fire Protection Engineers (SFPE), Bethesda, MD, 2008.

[5] SFPE, Piloted Ignition of Solid Materials Under Radiant Exposure, Engineering Guide, SFPE, Bethesda, MD, January, 2002.

[6] Gottuk, D. T. and White, D. A., Liquid Fuel Fires, Section 2-15, SFPE Handbook of Fire Protection Engineering, 4th Edition, Society of Fire Protection Engineers (SFPE), Bethesda, MD, 2008.

[7] Tewarson, A. and Kahn, M. M., Flame Propagation for Polymers in Cylindrical and Vertical Orientation, Twenty Second Symposium on Combustion, 1988.

k) As an additional refinement to the RAI Response Fire PRA model, the process used to assure that non-cable secondary combustibles are not missed in areas of the plant has been updated. Identification of fixed non-cable intervening combustibles is achieved through a combination of document reviews and plant walkdowns. A review of the PNP combustible loading analysis is performed to identify non-cable secondary combustibles that could impact the scenario development within a fire area. Reviews of drawings and walkdowns (where possible) are then performed to determine the impact of the secondary combustibles on the scenario development performed for fixed and transient ignition sources. Drawings include the Pre-Fire Plans, the Fire protection plans, and raceways drawings. The impacts are being incorporated into the modeling of the fire scenarios.

The scenarios are being updated and the numerical effects will be included in the RAI Response Fire PRA model.

l) Since the LAR submittal, the methodology used to determine when to consider horizontal and vertical cable tray propagation has been updated to address the issue identified in this RAI and other issues. A review of fire scenarios involving secondary combustibles has been performed and the ignition of these combustibles has been evaluated. The propagation and increase to the fire scenario zone of influence (ZOI) has been evaluated. See RAI Response FM 01, part g) for more details. The scenarios will be updated and the numerical effects will be included in the RAI Response Fire PRA model.

m) The methodology used for considering a fire (transient and fixed ignition source) located against a wall or in a corner is from the guidance provided in the Hughes Report, Generic Fire Modeling Treatments, Revision 0 [1]. Section 3.3.7, Guidance for Fuel Packages Positioned in a Corner and Wall, of the Hughes Generic Fire Modeling Treatments report, reads as follows:

1. If the fuel package is within 0.6 m (2 ft) of a wall, then double the heat release rate and assume that the fire is centered at the fuel package edge adjacent to the wall.

Page 20 of 227

2. If the fuel package is within 0.6 m (2 ft) of a corner, then quadruple the heat release rate and assume that the fire is centered at the fuel package corner nearest the wall corner.

References:

[1] Hughes Associates. Inc., Generic Fire Modeling Treatments, Revision 0, Jan.

2008.

n) In response to PRA RAI 01 r), the time to smoke detector activation is justified for the areas that credit automatic detection in the fire PRA.

In response to PRA RAI 01 u), the credit for automatic wet pipe systems in the fire PRA is described.

o) Non-abandonment fire scenarios for ignition sources other than the main control board (MCB) are analyzed using the Generic Fire Modeling Treatments (GFMTs).

These ignition sources include NUREG/CR-6850 bin 15 electrical cabinets. The zones of influence (ZOIs) for the electrical cabinets are determined and targets located within the ZOIs are postulated to fail. In addition, cables terminating at the ignition source are failed.

The fire PRA model is being updated to include MCR transient scenarios in the RAI Response Fire PRA model consistent with the treatment of transient fire scenarios discussed in PRA RAI 06.

p) The assumption that the damage times provided in Appendix H of NUREG/CR-6850 can be converted to a percent of damage function as a function of heat flux was made based on an interpretation of the available data. However, given the uncertainty of this assumption, a more accepted application of the times provided in NUREG/CR-6850 Appendix H is being applied in the RAI Response Fire PRA model. Refer to the response to PRA RAI 01q. Therefore, an assessment to the impact on CDF, LERF, CDF, and LERF is not provided in this RAI response.

q) The PNP senior civil engineer was consulted regarding the assumption that the engulfed column fails and considers the temperature rise of the adjacent column.

The PNP senior civil engineer provided evidence to support this assumption by stating that there is sufficient redundancy designed into a structure to account for other loading combinations, such as wind or seismic, in addition to the inherent structural margins that building codes require in order that design stresses remain well below the yield point of the materials. Also, steel structures are typically able to withstand stresses well into the inelastic (plastic) range before failure occurs.

He also agreed that this assumption has a somewhat conservative bias in that multiple, adjacent building column failures would be necessary to cause the entire structure to collapse during normal plant operation, absent other design basis external loads (or crane loads) or when Operating Floor Loads are near their limits, such as during a refueling outage when both Low Pressure Turbines are disassembled.

Page 21 of 227

Therefore, because sound engineering judgment has been provided regarding this assumption, additional technical justification for not having to consider structural collapse of the compartment as a result of the failure of one structural steel column is not required.

NRC Request FM RAI 02 American Society of Mechanical Engineers/American Nuclear Society (ASME/ANS)

Standard RA-S-2008, Standard for Level 1/Large Early Release Frequency Probabilistic Risk Assessments for Nuclear Power Plant Applications., Part 4, requires damage thresholds be established to support the FPRA. Thermal impact(s) must be considered in determining the potential for thermal damage of SSCs. Appropriate temperature and critical heat flux criteria must be used in the analysis. During the audit, the damage criteria used for cables, sensitive electronics and component failures due to smoke damage was discussed.

Provide the following information:

a) Describe how the installed cabling in the power block was characterized, specifically with regard to the critical damage threshold temperatures and critical heat flux for thermoset and thermoplastic cables as described in NUREG/CR-6850.

b) During the audit, the detailed calculations performed with MathCad, used to modify the non-suppression probability curves were discussed. One of the underlying assumptions of the analysis involves converting damage times presented for thermoplastic cables in Appendix H of NUREG/CR-6850 to percent damage as a function of heat flux. The following was stated: It is assumed that these times can be converted to a percent of damage function as a function of heat flux. This provides a representative means to apportioning the impact of the varying heat fluxes over time. This may be slightly non-conservative as the damage threshold may not be the same at each heat flux value.

Provide additional discussion of how the damage threshold might change in a given analysis when this assumption is applied. Provide a list of each fire area where this methodology was used.

c) During the audit, the licensee stated that, NUREG/CR-6850 recommends failure criteria for solid-state control components of 3 kW/m2 (versus 11 kW/m2 for IEEE-383 qualified cables and 6 kW/m2 for non-IEEE-383 qualified cables) be used for screening purposes. However, given that the enclosure would provide protection to the sensitive internal contents from external fire effects, it is reasonable to apply the same zone of influence established for cable damage.

The omission of the credit for the enclosure is judged to offset the non-conservatism of the damage threshold.

Page 22 of 227

Provide technical justification for using cable damage thresholds for temperature sensitive equipment located inside cabinets.

ENO Response a) Based on the age of PNP, the majority of the cables used during initial installation are thermoplastic cables. It is likely that more recent cable installations would have used a thermoset IEEE-383 qualified cable. However, the guidance in NUREG/CR-6850 indicates that in most situations of mixed cable types it is the thermoplastic criteria that should be applied.

Therefore, a critical damage threshold temperature of 205C and a critical heat flux of 6 kW/m2 were used for all instances in which cable damage was postulated as recommended for thermoplastic cable in NUREG/CR-6850.

b) The approach used in the development of the LAR model converted the times to damage listed in NUREG/CR-6850 Appendix H, Table H-8, to damage rates. The damage rates represented the average rate at which damage occurred for the given heat fluxes. These damage rates were utilized as the representative rate of damage for the heat fluxes.

While it is recognized that the rate of damage is not constant, variations in this rate were assumed to have minimal impact on the FPRA model results as this would be a second order impact on the developed non-suppression probabilities. The change depends on how the damage rate curve is shaped. As the behavior of this phenomenon was not known precisely, a constant rate of damage was assumed to provide a best estimate of the time available before electrical failures are assumed to occur.

The response to PRA RAI 01q provides an overview on how these average damage rates are used in the time to damage calculation. However, based on the potential concern identified here, the approach that will be utilized in the RAI Response Fire PRA Model will be to conservatively assume that the time available is equal to the time delay associated with the peak heat flux obtained (from the times provided in NUREG/CR-6850, Appendix H) as soon as the lower bound critical heat flux of 6 kW/m2 is reached. This does not substantially change the non-suppression probabilities utilized, but the numerical effects of utilizing this more conservative approach will be included in the RAI Response Fire PRA Model results.

Table 1 lists the Fire Areas where the damage accrual methodology is used.

Page 23 of 227

Table 1: Fire Areas Where the Damage Accrual Methodology is Used Fire Area Description Comment 02 Cable Spreading Updated scenarios for 16 ignition sources to utilize the Room revised non-suppression probabilities based on the more conservative approach for the RAI Response PRA Fire Model. Also, refer to the response to PRA RAI 07 for a discussion of the revised scenario development methodology applied in this area for the RAI Response model.

03 1D Switchgear This approach was used for 4 ignition sources in the Room LAR model but will be replaced with results of the FDS fire modeling analysis in the RAI Response Fire PRA Model.

13 Auxiliary Building This approach was used for three ignition sources in the LAR model but will no longer be used in the RAI Response Fire PRA Model. One Ignition source is no longer in the fire PRA as it was confirmed that is does not meet the ignition source counting criteria for motors.

The other two ignition sources are being updated to account for the impact of secondary combustibles on the zone of influence and due to the close proximity of secondary combustibles will not utilize this method.

15 Engineering Updated scenarios for 4 ignition sources to utilize the Safeguard Panel revised non-suppression probabilities based on the and Auxiliary more conservative approach for the RAI Response Fire Building Stairways PRA Model.

21 Electrical Updated scenarios for 2 ignition sources to utilize the Equipment Room revised non-suppression probabilities based on the more conservative approach for the RAI Response Fire PRA Model.

26 Southwest Cable Updated scenarios for 2 ignition sources to utilize the Penetration Room revised non-suppression probabilities based on the more conservative approach for the RAI Response Fire PRA Model.

c) The guidance in FAQ 13-0004, Rev. 1 [1] is to treat sensitive electronics mounted inside a cabinet that are not directly exposed to the convective and/or radiant energy of a fire as qualified up to the NUREG/CR-6850 heat flux damage threshold for thermoset cables (i.e., 11 kW/m2). The guidance is based on FDS simulations. The PNP fire PRA used the NUREG/CR-6850 damage threshold for thermoplastic cables (6 kW/m2) for targets. Therefore, the fire PRA treatment for sensitive electronics mounted in cabinets is acceptably conservative compared to the guidance in FAQ 13-0004, Rev. 1.

Reference:

[1] FAQ 13-0004, Revision 1, Clarifications on Treatment of Sensitive Electronics, ADAMS Accession Number ML13182A708, June 2013.

Page 24 of 227

NRC Request FM RAI 03 NFPA 805, Section 2.7.3.2, "Verification and Validation," states: "Each calculational model or numerical method used shall be verified and validated through comparison to test results or comparison to other acceptable models."

LAR Section 4.5.1.2, "Fire FPRA" states that fire modeling was performed as part of the FPRA development (NFPA 805 Section 4.2.4.2). Reference is made to LAR Attachment J, "Fire Modeling V&V," for a discussion of the verification and validation (V&V) of the fire models that were used.

Furthermore, LAR Section 4.7.3 "Compliance with Quality Requirements in Section 2.7.3 of NFPA 805" states "Calculational models and numerical methods used in support of compliance with 10 CFR 50.48(c) were verified and validated as required by Section 2.7.3.2 of NFPA 805."

Regarding the V&V of fire models:

a) It is stated on page J-3 of LAR Attachment J that CFAST does not use a fire diameter, therefore, it is possible to specify a fire that falls within the range of Froude numbers considered in the NUREG-1824 validation documentation.

Provide confirmation that this is true for all the CFAST model calculations or justify why CFAST can be used for Froude numbers outside the validated range.

b) It is stated on page J-3 of LAR Attachment J that [The] flame length ratio is normally met, but in the case of the largest fire sizes postulated, the flame height may reach or exceed the ceiling height. Because sprinkler actuation and thermal radiation to targets are not computed with the CFAST model, this parameter is not an applicable metric.

Provide a technical justification for the use of CFAST to model fires with flames that impinge on the ceiling.

c) During the audit, the licensee stated that a CFAST analysis was performed to evaluate the temperature rise of an exposed structural steel target from a defined lube oil pool fire source on the 590-ft elevation of the turbine building.

Provide a V&V basis for the use of CFAST for this application (calculation of structural steel target temperature) and include its reference in Attachment J of the LAR.

d) During the audit, the licensee discussed the use of CFAST to calculate the maximum possible fire size in an upper cubicle of switchgear cabinets in Switchgear Rooms 1-C and 1-D. The licensee also stated on page J-14 of LAR Attachment J that Appendix A of the 1-C Switchgear Room (FA4) [Hughes Page 25 of 227

Associates, Rev. 0, 2012a] report documents the validation basis for CFAST as applied in the 1-C Switchgear Room. Essentially, CFAST is used as a one-zone model to provide a mass and energy balance over a control volume and a bounding empirical model based on the equivalence ratio is used to determine the maximum heat release rate at a vent. This is the most limited use of CFAST and the application to the switchgear cubicles relies on the verification of the vent mass flows and the energy balance as provided in NIST-SP 1086 [2008]. Because the model is used in the most simplistic way possible, it is considered to be applied within in its validation basis.

Provide the validation basis for using the single zone approximation in CFAST to simulate fires in the upper cubicles of switchgear panels at PNP.

ENO Response a) The zone computer model permits the selection of two plume models: McCaffrey and Heskestad. The McCaffrey plume is selected in all PNP CFAST modeling since this is the plume model that was used in the NUREG-1824, Volume 5 [1]

validation of CFAST. When the McCaffrey plume is selected, the zone computer model CFAST does not use a fire diameter; thus, the determination of the appropriate fire Froude Number is based on the application of the CFAST results rather than on the fire model inputs. The fire scenarios evaluated in the PNP main control room abandonment calculation [2] using CFAST involve electrical panels and transient ignition sources that are typical of nuclear power plants and are consistent with the types of fire scenarios considered in the NUREG-1824, Volumes 1 and 5 V&V effort [1, 3]. The application of the fire modeling results in ignition sources that fall within the NUREG/CR-6850 [4] conditional probability distribution for transient and electrical panel ignition sources and are thus considered to be typical of those source fires used in NUREG-1824, Volumes 1 and 5 [1, 3] to validate the CFAST fire model.

Electrical panel and transient ignition source fire scenarios are also evaluated in various PNP plant areas using CFAST as contained in the Generic Fire Modeling Treatments report for hot gas layer development. Additional hot gas layer development calculations have been generated for the same types of ignition sources as well as electric motor and pump ignition sources as part of the response to FM RAI 01f and FM RAI 01g, which focuses on combination fixed ignition source - cable tray fire scenarios. The electrical panel, electric motor, pump, and transient ignition source fires evaluated for hot gas layer development are typical of nuclear power plants and comparable to the types of fire scenarios envisioned in the NUREG-1824, Volumes 1 and 5 V&V effort. Cable tray fires are also typical of nuclear power plants but were not specifically examined in the NUREG-1824, Volumes 1 and 5 V&V [1, 3] effort. However, it may be shown that the treatment in CFAST yields a conservative result relative to the output parameters that are applied. Additional discussion for each ignition source class is provided in the sub-sections that follow. It is also noted the updated control room abandonment calculation includes the applicable fire Froude Number discussion Page 26 of 227

provided in the sub-sections that follow.

Closed Electrical Panels There is no simple or obvious way to compute a meaningful fire Froude Number for closed electrical panels (i.e., NUREG/CR-6850 Appendix E, Cases 1, 2, 3, and 4 [4]). This is because the combustion primarily occurs within the panel and the transfer of heat and mass to the surrounding enclosure occurs across the panel vents and any gaps that may exist or form during the fire. The current method for evaluating closed electrical panel fires per NUREG/CR-6850 [4] and NUREG/CR-6850, Supplement 1 [6] is to assume an open configuration source fire with a base height equal to the panel height or 0.3 m (1 ft) below the panel top, depending on the panel configuration. This is a conservative alternative to modeling the fire conditions within the panel and the mass and energy flows between the panel and the surroundings. When using this method to bound the mass and energy transfer across the panel boundaries and thus into the thermal plume, it is assumed that the open configuration is such that the fire diameter produces a fire Froude Number within the NUREG-1824, Volume 1 [3] validation range. Essentially, the method for modeling closed electrical panel fires is to treat them as an open source fire that has a fire Froude Number that falls within the range considered by NUREG-1824, Volume 1 [3].

Transient Ignition Sources Transient ignition sources (NUREG/CR-6850, Appendix E Case 8 [4]) are located in the main control room as well as in other areas of the plant. The analysis in the main control room [2] considers the entire transient ignition source conditional probability distribution; whereas other areas consider only the 98th percentile heat release rate bin and a reduced heat release rate 69 kW (65 Btu/s) heat release rate bin.

The fire Froude Number for a transient fuel package fire may be computed using the following equation per NUREG-1934 [7]:

Fr (FM 03a-1) where Fr is the fire Froude Number, is the fire heat release rate modeled (kW

[Btu/s]), is the density of the ambient air (kg/m³ [lb/ft³]), is the heat capacity of the ambient air (kJ/kg-K [Btu/lb-°R]), is the ambient air temperature (K [°R]),

is the acceleration of gravity (9.81 m/s² [32.2 ft/s²]), and is the diameter of the fire (m [ft]).

The density is inversely proportional to the temperature via the following equation:

(FM 03a-2)

Page 27 of 227

where is the density (kg/m³ [lb/ft³]) and is the temperature (K [°R]). In addition, the heat capacity is nearly constant over the temperature ranges applicable to the target exposure, equal to 1 kJ/kg-K (0.24 Btu/lb-°R). Consequently, Equation FM 03a-1 may be simplified to the following:

Fr . (FM 03a-3) where all terms have been defined. In order to define a diameter for use in Equation FM 03a-3, a reasonable approximation of the area involved is necessary. When the heat release rate per unit area is known (rather than the actual fire area or diameter), the approximate fire area may be computed using the following equation:

(FM 03a-4) where is the plan burning area of the open panel ignition source (m² [ft²]), is the fire heat release rate modeled (kW [Btu/s]), and " is the heat release rate per unit area of the burning material (kW/m² [Btu/s-ft²]). The effective fire diameter may be computed assuming an axisymmetric source:

(FM 03a-5) where is the effective fire diameter (m [ft]) and A is the plan burning area (m²

[ft²]) for use in Equation FM 03a-3.

The plan heat release rate per unit area range for the transient fuels varies considerably as described in Section 3-1 of the SFPE Handbook of Fire Protection Engineering - 3rd Edition [8] given the large variation in the types and arrangement of the fuel packages. The heat release rate per unit area range is about 100 - 370 kW/m² (8.8 - 32.6 Btu/s) for transient materials that are loose or located in containers, based on the test considered in NUREG/CR-6850 [4], provided the material does not contain flammable or combustible liquids. The heat release rate per unit area for loose material alone is closer to 270 - 370 kW/m² (23.8 - 32.6 Btu/s-ft²) based on a sub-set of tests involving trash bags [9]. The loose material tests are applicable in the main control room abandonment calculation [2] and the hot gas layer evaluations such as those documented in the Generic Fire Modeling Treatments report [5], since explicit credit for the slower fire growth in a container is not credited. Contained transient fire scenarios, which credit the slower fire growth rate, are primarily applicable to the main control room abandonment calculation.

The approximate fire Froude Number for the fifteen bins listed in NUREG/CR-6850, Appendix E Case 8 [4] for transient ignition source fires is listed in Table FM 03a-1 for loose transient fuel packages having a plan heat release rate per unit area of 270 kW/m² (23.8 Btu/s-ft²) and Table FM 03a-2 for loose transient fuel Page 28 of 227

packages having a plan heat release rate per unit area of 370 kW/m² (32.6 Btu/s-ft²).

Table FM 03a Approximate Fire Froude Number for NUREG/CR-6850, Appendix E [4] Case 8 (Transient Fires) - 270 kW/m² (23.8 Btu/s-ft²) Plan Heat Release Rate Per Unit Area (Loose Transient Fuel Package)

NUREG/CR- Heat Release Area Diameter Fire Froude 6850 Heat Rate (m² [ft²]) (m [ft]) Number Release Rate (kW [Btu/s])

Bin 1 22 (21) 0.081 (0.88) 0.32 (1.06) 0.34 2 55 (52) 0.20 (2.2) 0.51 (1.67) 0.27 3 92 (87) 0.34 (3.7) 0.66 (2.16) 0.24 4 128 (121) 0.47 (5.1) 0.78 (2.55) 0.22 5 165 (156) 0.61 (6.6) 0.88 (2.89) 0.20 6 202 (191) 0.75 (8.0) 0.98 (3.20) 0.19 7 238 (226) 0.88 (9.5) 1.06 (3.47) 0.19 8 275 (261) 1.02 (11.0) 1.14 (3.73) 0.18 9 312 (296) 1.16 (12.4) 1.22 (3.98) 0.17 10 349 (331) 1.29 (13.9) 1.28 (4.21) 0.17 11 386 (366) 1.43 (15.4) 1.35 (4.43) 0.17 12 423 (401) 1.57 (16.9) 1.41 (4.63) 0.16 13 460 (436) 1.70 (18.3) 1.47 (4.83) 0.16 14 497 (471) 1.84 (19.8) 1.53 (5.02) 0.16 15 578 (548) 2.14 (23.0) 1.65 (5.42) 0.15 Bold values indicate a fire Froude Number that falls below the NUREG-1824, Volume 1 validation range of 0.4 - 2.4.

Page 29 of 227

Table FM 03a Approximate Fire Froude Number for NUREG/CR-6850, Appendix E [4] Case 8 (Transient Fires) - 370 kW/m² (32.6 Btu/s-ft²) Plan Heat Release Rate Per Unit Area (Loose Transient Fuel Package)

NUREG/CR- Heat Release Area Diameter Fire Froude 6850 Heat Rate (m² [ft²]) (m [ft]) Number Release Rate (kW [Btu/s])

Bin 1 22 (21) 0.059 (0.64) 0.28 (0.90) 0.50 2 55 (52) 0.15 (1.60) 0.44 (1.43) 0.40 3 92 (87) 0.25 (2.67) 0.56 (1.85) 0.35 4 128 (121) 0.35 (3.72) 0.66 (2.18) 0.32 5 165 (156) 0.45 (4.80) 0.75 (2.47) 0.30 6 202 (191) 0.55 (5.87) 0.83 (2.73) 0.29 7 238 (226) 0.64 (6.92) 0.90 (2.97) 0.28 8 275 (261) 0.74 (8.00) 0.97 (3.19) 0.27 9 312 (296) 0.84 (9.07) 1.04 (3.40) 0.26 10 349 (331) 0.94 (10.1) 1.10 (3.59) 0.25 11 386 (366) 1.04 (11.2) 1.15 (3.78) 0.25 12 423 (401) 1.14 (12.3) 1.21 (3.96) 0.24 13 460 (436) 1.24 (13.4) 1.26 (4.13) 0.24 14 497 (471) 1.34 (14.5) 1.31 (4.29) 0.23 15 578 (548) 1.56 (16.8) 1.41 (4.63) 0.22

Bold values indicate a fire Froude Number that falls below the NUREG-1824, Volume 1 validation range of 0.4 - 2.4.

Similarly, the approximate fire Froude Number for the fifteen bins listed in NUREG/CR-6850, Appendix E [4] Case 8 for transient ignition source fires is listed in Table FM 03a-3 for contained transient fuel packages. The fire diameters for the contained transient fire scenarios are based on the typical diameters for contained transient fuel packages considered in NUREG/CR-6850 Appendix E [4], which range from under 0.30 m (0.98 ft) to about 0.69 m (2.26 ft) [9].

Page 30 of 227

Table FM 03a Approximate Fire Froude Number for NUREG/CR-6850, Appendix E [4] Case 8 (Transient Fires) - Contained Transient Fuel Package.

NUREG/CR- Heat Release Fire Froude Number 6850 Heat Rate Small Diameter (0.30 Large Diameter Release Rate (kW [Btu/s]) m [0.98 ft]) Container (0.69 m [2.26 ft)]

Bin Container 1 22 (21) 0.39 0.05 2 55 (52) 0.97 0.13 3 92 (87) 1.63 0.21 4 128 (121) 2.26 0.30 5 165 (156) 2.92 0.38 6 202 (191) 3.57 0.47 7 238 (226) 4.21 0.55 8 275 (261) 4.87 0.64 9 312 (296) 5.52 0.72 10 349 (331) 6.17 0.81 11 386 (366) 6.83 0.89 12 423 (401) 7.48 0.98 13 460 (436) 8.14 1.06 14 497 (471) 8.79 1.15 15 578 (548) 10.23 1.34 Bold values indicate a fire Froude Number that outside the NUREG-1824, Volume 1 validation range of 0.4 - 2.4.

Tables FM 03a-1 through FM 03a-3 indicate that the fire Froude Number falls below the NUREG-1824, Volume 1 [3] fire Froude Number range of 0.4 - 2.4 in nearly all cases for the loose configuration and the low bins for the large diameter contained configuration. The fire Froude Number is high for the high bins associated with the small diameter contained configuration. In the case of the confined configuration, based on the data provided in NUREG/CR-6850 [4], the small diameter containers (0.3 m [1 ft] diameter) produce heat release rates that correspond to Bin 1, whereas the larger diameter containers produce heat release rates that can range all heat release rate bins. This means that, in practice, the characterization of all transient fire scenarios at PNP either results in a fire Froude Number that is within the NUREG-1824, Volume 1 [3] validation range or falls below the NUREG-1824, Volume 1 [3] validation range. When this occurs, the thermal plume that is expected from the ignition source fire could be wider than the range evaluated NUREG-1824, Volume 1 [3]. A wider thermal plume will have a greater entrainment rate than one associated with a similar heat release rate fire that has a smaller diameter. This means that the conditions relative to a source fire that falls within the validation range will be less severe both in terms of the concentration of combustion products and the temperature. Conversely, the hot gas layer descent time will be faster than a case that falls within the NUREG-1824, Volume 1 [3] validation range for the Fire Froude Number. In the case of the hot gas layer evaluations such as those provided in the Generic Fire Modeling Treatments report [5], the position of the hot gas layer is not a factor in determining the potential for target damage, so it may be asserted that conditions Page 31 of 227

associated with low fire Froude Number scenarios that may arise among various transient ignition source configurations are bound by the calculation results. In the case of the main control room abandonment calculation [2], a low fire Froude Number may yield non-conservative results if the hot gas layer descent time is the limiting constraint, the abandonment condition is based on visibility, and the visibility is significantly greater than the abandonment threshold at the time the hot gas layer height reaches the threshold value. A review of the temporal plots provided in the control room abandonment calculation [2] indicates that these conditions are generally not met for any scenario considered. In addition, a revised discussion on the fire Froude Number has been provided in the updated main control room abandonment calculation.

A number of conservative factors that would tend to increase the fire Froude Number (toward the range validated in NUREG-1824, Volumes 1 and 5 [1, 3]) for the transients are not explicitly accounted for in the approximate calculation presented in Tables FM 03a-1 through FM 03a-3. These include the potential for the transient material to be consolidated or contained (causing most fire Froude Numbers to fall within the NUREG-1824, Volume 1 [3] validation range) and for contained fires to have heat release rates that fall within intermediate heat release rate bins. Consequently, the application of the CFAST fire modeling results to transient fuel package fire scenarios at PNP is considered to either fall within the NUREG-1824, Volume 1 [3] validation range for the fire Froude Number or produce results that are more conservative than a comparable case that falls within the NUREG-1824, Volume 1 [3] fire Froude Number validation range. Note that additional discussion on the Froude Number validation range has been provided in the updated main control room abandonment calculation [2].

Electric Motors A NUREG/CR-6850, Appendix E, Case 7 [4] 98th percentile generic motor fire scenario is evaluated for hot gas layer effects at PNP. The characteristics are as follows:

Fire diameter: 0.61 m (2.0 ft); and Peak heat release rate: 69 kW (65 Btu/s).

The fire Froude Number as computed using equation FM 03a-3 is 0.21, which is lower than the NUREG-1824, Volume 1 [3] validation range. As is the case for the transient fuel package fires, when the hot gas layer temperature is the only output parameter of interest regardless of the hot gas layer position, it may be asserted that conditions associated with low fire Froude Number scenarios that may arise among various transient ignition source configurations are bound by the calculation results. Equation FM03a-3 may be used to determine the minimum fire diameter for an electric motor fire for which the hot gas layer results are conservative. This minimum fire diameter is 0.23 m (0.75 ft), which is significantly smaller than the types of electric motors at PNP for which fire scenarios are Page 32 of 227

postulated and indicates that the PNP applications are conservative.

Pumps A NUREG/CR-6850, Appendix E, Case 6 [4] 98th percentile generic pump fire scenario is evaluated for hot gas layer effects at PNP. The characteristics are as follows:

Fire diameter: 0.91 m (3.0 ft); and Peak heat release rate: 211 kW (200 Btu/s).

The fire Froude Number as computed using equation FM 03a-3 is 0.23, which is lower than the NUREG-1824, Volume 1 [3] validation range. As is the case for the transient fuel package fires, when the hot gas layer temperature is the only output parameter of interest regardless of the hot gas layer position, it may be asserted that conditions associated with low fire Froude Number scenarios that may arise among various transient ignition source configurations are bound by the calculation results. Equation FM03a-3 may be used to determine the minimum fire diameter for an electric motor fire for which the hot gas layer results are conservative. This minimum fire diameter is 0.35 m (1.1 ft), which is significantly smaller than the types of electric motors at PNP for which fire scenarios are postulated and indicates that the PNP applications are conservative.

Cable Trays The last type of source fire for which CFAST is used to calculate the compartment conditions at PNP involves combination fixed ignition source - cable tray fires (as secondary combustibles). The fire Froude Number for each of these scenarios is provided in 0021-0019-000-003 [10]. It is shown that the fire Froude Number for cable tray fire scenarios calculated using the cable tray width falls within the NUREG-1824 [3] validation range of 0.4 - 2.4 or is lower than the validated range.

Because the hot gas layer temperature is the only output parameter, the results are either within the validated range or are more conservative that an equivalent value within the validated range.

References:

[1] NUREG-1824, Volume 5, Verification and Validation of Selected Fire Models for Nuclear Power Plant Applications Volume 5: Consolidated Fire Growth and Transport Model, NUREG-1824 / EPRI 1011999, Salley, M. H. and Kassawara, R. P., NUREG-1824, Final Report, U.S. Nuclear Regulatory Commission, Nuclear Reactor Regulation (NRR), Washington, D. C., May, 2007.

[2] Report 0021-0019-000-001, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, October, 2013.

Page 33 of 227

[3] NUREG-1824, Volume 1, Verification and Validation of Selected Fire Models for Nuclear Power Plant Applications Volume 1: Main Report, NUREG-1824

/ EPRI 1011999, Salley, M. H. and Kassawara, R. P., NUREG-1824, Final Report, NRR, NRC, Washington, D. C., May, 2007.

[4] NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, Electric Power Research Institute (EPRI) 1011989 Final Report, Nuclear Regulatory Commission (NRC), Rockville, MD, September, 2005.

[5] 1SPH02902.030, Rev. 0, Generic Fire Modeling Treatments, Hughes Associates, Baltimore, MD, January, 2008.

[6] NUREG/CR-6850, Supplement 1, Fire Probabilistic Risk Assessment Methods Enhancements, EPRI 1019259, Technical Report, NRC, Rockville, MD, September, 2010.

[7] NUREG-1934, Nuclear Power Plant Fire Modeling Application Guide, Salley, M. H. and Kassawara, R. P., NUREG-1934/(EPRI-1019195, U.S.

NRC, Office of Nuclear Reactor Research, Washington, D. C., November, 2012.

[8] Babrauskas, V., Heat Release Rates, Section 3-1, SFPE Handbook of Fire Protection Engineering, 4th Edition, Society of Fire Protection Engineers (SFPE), Bethesda, MD, 2008.

[9] 0021-0019-000-004, Rev. 0, Supplemental Generic Fire Modeling Treatments: Transient Fuel Package Ignition Source Characteristics, Hughes Associates, Inc., Baltimore, MD, 2013.

[10] 0021-0019-000-003, Rev. 0, Evaluation of the Development of Hot Gas Layer Conditions in Generic Palisades Fire Compartments with Secondary Combustibles, Hughes Associates, Inc., Baltimore, MD, 2013.

b) The flame height to ceiling height ratio is a measure of the degree to which flames impinge on the ceiling surface. Flame impingement on a ceiling surface can affect the predictions of the ceiling jet temperature, the heat transfer to the ceiling surface, and the radiant heat flux at a specific target location. All three of these model output parameters are not used in the CFAST models developed for the PNP fire PRA.

The key model parameters used in the CFAST models developed for the PNP fire PRA are the hot gas layer temperature, and, in the case of the control room abandonment calculation [1, 2], the hot gas layer temperature, its height and its composition. The hot gas layer composition and the hot gas layer height are primarily functions of the fuel properties, the entrainment into the fire plume from the lower layer, and the overall mass balance among the various forced and natural ventilation flow paths as described in NIST-SP-1026 [3]. When evaluated Page 34 of 227

in a zone model, the entrainment into the hot gas layer from the lower layer is not directly affected by the flame extension under the ceiling, because the entrainment occurs from the base of the fire to the interface between the hot gas layer and the lower layer per NIST-SP-1026 [3]. Indirect effects could arise through changes in the predicted temperature and layer density, though these are secondary effects relative to the transfer of mass from the lower layer to the upper layer and through forced and natural ventilation flow paths. In addition, the hot gas layer temperature is the only parameter that is used in developing the hot gas layer tables such as those provided in the Generic Fire Modeling Treatments report [4]. Specifically, there is no hot gas layer height threshold that is used to develop the hot gas layer tables in the hot gas layer evaluations [4].

In terms of the temperature of the hot gas layer, a situation in which the flames impinge on the ceiling will result in relatively high heat fluxes from the fire to the wall boundary (see Section 2-14 of the SFPE Handbook of Fire Protection Engineering - 4th Edition [5]) as compared to areas exposed only to the hot gas layer. The CFAST models used by the PNP fire PRA do not use the heat transfer model between the ceiling jet and an adjacent space, and, therefore, conservatively bound the hot gas layer temperature relative to a case in which the additional boundary heat losses are included. Further, there are no CFAST models used by the PNP fire PRA that credit or predict the detection actuation time using the ceiling jet models in CFAST or predict the target heat flux or temperature response due to flame radiation. In addition, the floor surfaces are adiabatic, so that flame radiation from the ceiling flames to the floor is conservatively retained by the hot gas layer. Although CFAST does perform the ceiling jet computation, the results are not coupled with the output data that is used at PNP. This applies to the hot gas layer calculations such as those provided in the Generic Fire Modeling Treatments report [4] and the control room abandonment calculation [1, 2].

It is noted that the updated main control room abandonment report includes the applicable fire discussion on the flame length to ceiling height ratio provided in this RAI response [2].

References:

[1] Report 1SPH02902.066, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, September, 2009.

[2] Report 0021-0019-000-001, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, October, 2013.

[3] NIST-SP-1026, CFAST - Consolidated Model of Fire Growth and Smoke Transport (Version 6) Technical Reference Guide, Jones, W. W., Peacock, R.

D., Forney, G. P., and Reneke, P. A., NIST, Gaithersburg, MD, April, 2009.

Page 35 of 227

[4] 1SPH02902.030, Rev. 0, Generic Fire Modeling Treatments, Hughes Associates, Baltimore, MD, January, 2008.

[5] Lattimer, B. Y. , V., Heat Fluxes from Fires to Surfaces, Section 2-14, SFPE Handbook of Fire Protection Engineering, 4th Edition, Society of Fire Protection Engineers (SFPE), Bethesda, MD, 2008.

c) In lieu of providing a V&V of CFAST for this fire scenario, an additional analysis [2]

based on algebraic models is used to support the conclusion that a postulated Main Feedwater Pump oil fire would not result in failure of the target steel column.

NUREG-1805 FDTs heat flux models are used to estimate the radiant heat flux to the steel column from the oil fire. NUREG-1824 provides verification and validation for the FDTs heat flux models. NUREG-1934 provides guidelines using normalized parameters to ensure the models are used within their validation.

These models are also discussed in Section 3.3 of the Generic Fire Modeling Treatments and are included in LAR Attachment J, Table J-2.

The normalized parameters were used to validate the use of the FDTs models to calculate radiative heat flux for the Main Feedwater Pump oil fire scenario. The following parameters were not applicable or calculated outside the valid range and are further discussed below:

Ceiling Jet Radius Relative to the Ceiling Height: Per NUREG-1934, the ceiling jet ratio is applicable primarily when sprinkler or detector activation is of interest. In this scenario, the target is the steel column. The ceiling height is greater than the horizontal target distance from the centerline of the fire; therefore, the ceiling jet effects on the target are not a concern.

Equivalence Ratio Based on Opening Area: The open configuration of the Turbine Building makes the equivalency calculation challenging because of the natural and mechanical ventilation. Instead, a calculation was performed to determine if enough oxygen is available to support the lube oil spill volume and is consistent with NUREG-1934 Appendix F. Based on the calculation results it was concluded that there was enough oxygen in the Turbine Building to support the fire.

Target Distance to Fire Diameter: The target close proximity to the assumed edge of the pool fire results in a low radial distance ratio. There is uncertainty in the lube oil spill location, spill volume, pool depth, and spill area given the Turbine Building drains in the Main Feedwater Pump area.

Therefore, the pool depth and spill area are varied based on the spill volume.

The NUREG-1805 FDTs point source model and the solid flame model (i.e.,

detailed method of Shorki and Beyler) are used to calculate the radiative heat flux. LAR Attachment J details the extended verification and validation range of these models provided by the Society of Fire Protection Engineers [6].

The conclusion was that the point source model is conservative when the Page 36 of 227

predicted heat flux is less than 5 kW/m2 and the Method of Shokri and Beyler is conservative when the predicted heat flux is greater than 5 kW/m2. Using these two methods, the radiative heat flux to the target steel column is calculated and the most conservative is used.

The temperature rise is then calculated using an energy balance equation consistent with that used in the lube oil fire scenario in NUREG-1934 Appendix F.

Equation 1 is Equation F-8 from NUREG-1934 solved for the maximum column temperature.

T =(t*q)/(cs*(W/D)) + T0 Equation 1 where T is the maximum column temperature, t is the fire duration (sec.), q is the heat flux (kW/m2), cs is the specific heat (kJ/kg/°C), W/D is the column W/D ratio (kg/m2), and T0 is the ambient temperature.

Equation 2 is used to calculate the fire duration (t). Equation 2 is Equation F-2 from NUREG-1934.

t = */m Equation 2 where is the pool depth (m), is the density (kg/m3), and m is the burning rate (kg/m2/s)

The NUREG-1805 FDTs radiative heat flux models have verification and validation provided in NUREG-1824. The guidelines in NUREG-1934 were used to ensure the models were valid given the applicable parameters for the Main Feedwater Pump oil fire scenario. These calculations provided the heat flux input to Equation

1. Equation 2 provided the fire duration input to Equation 1. Given these input and the fire input parameters, Equation 1 was used to calculate the target steel column temperatures for the Main Feedwater Pump oil fire scenario. This calculation supports the conclusion that the target steel column will not fail because the maximum steel temperature was below the failure temperature of 538°C (1,000°F).

The response to this RAI is based on two radiative heat flux correlations; the point source model and the solid flame model (i.e., Shokri and Beyler). These correlations were discussed in Section 3.3 of the Generic Fire Modeling Treatments. These correlations are included in LAR Attachment J, Table J-2.

Therefore, an update to Table J-2 is not required for this RAI response.

References:

[1] NUREG-1805 Fire Dynamics Tools (FDTS): Quantitative Fire Hazard Analysis Methods for the U.S. Nuclear Regulatory Commission Fire Protection Inspection Program Supplement 1.

[2] Calculation 0247-07-0005.08, Rev. 2, Palisades Nuclear Plant Fire Probabilistic Risk Assessment Exposed Structural Steel Analysis, Entergy, Page 37 of 227

November, 2013.

[3] NUREG-1934 Nuclear Power Plant Fire Modeling Analysis Guidelines (NPP Fire MAG)

[4] NUREG-1824 Verification & Validation of Selected Fire Models for Nuclear Power Plant Applications

[5] Email from Alan Lyon at Entergy W/D fraction of the structural steel at Palisades 10/17/13

[6] SFPE Handbook of Fire Protection Engineering, Fourth Edition.

d) The CFAST analysis of the upper switchgear cubicle is a preliminary assessment of the types of fire that should be input into the FDS model. In particular, the maximum internal heat release rate is estimated and this is used as a means of estimating the maximum exterior heat release rate given the known upper limit on the equivalence ratio within an enclosure. The selection of a single zone fire model for assessing the largest size vent fire at a single switchgear cubicle was made using insights obtained from a version of Kawagoes equation for fully developed fires as applied to the upper cubicle vent configuration [1]:

1500 (FM 03d-1) where is the fire size within a fully developed enclosure (kW), is the opening area (m²), and is the opening height (m). The dimensions of an upper switchgear cubicle are as follows:

Length: 0.66 m (2.16 ft)

Width: 2.02 m (6.61 ft)

Average height: 0.6 m (1.98 ft)

In addition, there are two vents one near the top that has a flow area of about 0.02 m² (0.21 ft²) and one near the base that has a flow area of about 0.023 m² (0.25 ft²). The enclosure and vent dimension is not typical of a room enclosure for which CFAST is normally applied. Further, the objective of the analysis provided in Reports 1JMW20053.000-1, Rev. 0 and 1JMW20053.000-2, Rev. 0 is to identify the largest fire size that can be supported within the enclosure by the vents. This implies a fully developed fire condition (with an internal equivalence ratio approximately equal to unity) for which the Kawagoe equation is applicable. As reported in 1JMW20053.000-1, Rev. 0 and 1JMW20053.000-2, Rev. 0, the vents under these conditions are expected to support a fire on the order of 20 - 34 kW (19 - 32 Btu/s) as evaluated using equation FM 03a-1. Although the Kawagoe equation was originally developed for standard room enclosures, data is available that provides validation and basis for small enclosures:

Page 38 of 227

Thomas and Heselden [2] summarize the results of about four hundred experiments conducted in small enclosures having a base scale of 0.5 m (1.6 ft), 1.0 m (3 ft), and 1.5 m (5 ft) in various shapes and with different opening fractions. The application of the opening factor is used to predict burning rate, which is proportional to the heat release rate.

Utiskul [3] investigated the applicability of one-zone correlations, including the Kawagoe equation, to fires within relatively small enclosure (0.4 0.4 1.2 m [1.3 1.3 4 ft]) having various ventilation factors and source fire fuels. Utiskul [3] concluded that a one-zone model, or, equivalently a well-mixed model assumption, is applicable if flashover occurs (i.e.,

temperatures approach 450 - 500°C [842 - 932°F]), the fuel area is large compared to the compartment floor area, or the hot gas layer height is low

(~twenty percent of the enclosure height or lower).

The types of fires postulated in the upper switchgear cubicle are intended to represent fully-developed enclosure fire scenarios because an equivalence ratio of unity is sought, which represents the transition point to a ventilation controlled fire, a low hot gas layer height, and the onset of flashover. The results associated with CFAST scenarios that do not result in a fully-developed condition within the switchgear cubicle (i.e., equivalence ratio less than unity) are not explicitly used in the subsequent evaluation of the switchgear panel fires in Fire Areas 3 and 4 (Reports 1JMW20053.000-1, Rev. 0 and 1JMW20053.000-2, Rev. 0).

Furthermore, transient effects during the fire growth in which two-zone conditions may occur are not quantified in this type of analysis and do not affect the subsequent use of the CFAST model results. Because the test enclosures and ventilation factors considered by Thomas et al. [2] and Utiskil [3] include cases that are comparable to the upper switchgear cubicle, they provide a validation basis for the one-zone approach as applied to the upper switchgear cubicle or, more specifically, the use of the Kawagoe correlation to predict the fire size within the cubicle.

The intent of evaluating the upper switchgear cubicles with CFAST under a one-zone assumption in Reports 1JMW20053.000-1, Rev. 0 and 1JMW20053.000-2, Rev. 0 was to provide a more realistic assessment of the actual vent configurations. The use of CFAST as a one-zone model reduces the model calculation to a mass and energy balance across a control volume. In this case, the control volume is the upper switchgear cubicle and the mass flows are driven by thermally induced pressure differentials across the cubicle boundary. Using CFAST in this way, the maximum fire size that could be supported within the enclosure was found to be 37 kW (35 Btu/s), within ten percent of the 34 kW (32 Btu/s) predicted using the Kawagoe equation. Given the CFAST results are within ten percent of the value predicted by the Kawagoe equation, but slightly more conservative, the Kawagoe equation provides the validation basis for the one-zone CFAST application to the upper switchgear cubicle and the Thomas et al. and Utiskul [3] test data provide the validation basis for the Kawagoe equation as Page 39 of 227

applied to small enclosures.

References:

[1] Walton, W. D. and Thomas, P. H., Estimating Temperatures in Compartment Fires, Section 3-6, SFPE Handbook of Fire Protection Engineering, 4th Edition, Society of Fire Protection Engineers (SFPE), Bethesda, MD, 2008.

[2] Thomas, P. H. and Heselden, A. J. M., CIB Report No. 20, Fire Research Note 923/197, 1972.

[3] Utiskul, Y., Theoretical and Experimental Study of Fully-Developed Compartment Fires, University of Maryland Ph. D. Dissertation, 2006.

NRC Request FM RAI 04 NFPA 805, Section 2.7.3.3, "Limitations of Use," states: "Acceptable engineering methods and numerical models shall only be used for applications to the extent these methods have been subject to verification and validation. These engineering methods shall only be applied within the scope, limitations, and assumptions prescribed for that method."

Section 4.7.3, "Compliance with Quality Requirements in Section 2.7.3 of NFPA 805," of the Transition Report states that "Engineering methods and numerical models used in support of compliance with 10 CFR 50.48(c) were applied appropriately as required by Section 2.7.3.3 of NFPA 805."

Regarding the limitations of use:

a) Identify uses, if any, of the GFMTs (including the supplements), CFAST, and FDS outside the limits of applicability of the method and justify how the use of these fire modeling approaches were appropriate.

ENO Response a) There are three general categories in which fire models are applied at PNP: the application of the Generic Fire Modeling Treatments (GFMT) approach [1, 2, 3, 4, 5], the calculation of the abandonment times in the main control room [6], the detailed fire modeling evaluations of fire scenarios in Fire Area (FA) 3 and FA4 [7, 8], and the CFAST evaluation of the structural steel.

Listed below are the fire models used outside the model limitations for each category and their justification for use.

1. Generic Fire Modeling Treatments Approach (ZOI)

Page 40 of 227

There are six basic limitations that apply to the GFMT approach when determining a ZOI. The six limitations represent conditions or configurations for which the Generic Fire Modeling Treatment ZOI data may potentially be non-conservative if applied outside the particular limitation:

The application of the generic ZOI data in compartments in which the hot gas layer temperature exceeds 80°C (176°F)

The application of the generic ZOI data to fire scenarios in wall and corner configurations The application of the generic ZOI data for panel ignition sources with panels having plan dimensions greater than 0.9 0.6 m (3 2 ft)

The application of the generic ZOI data to scenarios that result in flame impingement to the ceiling The application of the generic hot gas layer data to configurations in which secondary combustibles (cable trays) are ignited Application of the GFMT CFAST fire modeling results Each of the six limitations as applicable to the Palisade fire PRA implementation are described below and justification is provided where a limitation is not met. Note that Supplement 2 to the Generic Fire Modeling Treatments report (PSA-ANO1-03-FM-03, Rev. 0) was developed to address a number of these limitations under various circumstances that arise at PNP; however, this document has been replaced with Report 0021-0019-000-002, Rev. 0 [4] for ZOI tables and Report 0021-0019-000-003, Rev. 0 [5] for HGL tables when secondary combustibles are involved. Report 0021-0019-000-003, Rev. 0 [4] and Report 0021-0019-000-003, Rev. 0 [5] were developed specifically to address limitations in the original GFMT approach with regard to secondary combustible materials.

ZOIs in Elevated Temperature Enclosures The revised fire scenarios will limit the use of ZOIs to situations in which the maximum HGL temperature is 80°C (176°F). If the temperature exceeds 80°C (176°F), a full room burnout condition will be conservatively assumed.

ZOIs in Wall and Corner Locations Transient ignition source fire scenarios are postulated in open, wall, and corner configurations. Although the original GFMT report is limited to open configurations, wall and corner effects are determined using the Image method, which postulates an equivalent plume by increasing the fire size and enclosure volume by a factor of two or four depending on the fire location [9]. Wall and corner effects for fixed ignition sources that are located within 0.61 m (2 ft) of a wall boundary will be included in the updated fire scenarios report. Specific ZOI and HGL data are Page 41 of 227

provided in the updated fire modeling documents [3, 4, 5] and will be used to determine the target sets for the applicable ignition sources.

ZOIs for Large Dimension Electrical Panels The original Generic Fire Modeling Treatment report and Supplement 2 to the Generic Fire Modeling Treatment report ZOI data was derived for panels having plan dimensions up to 0.9 0.6 m (3 2 ft). The dimensions primarily affect the extent of the horizontal component of the ZOI that is below the top of the panel.

This ZOI component is calculated from an energy balance at the panel surface, and the target exposure mechanism is a heated radiating vertical plane.

Consequently, changes in the panel dimensions affect the dimensions of the radiating plane, which in turn affects the geometry configuration factor between the target and the radiating plane. The lower horizontal ZOI dimension is the limiting horizontal ZOI dimension and is used in the fire PRA as the basis for determining the affected target set.

An approximate upper limit for the ZOI dimensions based on the conservative 0.9 0.6 m (3 2 ft) plan dimensions may be estimated by comparing against a limiting open panel configuration. In this case, the maximum heat transferred across one boundary would be given through the definition of the emissive power and a radiation area as follows:

, (FM 04-1) where , is the maximum heat that can be transferred across a vertical boundary of an electrical panel (kW [Btu/s]), is the area of the boundary (m²

[ft²]), and is the flame emissive power (kW/m² [Btu/s-ft²]). Assuming the maximum average flame emissive power over the panel boundary is 120 kW/m² (10.6 Btu/s-ft²) based on Section 3-10 of the SFPE Handbook of Fire Protection Engineering [10] and data provided in Combustion and Flame, No. 139, pp. 263-277 [11], the maximum heat that could be transferred across a vertical boundary via thermal radiation is about 235 kW (227 Btu/s) if the heat transferred across an open boundary is considered to be an upper limit on the boundary heat losses in any one direction. To link this heat loss to the postulated fire size, the radiant fraction is used, which is reasonably approximated as 0.3 for enclosure fires per Section 3-8 of the SFPE Handbook of Fire Protection Engineering [12]. Dividing the maximum boundary heat loss of 235 kW (223 Btu/s) by the radiant fraction (0.3) results in the largest fire size for which the lateral ZOI dimensions would be conservative, or 783 kW (742 Btu/s). This value exceeds the severe fire heat release rate used to characterize both the multiple bundle (717 kW [680 Btu/s]

based on the Bin 8 heat release rate) and single bundle (211 kW [200 Btu/s])

electrical panels. This result is based on a radiant fraction of 0.3; if a value at the upper end of the often cited range 0.3 - 0.4 is assumed per Section 3-8 of the SFPE Handbook of Fire Protection Engineering, the largest fire size for which the lateral ZOI dimensions would be conservative, or 588 kW (557 Btu/s). However, this would be based on all heat losses being directed toward the target. The Page 42 of 227

internal temperature during a fully developed enclosure fire would be greater than 600°C (1,112°F), which suggest the heat losses from all boundaries, except the open boundary, would be on the order of 110 kW (104 Btu/s). This means that the maximum total energy that could radiate toward the target via thermal radiation would be about 600 kW (253 Btu/s) 0.4 or 240 kW (227 Btu/s). This is comparable to the maximum boundary heat loss via thermal radiation (235 kW

[223 Btu/s]), which indicates the conclusion applies over a wider range of radiant fractions when the additional boundary heat losses are included. The limiting fire size (and plan dimension for the panels) for wall and corner locations is increased by a factor of two and four due to the symmetry planes assumed in the Image method and applies when the lower ZOI dimension is limiting. There are no electrical panel ignition sources evaluated at PNP using the GFMT approach with a heat release rate greater than 783 kW (742 Btu/s) in an open configuration, 1,566 kW (1,484 Btu/s) in a wall location, or 3,132 kW (3,300 Btu/s) in a corner configuration. Therefore, although the specific limitation in the GFMT report is exceeded, there is no adverse effect on the ZOI dimensions for the PNP applications.

Flame Height Limitation for ZOIs The original Generic Fire Modeling Treatment report limits the application of the ZOIs to situations in which the flames remain lower than the ceiling height.

Subsequent analysis presented in Report 0021-0019-000-003, Rev. 0 [4] and Report 0021-0019-000-004, Rev. 0 [5] indicates that the ZOIs remain conservative provided the ceiling jet temperature at the ZOI boundary remains less severe than the threshold damage temperature for the cable target. The minimum ceiling height above the fire base is listed in Report 0021-0019-000-004, Rev. 0 for transient ignition sources and in Report 0021-0019-000-003, Rev. 0 [4] for fixed ignition sources and ignition sources with secondary combustibles. The minimum ceiling height above the fire base is generally in the 0.91 - 1.5 m (3 - 5 ft) range, but it does vary with the scenario and in cases that evolve with time, the time after ignition. The implementation of the original GFMT ZOIs at PNP accounted for low ceiling height elevations above the fire base for several fire scenarios by increasing the horizontal ZOI dimension significantly. Although the original flame height limitation specified in the GFMT report [1] or the ceiling jet limitation specified in the revised ZOI reports [4, 5] is not met, the conservative adjustment to the horizontal ZOI dimensions accounts for the effects of flame impingement and flame extension beneath the ceiling boundary. The ZOI extension will be applied to the revised ZOIs for combination ignition source - secondary combustible fire scenarios as applicable when the fire PRA is updated.

ZOIs and Hot Gas Layer Temperatures for Scenarios with Secondary Combustibles The ZOIs for configurations involving secondary combustibles have been developed using the methods described in NUREG/CR-6850 [13] and NUREG/CR-7010, Volume 1 [14]. The ZOI and HGL tables are provided in Report Page 43 of 227

0021-0019-000-002, Rev. 0 [3] and Report 0021-0019-000-003, Rev. 0 [4]. Refer to responses to RAI FM 01f and RAI FM 01g for a discussion of the treatment of fire scenarios that involve secondary combustibles.

Application of GFMT CFAST Results The Generic Fire Modeling Treatments approach involves CFAST calculations for Generic enclosures that minimize the heat losses to the boundaries [1, 4]. The key CFAST model limits that apply to the ANO2 CFAST evaluations as identified in NIST SP 1026 and NUREG-1824, Volume 5 are as follows:

Maximum vent size to enclosure volume ratio should not exceed 2 m-1 (0.61 ft-1)

Maximum enclosure aspect ratio of five (length to width)

The approach adopted in both the generic enclosure analysis is to evaluate a range of ventilation fractions, from 0.001 to 10 percent of the enclosure boundary.

Given that the width is set equal to the length in both the generic evaluations, the maximum vent size to enclosure volume ratio is given by the following equation:

(FM 04-2) where is the enclosure width (m [ft]), and is the enclosure height. Based on the definition of the generic volume, the enclosure height is one-half the enclosure width, so that the vent size to enclosure volume ratio can only exceed 2 m-1 (0.61 ft-1) if the ceiling height is 0.2 m (0.7 ft) or less. Because the minimum ceiling height considered is 1.4 m (4.5 ft), this condition is necessarily met in the tabulated data. Further, there are no spaces at PNP for which the GFMT approach is applied that has an actual ceiling height that is lower than 0.2 m (0.7 ft), thus this limitation is met in practice as well.

A second CFAST model limitation of the Generic Fire Modeling Treatments approach [1, 4] relates to the maximum aspect ratio of an enclosure for which hot gas layer data is applied. The hot gas layer information is provided for enclosures having an aspect ratio up to five, per NUREG-1824, Volume 5, Section 3.2 [15]. In situations where the model is applied to enclosures having a larger aspect ratio, the behavior transitions to a channel flow typical of a corridor configuration.

Localized effects in the vicinity of the fire could be more severe than the average conditions throughout the enclosure length, and thus a non-conservative result could be generated. NUREG-1934 [16] describes a method to apply a fire model in a conservative manner under these conditions. This method involves the modification of the enclosure dimensions such that the application falls within the model limitation and the hot gas layer temperature results are conservative. This modification will be used for spaces that have aspect ratios greater than five during the fire PRA update.

Page 44 of 227

2. Main Control Room Abandonment Calculation The key CFAST model limits that apply to the PNP control room abandonment calculation [6] as identified in NIST SP 1026 [17] and NUREG-1824, Volume 5 [15]

are as follows:

Maximum vent size to enclosure volume ratio should not exceed 2 m-1 (0.61 ft-1)

Maximum enclosure aspect ratio of five (length to width)

The maximum vent sizes considered in the control room abandonment calculation consist of a single open door and minor boundary leakage areas [6]. The total vent size to enclosure volume ratio for this opening combination remains much less 2 m-1 (0.61 ft-1) and indicates this limitation is met for all CFAST evaluations.

There are four primary spaces used to evaluate fires in the control room: the control room proper and the three sub-enclosures used to represent the control boards. The approximate aspect ratio for each of these spaces is as follows:

Control room proper: 1.01 Sub-enclosure 1: 3.55 Sub-enclosure 2: 2.50 Sub-enclosure 3: 1.11 In addition, ancillary areas that include the supervisor office, the viewing gallery, and the mechanical equipment room have a maximum aspect ratio of 2.70. These aspect ratios are less than the CFAST limit of 5 and indicate this limitation is met for all CFAST evaluations in the control room abandonment calculation.

3. Detailed Evaluations of FA3 and FA4 The detailed evaluations of fire scenarios in FA3 [8] and FA4 [7] involve the use of CFAST and FDS. The CFAST evaluations use a one-zone model feature that is outside the explicit limitations specified in NIST SP 1026 [17]. Refer to response to RAI FM 01i and RAI FM 03d for discussion on the one-zone CFAST implementations.

The FDS applications falls within the specifications for the model usage provided in NUREG-1824, Volume 7, Section 2.9, including low speed flow, rectilinear geometry, specification of fire spread, and no gas phase suppression [7, 8]. In addition, the grid cell spacing in the vicinity of the fire provides sufficient model resolution given the guidelines provided in NUREG-1934 [7, 8, 16].

Page 45 of 227

4. Structural Steel Analysis The analysis of the structural steel heat flux has been revised and is based on a thermal radiation exposure using a point source model and a solid flame model.

See response to RAI FM 01q and FM 03c for details on the analysis revision.

References:

[1] 1SPH02902.030, Rev. 0, Generic Fire Modeling Treatments, Hughes Associates, Baltimore, MD, January, 2008.

[2] Hughes Associates, Supplemental Generic Fire model Treatments: Hot Gas Layer Tables, Revision F, Hughes Associates, Inc., Baltimore, MD, March 1, 2011.

[3] Report 0021-0019-000-002, Rev. 0, Combined Ignition Source - Cable Tray Fire Scenario ZOIs for Palisades Nuclear Power Plant Applications, Hughes Associates, Inc., Baltimore, MD, November, 2013.

[4] Report 0021-0019-000-003, Rev. 0, Evaluation of the Development and Timing of Hot Gas Layer Conditions in Generic Palisades Fire Compartments with Secondary Combustibles, Hughes Associates, Inc., Baltimore, MD, November, 2013.

[5] Report 0021-0019-000-004, Rev. 0, Supplemental Generic Fire Modeling Treatments: Transient Fuel Package Ignition Source Characteristics, Hughes Associates, Inc., Baltimore, MD, November, 2013.

[6] Report 0021-0019-000-001, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, October, 2013.

[7] EA-PSA-1CFIRECFD-11-01, Fire Model Analysis of Palisades FA4, Revision 0, Entergy, Jackson, MI, 2011.

[8] EA-PSA-1DFIRECFD-11-03, Fire Model Analysis of Palisades FA3, Revision 0, Entergy, Jackson, MI, 2011.

[9] NIST-GCR-90-580, Development of an Instructional Program for Practicing Engineers Hazard I Users, Barnett, J. R. and Beyler, C. L., National Institute of Standards and Technology, Gaithersburg, MD, July, 1990.

[10] Beyler, C. L., Fire Hazard Calculations for Large, Open Hydrocarbon Pool Fires, Section 3-10, The SFPE Handbook of Fire Protection Engineering, 4th Edition, Society of Fire Protection Engineers, Bethesda, MD, 2008.

[11] Munoz, M., Arnaldos, J., Casal, J., and Planas, E., Analysis of the Geometric and Radiative Characteristics of Hydrocarbon Pool Fires, Page 46 of 227

Combustion and Flame, No. 139, pp. 263-277, 2004.

[12] McGrattan, K. and Miles, S., Modeling Enclosure Fires Using Computational Fluid Dynamics (CFD), Section 3-8, The SFPE Handbook of Fire Protection Engineering, 4th Edition, P. J. DiNenno, Editor-in-Chief, NFPA, Quincy, MA, 2008.

[13] NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, Electric Power Research Institute (EPRI) 1011989 Final Report, Nuclear Regulatory Commission (NRC), Rockville, MD, September, 2005.

[14] NUREG/CR-7010, Volume 1, Cable Heat Release Rate, Ignition, and Spread in Tray Installations During Fire (CHRISTIFIRE) Phase 1: Horizontal Trays, NUREG/CR-7010, Vol. 1., McGrattan, K., Lock., A., Marsh, N.,

Nyden, Bareham, S., M., Price, M., Morgan, A. B., Galaska, M., and Schenck, K., Office of Nuclear Regulatory Research, NRC, Washington, DC.,

July, 2012.

[15] NUREG-1824, Verification & Validation of Selected Fire Models for Nuclear Power Plant Applications Volume 5: Consolidated Fire Growth and Smoke Transport Model (CFAST), NUREG-1824 / EPRI 1011999, Volume 5, Salley, M. H. and Kassawara, R. P., May, 2007.

[16] NUREG-1934, Nuclear Power Plant Fire Modeling Application Guide, Salley, M. H. and Kassawara, R. P., NUREG-1934/EPRI-1019195, U.S.

NRC, Office of Nuclear Reactor Research, Washington, D. C., November, 2012.

[17] NIST SP 1026, CFAST - Consolidated Model of Fire Growth and Smoke Transport (Version 6) Technical Reference Guide, Jones, W. W., Peacock, R. D., Forney, G. P., and Reneke, P. A., NIST, Gaithersburg, MD, April, 2009.

[18] NUREG-1824, Verification & Validation of Selected Fire Models for Nuclear Power Plant Applications Volume 7: Fire Dynamics Simulator (FDS),

NUREG-1824 / EPRI 1011999, Volume 7, Salley, M. H. and Kassawara, R.

P., May, 2007.

NRC Request FM RAI 06 NFPA 805, Section 2.7.3.5, "Uncertainty Analysis," states: "An uncertainty analysis shall be performed to provide reasonable assurance that the performance criteria have been met."

Section 4.7.3, "Compliance with Quality Requirements in Section 2.7.3 of NFPA 805," of Page 47 of 227

the LAR states that "Uncertainty analyses were performed as required by 2.7.3.5 of NFPA 805 and the results were considered in the context of the application. This is of particular interest in fire modeling and Fire PRA development."

Regarding the uncertainty analysis for fire modeling:

a) Describe how the uncertainty associated with the fire model input parameters was accounted for in the fire modeling analyses.

b) Describe how the model and completeness uncertainty was accounted for in the fire modeling analyses.

ENO Response a) PNP addressed uncertainties associated with fire model input parameters through the use of a conservative and bounding analysis. Sensitivity studies were conducted to validate the analysis was indeed conservative and bounding. Details on how input parameter uncertainties were accounted for in the FPRA model follow.

There are five document areas at PNP in which fire modeling parameter uncertainty is applicable:

The main control room (MCR) abandonment analysis (Report 0021-0019-000-001, Rev. 0 [1]);

The hot gas layer (HGL) tabulations as contained in Report 1SPH2902.030, Rev. 0 [2] and Report 0021-0019-000-003 [3];

The Zone of Influence (ZOI) tabulations as contained in Report 1SPH2902.030, Rev. 0 [2], Report 0021-0019-000-002, Rev. 0 [4], and Report 0021-0019-000-004, Rev. 0 [5];

Detailed fire modeling in Fire Area (FA) 3 and FA 4 [6, 7]; and The evaluation of the structural steel exposure [8].

MCR Abandonment Calculation The control room abandonment calculation [1] is structured to provide a reasonably conservative abandonment time for a given heat release rate input over a range of potential input parameter values. The MCR abandonment calculation provides baseline cases for nine forced and natural ventilation combinations and effectively provides a sensitivity assessment on these parameters. Specifically, for a given fire scenario considered in the fire PRA, the shortest abandonment time among the various natural ventilation configurations is selected. In order to ensure the analysis results are conservative relative to the uncertainty in other parameters, a fire modeling sensitivity analysis is provided in Attachment 2 to Report 0021-0019-000-001, Rev. 0 [1]. The sensitivity analysis is used to justify the selection of the input Page 48 of 227

parameter values for the baseline cases using both an absolute abandonment time variation criterion (fifteen percent) and a variation in the total probability of abandonment criterion (fifteen percent). The total probability of abandonment is defined as a product of the severity factor for a particular heat release rate bin and the probability of non-suppression summed over the applicable number of heat release rate bins.

The sensitivity analysis demonstrates that the parameter sensitivity may be grouped as follows over the range of parameter uncertainty:

The parameter does not significantly affect the analysis results over the potential range of values that could be assigned to the parameter (Sensitivity Group 1);

The parameter does affect the analysis results, but value selected for the baseline case is conservative (Sensitivity Group 2); and The parameter does affect the analysis results and the value selected for the baseline case is not conservative (Sensitivity Group 3).

A significant effect is defined as a fifteen percent variation in the probability of control room abandonment as summed over three heat release rate bins. The fifteen parameters are evaluated against baseline fire scenarios.

It is shown that only two parameters fall into the third group (Sensitivity Group 3):

the initial ambient temperature; and the fire base height for closed, multiple-bundle electrical cabinets that spread to adjacent cabinets in the MCR. In the case of the initial ambient temperatures, it is shown that the baseline scenarios are conservative or the effect is not significant for initial ambient temperatures up to about 34°C (93°F). The maximum design basis temperature for the control room is 32.2°C (90°F), so an initial temperature greater than 34°C (93°F) would represent an off-normal condition. The results of this analysis should therefore not be used if the postulated scenario would have an initial temperature greater than 34°C (93°F) in the main control room. In the case of the fire base height, some of the Main Control Boards (MCBs) in the control room can be treated as the half height configuration, the FPRA should treat any MCB scenarios that are considered closed, multi-bundle cabinets that spread to adjacent cabinets with an increased probability of abandonment of twenty-five percent greater than the baseline.

Based on of Attachment 2 to Report 0021-0019-000-001, Rev. 0 [1], the baseline results presented in the control room abandonment calculation are considered conservative with respect to uncertainty in the parameter values.

Hot Gas Layer Tables Page 49 of 227

The Generic Fire Modeling Treatments report [2] and Report 0021-0019-000-003, Rev. 0 [3] provide times at which the hot gas layer in a generic enclosure will exceed specified temperature thresholds. The computations are performed using the zone computer model CFAST, version 6.0.10 and Version 6.1.1. The methodology for computing the hot gas layer tables is described in detail in Section 6.3 and Appendix B of the Generic Fire Modeling Treatments report [2]. Essentially, CFAST is used to balance energy and mass flow through openings and the time at which the hot gas layer temperature reaches a threshold value is reported regardless of the hot gas layer height. The input parameters that have the greatest influence on the model results include the fire size, the enclosure geometry, the fuel properties, the opening characteristics, the boundary material properties, and the initial ambient temperature. Each of these elements is described.

Fire Size. The fire size is a prescribed input per NUREG/CR-6850 [9] or is specified with a particular set of input parameters and subject to the parameter constraints (ignition source - cable tray fire scenarios). The room geometry is selected in such a way as to minimize the heat losses to the boundaries and thus varies from volume to volume. Under this assumption, the height of the enclosure necessarily varies with the volume. However, a sensitivity analysis is conducted on the room enclosure shape (Section B.4.4 of the Generic Fire Modeling Treatments report [2]) and it is shown that minimizing the enclosure boundary surface area provides a bounding or nearly bounding result for a given enclosure volume when the length to width aspect ratio is varied from 1:1 to 1:5 in the cases considered. As the aspect ratio increases, a significant reduction in the temperature is observed indicating that spaces that deviate from a 1:1 aspect ratio have an increasing safety margin embedded in the hot gas layer temperature results.

Fuel properties. The selection of the fuel properties is evaluated in Sections B.4.1 and B.4.2 of the Generic Fire Modeling Treatments report [2]. Fuel properties are varied over a large range of potential values and the most adverse combination is selected to represent all fuels. In this case a relatively low soot yield material is used because it reduces the radiant heat losses from the hot gas layer to the enclosure boundaries and maximizes the hot gas layer temperature.

Ventilation. The opening characteristics are described in terms of a boundary fraction and are varied over a range of 0.001 - 10 percent in the baseline cases. The hot gas layer associated with the most adverse ventilation case is selected in the fire PRA among the reported ventilation conditions for a given fire size and enclosure volume. The key input parameter that is set is the ventilation geometry (length, width, and base height) given a vent fraction.

Section B.4.5 of the Generic Fire Modeling Treatments report [2] provides a sensitivity analysis on the effects of various vent orientations and placements on the predicted temperature. A total of fifty-four vent configurations were examined for the baseline enclosures. It is found that the bounding case can be one of three orientations: one in which the vent width is equal to the Page 50 of 227

enclosure width, located either at the ceiling or at the floor and one in which the vent height is equal to the enclosure height. All hot gas layer tables reported in the Generic Fire Modeling Treatments report and Report 0021-0019-000-003, Rev. 0 [3] are based on the most adverse hot gas layer condition among the three vent orientations and thus represent the bounding configuration for the vent geometry.

Boundary material properties. The boundary material properties are defined as concrete having the lowest thermal diffusivity reported among available data as described in Section B.4.3 of the Generic Fire Modeling treatments report. The thermal diffusivity of the selected concrete, defined as the thermal conductivity divided by the heat capacity and density, is 5.9 10-7 m²/s (6.3 10-6 ft²/s) and is about thirty percent lower than the value of 8.9 10-7 m²/s (9.5 10-6 ft²/s) recommended in NUREG-1805 [10]. This conservatively biases the results for the boundary materials, though it is shown in Section B.4.3 of the Generic Fire modeling Treatments report [2]

that the results are not conservative if they are applied to spaces bound with thermal insulation, lightweight concrete, or gypsum wallboard.

Initial ambient temperature. The initial ambient temperature is assumed to be 20°C (68°F) in the Generic Fire Modeling Treatments report [2] and in Report 0021-0019-000-003, Rev. 0 [3]. Although an ambient temperature of 20°C (68°F) is not a conservative and bounding assumption, the effect is readily bound by other conservative aspects of the model approach such as the enclosure geometry, ventilation effects, fuel properties, and hot gas layer position.

Significant conservatism is embedded in the CFAST model results by the specification of an adiabatic floor. Radiant heat losses from both the fire and the hot gas layer to the floor are not credited with reducing the hot gas layer temperature. This assumption is expected to conservatively bias the temperature predictions. Based on the overall conservative bias associated with the CFAST model parameters (collectively), the hot gas layer tables reported in the Generic Fire Modeling Treatments report [2] and in Report 0021-0019-000-003, Rev. 0 [3]

are considered conservative with respect to uncertainty in the parameter values.

ZOI Calculations The Generic Fire Modeling Treatments report [2], Report 0021-0019-000-002, Rev. 0 [4], and Report 0021-0019-000-004, Rev. 0 [5] provide ZOI dimensions for various ignition sources and combination ignition source - cable tray configurations for which fire PRA fire scenarios are developed. The tabulated ZOI dimensions are all based on the methodologies described in Generic Fire Modeling Treatments report [2], except the ZOI dimensions for the ignition source

-secondary combustible configurations include the physical offset associated with both the cable tray arrangement and the fire spread in the cable trays. The ZOI dimensions essentially consist of a vertical component derived from a plume Page 51 of 227

exposure correlation and one or more horizontal components, each derived from a radiant heat flux calculation.

The vertical plume calculation uses an empirical model that requires as inputs the fire size, the ambient temperature, and fire diameter. The fire size is an input parameter specified by NUREG/CR-6850 [9]. The fire diameter and ambient temperature are the primary parameters subject to uncertainty. In this case, the fire diameter in the original Generic Fire Modeling Treatments report [2] provides ZOI dimensions assuming a variable diameter (as characterized using the heat release rate and heat release rate per unit area). A heat release rate per unit area range between 200 kW/m² (17.6 Btu/s-ft²) and 1,000 kW/m² (88.1 Btu/s-ft²) is used for transient combustible materials and range up to 3,000 kW/m² (264 Btu/s-ft²) for electronic panels. The baseline ambient temperature assumed in the original Generic Fire Modeling Treatments report [2] is 20°C (68°F) with a maximum application limit of 80°C (176°F). The baseline ambient temperature selection in the Report 0021-0019-000-002, Rev. 0 [4] and Report 0021-0019-000-004, Rev. 0

[5] is varied from 20°C (68°F) to 80°C (176°F) and is thus not subject to assumption or uncertainty, at least within the limits of applicability.

The maximum effect of an elevated initial ambient temperature on the ZOI dimensions for transient fuel package fires is provided in Report 0021-0019-000-004, Rev. 0 [5] and in Report 0021-0019-000-002, Rev. 0 [4] for various ignition source - cable tray configurations. The ZOI dimension may change by about two to five percent when the ambient temperature is 40°C (104°F) and ten - twenty percent if the ambient temperature is 80°C (176°F) based on a various ignition sources and cable tray configurations evaluated in open, wall, and corner locations. This differential is expected to be readily bounded by the conservatisms that are embedded in the ZOI development. These conservatisms relative to a transient fuel package fire include the use of steady-state target damage thresholds, a fire diameter that maximizes the ZOI dimension, the use of a ZOI box rather than a cone, and the selection of the most adverse result among a range of methods.

An additional offsetting conservative factor for the panel fires relative to an elevated ambient temperature environment is the assumed heat release rate per unit area for the electronic panel fires for the vertical ZOI dimension is effectively 3,000 kW/m² (264 Btu/s-ft²). This means that the characteristic fire dimension for the 98th percentile panel fires is on the order of 0.26 - 0.48 m (0.9 - 1.6 ft). The characteristic dimension for the electronic panels as evaluated using the NUREG/CR-6850 [9] guidance would be based on the panel top surface area and will typically be on the order of 0.6 - 1.2 m (2 - 4 ft). This indicates a significant bias is introduced by assuming the fire plan area can occupy only a fraction of the panel top. An additional conservative bias is introduced in setting the base location of the vertical ZOI dimension. Per NUREG/CR-6850, Supplement 1 [11], the fire base height may be set 0.3 m (1 ft) below the panel top (if the panel does not have significant openings in the top). The vertical ZOI dimensions for the electronic panels reported in the Generic Fire Modeling Treatments report [2] use the panel Page 52 of 227

top as the base height reference for the vertical ZOI dimension. This introduces a uniform 0.3 m (1 ft) bias in all vertical ZOI dimensions for the electronic panels. As such, the vertical ZOI dimension is calculated using bounding input parameters when viewed collectively.

In the case of the horizontal ZOI dimension, the maximum distance as obtained using the more severe prediction among both a solid flame model and the Point Source Model (PSM). The ZOI dimensions for the ignition source - cable tray configurations are obtained in a similar manner, but include heat flux calculations for using the total heat release rate of all heat sources and the heat flux calculations using the sum of the ignition source and cable trays contributions. The PSM requires as input the fire size and the fire radiant fraction, which is assumed to be 0.4. The fire size is a prescribed input per NUREG/CR-6850 [9]. Based on SFPE Handbook of Fire Protection Engineering, Section 3-10 [12], the effective radiant fraction for conservative (but not bounding) results is 0.21. A bounding result is obtained when a safety factor of two is used. By assuming a radiant fraction of 0.4, an effectively bounding result is therefore obtained. The solid flame heat flux model requires the fire size and fire diameter as input parameters. The fire size is a prescribed input per NUREG/CR-6850 [9]. The fire diameter is varied via the heat release rate per unit area parameter. In this case, the most adverse fire diameter is intermediate with a heat release rate per unit area of about 350 -

400 kW/m² (30.8 - 35.2 Btu/s-ft²), depending on the specific case. The value that yields the maximum ZOI dimension is the value used in the analysis.

The electronic panel ignition source ZOIs have additional conservative margins by including an additional calculation that is more conservative than the approach suggested in NUREG/CR-6850, Supplement 1 [11]. Per NUREG/CR-6850, Supplement 1 [11], the fire base is located 0.3 m (1 ft) below the top of the panel and is typically modeled assuming the panel boundaries do not exist (open fire).

The horizontal ZOI dimensions developed in the Generic Fire Modeling Treatments report [2] include an upper horizontal ZOI dimension that is computed in this manner and a lower ZOI dimension that assumes internal flame impingement on the panel boundary. This flame impingement imposes a 120 kW/m² (10.6 Btu/s-ft²) heat flux on any internal boundary that radiates outward from a single side. The lower horizontal ZOI dimension is significantly larger than the upper fire plume base horizontal dimension, typically by a factor of two (compare Tables 5-16 and 5-17 in the Generic Fire Modeling Treatments report, for example). The fire PRA selects the most adverse horizontal ZOI dimension and thus incorporates this bias directly.

Based on the overall conservative bias associated with the input parameter, both the horizontal and vertical ZOI dimensions reported in the Generic Fire Modeling Treatments report [2], Report 0021-0019-000-002, Rev. 0 [4], and Report 0021-0019-000-004 Rev. 0 [5] are considered conservative with respect to parameter uncertainty.

Plant Specific Detailed Fire Modeling Page 53 of 227

Plant specific calculations are provided in EA-PSA-1CFIRECFD-11-01 [6] and EA-PSA-1DFIRECFD-11-03 [7]. The calculations involve FDS simulations in the 1-C and 1-D Switchgear Rooms to determine target failure times. The key parameter uncertainties in these evaluations consist of the ignition source and secondary combustible heat release rate, the fire location, and the ventilation conditions. The heat release rates for the ignition sources are characterized using a two or a three point treatment of the continuous conditional probability distribution provided in NUREG/CR-6850 [9] for each ignition source considered. The heat release rates for the cable trays follows the guidance provided in NUREG/CR-7010, Vol. 1 [13]

for fire spread and heat release rate per unit area with an additional conservative ignition criteria based on temperature imposed. As such, the heat release rate is at least as conservative as recommended in NUREG guidance, and may be more conservative for scenarios where cable trays ignite remotely from the ignition source due to temperature rather than a comparable case that only considers ignition based on continuous propagation. Uncertainty in the fire location is incorporated directly into the base cases via assessment of specific ignition sources. The uncertainty in the ventilation conditions is assessed through scenario sensitivity analysis that shows the baseline scenarios are conservative [6, 7].

Structural Steel Analysis The revised structural steel report [8] provides a parameter sensitivity analysis and demonstrates that the conclusions are not affected by the uncertainty in the input parameters.

References:

[1] Report 0021-0019-000-001, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, October, 2013.

[2] 1SPH02902.030, Rev. 0, Generic Fire Modeling Treatments, Hughes Associates, Baltimore, MD, January, 2008.

[3] Report 0021-0019-000-003, Rev. 0, Evaluation of the Development and Timing of Hot Gas Layer Conditions in Generic Palisades Fire Compartments with Secondary Combustibles, Hughes Associates, Inc., Baltimore, MD, November, 2013.

[4] Report 0021-0019-000-002, Rev. 0, Combined Ignition Source - Cable Tray Fire Scenario ZOIs for Palisades Nuclear Power Plant Applications, Hughes Associates, Inc., Baltimore, MD, November, 2013.

[5] Report 0021-0019-000-004, Rev. 0, Supplemental Generic Fire Modeling Treatments: Transient Fuel Package Ignition Source Characteristics, Hughes Associates, Inc., Baltimore, MD, November, 2013.

Page 54 of 227

[6] EA-PSA-1CFIRECFD-11-01, Fire Model Analysis of Palisades FA4, Revision 0, Entergy, Jackson, MI, 2011.

[7] EA-PSA-1DFIRECFD-11-03, Fire Model Analysis of Palisades FA3, Revision 0, Entergy, Jackson, MI, 2011.

[8] Calculation 0247-07-0005.08, Rev. 2, Palisades Nuclear Plant Fire Probabilistic Risk Assessment Exposed Structural Steel Analysis, Entergy, November, 2013.

[9] NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, Electric Power Research Institute (EPRI) 1011989 Final Report, Nuclear Regulatory Commission (NRC), Rockville, MD, September, 2005.

[10] NUREG-1805, Fire Dynamics Tools (FDTs) Quantitative Fire Hazard Analysis Methods for the U.S. Nuclear Regulatory Commission Fire Protection Inspection Program, Nuclear Regulatory Commission, Washington, DC, 2004.

[11] NUREG/CR-6850 Supplement 1, Fire Probabilistic Risk Assessment Methods Enhancements, EPRI 1019259, Technical Report, NUREG/CR-6850 Supplement 1, NRC, Rockville, MD, September, 2010.

[12] Beyler, C. L., Fire Hazard Calculations for Large, Open Hydrocarbon Pool Fires, Section 3-10, The SFPE Handbook of Fire Protection Engineering, 4th Edition, Society of Fire Protection Engineers, Bethesda, MD, 2008.

[13] NUREG/CR-7010, Volume 1, Cable Heat Release Rate, Ignition, and Spread in Tray Installations During Fire (CHRISTIFIRE) Phase 1: Horizontal Trays, NUREG/CR-7010, Vol. 1., McGrattan, K., Lock., A., Marsh, N.,

Nyden, Bareham, S., M., Price, M., Morgan, A. B., Galaska, M., and Scheneck, K., Office of Nuclear Regulatory Research, NRC, Washington, DC., July, 2012.

b) Fire model model and completeness uncertainty was not explicitly accounted for in all fire modeling evaluations or incorporated into the fire PRA at PNP.

However, the uncertainty associated with fire modeling model and completeness uncertainty is addressed through the use of a conservative and bounding analysis. There are five primary areas at PNP in which fire modeling model and completeness uncertainty is applicable:

The control room abandonment analysis (Report 0021-0019-000-001, Rev. 0

[1]);

The hot gas layer (HGL) tabulations as contained in Report 1SPH2902.030, Rev. 0 [2] and Report 0021-0019-000-003 [3];

Page 55 of 227

The Zone of Influence (ZOI) tabulations as contained in Report 1SPH2902.030, Rev. 0 [1], Report 0021-0019-000-002, Rev. 0 [4], and Report 0021-0019-000-004, Rev. 0 [5];

Detailed fire modeling in Fire Area (FA) 3 and FA 4 [6, 7]; and The evaluation of the structural steel [8].

MCR Abandonment Calculation. The MCR abandonment calculation (Report 0021-0019-000-001, Rev. 0 [1]) provides an assessment of the model uncertainty in Section A2.3 of Attachment 2 using the methods described in NUREG-1934 [9].

The uncertainty is assessed for a range of heat release rate bins associated with the primary baseline fire scenario ignition sources in control room area. Table A2-14 of Report 0021-0019-000-001, Rev. 0 [1] shows that the maximum probability the actual abandonment time would be fifteen percent or more lower than the predicted value is less than 16.75 percent for transient fire scenarios in the MCR.

The average probability for transient fires in the MCR is approximately 6.7 percent.

For electrical cabinet fires located in the MCR or the Sub-Enclosures (i.e., main control boards), the probability the actual abandonment time would be fifteen percent or more below the predicted value is nearly zero (less than 0.9 percent).

Based on the sensitivity analysis presented in Section A2.2 of Report 0021-0019-000-001, Rev. 0 [1], it may be concluded that the model and completeness uncertainties are bound by the conservative bias introduced by the parameter selection.

Hot Gas Layer Tabulations. The hot gas layer tables are computed using the zone computer model CFAST, Version 6.0.10 and 6.1.1 in the Generic Fire Modeling Treatments report [2] and in Report 0021-0019-000-003, Rev. 0 [3]. As described in the response to RAI FM-06(a), there are a significant number of parameters that are conservatively biased in the model, including the fuel properties, the combustion properties, the boundary properties, the adiabatic floor surface, and the vent configuration. In addition, for a given CFAST geometry, the fire PRA selects the most adverse scenario among a ventilation range between 0.001 and 10 percent of the boundary area.

The approximate effect of each of these parameters (except for the adiabatic floor surface) on the temperature results are provided in Appendix B of the Generic Fire Modeling Treatments report. For example, Figure B4-7 and B4-8 in the Generic Fire Modeling Treatments report [2] demonstrates that the effect of changing the thermal diffusivity of the boundary materials on the steady state temperature is roughly proportional to the change in the thermal diffusivity at least when centered on a value of 5.9 10-7 m²/s (6.3 10-6 ft²/s). Given that this value is about thirty percent lower than the thermal diffusivity recommended in NUREG-1805 [10] for normal weight concrete, a comparable test case would have a temperature reduction of about 60°C (108°F). This sensitivity alone is comparable to the temperature change necessary to reduce the model and completeness uncertainty to less than two percent as determined using the methods described in Page 56 of 227

NUREG-1934 [9] (i.e., the probability of exceeding the critical value due to model uncertainty is less than two percent when the hot gas layer temperature is about 60°C (108°F) lower than the critical value). When all conservatively biased input parameters are considered together, it is expected that the collective effect on the predicted temperature will result in a low probability of exceeding a threshold value at a tabulated time.

Consequently, it is concluded that fire model model and completeness uncertainty either would not contribute significantly to the risk uncertainty because it is sufficiently bound by the conservatisms in the CFAST hot gas layer analyses.

ZOI Calculations. The ZOI computations rely on a plume centerline temperature, an open source fire radiant heat flux computation, and a radiant heat flux computation from a heated panel or burning array of cables. The plume centerline temperature computation is shown in NUREG-1934 [9] and NUREG-1824, Volume 3 [11] to have a non-conservative bias and a relatively large standard deviation.

However, the application considered did not explicitly account for the hot gas layer temperature changes, which are the expected source of the bias and variation.

Similar plume correlations used by CFAST and MAGIC show a conservative bias and smaller variation. The application of the plume correlations is limited in the Generic Fire Modeling Treatments report [2] and Report 0021-0019-000-002, Rev.

0 [4], and Report 0021-0019-000-004, Rev. 0 [5] to 80°C (122°F) or less through the use of the modified critical heat flux, which is intended to adapt the models for elevated internal temperatures. Further, as discussed in the response to RAI FM-06a, the vertical plume ZOI dimension may have as much as a 0.3 m (1 ft) conservative bias embedded based on the assumed diameters and the base elevations relative to NUREG/CR-6850 [12] and NUREG/CR-6850, Supplement 1

[13] guidelines.

The horizontal ZOI dimensions are computed using a radiant heat flux model with the radiant fraction set to about two times the value recommended in the SFPE Handbook of Fire Protection Engineering, Section 3-10 [14]. Effectively, the radiant heat flux has a bias of two explicitly embedded in the calculation. The probability that the heat flux at a fixed location would exceed 5.7 kW/m² (0.5 Btu/s-ft²) given a prediction of 2.35 kW/m² (0.50 Btu/s-ft²) (i.e., removed conservative bias) may be computed using the methods described in NUREG-1934 [9] with the bias and normalized variance for the radiant heat flux models, which are 2.02 and 0.59. The resulting probability is nearly zero. In the case of the electronic panel fires, an additional margin is provided through the use of a conservative model beyond that required in NUREG/CR-6850 [12] and NUREG/CR-6850, Supplement 1 [13] for portions of the ZOI below the panel. Because fire PRA uses the most adverse horizontal ZOI dimension above or below the panel, this additional model introduces a second conservative factor.

Consequently, it is concluded that fire model model and completeness uncertainty either would not contribute the risk uncertainty or are bound by the conservatisms in the analysis, depending on the ZOI dimension considered.

Page 57 of 227

Plant Specific Detailed Fire Modeling. Plant specific calculations are provided in EA-PSA-1CFIRECFD-11-01 [6] and EA-PSA-1DFIRECFD-11-03 [7]. The calculations involve FDS simulations in the 1-C and 1-D Switchgear Rooms to determine target failure times. The methods of NUREG-1934 [9] may be used to show that model and completeness uncertainty are bound by the conservative input assumptions and the numerical rounding of the model results. Given a model bias of 1.02 and a normalized variance of 0.13 for target heating [9], a predicted temperature of 200°C (392°F) for the target has a fifteen percent probability of exceeding the critical temperature of 204°C (400°F). Assuming a linear target temperature response at the approximate time of the predicted target failure and a target failure time of sixty minutes, there is a fifteen percent probability that the target fails at fifty-nine minutes rather than sixty minutes. The results are implemented in the fire PRA using ten minute increments, so this would not affect the analysis results. In addition, a conservative treatment of the targets themselves includes the assumption of a fully exposed cable rather than a cable within a cable tray or conduit and an adverse target orientation relative to the fire.

These conservatisms would readily bound the significant uncertainty in the predicted damage time, which is one minute for a sixty minute predicted damage time. A similar assessment may be performed using a smaller probability of exceeding a critical temperature. For example, if five percent is selected rather than fifteen percent, the maximum effect on the damage time is on the order of five minutes for a sixty minute damage time. This is still within the rounding increment of the target damage times and the fire PRA would not be affected. The effect is dampened for targets that are predicted to fail sooner than sixty minutes.

Consequently, it is concluded that fire model model and completeness uncertainty either would not contribute the risk uncertainty or are bound by the conservatisms in the analysis, depending on the ZOI dimension considered.

The revised structural steel report [8] provides an assessment of the model and completeness uncertainty using the methods described in NUREG-1934 [9]. It is concluded that the results are not significantly affected by the model and completeness uncertainty and are bound by conservatism in the input parameters.

References:

[1] Report 0021-0019-000-001, Rev. 0, Evaluation of Control Room Abandonment Times at the Palisades Nuclear Station, Hughes Associates, Inc., Baltimore, MD, October, 2013.

[2] 1SPH02902.030, Rev. 0, Generic Fire Modeling Treatments, Hughes Associates, Baltimore, MD, January, 2008.

[3] Report 0021-0019-000-003, Rev. 0, Evaluation of the Development and Timing of Hot Gas Layer Conditions in Generic Palisades Fire Compartments with Secondary Combustibles, Hughes Associates, Inc., Baltimore, MD, November, 2013.

Page 58 of 227

[4] Report 0021-0019-000-002, Rev. 0, Combined Ignition Source - Cable Tray Fire Scenario ZOIs for Palisades Nuclear Power Plant Applications, Hughes Associates, Inc., Baltimore, MD, November, 2013.

[5] Report 0021-0019-000-004, Rev. 0, Supplemental Generic Fire Modeling Treatments: Transient Fuel Package Ignition Source Characteristics, Hughes Associates, Inc., Baltimore, MD, November, 2013.

[6] EA-PSA-1CFIRECFD-11-01, Fire Model Analysis of Palisades FA4, Revision 0, Entergy, Jackson, MI, 2011.

[7] EA-PSA-1DFIRECFD-11-03, Fire Model Analysis of Palisades FA3, Revision 0, Entergy, Jackson, MI, 2011.

[8] Calculation 0247-07-0005.08, Rev. 2, Palisades Nuclear Plant Fire Probabilistic Risk Assessment Exposed Structural Steel Analysis, Entergy, November, 2013.

[9] NUREG-1934, Nuclear Power Plant Fire Modeling Application Guide, Salley, M. H. and Kassawara, R. P., NUREG-1934/EPRI-1019195, U.S.

NRC, Office of Nuclear Reactor Research, Washington, D. C., November, 2012.

[10] NUREG-1805, Fire Dynamics Tools (FDTs) Quantitative Fire Hazard Analysis Methods for the U.S. Nuclear Regulatory Commission Fire Protection Inspection Program, Nuclear Regulatory Commission, Washington, DC, 2004.

[11] NUREG-1824, Verification & Validation of Selected Fire Models for Nuclear Power Plant Applications Volume 3: Fire Dynamics Tools (FDTS), NUREG-1824 / EPRI 1011999, Volume 3, Salley, M. H. and Kassawara, R. P., May, 2007.

[12] NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, Electric Power Research Institute (EPRI) 1011989 Final Report, Nuclear Regulatory Commission (NRC), Rockville, MD, September, 2005.

[13] NUREG/CR-6850 Supplement 1, Fire Probabilistic Risk Assessment Methods Enhancements, EPRI 1019259, Technical Report, NUREG/CR-6850 Supplement 1, NRC, Rockville, MD, September, 2010.

[14] Beyler, C. L., Fire Hazard Calculations for Large, Open Hydrocarbon Pool Fires, Section 3-10, The SFPE Handbook of Fire Protection Engineering, 4th Edition, Society of Fire Protection Engineers, Bethesda, MD, 2008.

Page 59 of 227

NRC Request Probabilistic Risk Assessment (PRA) RAI 01 Contrary to Regulatory Guide 1.200, An Approach for Determining the Technical Adequacy of Probabilistic Risk Assessment Results for Risk-Informed Activities, Rev.

2, it is not clear that all the peer review F&Os were resolved to bring them into alignment with Capability Category (CC) -II (Met) or justified why a lesser CC was acceptable. Clarify the following dispositions to fire F&Os and Supporting Requirement (SR) assessments identified in Attachment V of the LAR that have the potential to impact the FPRA results:

a) F&O CS-A9-01 against CS-A9:

Confirm that the supplemental analysis work discussed in the disposition to this F&O considered proper polarity hot shorts on ungrounded DC circuits up to and including two independent faults. Additionally, describe what portions of the cable analysis have yet to be integrated into the FPRA model, and discuss the quantitative impact of this exclusion on the risk estimates, clarifying the anticipated numerical changes referenced in Table 2, Supplemental Information to Table V-1, of the LAR supplement dated February 21, 2013 (ADAMS Accession No. ML13079A090).

b) F&O CS-B1-01 against CS-B1 and F&O CS-C4-01 against CS-C4:

Attachment 2 of EA-APR-95-004 (Rev. 5) notes that the coordination was not verified for some power sources and states that these coordination issues are being addressed through the PRA model. Provide additional clarification regarding the treatment of breaker coordination issues in the FPRA, particularly for those power sources for which coordination could not be verified or demonstrated.

c) F&O CS-C1-01 against CS-C1:

The disposition to this F&O states that the results of the data verification performed on cable routing have not been fully implemented into the FPRA model. Describe the results of the verification process, and discuss the quantitative impact of this exclusion on the risk estimates, clarifying the anticipated numerical changes referenced in Table 2, Supplemental Information to Table V-1, of the LAR Supplement dated February 21, 2013.

d) F&O ES-A2-01 against ES-A2 and ES-B4:

This F&O cites incomplete treatment of interlock and permissive circuits.

Describe the process utilized to review power supply, interfacing equipment (e.g.,

interlocks, permissives, autocontrol functions, etc.), instrumentation, and support system dependencies to identify additional equipment whose fire-induced failure, including spurious actuation, could result in a fire-induced initiating event or Page 60 of 227

adversely affect accident mitigating equipment. Evaluate the impact on the FPRA (i.e., core damage frequency (CDF), large early release frequency (LERF), delta

()CDF and LERF) of not completing the treatment of interlock and permissive circuits.

For self-approval, complete the on-going evaluation of instrumentation that could potentially affect accident mitigating equipment, and complete the associated cable tracing as appropriate. Provide revised risk results (i.e., CDF, LERF, CDF and LERF) based on this modeling update.

e) F&O ES-A5-01 against ES-A5:

Appendix B of the MSO Report states that the PCP seal module is currently undergoing an update for consistency with the latest CEOG guidance and to ensure the potential for multiple spurious operations failure modes due to specific fire scenarios can be captured. Provide additional information regarding this update and its status. Additionally, confirm whether related modeling changes have been integrated into the FPRA model; if not, discuss the quantitative impact.

f) F&O ES-C1-01 against ES-C1:

This F&O cites incomplete treatment of instrumentation needed to support operator actions. Describe how fire-induced instrument failure (including no readings, off-scale readings, and incorrect/misleading readings) is addressed in the human reliability analysis (HRA) credited in the FPRA. Include discussion of the success criteria assumed for this modeling and description of how instrumentation that is relied on for credited operator actions was identified and verified as available to a level of detail commensurate with the risk importance and quantification of the human error probabilities (HEPs). Evaluate the impact on the FPRA (i.e., CDF, LERF, CDF and LERF) of not completing the treatment instrumentation and associated cable tracing.

For self-approval, complete the on-going identification and mapping of instrumentation for credited human failure events (HFEs) in the FPRA as well as associated cable tracing. Provide revised risk results (i.e., CDF, LERF, CDF and LERF) based on this modeling update.

g) F&O ES-C2-01 against ES-C2 and F&O HRA-A3-01 against HRA-A3 and HRA-B4:

There appears to be no documented process for systematically identifying and defining HFEs that may result in an undesired operator response (i.e., error of omission or commission) to spurious cues and indication as recommended in Section 3.4.1 of NUREG-1921, EPRI/NRC-RES Fire Human Reliability Analysis Guidelines. Describe and further justify the process utilized to identify and model such actions on a fire scenario basis per HRA-A3 and HRA-B4.

Page 61 of 227

Additionally, according to Appendix H of the Model Development Report, the annunciator system is impacted by fires in almost all physical access units (PAUs); however, cable tracing was not performed. Discuss how this cable location uncertainty associated with the fire impact on alarms and annunciators is accounted for in the FPRA HRA, particularly regarding those HFEs for which the alarm or annunciator serves as the primary cue (as defined by Section 5.2.5.2 of NUREG-1921).

h) F&O FQ-C1-01 against FQ-C1:

Describe the process used to identify and evaluate HFE combinations, and further justify those assumptions identified as non-conservative in EA-PSA-FPRA-HEPDEP (Rev. 0) using guidance provided in Section 6.2 of NUREG-1921 for each branch of the dependency analysis decision tree (e.g., manpower, same location, etc.). Additionally, discuss how the timing (i.e., for simultaneous and sequential actions) and stress levels associated with HFE combinations were determined to support the dependency evaluation.

i) F&O FQ-E1-01 against FQ-E1:

This F&O cites incomplete presentation of dominant results. Describe the scope of the reasonableness review, including the number of non-significant cutsets reviewed and the criteria used to determine the extent to review them.

Additionally, discuss the assessment performed and the results obtained related to the determination and review of significant risk contributors to fire CDF and LERF, including the extent to which individual basic events (e.g., equipment failures, operator actions, common-cause failures, etc.) and FPRA-related parameters (e.g., hot short probabilities, non-suppression probabilities, etc.) are considered in the importance analysis.

j) F&O FSS-A1-01 against FSS-A1:

Section 6.1 of the Fire Scenario Development Report states that all motor control centers (MCCs) have been treated as closed, sealed and robust in which damage beyond the ignition source will not be postulated. However, this same section also notes that the walkdown team only addressed a selection of MCCs and appears to have limited their review to determining whether or not select MCCs were robustly secured (but not also well-sealed). Provide additional justification that all MCCs are both well-sealed and robustly secured FAQ 08-0042, Fire Propagation from Electrical Cabinets, (ADAMS Accession No. ML092110537, closure memo). In addition, discuss and further justify per FAQ 08-0042 any other electrical cabinets that were treated as closed, sealed and robust. If any do not meet these criteria, provide risk estimates (i.e., CDF, LERF, CDF and LERF) treating them as open cabinets.

Page 62 of 227

k) F&O FSS-B1-01 against FSS-B1:

The disposition to this F&O and Section 14.2 of the Fire Scenario Development Notebook note that the current FPRA model does not specifically identity fire scenarios that result in abandonment due to equipment damage. Clarify whether MCR abandonment due to loss of control or function is not modeled for fire areas or scenarios modeled in the FPRA, and provide additional justification as needed.

l) F&O FSS-B2-01 against FSS-B2:

Describe how CDF and LERF are estimated in MCR abandonment scenarios. In doing so, discuss how abandonment scenarios modeled in the PRA reflect the equipment, instrumentation, and/or control that may or may not be available following MCR abandonment. In addition, describe the HRA approach utilized for MCR abandonment scenarios, stating if "screening" values for post MCR abandonment actions are used or if detailed human error analyses have been completed for this activity. Include a discussion of those actions credited as being performed in the MCR prior to abandonment, confirming whether such actions were treated in accordance with guidance within NUREG-1921. Also, provide the results of the HFE quantification process described in Section 5 of NUREG-1921, including the following:

i. The results of the feasibility assessment of the operator action(s) associated with the HFEs, specifically addressing each of the criteria discussed in Section 4.3 of NUREG-1921.

ii. The results of the process in Section 5.2.7 of NUREG-1921 for assigning scoping HEPs to actions associated with switchover of control to an alternate shutdown location, specifically addressing the basis for the answers to each of the questions asked in the Figure 5.4 flowchart.

iii. The results of the process in Section 5.2.8 of NUREG-1921 for assigning scoping HEPs to actions associated with the use of alternate shutdown, specifically addressing the basis for the answers to each of the questions asked in the Figure 5-5 flowchart.

iv. The results of a sensitivity analysis that shows the impact on the PRA results (i.e., CDF, LERF, CDF and LERF) of using the resultant scoping HEPs for MCR abandonment scenarios.

Note that results of a detailed HRA quantification per Section 5.3 of NUREG-1921 may be provided as an alternative to items I.ii and I.iii above.

Page 63 of 227

m) F&O FSS-C3-01 against FSS-C3:

The resolution to F&O FSS-C3 provided in EA-PSA-FIREF&O-11-02 states that the fire impacts of an ignition source are in general bounded by assuming 98th percentile heat release rates (HRRs) from NUREG/CR-6850 and EPRI 1011989, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities, at time t=0. Section 7.4 of the Fire Scenario Development Report states that refinements to this approach have been made to incorporate fire growth utilizing the MathCAD approach discussed in Appendix E. Section 2.1.1 of the Fire Scenario Development Report notes that scenarios for fixed ignition sources in PAUs 03 and 04 (1C and 1D Switchgear Rooms) are based on three HRR bins representing the 49th, 81st, and 98th-percentile HRRs; however, Appendix E does not appear to be used. Section 9.6 indicates that scenarios in PAU 02 (Cable Spreading Room) are defined based on a grid system related to the arrangement of sprinkler heads. Given the variety of approaches to fire scenario development utilized by the FPRA, provide an overview of these approaches, clarifying how target sets, severity factors and non-suppression probabilities are developed and applied for each approach. In doing so, describe the treatment utilized for the growth stages of a fires HRR and the corresponding fire scenarios modeled in the FPRA.

n) F&O FSS-C4-01 against FSS-C4:

Section 8.3 of the Fire Scenario Development Notebook indicates that prompt suppression is credited for hot work fire scenarios using the welding suppression curve at a time of 5 minutes. Describe whether a fire watch is proceduralized for all hot work, and provide additional justification for using a time of 5 minutes in lieu of an estimated time to target damage as recommended by NUREG/CR-6850. Also, discuss whether or not any additional suppression credit is taken (e.g., automatic suppression system, delayed manual suppression, etc.).

o) F&O FSS-C5-01 against FSS-C5:

According to this F&O, no scenario was evaluated for conditions where the target damage criteria are those of sensitive electronics. Noting guidance within Appendix S of NUREG/CR-6850, provide additional justification for not modeling the potential impacts to sensitive electronics from fires in adjacent cabinets, including those within the MCR and PAUs for which detailed fire modeling was performed (e.g., 1C and 1D switchgear rooms).

In addition, although Appendix H of NUREG/CR-6850 recommends that vulnerability to transient fires be limited to cable vulnerability, Section H.2 recommends that if sensitive electronics can be impacted, then damage to and ignition of such components from transient fires should be considered. Describe how the impact on sensitive electronics from fire effects is modeled in the fire PRA.

Page 64 of 227

p) F&O FSS-C7-01 against FSS-C7:

Describe the dependencies that exist amongst credited suppression paths (including dependencies associated with recovery of a failed fire suppression system), and discuss how they were evaluated and modeled.

q) F&O FSS-D1-01 against FSS-D1:

The target damage time model programmed in MathCAD is used extensively in the fire PRA (e.g., including electrical cabinets, transients and high energy arc fault (HEAF) events) and appears to be an unreviewed analysis method. The final peer review report notes that the model development and application have not been independently reviewed [by the industry] and the applicability of the tool to scenarios is not sufficiently justified. Also, based on review of Appendix E in the Fire Scenario Development report, the approach contains numerous assumptions that lack sufficient bases. The documentation provided is also insufficient to understand its application.

A sensitivity analysis is provided in Section 7.3.1 of the Fire Risk Quantification and Summary report; however, this study only addresses one assumption of the approach (i.e., credit for time delay) and does not appear to replicate NUREG/CR-6850 methods. In addition, the sensitivity study is discussed as producing more conservative non-suppression probability (NSP) values; however, Table 7-4 documents values both above and below those of the original MathCAD analysis.

Describe the approach documented in Appendix E of Fire Scenario Development report, including identification and justification of assumptions used in applying the approach. In addition, explain and justify how the approach is implemented for each type (or case) of fire scenario.

r) F&O FSS-D2-01 against FSS-D2:

No fire detection analysis was conducted in support of the activation of fixed suppression systems or the time to smoke detection. Appendix E of the Fire Scenario Development Report appears to indicate that only simplified assumptions, independent of the context of individual fire scenarios, were made.

For example, PAUs with automatic detection assume a detection time of 0 minutes; otherwise, manual suppression is assumed to occur at 15 minutes.

Provide further technical justification for these detection times as well as the results (i.e., CDF, LERF, CDF, and LERF) of a sensitivity analysis characterizing the uncertainty associated with their selection.

s) F&O FSS-D4-01 against FSS-D4:

According to Section 7.2 of the Fire Scenario Development report, the transient fires were generally assumed to be one foot above floor level unless a ledge or permanent scaffolding was present. Given that transient fire might occur at a Page 65 of 227

higher elevation, provide additional justification for this assumption. Additionally, explain whether the risk estimates reported in Attachment W are based on the 317 kW (98th percentile) HRR for transient fires per NUREG/CR-6850.

t) F&O FSS-D7-01 against FSS-D7:

The basis for fire detection and suppression system unavailability provided in Section 10.1 of the Fire Scenario Development Report is limited to an interview with a fire protection engineer and appears to only discuss system unreliability.

The intent for CCII is to additionally require a review of plant records to determine if the generic unavailability credit is consistent with actual system unavailability.

Generic values reported in NUREG/CR-6850 only provide estimates of system unreliability and do not include maintenance contributions to unavailability, credit for manual actuation of the system, dependent failures, and plant-specific data.

Provide additional justification, such as evaluation of plant records, that generic estimates bound actual system unavailability.

u) F&O FSS-D8-01 against FSS-D8:

Section 10.1 of the Fire Scenario Development report notes that fire scenarios, in general, credit automatic suppression to prevent the formation of a hot gas layer (HGL) or propagation beyond the zone of influence (ZOI); however, it is not clear that the effectiveness of the fire suppression and detection systems credited in the context of each analyzed fire scenario are assessed as required by CC-II of SR FSS-D4.

Describe and justify how the effectiveness of the fire suppression and detection systems were addressed on a fire-scenario basis to assure their effectiveness as credited in the FPRA.

v) F&O FSS-F1-01 against FSS-F1:

Describe how the FPRA addresses the possibility of effects of oil pooling, flaming oil traversing multiple levels, and oil spraying from continued lube-oil pump operation. In addition, confirm that the analysis considered scenarios involving other high-hazard fire sources as present in the relevant PAUs (e.g., oil storage tanks, hydrogen storage tanks and piping, mineral oil-filled transformers, etc.).

w) F&O FSS-F3-01 against FSS-F3:

The catastrophic turbine/generator (T/G) fire frequency analysis presented in Section 3.2.2 of the Structural Steel Analysis report is not consistent with NUREG/CR-6850 guidance; namely, information from NUREG/CR-6850 appears not to have been correctly applied. Table O-2 of NUREG/CR-6850 lists a frequency (i.e., 1E-5/yr) in the conditional probability column that is actually meant to be used as the fire frequency (as opposed to a conditional probability) for catastrophic turbine building fires. Moreover, this frequency already accounts for the conditional probability for catastrophic T/G fires (i.e., 0.025) and the failure Page 66 of 227

of fixed suppression preventing catastrophic damage with a probability of 0.02.

Evaluate the impact of this correction on CDF, LERF, CDF and LERF.

Additionally, Section 3.2.2 of the Structural Steel Analysis report indicates that credit for manual suppression is taken. If a suppression credit associated with a catastrophic T/G fire is desired beyond that which is quantified in NUREG/CR-6850 (namely, failure of fixed suppression with a probability of 0.02), provide a basis for this credit based on the severity of such a fire as outlined in Table O-2 of NUREG/CR-6850 and in light of plant specific training and conditions, as appropriate. Note that the NSP curve presented in Appendix P of NUREG/CR-6850 for T/G fires does not reflect the severity of the catastrophic T/G fire scenario.

x) F&O FSS-G2-01 against FSS-G2:

A primary screening criterion used in the multi-compartment analysis (MCA) is based on the exposing or exposed PAU being incapable of forming a HGL; however, it is not clear how related criteria are applied. For example, MCA Screening Criterion 3.02 for PAU 13 appears to screen based on an approximated volume of the full PAU; however, based on Table D-1 of the Fire Scenario Development report, this PAU is highly compartmentalized, and some of the fire zones within have the potential to form HGLs (e.g., 13G, 13E, etc.).

Explain and further justify the screening criteria used for PAU combinations, including how the sub-volumes (e.g., fire zones) within PAUs were addressed.

Also, if an exposed PAU is determined to be unable to support a HGL, clarify whether or not additional components (e.g., cables) within the exposed area could be damaged due to hot gas impingement from an exposing PAU via its propagation pathways (e.g., door plumes).

y) F&O FSS-G4-01 against FSS-G4:

Section 1.0 of the Multi-Compartment Analysis Report states that the MCA methodology may differ slightly from that specified in NUREG/CR-6850, but that the overall intent remains the same. Provide further clarification on this statement, including the differences between methodologies.

Also, Assumption 1 of the Multi-Compartment Analysis Report states that a fire damper exists between adjacent PAUs, noting that the most limiting boundary was not always obvious from review of design documents. In light of this assumption, provide an overview of the criteria used to assign failure probabilities to passive barriers. Discuss how credited barriers were confirmed to be consistent with their demonstrated fire-resistance ratings, and describe the treatment of passive fire barrier features that do not have an established fire-resistance rating or those barriers with openings. Lastly, given the use of generic barrier failure probabilities, discuss how it is verified that there are no Page 67 of 227

plant-specific barrier problems identified by the plant fire protection staff that may result in a higher failure probability.

z) F&O FSS-G5-01 against FSS-G5:

A review of the fire hazards analysis (FHA) indicates that some doors (e.g.,

PAUs 3, 21, 33, etc.) and dampers (e.g., PAUs 22, 23, etc.) between PAUs are held open with fusible links. Table A-1 of the Multi-Compartment Analysis Report indicates that failure of the fusible link to function appropriately is similar to a valve failing to close. Justify the basis for this assumption and the assigned barrier failure probability used for associated scenarios in the MCA.

In addition, discuss, in general, how the effectiveness, reliability and availability of active fire barrier elements (e.g., normally open fire doors or dampers closed upon detection, water curtains, etc.) were evaluated in the MCA, including how potential random and fire-induced failures were addressed.

aa) F&O FSS-H2-01 against FSS-H2:

The Plant Partitioning and Fire Ignition Frequency Development Report states that the Bin 19 frequency (misc. hydrogen fires) is apportioned by linear feet. The frequency report for Fire Area 04 (1-C Switchgear Room) appears to suggest that the frequency is based on 1 foot of piping; however, Scenario 04_FC01-3 of the Fire Scenario Development Report states that the hydrogen line enters in about door from turbine, runs along path of raceways and enters into Reactor Building along with raceways. Clarify this discrepancy and how hydrogen piping was traced and the Bin 19 frequency apportioned. Additionally, discuss how targets for hydrogen fires were established.

bb) F&O HRA-A4-01 against HRA-A4:

Post-Initiator Operator Action Questionnaires, Recovery Action Feasibility Evaluations and Validation Forms are provided in Attachment F; however, the attachment states that several reviews are currently in progress or under revision. A review of Attachment F indicates a number of actions do not appear to have been addressed. As a result, clarify the extent to which talk-throughs with plant operations and training personnel have been performed to confirm that interpretation of current and planned procedures relevant to modeled actions is consistent with plant operational and training practices.

cc) F&O HRA-B3-01 against HRA-B3:

Describe the extent to which the definition of HFEs takes into account scenario context, including timing, procedural guidance, instrumentation, task complexity, path of travel, etc. Include discussion of how accident-sequence-specific timing of cues and the time window for successful completion are addressed, including the methods used.

Page 68 of 227

dd) F&O HRA-C1-01 against HRA-C1:

The HRA screening analysis does not follow the guidance in NUREG/CR-6850 or in NUREG-1921, which updates NUREG/CR-6850. Provide the results of a sensitivity analysis (i.e., CDF, LERF, CDF and LERF) using screening/scoping approaches in NUREG/CR-6850 and/or NUREG-1921.

For self-approval, upgrade the fire HRA to account for relevant fire-related effects by using detailed analyses for significant HFEs and conservative estimates (e.g.,

screening values) for non-significant HFEs in accordance with CC-II for SR HRA-C1-01.

ee) F&O HRA-D1-01 against HRA-D1:

Given that screening HEP values are utilized in the FPRA for both HFEs currently modeled in the internal events PRA (IEPRA) as well as fire response actions, explain how operator RAs are included in the FPRA to provide a realistic evaluation of significant accident sequences, or discuss why not meeting this SR at CC-II or greater is acceptable for transition. In particular, clarify actions taken to address the peer review observation that top core damage fire scenarios do not account for realistic RAs. Provide a description of the RAs added for the risk-significant fire scenarios, their method of HEP quantification, and the resulting HEPs.

ff) F&O HRA-D2-01 against HRA-D2:

Post-Initiator Operator Action Questionnaires, Recovery Action Feasibility Evaluations and Validation Forms are provided in Attachment F; however, the attachment notes that several reviews are currently in progress or under revision.

A review of Attachment F indicates numerous actions do not appear to have been documented. Describe how modeled RAs account for relevant fire effects, including any effects that may preclude a RA or alter the manner in which it is accomplished. In addition, discuss whether or not RAs carried over from the IEPRA, (e.g., recovery of off-site power) also address fire-related effects accordingly.

gg) F&O IGN-B4-01 against IGN-B4:

The descriptions provided for plant-specific fire events in Appendix A do not provide a sufficient level of detail to determine whether established criteria for classifying events as potentially challenging are met or not. For instance, Event

1. 12 is classified as not challenging; however, Appendix C of NUREG/CR-6850 notes that electrical fires that self-extinguish after plant personnel de-energize the impacted equipment are generally classified as potentially challenging given that the act of de-energizing is one mechanism of active intervention by plant personnel. Provide additional justification for each fire event classified as not challenging in Appendix A and a documented basis as to why each established criteria are met or not met.

Page 69 of 227

hh) F&O PP-B2-01 against PP-B2:

Discuss whether or not partitioning credits wall, ceiling, or floor elements that lack a fire-resistance rating. In particular, provide further justification for those plant locations that were not specifically addressed in the FHA and identified as new PAUs in the FPRA (e.g., cooling tower pump house, feedwater purity building, etc.). Describe how NUREG/CR-6850 guidance was followed.

ii) F&O PRM-B11-01 against PRM-B11 and related SRs PRM-B12, PRM-13 and PRM-C1:

The disposition to this F&O does not address deficiencies associated with unmet SRs PRM-B12, PRM-B13 and PRM-C1. Identify any FPRA plant response model (PRM) probability input values that either required reanalysis given the fire context or that were not included in the IEPRA, and explain whether a data analysis was performed consistent with the requirements of PRM-B12.

jj) F&O PRM-B3-01 against PRM-B3:

Describe the modeling changes made to include the DC power dependency for the primary coolant pump breaker trip function. In particular, discuss credit given for operator actions to trip pumps to prevent seal failure for cases in which DC power is lost. In doing so, address the impact and modeling of Modification S2-5 (see LAR Attachment S, Table S-2, Item S2-5).

kk) F&O PRM-B9-01 against PRM-B9:

This F&O indicates that failure to trip pressurizer heaters was not explicitly addressed. Describe the modeling changes made to include the failure to trip pressurizer heaters. In particular, discuss credit given to operator actions to trip heaters both in the MCR and locally.

ll) F&O SF-A1-01 against SF-A1:

The peer review identifies numerous deficiencies associated with seismic fire interactions analysis as identified on Pages C-35 and C-36 of the final peer review report; however, the disposition provided states that the analysis has not been updated. Demonstrate that the scope of work performed meets the objectives of the Standard and addresses the deficiencies identified by the peer review. Also, discuss why the treatment from the individual plant examination for external events (IPEEE) remains valid. Given that the (United States Geological Survey (USGS) updated the seismic hazard curves (see Safety/Risk Assessment Results for Generic Issue 199, Implications of Updated Probabilistic Seismic Hazard Estimates in Central and Eastern United States in Existing Plants),

provide at least a qualitative discussion of these updated effects.

mm) F&O UNC-A1-01 against UNC-A1:

Page 70 of 227

This F&O notes that only a limited number of parameter and modeling uncertainties and associated assumptions have been identified, and that the associated analysis was incomplete and not of sufficient detail to support a reasonable characterization. In addition, it does not appear that sources of uncertainty originating from the IEPRA were reviewed to determine their impact on the FPRA application.

Describe and justify how key assumptions and sources of uncertainty for the FPRA model were comprehensively identified, documented and characterized. In this description, identify criteria used to judge the importance of assumptions and whether any sensitivity studies were performed as a result. Evaluate the impact on FPRA CDF, LERF, CDF and LERF values reported in Attachment W for the key sources of uncertainty and other uncertainties (such as fire phenomenological modeling uncertainties) considered for SR UNC-A2. Include in this evaluation the impact of using NUREG/CR-6850 frequencies for all bins.

Note that this is in addition to the sensitivity analysis for selected bins in FAQ 08-0048, Revised Fire Ignition Frequencies, (ADAMS Accession No. ML092190457, closure memo) (see PRA RAI 20) in that uncertainties on all bins needs to be quantified.

nn) F&O UNC-A2-01 against UNC-A2:

According to this F&O and F&O UNC-A1-01, the state of knowledge correlation (SOKC) for FPRA-specific parameters has not been addressed. In addition, uncertainty intervals assigned to fire ignition frequencies, spurious actuation probabilities, severity factors and non-suppression probabilities are noted as not being based on acceptable methods and differ in part from recommendations provided within NUREG/CR-6850. Describe and further justify how parametric data uncertainty was characterized per relevant SRs and propagated. Include discussion on how the SOKC was evaluated for fire CDF and LERF. Identify FPRA-specific parameters (e.g., hot short probabilities, fire frequencies) that can appear in FPRA cutsets and how they were correlated.

ENO Response a) The supplemental analysis work discussed in the disposition to this F&O considered proper polarity hot shorts on ungrounded DC circuits up to and including two independent faults. The completed cable analysis has been integrated into the fire PRA model and is no longer excluded from the risk estimates. The numerical effect will be reflected in the base quantification of the RAI Response Fire PRA model.

b) The treatment of coordination issues in the FPRA for power supplies and their associated components is discussed below. Treatment of the lack of coordination in the Fire PRA model is also discussed. Note that in many cases the power supplies for which coordination could not be verified were identified during circuit analysis to support the fire model for PRA components not previously analyzed.

Page 71 of 227

The power supplies for which coordination was not verified identified fall into three categories. The first category includes power supplies determined to be important to compliance or risk. For these power supplies it was determined that coordination is required and they were included in modification table (S2).

Category two represents power supplies for which coordination could not be verified and are included in the Fire PRA model but determined to be low risk contributors. These power supplies are failed in the Fire PRA model quantification as discussed below. Category three represents power supplies for which coordination was not verified and they are not required to support the Fire PRA model. These power supplies were identified by circuit analysis for which the supported function is not included in the PRA model. Consequently category three power supplies do not require treatment within the Fire PRA model.

The bypass regulator (comprising EX-161, EY-01 Bypass Regulator; EY-01-51A, Inverter #1 ED-06 Bypass AC Source from EY-01; EY-01-52A, Inverter #2 ED-07 Bypass AC Source from EY-01; EY-01-53A, Inverter #3 ED-08 Bypass AC Source from EY-01; EY-01-54A, Inverter #4 ED-09 Bypass AC Source from EY-01) provides an alternate supply to the four preferred ac power panels. The bypass regulator can supply one preferred ac panel at a time. The analysis determined that while in the normal configuration (all breakers to preferred ac panels open) there is no coordination issue. With a breaker to a preferred ac panel closed the breaker to the panel does not coordinate with the incoming breaker to the regulator. The procedural guidance for completing the alignment of the bypass regulator to a preferred ac panel requires confirmation that a faulted condition does not exist or is cleared prior to completion of the alignment. The process accounts for the possibility that the loss of the primary source may have been a faulted condition which would then impact the bypass regulator and the lack of coordination. Verifying a faulted condition does not exist assures that the alignment will not occur under an existing condition that would require coordination. Based on this lack of coordination, in the Fire PRA model the bypass regulator is assumed failed in the cable spreading room where the bypass regulator, breakers and cables to the preferred ac panels are located.

ED-211 (125 Volt DC Distribution Panel), ED-212 (125 Volt DC Battery Charger) and ED-213 (125 Volt DC Station Battery - Feedwater Purity Building) were identified as not having coordination. This dc power is required for the control circuits for the feedwater purity air compressors (C-903 and C-903B) and the load centers (EB-90 and EB-91) that power the compressors. Based on this lack of coordination, in the Fire PRA model the dc power supply is assumed failed in the feedwater purity building.

Coordination was not verified for load center 14. However, the analysis states that the lack of coordination only exists with the load center in a non-standard alignment. The analysis results were based on the load center 14 cross-connected to load center 13. In this lineup with load center 13 supplying load center 14 loads, coordination could not be verified. However, the analysis states that with load center 14 in the normal alignment, coordination does exist. The Page 72 of 227

PRA model does not credit the cross-connect alignment to power the load center 14 loads. Therefore no impact was assigned in the Fire PRA model.

Coordination was not verified for lighting panel EL-22. EL-22 is the power supply to solenoid valves SV-2008 and SV-2010. These solenoids control the air supplies to control valves CV-2008 and CV-2010. The control valves are part of an automatic makeup of condensate from the demineralized water storage tank (T-939) to the condensate storage tank (T-2). Coordination was verified for lighting panel EL-03 which supplies EL-22. Based on the lack of coordination lighting panel El-22 is failed in the Fire PRA Model. Currently the solenoid valves listed above are the only components included in the model that are powered from the panel.

The four lighting panels discussed below are not in the Fire PRA Model. They are associated with components whose function is not required in the Fire PRA model.

Lighting Panel EL-08 was identified as not having coordination. The panel supplies radiation monitor RIA-2323 in the PRA model. The radiation monitor is required in response to steam generator tube rupture events and is not included in the Fire PRA Model. Since the event and its power supply are not included in the Fire PRA Model treatment was not addressed.

Lighting Panel EL-25 was identified as not having coordination. The panel provides power to solenoid valve SV-1414. The valve controls fuel oil makeup to the day tank (T-24) to diesel-driven fire pump P-9B. The PRA model does not require fuel oil makeup to the fire pump day tank. Since the event and its power supply are not included in the PRA Model treatment was not addressed.

Lighting Panel EL-33 was identified as not having coordination. The panel provides power to high pressure air compressor C-6C air dryer M-9C. The function of the air dryer is not required in the PRA model to support compressor operation for 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />. Since the event and its power supply are not included in the Fire PRA Model treatment was not addressed.

Transformer EX-47 which feeds Lighting Panel EL-34 was identified as not having coordination. The panel provides power to solenoid valve SV-5353. The valve controls fuel oil makeup to the day tank (T-40) to diesel-driven fire pump P-41.

The PRA model does not require fuel oil makeup to the fire pump day tank. Since the event and its power supply are not included in the PRA Model treatment was not addressed.

c) Cable routing data verification has been completed and will be implemented into the RAI Response Fire PRA Model. There is no longer an exclusion from the risk estimates. The numerical effect will be reflected in the base quantification of the RAI Response Fire PRA model.

d) The Full Power Internal Events (FPIE) PRA model includes a comprehensive treatment of power supply, interlock circuits, instrumentation, and support system Page 73 of 227

dependencies. The model was developed in accordance with systems analysis high level requirement SY-A of ASME/ANS combined standard RA-Sa-2009, and includes modeling of interlocks and permissive circuits to the level of detail of interfacing relays and contact pairs. The fire PRA uses this logic model.

The phase 1 peer review recognized the completeness of the PRA logic model, but was concerned that the Appendix R circuit analysis used by the fire PRA at that time may not use the same boundary conditions or be completed to the same level of detail. Also, at the time of the phase 1 peer review, the circuit reanalysis of Appendix R components using modern criteria had just begun. The reanalysis included review of the applicable schematic diagrams, single-line diagrams, control wiring diagrams, instrument loop diagrams, and vendor drawings associated with each component, as required, for the purpose of identifying interlocks and electrical circuit interfaces. In addition, the ongoing circuit analysis for components explicitly treated in the fire PRA included a complete review of Safety Injection Signal (SIS), Containment High Pressure (CHP), Containment High Radiation (CHR), Containment Isolation Signal (CIS) and Recirculation Actuation Signal (RAS) logic to identify potential adverse component actuations and assure complete treatment in the fire PRA. Although satisfied with the intent of these two efforts, there was insufficient progress at the time of the phase 1 peer review to assure that the SR was met.

This F&O was re-visited in more detail during the phase 2 and phase 3 Peer Reviews, satisfying the reviewers that the circuit analysis process captures the necessary circuits and interlocks. The F&O was not closed however, as the circuit analysis tasks were not fully completed. Subsequent to the phase 3 Peer Review and prior to the LAR submittal, the Appendix R circuit reanalysis and circuit analysis of new components were completed. During the course of the project, the circuit analysis results were integrated into the fire PRA using a series of data uploads. The final upload was not completed prior to the LAR submittal, therefore this F&O was considered open.

The evaluation of circuits that could potentially affect accident mitigating equipment, including the associated cable tracing is complete. The final upload has been completed and the circuit analysis data will be implemented in the RAI Response Fire PRA model. The numerical effect will be reflected in the base quantification of the RAI Response Fire PRA model. This new base quantification will provide the revised risk results. Therefore, no further action is necessary regarding self-approval.

The evaluation of circuits for instrumentation to support operator cues is addressed in RAI PRA 01f.

e) A review was conducted against the CEOG guidance regarding the PCP seal module update. The MSO report requires updating to reflect that the review did not require changes to the seal logic model. The indicated review was completed.

The PCP seal logic development was maintained consistent with the proposed Page 74 of 227

changes such that when the guidance was officially issued the seal failure logic was consistent with the guidance. Therefore no change to the seal logic was required in the Fire PRA Model as a result of the review. A model update was separately completed prior to the LAR submittal to address multiple spurious operation issues, a more complete treatment of pump trip logic and the proposed modification to install an alternate pump trip capability.

References:

[1] WCAP-15749-P, Revision 1, Guidance for the Implementation of the CEOG Model for Failure of RCP Seals Given Loss of Seal Cooling (Task 2083),

Combustion Engineering Owners Group (CEOG), December 2008.

f) Instrumentation needed to support modeled operator actions was identified through procedure reviews and operations input. The PNP NFPA-805 Project Team includes a previously licensed PNP SRO who has been reviewing the operator actions being credited in the fire PRA model for timing, crew requirements, procedural direction and instrumentation requirements. This instrumentation information was presented to currently licensed, on-shift personnel and training personnel for their review and concurrence. The reviews performed by Operations staff are documented in the HRA Notebook Volume 1.

Circuit analyses of the identified instrumentation for the dominant operator actions have been completed. The identified instrumentation and the circuit analysis results will be used in the determination of the viability of the operator actions.

That is, if sufficient instrumentation is not available for a given fire scenario, then the action will be assumed to be failed for that scenario. The numerical effect of these impacts on the human reliability analyses for operator actions credited in the fire analysis will be included in the base case results of the RAI Response Fire PRA Model.

Completion of the on-going identification and treatment of instrumentation for credited human failure events (HFEs) in the FPRA, as well as associated cable tracing (or equivalent) are required for self-approval.

g) Undesired Operator Actions The following describes the process for systematically identifying and defining human failure events that may result in an undesired operator response (i.e., error of omission or commission) to spurious cues and indication during fire scenarios as recommended in Section 3.4.1 of NUREG-1921 for the PNP fire PRA.

Undesired operator actions may be taken in response to spurious indications.

These actions are well intentioned but result in aggravating or complicating a scenario. As indicated in NUREG/CR-6850 (Volume 2, Section 2.5.5.2) it is unlikely a single fire can affect so many indications as to prevent the operator Page 75 of 227

knowing when to take the desired actions or fool the operator into taking an undesired action.

Based on this philosophy, the ASME fire PRA standard supporting requirement ES-C2 limits the number of spurious indications that must be considered as potentially affecting operator actions (either negatively affecting an action that is required or causing an operator to take an undesired action). Per the fire PRA standard for capability category II in ES-C2, each operator action needs to be examined from the standpoint that there are (a) none or (b) as many as one spurious indication occurring as a result of the fire.

This assumption is based on consideration of the typical redundancy and diversity available in nuclear plant indications as well as a practical constraint on resources required for tracing instrumentation cables. Per PNP procedure EN-OP-115 (Conduct of Operations), operators use diverse indications whenever possible.

Based on this procedure and the procedure reviews and simulator exercise described below, no new operator actions (errors of commission) are postulated in response to a single spurious indication failure.

Alarm response procedures (ARPs) were reviewed to identify potential undesired operator actions that can result from an annunciator or alarm. ARPs reviewed included those that involve equipment and systems modeled in the fire PRA.

The following assumptions were made to reasonably bound the number of undesired operator actions in accordance with capability category II of the fire PRA standard:

Actions that require multiple spurious indications on different parameters can be screened from consideration.

Actions that require multiple spurious indications on redundant channels can be screened from consideration.

Actions that include a proceduralized verification step can be screened from consideration if the verification will be effective given the fire scenario.

The ARP review did not identify any new operator actions (errors of commission) in response to a single spurious indication failure.

Emergency operating procedures (EOPs) were reviewed to identify all steps in which an undesired operator action can result. EOPs reviewed included those that operators could be expected to perform following a fire-induced initiating event.

The same screening assumptions used in the ARP review were applied for the EOP review.

The EOP review did not identify any new operator actions (errors of commission) in response to a single spurious indication failure.

Page 76 of 227

To confirm these results, PNP staff conducted a simulator exercise of operator responses to postulated instrumentation failures and spurious indications. The exercise was conducted over several days and included participation of three to four licensed plant operators. The exercise was conducted in the plant simulator area (the simulator was not running a specific transient, but the panels, instruments, and gauges of the simulated control room were energized).

During the exercise operators were asked in detail to respond to various sets of postulated conditions, formulated along the general guidelines of event is real, instruments are faulty and event is not real, instruments are faulty. The operators were asked to explain what they would do given certain indications, or lack of indications, in the control room. In no situation did the operators suggest taking (or not taking) a step that would have exacerbated the situation. No acts of commission were identified. The exercise description and results have been documented in Volume 1 of the HRA notebook NB-PSA-HR.

Treatment of Annunciators The following describes how the annunciator system is treated in the PNP fire PRA, particularly regarding the impact on human failure events.

Annunciators are not credited as primary cues in the PNP fire PRA since the annunciator system may be impacted by fires in almost all physical access units (PAUs) and detailed cable tracing was not performed. Note the potential for annunciator to result in undesired operator actions was discussed above.

Parameters and related indication (instrumentation, component status indication, etc.) providing cues for operator actions having the most significant impact on CDF and LERF were identified and classified as Primary, Secondary or Tertiary.

Primary indication is defined as the most readily available and likely for operators to monitor. Secondary and Tertiary indication is defined as being progressively more remote and/or less likely to be observed. Instrument cable mapping was then performed to determine what indications would be available for fires in various fire areas. Annunciators are not credited as cues due to their potential for being impacted by fires in many PAUs. The all inclusive fire response procedure developed for NFPA 805 implementation will include information regarding indication availability/reliability and the need for condition validation prior to taking action.

h) Reference [1] documents the processes used for and the results of the human reliability analyses performed for the full power internal events (FPIE) probabilistic risk assessment (PRA), as well as for the fire PRA (FPRA). Section 5.2 of Reference [1] describes the dependency analysis; the paragraphs below restate the dependency evaluation approach taken for the PNP FPIE and fire PRAs.

The dependency analysis evaluates the dependency between the multiple operator actions that occur in the accident sequences of the PNP risk Page 77 of 227

assessments. The human reliability analysis (HRA) developed human error probabilities (HEPs) as though they were independent of one another. It is known that a number of these operator actions appear in the same accident sequences. If dependencies exist between these operator actions, then the core damage frequency (CDF) may be higher than quantified in the accident sequence analysis.

This dependency analysis evaluated the post-initiator dependencies among operator actions credited in the PNP FPIE and fire PRAs and determined whether the impact of these dependencies on the overall core damage frequency is significant. Following identification of the most risk significant human error dependencies, conditional human actions are developed and incorporated explicitly in the PNP PSA fault trees.

The analysis identifies the combinations of human actions that would appear in the accident sequences initially assuming complete dependence for all HFE combinations and subsequently developing dependent failure probabilities and fault tree logic for combinations of human actions that would have a significant impact on CDF.

Process Used to Identify and Evaluate HFE Combinations The general steps used in this analysis to identify and evaluate HFE combinations are as follows:

1. Run the base model with the post-initiator action failure event probabilities set to 1.0.

The human error probabilities for all HFEs are set to 1.0 within the basic event database used for quantifying the PRA logic models (fault trees and event trees). The logic models are then quantified, using a truncation limit low enough to ensure that cut sets containing potentially important HFE combinations are generated; a truncation limit of 1E-8 is typically chosen. The limit is chosen to be as low as possible while still allowing the quantification to proceed; if the truncation limit is too low, the number of cut sets generated will be so numerous that the quantification process will fail. Combining these cut sets with those from the baseline PRA produces an equation that includes the original PRA results plus hundreds of thousands of additional cut sets that are functions of combinations of HFEs.

2. Identify the multiple human action combinations that appear in the cut sets.

The cutsets generated in Step 1 are used as input to identify HFE combinations. HFE combinations for PNP are identified using the EPRI HRA Calculator (Reference [2]). Several thousand combinations are identified using this approach. (Note: a combination consists of at least two HFEs within a single cutset.)

Page 78 of 227

3. Identify the risk significant combinations.

Using guidelines described in Reference [2] for the assessment of dependence among post-initiator HEPs, the dependent HEPs are calculated.

The HRA Calculator provides a default set of dependencies for each combination, given the information generated during the human reliability analyses performed for each HFE (such as information about location of action, cues, time windows, etc.; this information is provided by the HRA analyst for all HFEs, including those using screening values for their HEPs).

The default dependencies are reviewed and modified as necessary to reflect accident sequence-specific conditions that may vary from those assumed for the default HRA. Once the level of dependence is established, the HEPs for the conditional (dependent) HFEs are calculated using the equations contained in Reference [2] (e.g., for zero dependence, the conditional HEP is equal to the independent HEP; for complete dependence, the conditional HEP is equal to 1.0).

The assessment of the increase in CDF resulting from each of the human action combinations assessed for dependencies is performed with an R&R Workstation based code, DepHep3 (Reference [3]). Using an iterative approach, risk-significant combinations are identified and the dependencies within those combinations explicitly evaluated. DepHep3 performs a bounding assessment of the completeness of the dependency analysis by including the conditional events that have been explicitly developed and assuming all non-assessed combinations are completely conditional (i.e., complete dependence exists between all HFEs in the non-assessed combinations). After each quantification, additional potentially risk significant combinations are explicitly evaluated, and the quantification completed again, until the increase in CDF due to dependencies converges (i.e., the change in CDF for one quantification run and the next is approximately the same).

Using this approach, the change in CDF when HFE dependencies are incorporated is calculated. The CDF increase is attributed to only a handful of HFE combinations (as opposed to the several thousand that exist throughout the cutsets). The techniques used in the DepHEP 3 calculations show that the potential contribution from non-assessed combinations would only increase CDF by a small fraction assuming complete dependence. An analysis of the total increase in CDF across all accident sequences due to unevaluated combinations results in a potential increase of only a few percent of the total CDF. Demonstrating that a specific subset of HFE combinations accounts for the majority of the potential increase in CDF due to HFE dependencies, and that the total potential contribution from all non-assessed combinations is small, is a good indication that the evaluation performed, focusing on the risk significant combinations, is complete.

Page 79 of 227

Justifications for Assumptions Identified as Non-Conservative in EA-PSA-FPRA-HEPDEP Although there are assumptions identified as non-conservative in Section 5 of Reference [4], they are not necessarily exceptions to NUREG-1921. Each of the assumptions labeled as non-conservative in Reference [4] is listed here.

The justification for the assumption is also provided.

1) Reference [4], Section 5.1.3 - Adequate staff is available to complete all actions that might be required during a fire sequence, regardless of the number of actions (some of which may be simultaneous).

Basis: Staffing levels had not been finalized as of the preparation of this calculation note, and were under evaluation. The results of the FPRA (sequence analysis) are being used to identify important sequences in which multiple operator actions may be required in order to mitigate the impacts of the fire, and that the staffing levels will then be established to ensure an appropriate response to such sequences is achievable.

This assumption is considered valid given that the results of the FPRA are in fact being used to identify and assess potential changes to procedures, staff size, and/or fire-brigade assignments.

2) Reference [4], Section 5.1.4 states, When determining dependency levels, the impact of stress is considered to be incorporated into each individual HFE comprising the combination under evaluation. No additional stress is assumed as a result of the combination itself.

Basis: A decision tree is used in the determination of the dependency between actions. The HRA Calculator employs a decision tree that has as its headings the categories associated with cues, timing, adequate resources, location, and stress. Moderate or high stress is assumed by the calculator for a combination if either of the actions being assessed for the combination have moderate or high stress levels assigned as part of the examination of performance shaping factors completed during the human reliability analysis (HRA) for each independent action. In this case, the moderate or high stress branch of the decision tree would be selected, and the dependency level between two HFEs increased over the level that would be assigned if both HFEs had low stress as individual actions. This increases the conditional (dependent) HEP over what it would be if the low stress branch were selected. For HFEs in the combination for which individual HEPs have been developed assuming moderate or high stress, there appears to be nothing that warrants an additional increase to the HEPs over what is already incorporated in the individual HRAs. In fact, such consideration appears to double count the impacts of stress, resulting in a higher conditional HEP than should be used.

Page 80 of 227

NUREG-1921 states that Stress is a culmination of all other performance shaping factors. These may include preceding functional failures and successes, preceding operator errors or successes, availability of cues and appropriate procedures, workload, environment (heat, humidity, lighting, atmosphere, and radiation), requirement and availability of tools or parts, accessibility of locations. In general, stress is considered high for loss of support system scenarios or when the operators need to progress to functional restoration or emergency contingency action procedures -

the closer they get to exhausting procedural options, the higher the stress. NUREG-1921 does not explicitly suggest adding additional stress multipliers during the performance of dependency analyses.

For the FPRA, screening values were employed for the HEPs. These values were chosen by the PNP PRA analysts and influenced by their judgment and knowledge of the complexity, timing, and other performance affecting variables that may exist during fire scenarios. Stress is implicitly included as one of the parameters that may affect an operators ability to perform a given action.

3) Reference [4], Section 5.1.5 states, When determining dependency levels, actions are assumed to take place in different locations if the locations are physically distinct and separated by distance, even if in the same room. Thus, an action taking place at the north end of a room is considered as a different location from an action taking place at the south end; an action taking place at panel EC-03 in the control room is considered to be at a different location from an action taking place at panel EC-12, even though both actions are taking place in the control room.

Basis: For this analysis, it is assumed that actions taking place at different panels, or in different areas of a room, will provide sufficient decoupling to reduce the potential for dependency. This is in contrast to the treatment of location in NUREG-1921. NUREG-1921 states ((L)ocation refers to the room or general area where the crew members are located. For example, the control room is a location - location is not differentiated down to individual panels in the control room).

Taken in its extreme the NUREG-1921 approach implies that a single error in the control room, for example, negates the possibility of success of all subsequent actions that are also taken in the control room. In other words, there would be high or complete dependence between all actions in the control room following the failure of one action. This is not a realistic approach to operations in a control room, and thus this aspect of the NUREG-1921 approach as it regards location of actions in the control room was not applied.

Page 81 of 227

Timing and Stress Levels Associated with HFE Combinations Time windows for determining when actions could be effectively implemented are based on combinations of deterministic analyses (e.g., MAAP runs) and, when it exists, experience (e.g., operating experience following transients that have occurred at the plant). Reference [5] contains a listing of MAAP analyses utilized for this evaluation.

MAAP runs were supplemented by walk-downs and interviews, especially for response times and manipulation times. A walk-down of the PNP simulator was performed as part of this task during the 2009 Update. Operator interviews were also conducted to address risk significant actions during the 2009 update. The interview insights are documented in the applicable post-initiator action HEP bases in Appendix A of Reference [1]. Additionally; the post-initiator HEP calculations were reviewed in detail by a former PNP SRO license holder. The time estimates for credited actions are based on the analysts understanding of the procedural steps and information collected from interviews with knowledgeable plant staff. References [6] through [9]

document the walk-down, operator interviews, and SRO review. Interviews and reviews are continuing to support the updated FPIE and fire PRAs.

For the FPRA included in the LAR submittal, screening values are used exclusively for all HFEs; the assignment of the screening values used for the HEPs took into account parameters such as complexity of the required action, timing, location, and stress. When considering stress levels for individual HFEs as part of the assignment of screening values, consideration was given to the guidance provided in the HRA Calculator (Reference [2]), in which work-load and other performance shaping factors are considered. When detailed HRAs are performed, the stress level may be modified (higher or lower) following review by the PRA analysts, taking into account sequence-specific parameters that may differ from the issues incorporated into the default levels determined using Reference [2].

References:

[1] NB-PSA-HR Vol 1, Rev. 4, Human Reliability Analysis Notebook Volume 1 (Post Initiator Operator Actions)

[2] The EPRI HRA Calculator Software Users Manual, Version 4.21, EPRI, Palo Alto, CA, and Scientech, a Curtiss-Wright Flow Control company, Tukwila, WA: 2007. Software Product ID #: 1022814.

[3] Dependent Human Error Analysis Program DEPHEP 3 User Manual, Applied Reliability Engineer, Inc.

[4] EA-PSA-FPRA-HEPDEP, Fire Probabilistic Risk Assessment Human Failure Event Dependency Analysis, Rev. 0.

Page 82 of 227

[5] PLP0247-07-0004-01, Palisades Nuclear Plant Thermal Hydraulic MAAP Calculations, Revision 2.

[6] Palisades Nuclear Plant Simulator Walkdown, August 27, 2009

[7] Palisades Nuclear Plant Operator Interviews, August 26, 2009

[8] Palisades Nuclear Plant Operator Interviews, August 27, 2009

[9] Palisades Nuclear Plant Post-initiator HEP Calculation Reviews, April-September 2009 i) Cutset reviews were performed at the scenario level during quantification. This included a mix of both dominant and non-dominant cutsets. Criteria described in Entergy PRA guidelines were employed in this assessment. The aggregate results were then reviewed for reasonableness based on the dominant contributors to CDF and LERF. The presentation of dominant results was completed after the final phase of the peer review in Section 6 of the Fire Risk Quantification and Summary report used to support the LAR model [1]. This included the following results:

a discussion of the top fire scenario contributors to CDF and LERF, PAU contributions to CDF and LERF, significant equipment failure contributions to CDF and LERF, and significant Human Failure Event (HFE) contributions to CDF and LERF.

The importance analysis for CDF and LERF does include the individual contributions from common-cause failures, hot-short probabilities, and non-suppression probabilities, but these were not formally presented in the Fire Risk Quantification and Summary report.

The presentation of dominant results from the aggregate cutsets will be repeated in the revision to the Fire Risk Quantification and Summary report to support the RAI Response Fire PRA model. In addition to the dominant contributors by scenario, PAU, equipment failure, and HFEs, the presentation will be expanded to include equipment common cause failures and FPRA-related parameters (hot short probabilities, non-suppression probabilities, and severity factors). A complete listing of CDF and LERF importance measures from the aggregate CDF and LERF cutsets will also be provided.

References:

[1] 0247-07-0005-01, Revision 1, Palisades Nuclear Plant Fire Probabilistic Risk Assessment Fire Risk Quantification and Summary, ERIN Document, November 2012.

Page 83 of 227

j) The referenced statement: All Motor Control Centers (MCC) have been treated as closed, sealed and robust in which damage beyond the ignition source will not be postulated refers to MCCs in plant locations where fire scenarios were developed.

MCCs located in areas that were treated by Physical Analysis Unit (PAU) bounding scenarios are not explicitly evaluated. In addition, there are two non-MCC cabinets that were also treated as sealed and robust; these cabinets are located in the Cable Spreading Room (CSR), PAU 2.

Consistent with FAQ 08-0042, the PNP Fire PRA Model treated cabinets as sealed and robust provided they met the following criteria:

No ventilation openings o Doors do not require rubber gaskets as a tight fit will sufficiently restrict the passage of air, thus limiting the size of a potential fire Cable penetrations into top or sides are sealed such that they do not readily allow for the passage of air o Seals do not need to be fire-rated o House-keeping seals are sufficient Doors must be attached and anchored at multiple points such to limit the amount of warping that is possible o Multiple mechanical fasteners, such as thumb screws, are sufficient o Twist-handle top-and-bottom door latches with an additional anchor point in the middle of the door are considered sufficient as they provide 3 connection points on the opening side of the door.

o Simple twist-handle top-and-bottom door latches without an additional anchor in the middle are not sufficient Table 1 lists the MCCs and other electrical cabinets treated as closed, sealed, and robust. Table 2 lists the MCCs that were not treated as closed, sealed, and robust, and therefore do not require additional walkdowns or further justification.

Page 84 of 227

Table 1: Electrical Cabinets Treated as Sealed and Robust FIS Description Comp Type PAU ED-10 125 V DC Bus No.1 (Includes ED-11) MCC 2 ED-20 125 V DC Bus No.2 (Includes ED-21) MCC 2 EB-01 MCC 1 MCC 2 EB-02 MCC 2 MCC 2 EB-03 MCC 3 MCC 23 EB-04 MCC 4 MCC 23 EB-05 MCC 5 MCC 9 EB-06 MCC 6 MCC 9 EB-07 MCC 7 MCC 15 EB-08 MCC 8 MCC 15 EB-09 MCC 9 MCC 14 EB-10 MCC 10 MCC 23 EB-21 MCC 21 MCC 2 EB-22 MCC 22 MCC 3 EB-23 MCC 23 MCC 2 EB-24 MCC 24 MCC 2 EB-25 MCC 25 MCC 21 EB-26 MCC 26 MCC 21 EC-09 Status Interface Panel EP 2 TCAC 2 Terminal Box/Junction Box EP 2 Page 85 of 227

Table 2: MCCs Not Treated as Sealed and Robust FIS Description Comp PAU Notes Type EB-81 MCC 81 MCC 17 PAU is analyzed by a single PAU EB-82 MCC 82 MCC 17 bounding scenario. Therefore, the MCCs are not treated as sealed and robust.

MCCs are included in PAU bounding scenario.

EB-18 VRS MCC MCC 27 PAU is analyzed by a single PAU EB-79 MCC 79 MCC 27 bounding scenario. Therefore, the MCCs are not treated as sealed and robust.

EB-80 MCC 80 MCC 27 MCCs are included in PAU bounding scenario.

EB-92 MCC 92 MCC 39 The analysis for PAU 39 is being revised EB-93 MCC 93 MCC 39 in the RAI Response Fire PRA Model to be a PAU bounding scenario. Therefore, EB-94 MCC 94 MCC 39 the MCCs will not be treated as sealed EB-95 MCC 95 MCC 39 and robust. MCCs are included in PAU EB-96 MCC 96 MCC 39 bounding scenario.

EB-97 MCC 97 MCC 39 As shown in Table 1, the MCCs treated as closed, sealed and robust are ED-10 and ED-20, EB-01 through EB-10, EB-21 through EB-24, and EB-25 through EB-

26. Note that the MCCs within the 4 groups listed are of the same construction as other MCCs within the group.

There are also 2 non-MCC panels treated as closed, sealed and robust in the PNP LAR model. These two panels (EC-09 and TCAC-2) are located in the CSR.

A walkdown (external) of each of the MCCs and cabinets listed in Table 1 was performed. In addition, a representative selection of MCCs was opened by a walkdown team to confirm the robust locking mechanisms used to secure the panel doors and sides. Further justification for the treatment of the equipment as closed, sealed and robust for each of the groups in Table 1 is provided below:

ED-10 and ED-20 are DC MCCs o No ventilation openings are present.

o All the cable penetrations have house-keeping seals.

o Each cubicle door is fastened with multiple screws and therefore meets the definition as robust.

o The cubicle doors do not contain rubber gaskets, however if a gap between the door and the face of the MCC exists, it is of a minimal size Page 86 of 227

that the flow of air into the MCC is significantly small.

EB-01 through EB-10 are 480 VAC MCCs o No ventilation openings are present.

o All the cable penetrations have house-keeping seals.

o Each cubicle door is fastened with multiple screws and therefore meets the definition as robust.

o The cubicle doors do not contain rubber gaskets, however if a gap between the door and the face of the MCC exists, it is of a minimal size that the flow of air into the MCC is significantly small.

EB-21 through EB-24 are 480 VAC MCCs o No ventilation openings are present.

o All the cable penetrations have house-keeping seals.

o Most cubicle doors are fastened with multiple screws and therefore meet the definition as robust.

o Cubicle doors that are fastened with only one screw are much smaller than a typical control cabinet door. Therefore it is judged that the likelihood of the door warping enough to allow sufficient air flow is still smaller than the warping that could occur on a multi-point fastened typical control cabinet.

o The cubicle doors do not contain rubber gaskets, however if a gap between the door and the face of the MCC exists, it is of a minimal size that the flow of air into the MCC is significantly small.

EB-25 through EB-26 are 480 VAC MCCs o No ventilation openings are present.

o All the cable penetrations have house-keeping seals.

o Most cubicle doors are fastened with multiple screws and therefore meet the definition as robust.

o Cubicle doors that are fastened with only one screw are much smaller than a typical control cabinet door. Therefore it is judged that the likelihood of the door warping enough to allow sufficient air flow is still smaller than the warping that could occur on a multi-point fastened typical control cabinet.

Page 87 of 227

o The cubicle doors do not contain rubber gaskets, however if a gap between the door and the face of the MCC exists, it is of a minimal size that the flow of air into the MCC is significantly small.

TCAC-2 is a large terminal box in the CSR o No ventilation openings are present.

o Only conduits penetrate the cabinet.

o The doors are attached by robust piano hinges on one side and appear very rigid.

o It appears that there are fasteners at the top and bottom of the panel door as well as a pad lock in the middle of the doors o No visible gaps around the doors were visible EC-09 is a relatively small cabinet that appears to only contain telecommunication wiring o No ventilation openings are present.

o Only conduits penetrate the cabinet.

o Combustible loading in the panel is considered low given the quantity and type of cables present.

o The door opens by pulling up on the handle which causes rods at the top and bottom of the door to retract.

It is recognized that FAQ 08-0042 states that simple twist-handle style top-and-bottom door latches are not sufficient to contain a fire within a panel. Substantial warping of the door face may occur due to the heat of the fire. However, given the low combustible loading in the cabinet, the top-and-bottom door latches are considered to be acceptable as the heat of the fire would likely be sufficiently lower than a typical multi-bundle electrical cabinet o No visible gaps around the doors were visible In summary, as all the cabinets treated as closed, sealed and robust were judged to meet the criteria established in FAQ 08-0042, risk estimates treating cabinets as open is not required.

k) Main Control Room abandonment due to loss of control or function is not modeled in the FPRA. Main Control Room abandonment is postulated in scenarios for which Main Control Room habitability is lost.

Page 88 of 227

l) Part l): MCR Abandonment CDF and LERF Fires in the control room have the potential to challenge habitability or visibility due to excessive heat or smoke generation. The abandonment analysis determines which scenarios require evacuation of the control room. Multiple control room abandonment scenarios are quantified for the fire PRA.

CDF and LERF for the PNP control room abandonment scenarios are calculated directly and not estimated. Calculation of CDF and LERF for control room abandonment scenarios is performed as in other quantified scenarios. Scenario-specific fire ignition frequencies are combined with scenario-specific non-suppression factors, severity factors and conditional core damage probabilities (and conditional large early release probabilities).

The calculated CCDP (and CLERP) values for each scenario reflect the loss of fire-affected equipment and instrumentation in the same manner as quantified scenarios. Operator actions from the control room that rely on operation of equipment and controls impacted by the fire scenario are not credited and considered failed. Operator actions from the control room that rely on equipment and controls that are not impacted by the fire scenario but that cannot be performed prior to the time of control room abandonment are also not credited and considered failed. Operator actions outside the control room, for example, the operation of the turbine driven auxiliary feedwater pump from the alternate shutdown panel, are credited unless fire effects precluded completion of the action (i.e., component to be manipulated is in the fire or pathway(s) to the component are fire affected preventing accessibility).

The PNP alternate shutdown strategy involves operating the turbine driven AFW pump (TDAFW) at the alternate shutdown panel. The structure of the PNP PRA model inherently credits local operation of the TDAFW pump if both the motor driven AFW pumps and remote operation of the TDAFW pump are impacted by a fire scenario. However, each scenario is reviewed to ensure that proper credit for the alternate shutdown method is present in the scenario. Some scenarios include failures that preclude the use of the alternate shutdown strategy, such as primary coolant pump seal LOCAs. The PRA model logic is constructed in such a way to account for these dependencies including representation of the operator action associated with using the TDAFW pump at the alternate shutdown panel.

For the LAR, a screening HRA approach was utilized for pre- and post-abandonment operator actions not already considered failed. The RAI Response Fire PRA Model will be quantified utilizing the HRA approach in NUREG-1921 for pre- and post-abandonment operator actions not already considered failed.

NUREG-1921 screening and scoping human error probabilities are utilized for pre-and post-abandonment operator actions from the control room except the actions to trip primary coolant pumps, the action to trip charging pumps, and the action to trip pressurizer heater breakers. These actions are required to be completed prior Page 89 of 227

to control room abandonment. Given the relative importance of these actions to the quantification result, detailed HEPs were developed to avoid the conservative penalties resulting from the use of screening and scoping HEPs.

NUREG-1921 screening and scoping human error probabilities are utilized for pre-and post-abandonment operator actions occurring outside the control room except the action to refill the emergency diesel generator fuel oil day tanks. This action is required to be completed prior to 18 hours2.083333e-4 days <br />0.005 hours <br />2.97619e-5 weeks <br />6.849e-6 months <br /> after the fire event initiation. Given this long system window, a detailed HEP was developed to avoid the conservative penalties resulting from the use of screening and scoping HEPs.

Part I.i) HFE Feasibility Assessment Results The feasibility assessment of operator actions associated with HFEs has been completed for important HFEs. Each of the criteria in Section 4.3 of NUREG-1921 has been addressed as described in PRA RAI 01cc. A status of complete means reviews, comment incorporation and signature concurrence on the forms described below have been completed. Feasibility assessments associated with HFEs that are not important to risk have not completed all phases of feasibility documentation. HFEs with this status are not credited in the fire PRA.

Fire-specific Post-Initiator Operator Action Questionnaires (P-IOAQs), Recovery Action Feasibility Evaluations and HEP Validation forms were developed by previously SRO licensed personnel, and reviewed and approved by current SRO licensed Operations personnel and the Assistant Operations Manager - Training.

Operations reviews included verification that listed cues, procedure use, manpower requirements and performance time were reasonable, and that the actions were feasible. Training reviews included verification that listed training was correct and P-IOAQ information reflected training expectations. Acceptable Operations and Training review comment incorporation is documented by signatures on the HEP Validation forms.

Only actions determined to be feasible without qualification or feasible pending deficiency resolution are credited and assigned scoping or detailed HEP values.

An example of feasible pending deficiency resolution would be when a procedure needs to be written/modified to accomplish an action as is the case for the operation of the planned new auxiliary feedwater pump. Detailed descriptions of P-IOAQ, Feasibility Evaluation and Validation content, and the review process are provided in PRA RAI 01cc.

Part l.ii)

The operator action associated with transfer of control to the alternate shutdown panel is the action to transfer TDAFW pump control to the alternate shutdown panel. The action is part of the HFE identified in the fire PRA model as AFW-PMOE-EC-150-FR [OP FT XFR AFW PP START TO C-150 PNL (LOC) (HEP)].

Page 90 of 227

This HFE will be assessed via the scoping method of NUREG-1921, Section 5.2.7, for the RAI Response Fire PRA model. The table below describes the results of the process in Section 5.2.7 of NUREG-1921 for assigning scoping HEPs and specifically addresses the basis for the answers to each of the questions asked in the Figure 5.4 flowchart.

Scoping HEP for AFW-PMOE-EC-150-FR per NUREG-1921 Section 5.2.7 ID Flowchart ID Text Answer Basis D22 Has the fire been No Cues are generally expected to be suppressed before the cue received prior to suppression of the fire.

is received? Specific cues include: control room annunciators for low steam generator level; AFW flow indicators; SG narrow range level indicators; and SG wide range level indicators. Also, inadequate heat removal could be determined using PCS temperature indication. Also, local AFW flow indicators are available in the CCW room.

Also, general cues (such as loss of control or indication or smoke and/or ambient temperature conditions in the control room) can preempt other specific cues and lead the Shift manager to call for control room abandonment.

Given abandonment would generally occur prior to fire suppression, and that control room fires are considered challenging, the cue is considered to occur before the fire has been suppressed.

D26 Are both conditions met: 1) Yes Fires resulting in the need for control the area is accessible and room abandonment and operation of TD

2) there is no fire in the AFW pump from the alternate shutdown vicinity of the action? panel are not expected at the location of the alternate shutdown panel or on the path to this location from the control room.

Multiple ways to exit the control room and reach the alternate shutdown panel exist. The action does not take place in the direct vicinity of the fire.

Page 91 of 227

Scoping HEP for AFW-PMOE-EC-150-FR per NUREG-1921 Section 5.2.7 ID Flowchart ID Text Answer Basis D27 Is the time available (Tavail) Yes Tsw is 150 minutes based on greater than 30 minutes? unrecoverable loss of decay heat removal. Tdelay is 20 minutes and represents the time to complete EOP-1.0 or the time to abandonment. Tcog is 2 minutes and represents the time to detect, diagnose, and decide to leave the control room. Texe is 20 minutes and represents travel time and manipulation time (Note this is conservative with respect to the walk-through which indicated 10 minutes).

Therefore, Tavail is 130 minutes.

D33 Is the execution complexity No Execution involves positioning two high? power supply breakers to the on position and manipulating two handswitches to disconnect control room circuits and bring the alternate shutdown panel on-line. This action is covered in ONP 25.2 training, which is part of the 2 year training plan. Two JPMs are provided to place the alternate shutdown panels in service:

PL-OPS-ONP-005J and PL-OPS-ONP-006J.

D34 Is there smoke or other No See basis for D26. Hazardous hazardous elements in the chemicals are not stored in the vicinity vicinity? of the south west cable penetration room.

This results in HEP Lookup Table X.

HEP Lookup Time Margin HEP HEP Label Table X 100% 0.02 EXCR27 Time margin is 490%. See basis for D27. Therefore, time margin is greater than 100% and the scoping HEP is 0.02.

Note: Off Normal Procedures (ONPs) have recently been renamed/converted to Abnormal Operating Procedures (AOPs) at PNP. HRA documentation has not yet been updated.

Note: An additional operator action considered associated with the transfer of control to the alternate shutdown panel is the action to open the shunt trip Page 92 of 227

breakers to increase battery life and help maintain dc power. This action is identified in the fire PRA model as EDC-C1OA-72-0102-FR [OP FTO SHNT TRIPS (72-01 & 72-02) TO SHED DC LOADS (LOC) (HEP)]. This action will utilize a screening value HEP of 1.0, and therefore, would not be credited in the RAI Response Fire PRA Model.

Part l.iii)

The operator action associated with the use of the alternate shutdown panel is the action to operate the TDAFW pump from the alternate shutdown panel. The action is also part of the HFE identified in the fire PRA model as AFW-PMOE-EC-150-FR

[OP FT XFR AFW PP START TO C-150 PNL (LOC) (HEP)].

This HFE will be assessed via the scoping method of NUREG-1921, Section 5.2.8, for the RAI Response Fire PRA Model. The table below describes the results of the process in Section 5.2.8 of NUREG-1921 for assigning scoping HEPs and specifically addresses the basis for the answers to each of the questions asked in the Figure 5.5 flowchart.

Page 93 of 227

Scoping HEP for AFW-PMOE-EC-150-FR per NUREG-1921 Section 5.2.8.

ID Flowchart ID Text Answer Basis D40 Are all the necessary cues Yes For control room abandonment for required actions scenarios, general cues (such as loss of protected? control or indication or smoke and/or ambient temperature conditions in the control room) can preempt other specific cues and lead the Shift manager to call for control room abandonment.

For control room abandonment and non-control room abandonment scenarios, specific cues include: control room annunciators for low steam generator level; AFW flow indicators; SG narrow range level indicators ; and SG wide range level indicators. Also, inadequate heat removal could be determined using PCS temperature indication. Also, local AFW flow indicators are available in the CCW room.

Since steam generator level and pressure instrumentation is required to be protected per the Appendix R safe shutdown scheme per NRC Information Notice 84-09, the necessary cues are presumed protected.

D41 For the given action, do the Yes Procedures related to diagnosing the procedures match the action to operate TD AFW pump from the scenario? alternate shutdown panel include: EOP-1.0, EOP-3.0, EOP-9.0, ONP-25.1 and ONP 25.2, and if annunciators are available, ARP-1 and ARP-5.

These procedures generally match the expected range of the pattern of cues for scenarios in which operation of the TDAFW pump from the alternate shutdown panel is required.

D42 Is one of the following Yes The procedure for operating the TDAFW conditions met: 1) there are pump from the alternate shutdown panel procedures for executing is ONP 25.2.

the action or 2) it is skill-of-the-craft?

D43 Are both conditions met: 1) Yes Fires resulting in the need for control the area is accessible and room abandonment and operation of TD

2) there is no fire in the AFW pump from the alternate shutdown vicinity of the action? panel are not expected at the location of the alternate shutdown panel or on the path to this location from the control Page 94 of 227

Scoping HEP for AFW-PMOE-EC-150-FR per NUREG-1921 Section 5.2.8.

ID Flowchart ID Text Answer Basis room.

Multiple ways to exit the control room and reach the alternate shutdown panel location. The action does not take place in the direct vicinity of the fire.

D44 Is the time available (Tavail) Yes Tsw is 150 minutes based on greater than 30 minutes? unrecoverable loss of decay heat removal. However, the Tdelay, Tcog and Texe from the transfer action above is considered to reduce Tsw to 108 minutes. Tdelay is still conservatively taken as 20 minutes and represents the time to complete EOP-1.0 or the time to abandonment. Tcog is 2 minutes and represents the time to detect, diagnose, and decide to leave the control room.

Texe is 20 minutes and represents travel time and manipulation time (Note this is conservative with respect to the walk-through which indicated 10 minutes).

Therefore, Tavail is 88 minutes.

D49 Is the execution complexity No Execution involves positioning two power high? supply breakers to the on position and manipulating two handswitches to disconnect control room circuits and bring the alternate shutdown panel on-line. This action is covered in ONP 25.2 training, which is part of the 2 year training plan. Two JPMs are provided to place the alternate shutdown panel in service: PL-OPS-ONP-005J and PL-OPS-ONP-006J.

D50 Is there smoke or other No See basis for D43. Hazardous chemicals hazardous elements in the are not stored in the vicinity of the south vicinity? west cable penetration room.

This results in HEP Lookup Table AG.

HEP Lookup Time Margin HEP HEP Label Table AG 100% 0.04 ASD15 Time margin is 300%. See basis for D44. Therefore, time margin is greater than 100% and the scoping HEP is 0.04.

Note: Off Normal Procedures (ONPs) have recently been renamed/converted to Abnormal Page 95 of 227

Scoping HEP for AFW-PMOE-EC-150-FR per NUREG-1921 Section 5.2.8.

ID Flowchart ID Text Answer Basis Operating Procedures (AOPs) at PNP. HRA documentation has not yet been updated.

Part l.iv)

The LAR fire PRA model utilized a screening value for AFW-PMOE-EC-150-FR of 0.1. Since this operator action is comprised of both actions to transfer control to the alternate shutdown panel and to operate the TDAFW pump from the alternate panel, the RAI Response Fire PRA Model will utilize the aggregate value that results from the logical combination (OR) of the NUREG-1921 scoping methods, that is, a value of 0.06.

Since the NUREG-1921-based scoping value is higher than the screening value used in the LAR, there is a beneficial impact on the PRA results using NUREG-1921 and a sensitivity study is not required.

The numerical effect of using the resultant scoping HEPs for control room abandonment scenarios will be reflected in the base quantification of the RAI Response Fire PRA model.

m) There are seven (7) different approaches applied in the PNP fire PRA model when defining postulated scenarios and the associated fire PRA target sets:

1. PAU Bounding Scenarios

2. Cable Tray and Junction Box Fires Scenarios per the guidance in FAQ 13-0005 [1] and FAQ 13-0006 [2]

3. Fire scenario probability and targets identified based on the Hughes report Generic Fire Modeling Treatments [3]

4. Fire scenario probability and targets based on the Hughes report Generic Fire Modeling Treatments refined with fire growth, damage time, and manual non-suppression probability using the Mathcad software application of the Generic Fire Modeling Treatments

5. Cable Spreading Room Grid approach which credits the automatic wet pipe suppression system in the Cable Spreading Room using the results of the Hughes report Generic Fire Modeling Treatments

6. Cable Spreading Room Grid approach which credits the automatic wet pipe suppression system in the Cable Spreading Room using the results of the Hughes report Generic Fire Modeling Treatments refined using the Mathcad software to incorporate fire growth, damage time, and manual non suppression probability, Page 96 of 227

7. Detailed fire modeling using FDS [4] in the 1C and 1D switchgear rooms to determine the time to target damage for cables of interest An overview for each of the approaches listed above is provided in the following sections to clarify how target sets, severity factors and non-suppression probabilities are developed and applied. Additionally, the treatment utilized for the growth stages of a fires HRR and the corresponding fire scenarios modeled in the FPRA are discussed.

1. PAU Bounding Scenarios

a. This approach is applied to PAUs that did not require scenario refinement in the fire PRA.

b. Target Sets

i. PAU bounding scenarios fail cable trays, conduits, and equipment in the PAU vulnerable to fire induced failure.

c. Severity Factors and Non-Suppression Probabilities

i. No severity factors or non-suppression probabilities are developed or applied to these scenarios.

d. Growth Stage of the fires HRR

i. The growth stage of the fires HRR is not considered as fires are considered to fail targets in the PAU with the full ignition frequency of the ignition sources.

2. Cable Tray and Junction Box Fires Scenarios per the guidance FAQ 13-0005 and FAQ 13-0006

a. This approach is applied to PAUs quantified in the fire PRA model. PAUs excluded from the Global Analysis Boundary and qualitatively screened PAUs are not included.

b. Target Sets

i. According to FAQ 13-0005 and FAQ 13-0006 the target sets are defined to be limited to the ignition source itself.

c. Severity Factors and Non-Suppression Probabilities

i. No severity factors or non-suppression probabilities are developed or applied to these scenarios.

d. Growth Stage of the fires HRR Page 97 of 227

i. The growth stage of the fires HRR is not considered as the extent of damage is limited the ignition source per the FAQ guidance.

3. Generic Fire Modeling Treatments

a. This approach is applied to the PAUs that are not treated as PAU Bounding Scenarios and for which additional refinement beyond the Generic Fire Modeling Treatments was not considered necessary.

b. Target Sets

i. The targets set for these scenarios is based on the 98th percentile HRR zones of influence (ZOI) as documented in the Generic Fire Modeling Treatments.

ii. An additional target set may be identified based on a lower percentile HRR ZOI and included as an additional scenario.

iii. Target sets for ignition sources involving combustible liquid are initially based on a fire involving 100% of the liquid. In most cases this is further refined to identify targets set for fires involving only 10% of the liquid and when applicable the target set for a very small amount of liquid is identified (Pumps and Main Feedwater Pumps).

iv. The analysis will be updated in the RAI Response Fire PRA Model to also increase the ZOI for scenarios involving ignition of secondary combustibles. The updated ZOIs are based on different time intervals and therefore the target sets vary based on which time interval is represented in a scenario.

c. Severity Factors and Non-Suppression Probabilities

i. No severity factors or non-suppression probabilities are applied to these scenarios except for the situations outlined below.

1. Severity factors are applied to scenarios if a target set other than that corresponding to the 98th percentile HRR is identified and modeled. This is consistent with the guidance in NUREG.CR-6850 Appendix E [5].

2. For ignition sources containing combustible liquid for which multiple size liquid spills are postulated, severity factors are applied consistent with the guidance in NUREG/CR-6850 [5],

Supplement 1 to NUREG/CR-6850 [6], and the Recent Fire PRA Methods Review Panel Decisions and EPRI 1022993, "Evaluation of Peak Heat Release Rates in Electrical Cabinet Fires" [7].

3. Severity factors and non-suppression probabilities are sometimes Page 98 of 227

applied to scenarios involving secondary combustibles. In these cases, severity factors are applied to reflect the fraction of fires that can ignite the secondary combustibles compared to those that cannot. For the range of HRRs that can ignite the secondary combustibles, non-suppression probabilities are applied to scenarios to represent the likelihood that the fire is suppressed within the time period corresponding to the modeled target sets.

d. Growth Stage of the fires HRR

i. The growth stage of a fire is not considered within this approach (except as described below for scenarios involving secondary combustibles). The target set is assumed to fail at t=0.

ii. The growth stage of a fire is also not considered for ignition of combustible liquids.

iii. However, in scenarios involving ignition of secondary combustibles, the growth stage of the fires are considered when the time based ZOIs are calculated.

4. Generic Fire Modeling Treatments Refined with Mathcad to include fire growth, damage time, and non-suppression probability.

a. This approach is applied to a small number of ignition sources in PAUs 02, 15, 21, and 26. The response to RAI FM 02b provides more information on where this approach is applied.

b. Target Sets

i. The initial target set for these scenarios is based on the 98th percentile HRR zone of influence (ZOI) as documented in the Generic Fire Modeling Treatments.

ii. This approach also involves identification of a smaller target set based on a reduced ZOI. The dimensions of the reduced ZOI are based on the distance to a particular target (generally the closest target that doesnt terminate at the ignition source). This is modeled as an additional scenario in the fire PRA.

c. Severity Factors and Non-Suppression Probabilities

i. The severity factor and non-suppression probabilities calculated for this approach are applied in composite NSP terms. The details on the calculation of the composite NSP terms are included in the response to RAI PRA 01q. The following bullets provide a high level summary of the approach.

Page 99 of 227

ii. The HRR distribution is separated into bins of varying severity and likelihood.

iii. For each HRR bin the time to damage of a target located just beyond the reduced ZOI is calculated.

iv. A non-suppression probability is calculated based on each of these times to damage.

v. The product of the SF and NSP for each HRR bin is calculated.

The summation of these values provides the composite NSP value that any resulting fire will not be suppressed prior to damaging the target just beyond the reduced ZOI. This is applied as the NSP for the scenario that models the target set corresponding to the 98th percentile HRR ZOI.

vi. The complement of that term is applied as the NSP for the target set defined by the reduced ZOI.

d. Growth Stage of the fires HRR

i. The growth stage of the fire is analyzed when the time to target damage is calculated. For this approach, the detail on how the growth stage is analyzed is included in the response to RAI PRA 01q.

5. Cable Spreading Room Grid

a. This approach is applied to floor based transient fires in the cable spreading room. Additional details on this approach are included in the response to RAI PRA 07.

b. Target Sets

i. The target sets are based on a grid whose intersections were defined by the location of the wet-pipe sprinkler heads.

ii. The cable tray and conduit targets are assigned to sections of the CSR defined by the grid lines.

iii. A transient fire scenario is postulated at the intersection points of the grid lines and includes the 4 sections that touch each intersection point iv. This results in overlapping scenarios to ensure that all credible target sets are analyzed.

v. An additional target set consisting of the entire PAU is included in Page 100 of 227

a scenario that postulates failure of the wet-pipe suppression system.

c. Severity Factors and Non-Suppression Probabilities

i. No severity factors or non-suppression probabilities are developed or applied to these scenarios with the exception of the scenario that models failure of the wet-pipe suppression system.

d. Growth Stage of the fires HRR

i. The growth stage of the fires HRR is not considered. All targets in the analyzed target set are assumed to fail at t=0.

6. Cable Spreading Room Grid refined with Mathcad to include fire growth, damage time, and non-suppression probability.

a. This approach is applied to the fixed ignition sources in the cable spreading room. Additional details on this approach are included in the response to RAI PRA 07.

b. Target Sets

i. The target sets are based on a grid whose intersections were defined by the location of the wet-pipe sprinkler heads.

ii. The cable tray and conduit targets are assigned to sections of the CSR defined by the grid lines.

iii. The initial target set for a fixed ignition source is based on the grids affected by the 98th percentile HRR horizontal ZOI from the Generic Fire Modeling Treatments iv. Refinements are applied by analyzing smaller target sets (less impacted grid sections) that correspond to reduced horizontal ZOIs in additional scenarios.

v. Another level of refinement is applied by analyzing a target set based on the distance to the closest vertical target. This utilizes much of the same techniques as described in Approach 4.

vi. An additional target set consisting of the entire PAU is included in a scenario that postulates failure of the wet-pipe suppression system.

c. Severity Factors and Non-Suppression Probabilities

i. Initially no SFs or NSPs are applied when the initial target set is Page 101 of 227

identified.

ii. Severity Factors for the first refinement are calculated based on the likelihood of a fire that can reach the grid sections included in the target set. For example, if it requires the 75th percentile fire or larger to impact the largest target set, a SF of 0.25 is applied to this scenario. The smaller target set has a SF of 0.75 applied.

iii. In the second refinement where a reduced ZOI is analyzed based on the distance to the closest vertical target, a composite NSP is calculated using the same techniques described in Approach 4.

This is performed separately for each fire size range analyzed previously. The NSPs calculated are applied to each scenario and the sum of the complementary values is applied to a new scenario that impacts the smallest target set identified.

iv. The failure of the wet-pipe suppression system is also included in the previous analysis and contributes to the likelihood that the entire PAU is damaged.

d. Growth Stage of the fires HRR

i. The growth stage of the fire is analyzed when the time to target damage is calculated. For this approach, the detail on how the growth stage is analyzed is included in the response to RAI PRA 01q.

7. Detailed fire modeling using FDS in 1C and 1D switchgear rooms

a. This approach was applied to all fixed ignition sources in PAUs 03 and 04 as well as a subset of the transient fire scenarios in these PAUs.

The remaining transients were analyzed using Approach 3.

b. Target Sets

i. The results of the FDS fire model provide times to target damage for select HRR values ii. Targets sets are identified by selecting targets that fail before or during various time periods

c. Severity Factors and Non-Suppression Probabilities

i. Severity Factors are calculated based on the range of fires represented by the select HRR values analyzed in the FDS models

1. A severity factor of 0.49 is applied to all scenarios using Page 102 of 227

results obtained from the 49th percentile HRR cases. (0.49-0.0)

2. A severity factor of 0.32 is applied to all scenarios using results obtained from the 81st percentile HRR cases. (0.81-0.49)

3. A severity factor of 0.19 is applied to all scenarios using results obtained from the 98th percentile HRR cases. (1.0-0.81)

4. The upper 2 percentile of the HRR distribution is represented by the 98th percentile fire results.

ii. The non-suppression probabilities are calculated based on the time periods analyzed.

1. Assuming 4 time periods are analyzed, 0-20, 20-40, 40-60, and beyond 60 minutes, the NSP terms are calculated as follows. A term equal to 0.102 is used for illustrative purposes. A floor value of 0.001 was also applied.

a. NSP0-20 = e-*0- e-*20 = 1.000 - 0.130 = 0.870

b. NSP20-40 = e-*20- e-*40 = 0.130 - 0.017 = 0.113

c. NSP40-60 = e-*40- e-*60 = 0.017 - 0.001 = 0.016

d. NSP60-inf = e-*60 = 0.001

d. Growth Stage of the fires HRR

i. The growth stage of the fires HRR is explicitly considered in the FDS model.

ii. For each peak HRR analyzed the fire growth curve was included in the model used to calculate the times to damage.

References

[1] Cable Fires Special Cases: Self-Ignited and Caused by Welding and Cutting, NRC Comments on FAQ 13-0005, ADAMS Accession Number ML13150A242, May 2013.

[2] Modeling Junction Box Scenarios in a Fire PRA, FAQ 13-0006, Revision 0, ADAMS Accession Number ML13183A496, July 2013.

[3] Hughes Associates. Inc., Generic Fire Modeling Treatments, Revision 0, January 2008.

Page 103 of 227

[4] Verification and Validation of Selected Fire Models for Nuclear Power Plant Applications: Fire Dynamics Simulator (FDS), NUREG-1824, Volume 7, May 2007.

[5] EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities, EPRI 1011989, NUREG/CR-6850, Final Report, September 2005.

[6] Fire Probabilistic Risk Assessment Methods Enhancements, EPRI 1019259, NUREG/CR-6850 Supplement 1, Technical Report, September 2010.

[7] Recent Fire PRA Methods Review Panel Decisions and EPRI 1022993, "Evaluation of Peak Heat Release Rates in Electrical Cabinet Fires."

ADAMS Accession number ML12171A583, June 2012.

n) The referenced statement provided the latitude to credit prompt suppression for hot work fires, but this was not implemented in the fire PRA model. The hot work fire scenarios in the fire PRA were reviewed and it was confirmed that prompt suppression was not credited. That is, the welding suppression curve at a time of five (5) minutes was not applied for prompt suppression by a fire watch. Manual and automatic suppression were credited as applicable on a scenario basis consistent with the approaches discussed in the response to PRA RAI 01m.

o) NUREG/CR-6850 does not provide guidance for the classification of sensitive electronics. Therefore, the guidance in FAQ 13-0004, Rev. 1 [1] is used to classify sensitive electronics. FAQ 13-0004 states, The following is provided as additional guidance for identifying the scope of plant equipment to be treated using the lower damage threshold specified in Section H.2 of NUREG/CR-6850.

• Electro-mechanical devices are not considered sensitive electronics.

• Integrated circuits employing any of the variants of pin-grid arrays should be treated as sensitive electronics unless they satisfy the item below.

• Sensitive electronic components that are mounted inside a control panel (cabinet) such that the cabinet walls, top, front and back doors shield the component from the radiant energy of an exposure fire may be considered qualified up to the heat flux damage threshold for thermoset cables, provided that:

The component is not mounted on the surface of the cabinet (front or back wall/door) where it would be directly exposed to the convective and/or radiant energy of an exposure fire.

The presence of louvers or other typical ventilation means do not invalidate the guidance provided for here.

Page 104 of 227

Appendix S of NUREG/CR-6850 [2] provides guidance for the treatment of sensitive electronics in adjacent cabinets. NUREG/CR-6850 Appendix H provides guidance for the treatment of sensitive electronics with respect to damage criteria.

Sensitive Electronics in Adjacent Cabinets NUREG/CR-6850 Appendix S guidance is used to model the potential impacts to sensitive electronics from fires in adjacent cabinets. Appendix S recommends assuming a loss of function in adjacent cabinets if there is not a double wall with an air gap. Appendix S recommends assuming damage to sensitive electronic occurs at 10 minutes if there is a double wall with air gap. The treatment of sensitive electronics in adjacent cabinets in the RAI Response Fire PRA Model will be updated to be consistent with NUREG/CR-6850 Appendix S guidance.

Sensitive Electronic Damage Criteria The treatment of sensitive electronic vulnerability to transient fires is consistent with the industry guidance. The guidance in NUREG/CR-6850 Section H.2 is to treat sensitive electronics with damage criteria of a temperature of 65°C and heat flux of 3kW/m2. The subsequent guidance provided in FAQ 13-0004, Rev. 1 is to treat sensitive electronics mounted inside a cabinet that are not directly exposed to the convective and/or radiant energy of a fire as qualified up to the heat flux damage threshold for thermoset cables (i.e., 11 kW/m2). Therefore, the guidance in FAQ 13-0004, Rev. 1 is applied for sensitive electronics not directly exposed.

Sensitive electronics that may be directly exposed to the convective and/or radiant energy of a fire are treated consistent with the NUREG/CR-6850 Section H.2.

That is, the damage criteria of a heat flux of 3 kW/m2 and temperature of 65°C are applied.

References:

[1] FAQ 13-0004, Revision 1, Clarifications on Treatment of Sensitive Electronics, ADAMS Accession Number ML13182A708, June 2013.

[2] EPRI 1011089 - NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities, August 2005.

p) NUREG/CR-6850 Appendix P Section P.1.4 [1] provides guidance for identifying and analyzing dependencies in credited suppression paths. The following dependencies are identified as potentially being important:

Dependencies between automatic detection and suppression Dependencies between active fire barriers and automatic suppression Dependencies between safe shutdown capability and automatic suppression Page 105 of 227

Dependencies between manual and automatic suppression Dependencies between manual fire detection and suppression Each of these dependencies and the treatment in the fire PRA is discussed below.

Dependencies between automatic detection and suppression The fire PRA does not credit automatic suppression systems that are actuated by an automatic detection system. Therefore, this dependency is not applicable to the fire PRA.

Dependencies between active fire barriers and automatic suppression The fire PRA only credits area automatic wet pipe suppression systems. These credited systems do not rely on active fire barriers to be effective. Therefore, this dependency is not applicable to the fire PRA.

Dependencies between safe shutdown capability and automatic suppression Dependency between safe shutdown capability and automatic suppression consist of the fire protection system being credited for vessel injection or heat removal and the fire protection system being relied upon to suppress a fire. The fire PRA does not credit the fire protection system for vessel injection. Therefore, this dependency is not applicable to the fire PRA. The FPS is credited for backfill of the CST and long-term operation of P-8C as well as P-8A and P-8B as described in Section 4.2.1.2 of the NFPA 805 LAR. Credit for use of the FPS in this manner would be well after any initial fire suppression activities were required such that no direct dependency needs to be modeled in the fire PRA.

Dependencies between manual and automatic suppression The fire PRA credits the fire protection system for water supply for manual suppression activities with a hose stream and automatic wet pipe suppression systems. The dependency between these suppression activities is the fire pumps providing water supply.

The fire PRA applies the NUREG/CR-6850 generic wet pipe system failure probability of 0.02. The fire PRA applies the NUREG/CR-6850 manual non-suppression minimum floor probability of 1E-3. Given these probabilities, the minimum probability credited in the fire PRA for suppression failure is 2E-5 (0.02

• 1E-3 = 2E-5).

The PRA logic model includes treatment of the dependencies within the fire water supply system which includes common cause failure of the fire pumps. The common cause failure of the three pumps to start or run is included with a probability of 5.69E-6. This probability is less than the applied minimum total non-suppression probability of 2E-5 when manual and automatic suppression activities Page 106 of 227

are credited.

The fire protection system dependencies included in the PRA logic are modeled with a failure probability less than that applied for manual and automatic suppression failure. Therefore, the fire protection system dependency with respect to supplying water for manual and automatic suppression is adequately treated in the fire PRA.

Dependencies between manual fire detection and suppression The manual non-suppression probabilities applied in the fire PRA include the dependencies between manual detection and suppression. The manual non-suppression probabilities are calculated consistent with the guidance in NUREG/CR-6850 as clarified by FAQ 08-0050 [2]. That is, the calculated manual non-suppression probabilities include delay time for manual detection. First, the fire PRA credits manual suppression activities to prevent fire damage to the first target. Then the fire PRA credits manual suppression to prevent assumed damaging hot gas layer temperatures in the room. These manual suppression activities are treated in the fire PRA as being completely dependent on the previous manual suppression activity. That is, the applied manual non-suppression probability of preventing an assumed damaging hot gas layer in the room is calculated as the conditional probability that manual suppression failed to prevent target damage.

References:

[1] EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities, EPRI 1011989, NUREG/CR-6850, Final Report, September 2005.

[2] Fire Probabilistic Risk Assessment Methods Enhancements, EPRI 1019259, NUREG/CR-6850 Supplement 1, Technical Report, September 2010.

q) The purpose of the approach documented in Appendix E of the Fire Scenario Development report is to determine the likelihood that a fire results in endstates of varying severity. The likelihood calculations consider the non-suppression probabilities (NSPs) and severity factors (SFs) and provide a composite likelihood.

In order to calculate the NSPs, for the range of fire sizes considered, the corresponding times to damage are needed. These times are calculated for a target located a given distance away from the ignition source.

PNPs approach is to first determine the peak heat fluxes that would be experienced at the target. The Generic Fire Modeling Treatments provide the distances for which 3 different heat fluxes are experienced for a steady-state fire having a peak heat release rate (HRR) equal to the upper bound values of the 15 bins used to represent various HRR distributions from NUREG/CR-6850. A fourth heat flux of 120 kW/m2 is postulated at a distance equal to half the flame height for the corresponding fire size. The GFMTs state that this is a reasonable upper bound for open flames [Beyler, 2002; SFPE, 2004]. The GFMTs also state, The Page 107 of 227

incident heat flux at a target and the critical separation distance are maximized when the target is vertical and it is located at the flame mid-height [Beyler, 2002; Siegel et al., 2002]. Using these data points a heat flux vs. distance curve is generated for each HRR distribution bin. A peak heat flux at the specified distance is then calculated for each peak HRR.

The second step in the process is to represent the heat flux at the target as a function of time. An assumption is made that the heat flux at a given target will grow proportionally to the HRR of the ignition source. According to NUREG/CR-6850 Appendix G, the HRR (Q) is represented by the following function where represents the time to peak HRR: Q(t)=Min[Qpeak,Qpeak*(t/)2]. Therefore, after the peak heat flux (qpeak) values are determined the heat flux as a function of time is represented by the following equation: q(t)=Min[qpeak,qpeak*(t/)2].

The third step in the process is to determine the time at which the target is damaged. The conservative treatment would be to assume target damage when the critical heat flux (6kW/m2) is reached. However, as discussed in Appendix H of NUREG/CR-6850 and in the response to FM RAI 02b, it is understood that there is a delay before the target is damaged. One approach is to assume a linear damage accrual based the following interpretation of NUREG/CR-6850 Table H-8.

The application of this method is described below.

Linear Damage Accrual Method The rate of damage (DR) as a function of external heat flux (q) is developed based on an exponential curve fit of the external heat flux values and the inverse of the time to failures from Table H-8 in NUREG/CR-6850. DR=4.702E-3*e^(0.332*q)+0.034.

Table 1: Data based on Table H-8 in NUREG/CR-6850 External Heat Time to Failure Inverse of Time DR Flux (minutes) [t] (damage fraction/minute) (damage (kW/m2) [q] fraction/minute)

<6 No damage 0 0 6 19 0.053 0.055 8 10 0.100 0.097 10 6 0.167 0.173 11 4 0.250 0.230 14 2 0.500 0.547 16 or greater 1 1.000 0.973 As an example, if a target experiences a maximum external heat flux of 11 kW/m2 when the source fire reaches its peak HRR, the external heat flux as a function of time would be represented by the equation q=11*(t/12)^2. This is based on =12 minutes for an electrical fire. The damage rate at each time interval is then calculated. The example below uses 1 minute time steps for illustration purposes.

The MathCad file utilizes smaller time steps to arrive at a more precise answer. In Page 108 of 227

any event, the damage accrual is calculated by assuming the rate of damage is applicable over the time interval. Therefore, in the example below, damage accrual starts at 9 minutes. The resulting time to full damage accrual happens just after 14 minutes (when the accrued damage fraction exceeds 1.0). This is approximately 5 minutes after the critical heat flux of 6 kW/m2 is reached and only 2 minutes after the maximum heat flux of 11 kW/m2 is reached. These times are shorter than the time delays Appendix H describes for steady-state heat fluxes of 6 and 11 kW/m2. This approach was deemed to provide a reasonable representation to the time to electrical failure for non-steady-state conditions.

Table 2: Damage Accrual Time External Heat Flux Damage Accrued Damage (Minutes) (kW/m2) Rate Fraction 1 0.08 N/A N/A 2 0.31 N/A N/A 3 0.69 N/A N/A 4 1.22 N/A N/A 5 1.91 N/A N/A 6 2.75 N/A N/A 7 3.74 N/A N/A 8 4.89 N/A N/A 9 6.19 0.07 0.07 10 7.64 0.09 0.16 11 9.24 0.14 0.30 12 11.00 0.22 0.51 13 11.00 0.22 0.73 14 11.00 0.22 0.95 15 11.00 0.22 1.16 As discussed, this calculation is performed for each of the 15 HRR bins. Each HRR bin results in a different maximum heat flux at the selected target distance and therefore the times to target damage vary. The corresponding NSP values are calculated for each HRR bin and multiplied by the likelihood of the HRR bin (the SF). The summation of the NSP/SF products provides the overall probability that any size fire will result in failure to suppress the fire. The complementary value represents the likelihood that the fire is suppressed prior to damaging the target.

As scenarios are modeled for both endstates, it is expected that if it was assumed that the time to damage occurs when the critical heat flux is reached (as was done in the sensitivity analysis provided in Section 7.3.1 of the Fire Risk Quantification and Summary report), the NSP/SF values representing failure to suppress the fire will increase and those NSP/SF values representing successful suppression would decrease. This is why the sensitivity analysis documents values both above and below those of the original MathCAD analysis.

Page 109 of 227

However, due to the uncertainty surrounding the use of the linear damage accrual method, a more conservative approach for the NFPA-805 application will be used in the RAI Response Fire PRA Model. This approach is described below.

Time Delay Method In place of using the damage accrual method, the RAI Response Fire PRA model will implement a more conservative interpretation of the information provided in Appendix H of NUREG/CR-6850 for the NFPA-805 LAR. The first two steps in the process described above do not change, however the time to damage is determined differently. The time to damage is estimated by applying a time delay from the time the critical heat flux is reached. The time delay value applied is based on the peak heat flux reached for the fire being analyzed. Therefore, using the example from above, a time delay of 4 minutes would be applied since the peak heat flux reached at the target is 11 kW/m2. The time the target reaches 6kW/m2 was calculated to be at approximately 9 minutes. Therefore, the time to damage using this method is estimated at 13 minutes (as compared to 14 minutes obtained from the damage accrual method shown above).

The rest of the NSP/SF calculation is also performed in the same manner and therefore multiple NSP/SF values are calculated to represent the various endstates possible from a fire at the ignition source. The input parameters change depending on the type of fire scenario analyzed. The time to peak HRR () and the value used in the NSP equation vary depending on whether a transient fire or electrical cabinet fire is analyzed. Additionally, the HRR bins, severity factors, and the corresponding heat flux distances vary for different ignition source types including different types of electrical cabinet configurations.

The results of using the time delay method will be included in the RAI Response Fire PRA.

r) Automatic detection systems equipped with smoke detectors credited in the fire PRA are listed in Table 1 below. No automatic detection systems in support of the activation of fixed suppression systems were credited in the fire PRA.

NUREG-1805 Chapter 11 [1] provides guidance for estimating smoke detector response times with the Method of Alpert, Method of Mowrer, and Method of Milke.

A number of assumptions and limitations exist when these methods are used to calculate smoke detector response time. Additionally, verification and validation of these methods were not provided in NUREG-1824 [2]. However, these methods can be used to understand the range of response times for smoke detectors and can be used to support an estimate of smoke detector response time in the fire PRA. The NUREG-1805 FDTs Smoke spreadsheet 10_Detector_Activation_Time_Sup1_SI.xls is used to calculate the smoke detector response times for a range of the NUREG/CR-6850 [3] heat release rates (HRRs) for each of the three methods.

Page 110 of 227

Use of the FDTs to estimate automatic detection time is also consistent with the guidance in Inspection Manual Chapter 0609, Appendix F, Task 2.7.1 - Fire Detection Analysis [4]. This appendix provides guidance for estimating automatic detection times using the correlations in NUREG-1805 FDTs. This guidance recommends calculating the detector activation time in seconds and rounding up to the nearest minute.

The FDTs input parameters altered were those in the yellow cells, which include HRR, radial distance to the detector, height of ceiling above the top of the fire, and ambient air temperature. These inputs parameter values were chosen as follows:

HRR - varied based on the NUREG/CR-6850 case and bin.

Radial Distance to the Detector -15 feet (4.6 m) was taken as the maximum separation distance based on midpoint between detectors which are typically located 30 feet apart (NFPA 72 code conformance review in PLP-RPT 00053, Rev. 0).

Height of Ceiling Above the Fire - For NUREG/CR-6850 Case 3 and 4 fires the separation distance was selected to be one foot below the top of a typical seven-foot electric cabinet and the ceiling. The ceiling was chosen to be 16 feet which is representative of the 1C and 1D Switchgear Rooms and is higher than other significant areas where smoke detectors were credited.

Therefore, the separation distance was 10 feet (3.05 m). For NUREG/CR-6850 Case 8 fires (transients) the distance was taken to be the distance between the floor and ceiling. Again, a ceiling height of 16 feet (4.9 m) was applied.

Ambient Air Temperature - Selected to be 20°C which is consistent with the fire modeling performed.

Given the input parameter values above and the default parameter values in the FDTs, estimated smoke detector response time was calculated using each of the three methods. Table 2 presents the estimated response times for NUREG/CR-6850 Case 3 HRRs (NUREG/CR-6850 Table E-4, Single Bundle Cabinets with Unqualified Cable HRRs). Table 3 presents the estimated response times for NUREG/CR-6850 Case 4 HRRs (NUREG/CR-6850 Table E-5, Multi Bundle Cabinets with Unqualified Cable HRRs). Table 4 presents the estimated response times for NUREG/CR-6850 Case 8 HRRs (NUREG/CR-6850 Table E-9, Transient HRRs).

Based on the FDTs results, the smoke detector response time is expected to be within seconds for a range of NUREG/CR-6850 HRRs modeled in the fire PRA in which automatic detection is credited. While these calculations were performed for steady state HRRs, the estimated detection times within one minute is consistent for the range of NUREG/CR-6850 HRRs from the lowest HRRs up to the 98th percentile HRRs. Therefore, for the range of HRRs that are prescribed to Page 111 of 227

be analyzed, the plant automatic detection is expected within one minute during the fire growth phase.

In the absence of verification and validation of these calculations in NUREG-1824, the PNP fire PRA uses the guidance in Inspection Manual Chapter 0609, Appendix F, Task 2.7.1 and rounds up to the nearest minute. As such, the RAI Response Fire PRA Model will apply a one minute time delay for credited automatic detection systems.

For manual detection, the PNP fire PRA uses a time delay of 15 minutes when credited. The 15 minute detection time is consistent with guidance in Inspection Manual Chapter 0609, Appendix F, Attachment 8, Task 2.7.1 - Fire Detection Analysis [4]. This attachment provides guidance for detection times associated with different types of detection, one of which is detection by general plant personnel. Page F8-3, bullet 2 under Detection by General Plant Personnel states, In the absence of any other means of detection, a maximum fire detection time of 15 minutes will be used. Therefore, a time delay of 15 minutes for manual detection is consistent with the available guidance. To ensure that the application of the 15 minute time to manual detection was reasonable, a review of rooms modeled in the fire PRA was performed. The manual detection time was applied to rooms frequently occupied by operators, security personnel, or other plant staff.

Table 1:

Automatic Detection Systems Credited in the Fire PRA Fire Area Fire Area Description Automatic Detection 1 Control Room Complex Smoke 2 Cable Spreading Room Smoke 3 1-D Switchgear Room Smoke 4 1-C Switchgear Room Smoke 13 Auxiliary Building 590 Corridor Smoke 15 Engineering Safeguards Panel Room Smoke 21 Electrical Equipment Room Smoke 26 Southwest Cable Penetration Room Smoke Page 112 of 227

Table 2 Estimated Smoke Detector Response Time for NUREG/CR-6850 Case 3 NUREG/CR-6850 Bin HRR (kW) Response Time (sec.)

Method of Alpert Method of Method of Milke Mowrer 1 26 n/a 4 28 2 53 n/a 3 8 3 79 15 3 4 4 106 9 2 3 5 132 7 2 2 6 158 5 2 1 7 185 4 2 1 8 (98%) 211 4 2 1 Table 3 Estimated Smoke Detector Response Time for NUREG/CR-6850 Case 4 NUREG/CR-6850 Bin HRR (kW) Response Time (sec.)

Method of Alpert Method of Method of Milke Mowrer 1 53 n/a 3 8 2 106 9 2 3 3 158 5 2 1 4 211 4 2 1 5 264 3 2 1 6 317 3 2 <1 7 369 2 2 <1 8 422 2 1 <1 9 (98%) 464 2 1 <1 Table 4 Estimated Smoke Detector Response Time for NUREG/CR-6850 Case 8 NUREG/CR-6850 Bin HRR (kW) Response Time (sec.)

Method of Alpert Method of Method of Milke Mowrer 1 37 n/a 4 142 2 74 n/a 3 44 3 111 n/a 2 23 4 148 14 2 14 5 185 9 2 10 6 222 7 2 7 7 258 6 2 6 8 295 5 2 4 9 (98%) 317 4 2 4 Page 113 of 227

References:

[1] Fire Dynamics Tools (FDTs) Quantitative Fire Hazard Analysis Methods for the U.S. Nuclear Regulatory Commission Fire Protection Inspection Program, NUREG-1805 Chapter 11, December 2004.

[2] Verification and Validation of Selected Fire Models for Nuclear Power Plant Applications, NUREG-1824, May 2007.

[3] EPRI 1011089 - NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities, August 2005.

[4] NRC Inspection Manual Chapter 0609, Appendix F, Attachment 8, Task 2.7.1

- Fire Detection Analysis, February 2005.

s) The risk estimates reported in Attachment W are based on the 317 kW (98th percentile) HRR for transient fires per NUREG/CR-6850.

Transient fires are postulated to occur at the floor or on the surface of a feature such as a scaffold, grating or pedestal. The location of transient fires one foot above the floor is a convention used during scenario development to account for the height of the burning fuel package. This height establishes the base of the transient fire and is used as the reference point for the fires zone of influence.

While typical transient fires involve a fuel package of negligible height (e.g.

extension cords), a height of one foot was selected to address less likely transient scenarios involving fuel packages such as plastic bags supported by scissor stands as well a trash bags located on the floor. Trash bags and loose accumulations of material, though they initially may be ignited at their tops, reach a peak heat release rate when fully involved. In this case, the entrainment base is at the floor, and the correct zone of influence base is the floor. In addition, plastics will melt and the fire will pool at the floor level. The use of one foot also addresses fires in small rigid trash containers without lids that do not collapse. In response to this RAI, an additional assessment was conducted to validate that the types of receptacles used within the protected area are small (less than about a foot) or covered. It was noted that there were taller (up to 30), uncovered, trash cans in use in the main control room and in specific locations within the power block. These areas are:

Main Control Room (PAU 01, Rooms 320-325)

Electrical Maintenance shop, Chemistry lab, and associated office areas(PAU 23, Rooms 126 through 128)

Restroom, Locker Room, and former Work Control Complex (PAU 23, Rooms 240-245)

Technical Support Center (PAU 33, Room 320A)

Page 114 of 227

In order to properly address potential transient fires that could occur at the top of these containers, the assumed transient location was changed to 30above the floor for these specific plant locations. Transient scenarios developed in these areas using the revised criterion have been integrated into the fire PRA model and are include in the revised risk estimates. The numerical effect will be reflected in the base quantification of the RAI Response Fire PRA Model.

t) The PNP fire impairment list associated with detection and suppression systems has been reviewed and no impairments of credited detection or suppression systems that would impact unavailability estimates were identified over the past 9 years. Therefore, it is believed that the unavailability of the systems would be small comparison to the unreliability values, such that use of the generic unreliability values is acceptable. Plant Records (e.g. Work Orders, Operations Logs, etc.) are in the process of being reviewed to determine more precise unavailability values for credited detection and suppression systems. The unavailability values will be accounted for in the results of the RAI Response Fire PRA Model.

u) The applicable requirement from the PRA Standard is FSS-D8 which reads as follows:

INCLUDE an assessment of fire detection and suppression systems effectiveness in the context of each fire scenario analyzed.

The note for the FSS-D8 supporting requirement consists of the following:

Fire detection or suppression system effectiveness depends on, at a minimum, the following:

(a) system design compliance with applicable codes and standards, and current fire protection engineering practice (b) the time available to suppress the fire prior to target damage (c) specific features of physical analysis unit and fire scenario under analysis (e.g., pocketing effects, blockages that might impact plume behaviors or the visibility of the fire to detection and suppression systems, and suppression system coverage), and (d) suitability of the installed system given the nature of the fire source being analyzed Automatic detection systems credited in the fire PRA are equipped with smoke detectors and are listed by area in Table 1 below. Automatic suppression systems credited in the fire PRA are equipped with automatic wet pipe suppression systems. These are also listed in Table 1. These areas have been assessed per the FSS-D8 guidance as described below. The credited automatic detection systems are discussed followed by a discussion of the credited automatic Page 115 of 227

suppression systems.

Automatic Detection The automatic smoke detectors are credited in the fire PRA for manual suppression activities. LAR Table 4-3 identifies the credited detection systems in Table 1 as meeting the requirements of NFPA 805 or as requiring a modification.

The response to PRA RAI 1r concludes that the smoke detectors would actuate within one minute for the range of NUREG/CR-6850 heat release rates postulated in the fire PRA. The areas in which smoke detectors are credited do not contain physical features such as beam pockets that would change this conclusion. Cable trays located above postulated ignition sources are not considered obstructions to the smoke detectors, because smoke would be readily transported around the cable trays. Given this assessment, the automatic smoke detection systems are effective in the context of the fire scenarios in which they are credited. This is consistent with the requirements of FSS-D8.

Automatic Suppression The automatic wet pipe suppression systems are credited in the fire PRA to a) Control the fire such that a damaging full room hot gas layer does not form b) Prevent large fire spread in the Cable Spreading Room c) Prevent widespread damage inside the building and to maintain structural integrity of the building LAR Table 4-3 identifies the credited wet pipe systems in Table 1 as meeting the requirements of NFPA 805. These wet pipe systems are installed commensurate with the hazard and based on a review of plant drawings are modeled with an actuation temperature of 74°C (165°F). In the fire PRA, a full room hot gas layer is assumed when the room temperature is 80°C. Therefore, for the full room damaging fire scenarios the wet pipe systems will be expected to actuate prior to the assumed full room damage. With successful activation of the wet pipe suppression system, the system is credited to control the fire.

The fire PRA analyzed the Cable Spreading Room based on the wet pipe system sprinkler head locations using a grid approach. Sprinkler heads are located a maximum of 10 feet apart at the ceiling. Successful activation of sprinklers within the zone of influence or within the grid, is credited in the fire PRA to prevent damage to grids not included in the ignition source zone of influence.

The fire PRA included credit for the Turbine Building wet pipe suppression system to mitigate a Turbine Generator fire. The suppression system was credited to prevent widespread damage inside the building and to maintain structural integrity of the building consistent with the guidance in NUREG/CR-6850.

Page 116 of 227

Given this assessment, the automatic wet pipe suppression systems are effective in the context of the fire scenarios in which they are credited. This is consistent with the requirements of FSS-D8.

Table 1: Automatic Detection and Suppression Systems Credited in the Fire PRA Fire Fire Area Description Automatic Detection Automatic Area Suppression 1 Control Room Complex Smoke N/A 2 Cable Spreading Room Smoke Wet Pipe 3 1-D Switchgear Room Smoke Wet Pipe 4 1-C Switchgear Room Smoke Wet Pipe 13 Auxiliary Building 590 Corridor Smoke N/A 15 Engineering Safeguards Panel Room Smoke N/A 21 Electrical Equipment Room Smoke Wet Pipe 23 Turbine Building N/A Wet Pipe 26 Southwest Cable Penetration Room Smoke Wet Pipe v) ASME/ANS RA-Sa-2009 Standard SR FSS-F1 requires that it be determined if locations with exposed structural steel and a high hazard source are identified. If locations are identified that satisfy the two conditions then fire scenarios should be selected that could damage, including collapse, the exposed structural steel. Note 1 to SR FSS-F1 states:

The prototypical fire scenario leading to failure of structural steel would be catastrophic failure of the turbine itself (e.g., a blade ejection event) and an ensuing lube-oil fire. For the lube-oil fire, the possibility of effects of pooling, the flaming oil traversing multiple levels, and spraying from continued lube-oil pump operation should be considered. However, the analysis should also consider scenarios involving other high-hazard fire sources as present in the relevant physical analysis units (e.g., oil storage tanks, hydrogen storage tanks and piping, mineral oil-filled transformers).

The Fire PRA model addresses the possibility of effects of pooling, the flaming oil traversing multiple levels, and spraying from continued main turbine lube-oil pump operation consistent with the guidance in NUREG/CR-6850 Appendix O.2.3. That is, severe T/G oil fires are assumed to cause widespread damage because of the potential of oil pooling, flaming oil traversing multiple levels, and spraying from continued lube oil pump operation.

Additionally, the Fire PRA model includes scenarios involving other high hazard sources consistent with the guidance in NUREG/CR-6850. Hydrogen storage tanks and piping are treated consistent with the guidance in NUREG/CR-6850 Appendix N. Oil storage tanks, mineral oil filled transformers, and other high hazard fires are treated consistent with the guidance in NUREG/CR-6850 Appendix E, G, and O.

Page 117 of 227

w) In the RAI Response Fire PRA Model, the calculation for the T/G catastrophic fire will be updated to include the excitor frequency that was previously excluded in the LAR model based on a different interpretation of the information provided in Table O-2 of NUREG/CR-6850. The incorporation of the excitor frequency is required to match the 1.0E-5/yr catastrophic turbine/generator (T/G) fire frequency provided in Table O-2 of NUREG/CR-6850 as noted in the RAI.

An initial calculation using the generic NUREG/CR-6850 was performed to validate the calculation approach. Then subsequent calculations were performed using plant specific Bayesian updated NUREG/CR-6850 values, and plant specific Bayesian updated values from EPRI 1019259.

The updated results for the T/G catastrophic fire scenario will be included in results of the RAI Response Fire PRA Model. As the updated T/G catastrophic fire will be included in the base case results, as well as the VFDR calculations, an evaluation of the impact of this correction on CDF, LERF, CDF and LERF is not necessary.

No credit for suppression beyond that which is quantified in NUREG/CR-6850 (namely, failure of fixed suppression with a probability of 0.02) is applied in the updated T/G catastrophic fire frequency calculations. Therefore, no additional basis for the credited suppression probability is required.

x) The multi-compartment analysis (MCA) is being updated to apply the screening criteria based on compartments (i.e., sub-volumes) instead of PAUs. The numerical effects of these changes will be incorporated into RAI Response Fire PRA Model.

The screening criteria applied to compartments in the MCA are consistent with the guidance in NUREG/CR-6850 Section 11.5.4.

NUREG/CR-6850 Section 11.5.4.2 Step 2.c: First Screening - Qualitative:

The failures postulated by the target set of the exposed compartment does not increase the set of failures postulated by the full target set in the exposing compartment.

NUREG/CR-6850 Section 11.5.4.3 Step 3.c: Second Screening - Low Fire Load Exposing Compartments: The HRRs for the ignition sources and ignited secondary combustibles in the exposing zone are not capable of forming a damaging hot gas layer (HGL) in the exposing and exposed compartments. Compartments that are outside or have open communication with the outside are also screened based on this criterion.

NUREG/CR-6850 Section 11.5.4.4 Step 4.c: Third Screening Frequency of Occurrence: The likelihood that the ignition sources in the exposing compartment can produce a damaging HGL in the exposing and exposed compartments is calculated. The ignition frequencies, severity factors, non-Page 118 of 227

suppression probabilities, and applicable barrier failure probabilities are multiplied together and then totaled for the exposing compartment to determine the likelihood of the MCA interaction. The interaction was screened if the likelihood was less than 1E-7.

NUREG/CR-6850 Section 11.5.4.5 Step 5.c: Fourth Screening CDF Based:

MCA interactions were also screened if the CDF was below 1E-7. This is consistent with the Step 4.c screening performed except that consideration of the CCDP was included.

The screening frequency criterion used in MCA steps 4.c and 5.c was 1E-7/yr.

This screening frequency criterion was used because it conforms to SR QNS-C1 Cat II of the ASME PRA standard [2]. Use of a low screening CDF criterion of 1E-7/yr ensures that all highest risk PAU combinations are retained for further analysis.

For exposed compartments determined unable to support a HGL in the multi-compartment interaction analyzed, additional components (e.g., cables) within the exposed area were not considered damaged due to hot gas impingement from an exposing compartment via propagation pathways. This is consistent with the screening analysis described in NUREG/CR-6850.

NUREG/CR-6850 Section 11.5.4 states, The main objective of this task is to evaluate the risk associated with multi-compartment fire scenarios. The risk is evaluated by performing a multi-compartment screening and a detailed evaluation of fire scenarios in the unscreened compartments. [1]

NUREG/CR-6850 Section 11.5.4.6 states, Those scenarios that do not screen out in the preceding steps may be analyzed using the same methods for single compartments. The same set of steps may be followed. In this case, the target set should include items from the exposed compartments.

[1]

Hot gas impingement from an exposing compartment was considered when detailed scenario development was performed for unscreened multi-compartment interactions.

References:

[1] NUREG/CR 6850 (2005), EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, EPRI 1011989 Final Report, NUREG/CR-6850, Nuclear Regulatory Commission, Rockville, MD, September, 2005.

[2] ASME/ANS RA-Sa-2009, Standard for Level1/Large Early Release Frequency Probabilistic Risk Assessment for Nuclear Power Plant Applications Page 119 of 227

y) The Multi-Compartment Analysis (MCA) will be updated in the RAI Response Fire PRA Model as described in PRA RAI 01x to be consistent with NUREG/CR-6850 guidance. The presence of dampers between compartments was conservatively assumed between all compartments instead of confirming which interactions involved dampers. In the updated MCA, the most limiting barrier (e.g. non-rated barrier, door, damper, or wall) has been identified and therefore the assumption is no longer applied.

The failure probabilities assigned to credited passive barriers are consistent with NUREG/CR-6850 guidance. Fire rated doors, dampers, walls, and penetrations are assigned the generic failure probabilities provided in NUREG/CR-6850 Table 11-3. Non-rated fire area passive barriers are also assigned the generic failure probabilities in Table 11-3, because these barriers are analyzed as a part of the transition to NFPA-805 and found to be adequate for the hazard. Other non-rated barriers are assigned a value of 0.1. See Section 4.8.1 of the PNP LAR for the results of the fire area review which discusses the assessment of credited fire area barriers.

Plant-specific barrier problems are identified by fire protection staff through the PNP station corrective action process. Interviews with the fire protection staff revealed no known outstanding plant specific issues. A review of condition reports from 2008 to 2013 related to "fire barriers", "fire dampers", and "fire doors" confirmed the assessment from the interviews that there are no known outstanding plant-specific issues.

z) The MCA screening fire door barrier failure probability is being updated to be consistent with the guidance of NUREG/CR-6850 Section 11.5.4.4. The guidance states, For scenarios that include failure of a barrier in the open position, barrier failure probability should be estimated. Table 11-3 includes the recommended generic failure probability of 7.4E-3 for fire doors. Therefore, the conditional failure probability applied in the MCA screening for active fire doors failing in the open position is 7.4E-3. The numerical effects of these changes will be incorporated into RAI Response Fire PRA Model.

The active fire barrier elements credited in the fire PRA are limited to normally open fire doors and normally open fire dampers. Other active barriers (e.g. water curtains) are not credited barrier elements. The active fire doors and fire dampers effectiveness is established by the applicable fire resistance rating [2]. Random failures of the normally open fire doors and dampers have been evaluated in the MCA per NUREG/CR-6850 as discussed above. Normally open fire dampers and fire doors fail in the desired position when exposed to a fire. Therefore, fire induced failure is not applicable to these active fire barriers.

References

[1] NUREG/CR 6850 (2005), EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, EPRI 1011989 Page 120 of 227

Final Report, NUREG/CR-6850, Nuclear Regulatory Commission, Rockville, MD, September, 2005.

[2] PNP Plant Fire Hazards Analysis Report (FHAR) Rev 7.

aa) Section 6.5.7.3 of NUREG/CR-6850 provides guidance on establishing an ignition frequency for large systems that include a complex of components such as miscellaneous hydrogen (bin 19). NUREG/CR-6850 states that a geometric factor may be applied to adjust bin frequency in compartments where components could be risk-significant. NUREG/CR-6850 also states, in place of a geometric factor, the analyst may count the various components of the complex and rate them by an ad-hoc scheme that discriminates by the relative likelihood of ignition.

The miscellaneous hydrogen frequency apportioning was reviewed and will be updated in the RAI Response Fire PRA Model to explicitly use the length of pipe and quantity of other components in the calculation. To discriminate the various components by the relative likelihood of ignition, a count of one (1) is given to each foot of pipe and a count of 100 is given to each component (i.e., tank and valves).

The count is based on the available failure rates in NUREG/CR-6928 that are most applicable. NUREG/CR-6928 includes failure rates for pipe leaks per foot-hour and valve leaks per hour. The approximate difference between these is 100; therefore, a foot of pipe is given a count of one (1) and a component (i.e., tank or valve) is given a count of 100.

Reviews of drawings as well as plant walkdowns of the hydrogen piping are performed to determine the approximate length of pipe and quantity of valves and tanks by room.

The target sets for hydrogen fires are identified consistent with the methodology described in section N.2.4 of NUREG/CR-6850. For smaller PAUs, the contents are postulated to fail. For larger PAUs, such as the turbine building, a zone of influence is applied and targets in the locations (i.e., rooms) in which the hydrogen piping is routed are failed. The numerical effect of these changes will be reflected in the base quantification of the RAI Response Fire PRA Model.

bb) The HRA analysis will be updated to follow the guidance in NUREG-1921 for screening, scoping and detailed human error probabilities. The numerical effect will be reflected in the base quantification of the RAI Response Fire PRA Model.

Talk-throughs (reviews in detail) with plant operations and training personnel are being performed to confirm that interpretation of current and planned procedures relevant to modeled actions is consistent with plant operational and training practices.

The extent to which talk-throughs with plant operations and training personnel have been completed is discussed in the response to PRA RAI 28b. Detailed Page 121 of 227

descriptions of P-IOAQ, Feasibility Evaluation and Validation content and the review process are provided in RAI 01cc.

cc) The discussion below describes how the definition of human failure events takes into account scenario context, timing, procedural guidance, instrumentation, task complexity, path of travel, etc., in the development of the feasibility context and narrative. Also discussed are the methods for accident-sequence-specific timing cues timing and the time window for successful completion.

The HRA analysis will be updated to follow the guidance in NUREG-1921 for screening, scoping and detailed human error probabilities. The numerical effect will be reflected in the base quantification of the RAI Response Fire PRA Model.

The definition of human failure events for the fire PRA takes into account scenario context, including timing, procedural guidance, instrumentation, task complexity, path of travel, etc. Screening, scoping and detailed HEP values will be utilized in the RAI Response Fire PRA Model for both fire response actions and actions that were carried over from the FPIE PRA. Since NUREG-1921 methods will be utilized, accident-sequence-specific timing of cues and the time window for successful completion (e.g., time margin) will be addressed as per NUREG-1921.

Post-Initiator Operator Action Questionnaires (P-IOAQs), Recovery Action Feasibility Evaluations and HEP Validation forms were developed by previously SRO licensed personnel, and reviewed and approved (when complete) by currently SRO licensed Operations personnel and the Assistant Operations Manager - Training.

Operations reviews included verification that listed cues, procedure use, manpower requirements and performance time were reasonable, and that the actions were feasible. Training reviews included verification that listed training was correct and P-IOAQ information reflected training expectations.

Initial action timing feasibility was evaluated by comparing system window time (Tsw) to execution time (Texe). Development of a new all inclusive Fire Response procedure addressing both reactor trip and fire actions, and subsequent confirmation of potential staffing size increases and additional resource needs, are implementation items (see LAR Attachment S, Table S-3, Items 1 and 3).

The criteria in NUREG-1921 Section 4.3 are addressed as indicated by bold italics corresponding to the specific Section 4.3 criterion.

The following paragraphs describe P-IOAQ content.

The P-IOAQ General Comments/Information section provides the following information:

Manpower Requirements (Sufficient Manpower): A table showing current required minimal staffing, and personnel required to perform the subject Page 122 of 227

action is provided. Information regarding Fire Brigade response and Staff Augmentation resulting from Emergency Plan implementation is also included.

Event Assumptions: Assumptions are provided describing initial plant condition, fire effects requiring the action to be taken, and successful action completion criteria.

Procedure Sequence (Proceduralized and Trained Actions): Using information from the event assumptions, a procedure usage sequence leading to required action performance is given. Except for actions involving proposed modifications, current procedures were used to demonstrate direction for operators exists. Multiple procedures are currently required to perform most actions. For example, if a fire requiring reactor trip occurs, operators perform Off Normal Procedure ONP-25.1 Fire Which Threatens Safety-Related Equipment and Emergency Operating Procedure EOP-1.0 Standard Post-Trip Actions, and may be directed to ONP-25.2 Alternate Safe Shutdown Procedure for mitigating actions. A new all inclusive fire response procedure addressing reactor trip and fire recovery actions will be developed as part of NFPA 805 implementation.

Accessibility (Accessible Location): Locations of plant equipment required for action performance and associated access routes are identified for use in determining action successful completion feasibility.

Fire Areas of Concern: The six plant fire areas with highest probability for requiring action and in which failure of the action would be of greatest concern are listed (Fire Areas 1, 2, 3, 4, 13 and 23).

Miscellaneous Event Specific Details: Information such as electrical safety requirements, additional instrumentation availability, proposed modification descriptions, etc. is provided as necessary for specific actions.

The P-IOAQ Items Addressed by PSA Engineer section provides the event name, description of activity, sequences where the action applies, action location (in or out of Control Room), and special conditions that may apply for out of Control Room actions.

The P-IOAQ Questions for Operator section provides the following information:

Signals (cues) (Primary Cues Available/Sufficient) that lead to action performance: cues include annunciators, indication, noise, procedures, etc.

Applicable procedures (Proceduralized and Trained Actions): procedures include Emergency Operating Procedures (EOPs), Off Normal Procedures (ONPs), General Operating Procedures (GOPs), System Operating Procedures (SOPs), Alarm and Response Procedures (ARPs), Emergency Page 123 of 227

Implementing Procedures (EIs), Administrative Procedures (ADMINs).

Personnel required (Sufficient Manpower): includes all personnel required to perform the task (e.g. a second person (Safety Observer) is required when manually operating circuit breakers).

Other concurrent activities (Sufficient Manpower): other activities which may be distracting or competing for resources.

Performance time (Sufficient Time): the time to perform the action including briefing, travel and obtaining required tools, safety equipment, etc.

Special considerations (Equipment and Tools Available and Accessible):

items needed to enable performance (e.g. keys, safety equipment, ladders).

Component identification: potential for a component to be misoperated due to not being clearly labeled or in close proximity to similar components (e.g. one of several identical adjacent control switches).

Verification of intended response (Primary Cues Available/Sufficient):

indications available to verify the correct action was taken.

Training type and frequency (Proceduralized and Trained Actions):

training content, type and frequency Recovery Action Feasibility Evaluation forms document feasibility for all out of Control Room actions per the criteria in FAQ 07-0030. The following areas are evaluated explicitly, with overlap with those addressed in the P-IOAQ:

o Demonstrations: The proposed recovery actions should be verified in the field to ensure the action can be physically performed under the conditions expected during and after the fire event.

o Systems and Indications (Primary Cues Available/Sufficient): Consider availability of systems and indications essential to perform the recovery action.

o Communications (Equipment and Tools Available and Accessible):

The communications system should be evaluated to determine the availability of communication, where required for coordination of recovery actions.

o Emergency Lighting (Accessible Location): The lighting (fixed and/or portable) should be evaluated to ensure sufficient lighting is available to perform the intended action.

o Tools-Equipment (Equipment and Tools Available and Accessible):

Any tools, equipment, or keys required for the action should be available Page 124 of 227

and accessible. This includes consideration of SCBA and personal protective equipment if required.

o Procedures (Proceduralized and Trained Actions): Written procedures should be provided.

o Staffing (Sufficient Manpower): Walk-through of operations guidance (modified, as necessary, based on the analysis) should be conducted to determine if adequate resources are available to perform the potential recovery actions within the time constraints (before an unrecoverable condition is reached), based on the minimum shift staffing. The use of essential personnel to perform actions should not interfere with any collateral industrial fire brigade or control room duties.

o Actions in the Fire Area (Accessible Location): When recovery actions are necessary in the fire area under consideration or require traversing through the fire area under consideration, the analysis should demonstrate that the area is tenable and that fire or fire suppressant damage will not prevent the recovery action from being performed.

o Time (Sufficient Time): Sufficient time to travel to each action location and perform the action should exist. The action should be capable of being identified and performed in the time required to support the associated shutdown function(s) such that an unrecoverable condition does not occur. Previous action locations should be considered when sequential actions are required.

o Training (Proceduralized and Trained Actions): Training should be provided on the post-fire procedures and implementation of the recovery actions.

o Drills (Proceduralized and Trained Actions): Periodic drills that simulate the conditions to the extent practical (e.g., communications between the control room and field actions, the use of SCBAs if credited, the appropriate use of operator aids).

o

Conclusion:

When all individual feasibility criteria are met, the conclusion is feasible without qualifications. When individual feasibility evaluation criteria are currently not met, a comment describing the deficiency is provided and resolution information is given in the conclusion comment section. For example, for the operator action to locally open manual valves to open ADVs, the Procedure criteria is not met (since the modification has not yet been implemented) and the following comment is provided, There currently is no procedural guidance to locally-manually operate the ADVs. (A draft procedure change will be included in ONP-TBD.) The conclusion comments then include Identified modification and procedure changes will adequately address deficiencies.

Page 125 of 227

Note: NUREG-1921 Section 4.3 criterion of Relevant Components Are Operable is not directly correlated with any of the items above. If the fire has damaged equipment such that it will not function even if the operator takes the appropriate action, the operator action should not be credited with restoring the function. For the PNP fire PRA, this fire-induced failure is captured directly in the function/component logic of the fire PRA model. Separate failure of the operator action is not required since functionally the action will not succeed even for a successful operator action. For example, if an auxiliary feedwater pump is directly affected by fire, no credit for local operation of the pump will be taken regardless of the success or failure of the operator action.

Lastly, HEP Validation forms document P-IOAQ and Feasibility Evaluation review and comment incorporation. The following areas are specifically addressed in the validation forms:

Could listed cue items reasonably be expected to initiate the desired action?

Is procedure selection/usage described acceptable, and could it reasonably be expected to be followed?

Are manpower and performance time acceptable?

Is the action feasible?

Is the description and frequency of training accurate?

Acceptable Operations and Training review comment incorporation is documented by signatures on the HEP Validation forms.

dd) The HRA analysis is being updated to follow the guidance in NUREG-1921 for screening, scoping and detailed human error probabilities. The numerical effect will be reflected in the base quantification of the RAI Response Fire PRA Model.

The RAI Response Fire PRA Model will provide CDF, LERF, CDF and LERF values based on HEPs developed using the approaches specified in NUREG-1921; therefore, a sensitivity study will not be provided.

The RAI response fire HRA accounts for relevant fire-related effects by using detailed HEP analyses for more risk significant HFEs and using screening and scoping analyses for less risk significant HFEs (or HFEs in with greater uncertainty) per NUREG-1921.

Use of detailed analyses for significant HFEs and conservative estimates (e.g.,

screening values) for only non-significant HFEs in accordance with CC-II for SR HRA-C1-01 is required for self-approval.

ee) The HRA analysis is being updated to follow the guidance in NUREG-1921 for screening, scoping and detailed human error probabilities. The numerical effect will be reflected in the base quantification of the RAI Response Fire PRA model.

Page 126 of 227

Screening, scoping and detailed HEP values will be utilized in the RAI Response Fire PRA Model for both HFEs currently modeled in the FPIE PRA as well as for fire response actions.

The HFEs and associated HEPs included in the RAI Response Fire PRA Model through the process described below will provide a more realistic evaluation of accident sequences. While not meeting CC-II for HRA-D1, the evaluation will provide more realism, retain conservatism in the screening and scoping HEPs as prescribed by NUREG-1921, and is considered acceptable for transition to NFPA 805.

The process used to determine which HFEs get screening, scoping or detailed HEPs involved iterative quantification of the base risk model, examining importance measures for the HFEs, and adjusting HEPs in order to bin HFEs into three categories. The first category did not significantly impact risk results and was therefore targeted for screening HEPs. The second category moderately impacted risk results and was therefore targeted for scoping HEPs. The third category more highly impacted risk results and was therefore targeted for detailed HEPs. During this process, interim, surrogate HEPs were used to help bin HFEs appropriately.

Final quantification of the RAI Response Fire PRA Model will eliminate interim HEPs and replace the values with the NUREG-1921-based values, either screening, scoping or detailed (i.e., with the EPRI HRA calculator).

In this way, HFEs that are not as significant to accident sequences are assigned screening HEPs per NUREG-1921. HFEs that are moderately significant to accident sequences are assigned scoping HEPs per NUREG-1921. HFEs that are more significant to accident sequences are assigned detailed HEPs per NUREG-1921 (i.e., the EPRI HRA calculator). In this manner, realism is established for HFEs that are important to dominant accident sequences.

The peer review observation that top core damage fire scenarios do not account for realistic recovery actions was an early-phase peer review comment that has now been addressed. The RAI Response Fire PRA Model will reflect the treatment discussed above and realism is established via the development of HEPs based on their importance.

For the RAI Response Fire PRA Model, top core damage fire scenarios would include detailed HEPs. For example, top core damage fire scenarios involving failure to trip primary coolant pumps, failure to trip charging pumps due to fire-induced spurious charging pumps starts or pressurizer level control system failures, failure to trip pressurizer heater breakers, and failure to re-fill emergency diesel generator fuel oil day tanks are treated with detailed HEPs to provide a more realistic evaluation.

The resulting detailed HEP evaluations using the EPRI HRA Calculator that are important to top core damage fire scenarios will be reflected in the final quantification and documented in the HRA notebook NB-PSA-HR, Human Page 127 of 227

Reliability Safety Assessment - Post-Initiator Operator Actions, Volume 1.

ff) The HRA analysis will be updated to follow the guidance in NUREG-1921 for screening, scoping and detailed human error probabilities. The numerical effect will be reflected in the base quantification of the RAI Response Fire PRA Model.

Screening, scoping and detailed HEP values will be utilized in the RAI Response Fire PRA Model for both fire response actions and actions that were carried over from the FPIE PRA. Since NUREG-1921 methods will be utilized, fire-related effects will be addressed as per NUREG-1921. Recovery of coincident loss of off-site power is not credited in the fire PRA.

The feasibility assessments considered relevant fire effects that may alter the manner in which the actions will be accomplished. These fire effects will be considered for both fire response actions and actions that were carried over from the FPIE PRA. Detailed descriptions of P-IOAQ, Recovery Action Feasibility Evaluation and Validation content, and the review process are provided in PRA RAI 01cc.

Fire effects that may preclude completion of an action (e.g., location and pathway) were evaluated separately as documented in the model development report PAU to HEP matrix (see model development report Appendix E - HEP Location Matrix).

This matrix documents which actions are not credited due to the action location being in the area of the fire or where the path to the action location is precluded due to the fire.

gg) Twelve plant specific fire events were identified for the time period of January 1, 2001 through December 31, 2011. Eleven of these events occurred within the fire PRA global analysis boundary. It was later confirmed that only ten were in the global analysis boundary. Prior to the LAR submittal, these were evaluated with respect to the criteria for potentially challenging fires provided in NUREG/CR-6850 Appendix C. This was presented in Appendix A of Report 0247-07-0005.02 as referenced in this RAI. Two events (event numbers 1 and 2) were classified as potentially challenging. Events numbered 3 through 12 were classified as non-challenging. The non-challenging events were re-assessed in response to this RAI. For each of these events, justification for the classification as non-challenging is provided below based on an assessment of the established criteria.

This re-assessment confirms that none of these events are potentially challenging.

Event No. 3 Date: 9/19/04 Plant Status: Outage Location: Reactor Containment Building [PAU 14]

Page 128 of 227

Summary During a plant outage, the main hook brake drum of the Polar Crane motor caught on fire while the operator was lowering the main hook. Operation of the main hook was discontinued. The breaker for the polar crane was opened. By the time the crane was accessed, the fire had self extinguished. A CO2 extinguisher was used in short bursts to cool the brake assembly. Fire duration was not identified. It was detemined that a failed relay caused the brakes to close onto the drum causing it to overheat. This is not considered a potentially challenging event based on the following evaluation.

Screening Assessment This event occurred in the power block during a plant outage. The fire was associated with the polar crane, which is precluded from service during power operation.

Conclusion This event could not have occurred during power operation or low power operation, it is screened from further consideration and considered to be non-challenging.

Event No. 4 Date: 4/18/06 Plant Status: Outage Location: Turbine Building [PAU 23]

Summary During a plant outage, sparks from grinding on exhaust piping for Main Feedwater Pump P-1B migrated under herculite igniting a canvas tool bag. A water pump can was used to extinguish the bag. The fire duration was one minute.

Screening Assessment This fire occurred in the turbine building during a plant outage. The event could potentially have occurred during power operation and therefore is not screened from further consideration on the basis of location or plant operating mode.

Objective Criteria Group 1

• Suppression - Manual suppression was limited to a water can and no automatic suppression was used. This criterion is not met.

Page 129 of 227

• Damage to additional components - No components outside the boundaries of the ignition source were affected. This criterion is not met.

• Combustible materials - No additional combustible materials beyond the canvas bag were ignited. This criterion is not met.

Objective Criteria Group 2

• Automatic Detection - No automatic detection system was actuated. This criterion is not met.

• Plant trip - No plant trip was experienced. This criterion is not met.

• Cost of damage - The damage was limited to a replacement of the canvas bag which was estimated at $35. As this is less than $5000, this criterion is not met.

• Burning duration - The fire was extinguished in one minute. As this is less than ten minutes, this criterion is not met.

Subjective assessment and conclusion No objective criteria were met and there was no indication that the fire was self-sustaining or might have impacted equipment or materials beyond the ignition source. Further, this event occurred during grinding, an activity during which equipment and combustibles are protected, and additional controls, such as the fire watch that discovered the fire, are in place. This event is classified as non-challenging.

Event No. 5 Date: 4/21/06 Plant Status: Outage Location: Yard [PAU 41]

Summary During a plant outage, a ceiling exhaust fan in a temporary trailer ignited. The trailer was located in the yard. The breaker for the fan was opened and fire self extinguished within five minutes. This is not considered a potentially challenging event based on the following evaluation.

Screening Assessment This fire occurred in the yard area during a plant outage. The event could potentially have occurred during power operation and therefore is not screened from further consideration on the basis of location or plant operating mode.

Page 130 of 227

Objective Criteria Group 1 Suppression - The incident report cites a CO2 extinguisher but does not indicate that is was discharged. No automatic suppression was used. This criterion is not met.

Damage to additional components - No components outside the boundaries of the ignition source were affected. The breaker was opened which could be considered active intervention by plant personnel to extinguish the fire.

However, in this case, the ignited component is a small fan, a fire which is not expected to propagate beyond the source regardless of intervention.

Also, components of this type and size are not counted as ignition sources per NUREG/CR-6850 This criterion is not met.

Combustible materials - No combustible materials outside the boundaries of the ignition source were ignited. This criterion is not met.

Objective Criteria Group 2 Automatic Detection - No automatic detection system was actuated. This criterion is not met.

Plant trip - No plant trip was experienced. This criterion is not met.

Cost of damage - The damage was estimated at less than $100. As this is less than $5000, this criterion is not met.

Burning duration - The fire self extinguished within five minutes. As this is less than ten minutes, this criterion is not met.

Subjective assessment and conclusion No objective criteria were met and there was no indication that the fire was self-sustaining or might have impacted equipment or materials beyond the ignition source. The breaker was opened which could be considered active intervention by plant personnel to extinguish the fire. However, in this case, the ignited component is a small fan, a fire at which is not expected to propagate beyond the source regardless of intervention. Further, the type of equipment is not a piece of plant equipment, nor is it of a type and size that would be considered an ignition source in the fire PRA. This event is classified as non-challenging.

Event No. 6 Date: 2/9/09 Plant Status: At Power Location: Auxiliary Building [PAU 17]

Page 131 of 227

Summary The Fire Brigade was dispatched to investigate a report of arcing, sparking and black smoke coming from a disconnect switch on the Spent Fuel Handling Machine Bridge. The event duration was less than a minute, and there was no fire. This is not considered a potentially challenging event based on the following evaluation.

Screening Assessment This fire occurred in the power block while the unit was at power and therefore is not screened from further consideration on the basis of location or plant operating mode.

Objective Criteria Group 1

• Suppression - No manual or automatic suppression were used. This criterion is not met.

• Damage to additional components - No components outside the boundaries of the ignition source were affected. This criterion is not met.

• Combustible materials - No additional combustible materials were ignited.

This criterion is not met.

Objective Criteria Group 2

• Automatic Detection - No automatic detection system was actuated. This criterion is not met.

• Plant trip - No plant trip was experienced. This criterion is not met.

• Cost of damage - The damage is assessed to be less than $5000. This criterion is not met.

• Burning duration - No burning occurred. This criterion is not met.

Subjective assessment and conclusion No objective criteria were met and there was no indication that an actual fire event occurred. This event is classified as non-challenging.

Event No. 7 Date: 3/24/09 Plant Status: At Power Location: Dry Fuel Storage Pad [PAU 48/55]

Page 132 of 227

Summary A small fire developed when an employee put a cigarette butt into a metal cigarette container that lacked a bottom due to corrosion. The fire was observed by security personal via observation cameras. The fire was limited to the leaves at the base of the perimeter fence and was extinguished by plant personnel at the scene by stomping out the burning leaves. This is not considered a potentially challenging event based on the following evaluation.

Screening Assessment This event occurred at the Dry Fuel Storage Pad during power operation. This location is not included in the fire PRA global analysis boundary.

Conclusion This event occurred in an area not associated with either the nuclear or power generation blocks of the plant. The fire could not impact safe plant operation and is therefore screened from further consideration and considered to be non-challenging.

Event No. 8 Date: 4/3/09 Plant Status: Outage Location: Turbine Building [PAU 23]

Summary During a plant outage, sparks from welding migrated to the space between the condenser pit and wall causing an oil soaked board to smolder and ignite. The fire watch extinguished the fire. This is not considered a potentially challenging event based on the following evaluation.

Screening Assessment This fire occurred in the turbine building during a plant outage. The event could potentially have occurred during power operation and therefore is not screened from further consideration on the basis of location or plant operating mode Objective Criteria Group 1

• Suppression - No automatic suppression was used. From the event description, it is assumed that limited manual suppression was used. This criterion is not met.

Page 133 of 227

• Damage to additional components - No components outside the boundaries of the ignition source were affected. This criterion is not met.

• Combustible materials - No additional combustible materials beyond the board. This criterion is not met.

Objective Criteria Group 2

• Automatic Detection - No automatic detection system was actuated. This criterion is not met.

• Plant trip - No plant trip was experienced. This criterion is not met.

• Cost of damage - There was no damage to plant equipment. This criterion is not met.

• Burning duration - The fire was immediately extinguished upon discovery.

The fire was detected 25 minutes after completion of the welding activity. As the time from ignition to extinguishment cannot be determined, it is assumed that the time is greater than ten minutes, and this criterion is assumed to be met.

Subjective assessment and conclusion Only one group two objective criterion was met (by assumption), which is insufficient basis to warrant this event to be categorized as a potentially challenging fire, and there was no indication that the fire was self-sustaining or might have impacted equipment or materials beyond the burning piece of wood.

Further, this event occurred during welding, an activity during which equipment and combustibles are protected, and additional controls, such as the fire watch that discovered and extinguished the fire, are in place. This event is classified as non-challenging.

Event No. 9 Date: 7/31/10 Plant Status: At Power Location: Warehouse [PAU 49]

Summary A lighting ballast ignited causing the light cover to melt and drip onto cardboard located below. The fire occurred in a warehouse that is not located within the global analysis boundary of the fire PRA. This is not considered a potentially challenging event based on the following evaluation.

Page 134 of 227

Screening Assessment This event occurred in a warehouse location during power operation. This location is not included in the fire PRA global analysis boundary.

Conclusion This event occurred in an area not associated with either the nuclear or power generation blocks of the plant. The fire could not impact safe plant operation and is therefore screened from further consideration and considered to be non-challenging.

Event No. 10 Date: 10/17/10 Plant Status: Outage Location: Turbine Building [PAU 23]

Summary A lockout relay located in EA-14 began to smoke and ignited upon completion of installing a temporary jumper. The control room was notified and the fire was immediately extinguished. This is not considered a potentially challenging event based on the following evaluation.

Screening Assessment This fire occurred in the turbine building during a plant outage. The event could potentially have occurred during power operation and therefore is not screened from further consideration on the basis of location or plant operating mode.

Objective Criteria Group 1 Suppression - No automatic suppression was used. There is no indication that extighuishers were used. This criterion is not met.

Damage to additional components - No components outside the boundaries of the ignition source were affected. This criterion is not met.

Combustible materials - No combustible materials outside the boundaries of the ignition source were ignited. This criterion is not met.

Objective Criteria Group 2 Automatic Detection - No automatic detection system was actuated. This criterion is not met.

Page 135 of 227

Plant trip - No plant trip was experienced. This criterion is not met.

Cost of damage - The damage was limited to the failed relay. As this is assessed to be less than $5000, this criterion is not met.

Burning duration - The fire was immediately extinguished. This is assumed to be less than ten minutes, therefore, this criterion is not met.

Subjective assessment and conclusion No objective criteria were met and there was no indication that the fire was self-sustaining or might have impacted equipment or materials beyond the ignition source. This event is classified as non-challenging.

Event No. 11 Date: 10/20/10 Plant Status: Outage Location: Reactor Containment Building [PAU 14]

Summary During preparations for Reactor Vessel Head Lift to set the head back onto the Reactor Vessel Flange, a light hanging in the cavity caught fire. The fire was only at the light itself, and did not spread to surrounding material. The fire self-extinguished after a few moments. The light was unplugged and the control room was notified..

Screening Assessment This fire occurred in the Reactor Containment Building during a plant outage. The event could potentially have occurred during power operation and therefore is not screened from further consideration on the basis of location or plant operating mode.

Objective Criteria Group 1 Suppression - No manual or automatic suppression was used. This criterion is not met.

Damage to additional components - No components outside the boundaries of the ignition source were affected. This criterion is not met.

Combustible materials - No combustible materials outside the boundaries of the ignition source were ignited. This criterion is not met.

Page 136 of 227

Objective Criteria Group 2 Automatic Detection - No automatic detection system was actuated. This criterion is not met.

Plant trip - No plant trip was experienced. This criterion is not met.

Cost of damage - The damage was limited to the light fixture. As this is assessed to be less than $5000, this criterion is not met.

Burning duration - The fire self-extinguished. This is assumed to be less than ten minutes, therefore, this criterion is not met.

Subjective assessment and conclusion No objective criteria were met and there was no indication that the fire was self-sustaining or might have impacted equipment or materials beyond the ignition source. This event is classified as non-challenging.

Event No. 12 Date: 5/2/11 Plant Status: At Power Location: Turbine Building [PAU 23]

Summary The fire alarm sounded and the Fire Brigade dispatched for a fire on EX-107, the standby power transformer for data logger system inverter EY-210. EX-107 was de-energized and 8 minutes after the fire was reported, it was extinguished. The fire self-extinguished and off-site assistance was not required. This event is not considered potentially challenging based on the evaluation below.

Screening Assessment This fire occurred in the turbine building while the unit was at power and therefore is not screened from further consideration on the basis of location or plant operating mode.

Objective Criteria Group 1 Suppression - No manual or automatic suppression were used. This criterion is not met.

Damage to additional components - No components outside the boundaries of the ignition source were affected. The breaker was opened which could be Page 137 of 227

considered active intervention by plant personnel to extinguish the fire.

However, in this case, since the ignited component is a dry transformer having limited combustible material, a fire is not expected to propagate beyond the source regardless of intervention. If the fire was potentially challenging it is likely that the breaker would have tripped This criterion is not met.

Combustible materials - No combustible materials outside the boundaries of the ignition source were ignited. This criterion is not met.

Objective Criteria Group 2 Automatic Detection - No automatic detection system was actuated. This criterion is not met.

Plant trip - No plant trip was experienced. This criterion was not met.

Cost of damage - The damage was limited to replacement of the transformer and is considered to be less than $5000. This criterion is not met.

Burning duration - The fire self extinguished within eight minutes. This criterion is not met Subjective assessment and conclusion No objective criteria were met and there was no indication that the fire was self-sustaining or might have impacted equipment or materials beyond the ignition source. The breaker was opened which could be considered active intervention by plant personnel to extinguish the fire. However, in this case, since the ignited component is a dry transformer, a fire is not expected to propagate beyond the source regardless of intervention. This event is classified as non-challenging.

hh) Partitioning of the plant into physical analysis units (PAUs) is based on the PAUs being well defined areas that contain the damaging effects of fire by means of rated barriers, non-rated barriers, active fire protection features, and spatial separation. This is consistent with the discussion of sample criteria and assumptions provided in appendix A of NUREG/CR 6850.

For locations addressed in the fire hazards analysis, fire PRA physical analysis units (PAUs) directly align to Appendix R Fire Areas and credit the same barriers.

The FHA defines fire areas as follows:

A fire area as used in Appendix R is defined as an area sufficiently bounded to withstand the hazards associated with the area and, as necessary, to protect important equipment within the area from a fire outside the area. Fire area boundaries need not be completely sealed floor-to-ceiling, wall-to-wall boundaries.

However, all unsealed openings should be identified and considered in evaluating Page 138 of 227

the effectiveness of the overall barrier. Where fire area boundaries are not wall-to-wall, floor-to-ceiling boundaries with all penetrations sealed to the fire rating required of the boundaries, an evaluation must be performed to assess the adequacy of fire boundaries to determine if the boundaries will withstand the hazards associated with the area. This analysis must be performed by at least a fire protection engineer and, if required, a systems engineer.

Appendix R Fire Area boundaries generally consist of rated barriers. Exceptions are evaluated by GL86-10 calculations to assure that the non-rated barriers are commensurate with the expected fire hazards.

Six new PAUs were also identified to address plant locations in the fire PRA that were not specifically addressed in the FHA. These are discussed in more detail below:

PAU 38 - Cooling Tower Pump House This PAU contains equipment credited in the fire PRA. It is a stand-alone structure of substantial construction removed from other plant buildings. The perimeter walls contain the damaging effects of fire. Further, there are no buildings or equipment in the general vicinity to be exposed, should the perimeter walls fail.

PAU 39 - Feedwater Purity Building This PAU contains equipment credited in the fire PRA. Except for a common barrier with the turbine building, it is a stand-alone structure of substantial construction. Modification S2-20 will replace this common barrier with a rated fire barrier. The barrier will be installed between rooms 254 (Cable and Pipe Gallery -

Turbine Building) and 839 (Boiler Room - Feedwater Purity). This modification upgrades the barrier to be in compliance with the NFPA code.

PAU 40 - Switchyard This PAU contains equipment credited in the fire PRA. The switchyard is an outdoor area located outside of the protected area fence away from the Yard and plant buildings. The spatial separation (~0.5 miles) between this PAU and any other plant equipment is the basis for designation of the switchyard as a PAU.

PAU 41 - Yard This PAU contains equipment credited in the fire PRA. The Yard is the outdoor area within protected area fence. The Yard is separated from the Switchyard by spatial separations and separated from other PAUs by the substantial external barriers of those PAUs.

PAU 42 - Administration Building This PAU does not contain equipment or cables credited in the fire PRA. However, Page 139 of 227

rooms located in this building, share a common wall with other PAUs. Rooms 330 and 330A share a common wall with PAUs 15, 17, and 19 on the 625 elevation.

Fire barriers in PAU 15 consist of three hour fire walls, and three hour rated fire doors. The door between the boronometer room and the PAU 15 is unrated. Fire barriers in PAU 17 and 19 also consist of three hour fire walls on the 625 elevation. These fire barriers surrounding the Administration building will adequately prevent fires in the Administration building from damaging equipment and cables located in the adjacent PAUs.

PAU 43 - Service Building The Service Building and Service Building Expansion, although technically separate buildings, are considered one PAU for the purposes of the fire PRA. The buildings share a common wall and neither contains equipment or cables credited in the fire PRA. This location is included in the global analysis boundary, as the Service Building shares a wall with the Auxiliary Building (PAU 27). Per the Fire Hazard Analysis, the Auxiliary Building walls and floors are of reinforced concrete construction judged to have a minimum of three hour fire rating which adequately prevents fires in the Service Building from damaging equipment or cables located in the Auxiliary Building.

ii) The data analysis notebook [1] was developed to support both the FPIE PRA model and the Fire PRA model. Other than adjustments to the HEP values, no probability input values required reanalysis given the fire context and no probability input values were excluded in the IEPRA. As noted in the final peer review report, The Fire PRA plant response model was reviewed with very few findings. There were no technical F&Os on the scope or content of the PRM model itself. The F&Os assigned to PRM were either a) cross-referenced from other tasks [HRA and ES] or b) were for incomplete documentation. As such, the open portion of the F&O at the time of the LAR submittal was limited to the HRA portion of the F&O (PRM-B11). Related open F&Os were identified in Table V-1 of the LAR submittal (i.e., HRA-A4-01, HRA-B3-01, HRA-C1-01, HRA-D2-01, and HRA-E1-01). The data analysis was performed consistent with the requirements of PRM-B12, and other portions were consistent with PRM-B13 and PRM-C1.

References:

[1] PNP Probabilistic Safety Assessment Notebook NB-PSA-DA, Rev. 6, Data Analysis Notebook, November 2012.

jj) Modification S2-5 will provide an alternate means of tripping the PCPs from the Control Room following loss of the normal DC power source to the breakers.

At the time of the peer review model changes were being developed to include details of the current primary coolant pump trip capability from the control room.

The changes did not at that time include the dc supplies necessary to support the Page 140 of 227

pump trip circuits. The dc power dependency was added to the model prior to the LAR submittal.

The modeling is based on detailed fault tree modeling of the dc system and the required interconnections to supported equipment. The logic to support the dc panels that supply the pump control circuits was already developed in the model.

Completion of the model entailed connecting the correct dc panel to the pumps via fuse and breaker connections.

Credit for tripping primary coolant pumps is based on the ability to trip the pumps from the control room. On loss of the required dc power supply or other fire induced circuit faults the ability to trip the pumps from the control room is not available. Consequently consideration was given to crediting locally tripping the pumps at the switchgear.

However, in the plant this would require internally accessing each breaker cubicle.

Since the seal failure model was based on the owners group consensus model reviewed by the NRC, the time window to complete the action is limited to twenty minutes from the loss of seal cooling. The feasibility assessment determined that there was not sufficient time to complete the action given the time required to diagnose the condition, conduct a pre-job brief, don required safety equipment, travel to the breaker location, and perform the required actions. On loss of dc power to the current control room controls, the affected pumps cannot be tripped from the control room in time and local operation cannot be performed in time to prevent a seal LOCA.

Therefore, modification S2-5 was developed to provide an alternative means of tripping the pumps from the control room. The proposed modification would include trip capability at an alternate control room location separate from the current controls. It will include cable routing that provides separation from the routing of the current control cables and an independent dc power supply or alternate routing of cables to the control room from an existing dc power source.

The RAI Response Fire PRA Model will include limited modeling of the alternate controls and dc power to source to allow quantification of the expected risk reduction from the modification. Since the modification only provides alternate controls to complete the action, the existing human failure event to accomplish the pump trips is common to both methods. The impact of the modification was determined to be a significant reduction in risk.

kk) The model was updated to include events representing spurious operation of the proportional heater banks not in service, failure of pressurizer spray operation, fire induced faults that prevent de-energizing heater banks, and operator failure to de-energize heaters, both locally and from the control room. The operator actions to locally trip the heaters and to trip the heaters from the control room were credited in the LAR model with screening human error probabilities. The action from the control room was credited with a value of 1E-02. The local action was credited with Page 141 of 227

a value of 1E-01. The HRA analysis is being updated to follow the guidance in NUREG-1921 for screening, scoping and detailed human error probabilities.

ll) The content and organization of the Seismic-Fire Interaction Report (Report 0247-07-0005.05) has been updated to address the findings from the Fire PRA Peer Review documented in July, 2011. The revision used the Fire PRA Standard, ASME/ANS RASa2009 Part 4 [1], as a guideline to ensure that all requirements of the standard are met with respect to the Seismic Fire portion of the standard (section 4-2.11). The Seismic-Fire Interaction Report addresses each supporting requirement for HLR-SF-A, and the report itself fulfills the requirement of HLR-SF-B. Each of the supporting requirements is addressed individually in the report. No significant deficiencies were noted during the reviews and revision of the report.

The deficiencies noted by the Peer Review team [2] include the following. Both of these deficiencies have been addressed in the revision to the Seismic-Fire Interaction Report.

o The Fire PRA needs to document the results of the seismic/fire interaction assessment in a manner that facilitates Fire PRA applications, upgrades, and peer reviews.

As noted above, the Seismic-Fire Interaction Report was revised to address each supporting requirement for HLR-SF. The revision included additions and reorganization of the report to facilitate Fire PRA applications, upgrades, and peer reviews.

o The current seismic fire interactions analysis relies on the IPEEE study. The report needs to demonstrate that the scope of that work meets the objectives of the Standard and that plant changes since the work was performed do not compromise the conclusions.

Additional information has been added to the Seismic-Fire Interaction Report to demonstrate that the scope of the IPEEE study meets the objectives of the FPRA Standard. The current validity of the IPEEE analysis is discussed in detail below.

Regarding the IPEEE and why it remains valid, the IPEEE [3] was issued May 22, 1996 and accurately reflected the then Plant design. A review of the current PNP FSAR and discussions with Plant personnel determined the following:

o No plant changes have been made that would impact the seismic qualification of any building or area evaluated as part of the IPEEE.

Therefore the assumptions and assessments made in the IPEEE seismic evaluations remain valid and can be used in answering the supporting requirements for the Seismic Fire section of the Fire PRA Standard.

o No changes have been made to the Fire Water System that would improve Page 142 of 227

or diminish the capabilities of the system as it was evaluated in the IPEEE.

There has been no expansion of the system to support firefighting capabilities in additional areas and therefore no additional points for failure during a seismic event. No new automatic initiations of any deluge system have been added or changed and it can be expected that the IPEEE still accurately documents the impact of a seismic event on the Fire Water System. Therefore the assumptions and assessments made in the IPEEE with regard to the Fire Water System remain valid and can be used in answering the supporting requirements for the Seismic Fire section of the Fire PRA Standard.

o No changes have been made to the Fire Detection System that would improve or diminish the response of the system as documented in the IPEEE, including the addition of any automatic initiations of the Fire Protection System. Therefore the assumptions and assessments made in the IPEEE with regard to the Fire Detection System remain valid and can be used in answering the supporting requirements for the Seismic Fire section of the Fire PRA Standard.

o No new Halon, CO2, water or other fire protection systems have been added to the Plant that would require update of the analysis in the IPEEE.

Based on the above discussions regarding changes to PNP since the IPEEE was issued, it is concluded that the assumptions, analysis and system responses documented in section 3 and section 4 of the IPEEE remain valid and can be used in answering the supporting requirements for the Seismic Fire section of the Fire PRA Standard.

Finally, concerning the effects of the updated USGS seismic hazard curves on the results of the Seismic-Fire Interactions analysis. The PNP IPEEE was issued May 22, 1996 and the seismic PRA was based upon the 1994 LLNL seismic hazard curves. Using these curves and the Plant design, the following results were determined for PNP:

o HCLPF - 0.217g o SSE - 0.2g o Mean CDF - 8.88E-06/yr Referencing the Safety Risk Assessment for NRC GI-199 [4], Table D-3, the weakest link model using the 1994 LLNL seismic hazard curves resulted in a CDF of 1.0E-05.

Table D-1 of NRC GI-199 lists the postulated core damage frequencies using the updated 2008 USGS Seismic Hazard Curves. The weakest link model using the USGS curves is 6.4E-06, a reduction of almost 40%. Therefore, it can be Page 143 of 227

concluded that the reduction in the overall seismic hazard demonstrated by the updated curves will result in a lower mean CDF at PNP.

Because the contribution to risk from seismic events has decreased, the original conclusions of the IPEEE bound any changes to seismic and total CDF and LERF resulting from use of the updated USGS hazard curves. Therefore, the related information found in attachment W of the LAR remains acceptable (i.e., seismic CDF is < 1E-05).

Because no change to the DBA seismic event (SSE) is required based on the updated seismic hazard curves, and no change to plant fragilities is indicated, the overall conclusions in the IPEEE regarding seismic-fire interactions remain accurate and acceptable.

References:

[1] ASME/ANS RA-S- 2009, Standard for Level 1/Large Early Release Frequency Probabilistic Risk Assessment for Nuclear Power Plant Applications, the American Society of Mechanical Engineers and the American Nuclear Society, approved by the American National Standards Institute on February 2, 2009.

[2] PNP Fire PRA Peer Review to Requirements in Part 4 of the ASME/ANS Standard for Level 1/Large Early Release Frequency Probabilistic Risk Assessments for Nuclear Power Plant Applications, Report 17825-1, July 2011.

[3] PNP Electric Generating Plant Individual Plant Examination of External Events (IPEEE), Revision 1, May 1996.

[4] U.S. NRC, Generic Issue 199 (GI-199) Implications of Updated Probabilistic Seismic Hazard Estimates In Central And Eastern United States on Existing Plants Safety/Risk Assessment, August 2010.

mm) Key assumptions and sources of uncertainty for the FPRA model were comprehensively identified, documented and characterized by examining NUREG/CR-6850 Tasks 1 through 14, and describing key sub-tasks, identifying primary sources of uncertainty and characterizing the sensitivity of the results to the identified sources of uncertainty for each.

The approach relied on recognition of the following fire PRA developmental realities.

The development of a risk assessment inherently results in the introduction of uncertainty in the analysis results.

In general, the sources of uncertainty for each of the FPRA development tasks are discussed in an industry reference document, NUREG/CR-6850.

Page 144 of 227

Much of the potential burden associated with this task can be mitigated through careful selection of input parameters.

o For example, to the extent a specific analysis could accommodate conservative treatment while still satisfying the underlying goal of producing relatively realistic results, the analysis tended to do so.

Consequently, the treatment of uncertainty and sensitivity can be primarily limited to those fire scenarios where the refinements described in NUREG/CR-6850 Tasks 8 through 12 were applied.

o These scenarios will have been those that otherwise would have been notable risk contributors. Other fire scenarios that are subject to less aggressive refinements are expected to maintain a degree of conservatism so that their treatment would more closely resemble that of an upper bound analysis.

This approach embodied criteria for judging the importance of key assumptions and sources of uncertainty including: estimated impact on overall CDF and/or LERF, estimated impact on risk significant scenarios (i.e., those where refinements were utilized), degree of conservatism in the underlying deterministic analyses (e.g., heat release rates, zones of influence, damage thresholds, etc.),

elements of the fire PRA or PRA typically proven or judged to be uncertain and important.

Sensitivity studies were performed on: non-suppression probabilities, human error probabilities, fire ignition bin frequencies (in addition to the sensitivity analysis required by the use of NUREG/CR-6850 Supplement 1 (EPRI) ignition frequencies for all bins), and assumed cable routings.

The sensitivity analysis noted above will be re-performed with the base RAI Response Fire PRA Model. The results of these sensitivity cases will be included in the updated revision to the Fire PRA Fire Risk Quantification and Summary Notebook for the RAI Response Fire PRA Model. The results of the uncertainty analysis considered for SR UNC-A2 will also be summarized in RAI Response Fire PRA Model (refer to the response to RAI 01nn).

It is not necessary to perform these same set of sensitivities for every CDF and LERF evaluation included in Attachment W to gain appropriate insights. Based on the results of the sensitivity studies to be performed on the base RAI Response Fire PRA Model, a characterization of the impact on the CDF and LERF evaluations will be provided in the RAI Response Fire PRA Model results.

nn) The RAI Response Fire PRA Model will be populated with uncertainty interval information for the FPRA parameters listed below.

Fire ignition frequencies Page 145 of 227

Spurious Actuation Probabilities Severity Factors Non-suppression Probabilities Correlation groups are being established for the ignition frequency bins, spurious actuation probability values for common cable hot short failure modes, and common severity factors used for pump oil fires. Equations within the CAFTA basic event database file are used when multiple ignition frequency bins or spurious actuation probability values are included in a single basic event. The use of equations allows the individual uncertainty intervals to remain correlated while retaining the same number of basic events. Although uncertainty interval information is provided, a correlation group for other severity factors and the non-suppression probabilities is not provided as these represent unique values that are derived for each scenario (and therefore do not appear together in any cutsets and as such would not significantly contribute to the SOKC impact on the FPRA results).

The results of the uncertainty analysis taking into account the SOKC for these fire PRA parameters will also be summarized in the RAI Response Fire PRA Model results.

NRC Request PRA RAI 02 The American Society of Mechanical Engineering/American Nuclear Society (ASME/ANS) PRA Standard and Regulatory Guide 1.200, Rev. 2, provide guidance for the technical adequacy, including supporting requirements and peer reviews. Section 2.4.3.3 of NFPA 805 states that the probabilistic safety assessment (PSA) (PSA is also referred to as PRA) approach, methods, and data shall be acceptable to the authority having jurisdiction (AHJ). RG 1.205, Risk- Informed, Performance-Based Fire Protection for Existing Light-Water Nuclear Power Plants, Rev. 1, provides guidance for use in complying with the requirements promulgated for risk-informed, performance-based fire protection programs that meet the requirements of 10 CFR 50.48(c) and the referenced 2001 Edition of NFPA 805. Regulatory Guide 1.205 identifies NUREG/CR-6850 as documenting a methodology for conducting a FPRA and endorses, with exceptions and clarifications, NEI 04-02, Guidance for Implementing a Risk-Informed, Performance-Based Fire Protection Program Under 10 CFR 50.48(c), Rev. 2, as providing methods acceptable to the NRC for adopting a fire protection program consistent with NFPA-805. The following additional information is requested in order for the staff to complete its review:

Summarize any FPRA development completed since the final peer review as well as any development or reviews not yet complete and their significance to the NFPA 805 LAR.

Page 146 of 227

ENO Response The final phase of the Fire PRA peer review was conducted during the week of March 21, 2011. The initial LAR submittal was made on December 12, 2012. This RAI response focuses on changes made to the Fire PRA model between March 21, 2011 and September 2012, which is when the Fire PRA model was frozen for the December 2012 LAR submittal.

A summary of changes to the Fire PRA model between March 21, 2011 and September 2012 are provided below:

Updated coincident LOOP probability to resolve FPIE F&Os.

Updated consequential ISLOCA, PCP-SEAL and VSBLOCA event trees to resolve fire peer review finding ES-A3-01.

Updated various modeling logic and data to address MSOs per findings ES-A5-01, ES-D1-01, PRM-B3-02, PRM-B5-01, and PRM-B9-01.

Updated base model HRA logic and HEPs to address findings ES-C1-01, ES-C2-01, HRA-B2-01, and HRA-D1-01.

Updated PCP DC power dependency logic per finding PRM-B3-01.

Updated signal interlock logic (CHP, CHR, SIS) to address finding ES-A2-01.

Added final recovery actions and proposed plant modifications to the model to resolve finding FQ-A4-01.

To address the latest methodology for pressurizer safety valve failure per NUREG-7037, a new transfer event tree and fault tree logic was added to the base fire model.

Additional changes to the cable mapping and fire scenario development aspects included the following:

Incorporated additional cable data to address findings CS-A9-01 and CS-C1-01.

However, the results were not fully implemented at the time of the LAR submittal.

This effort has subsequently been completed. Refer to the responses to PRA RAIs 01a and 01c.

Performed a second assumed routing expert panel and incorporated results into the cable mapping tables to address finding FSS-A3-01.

Updated the main control room abandonment scenarios to address finding FSS-B2-01.

Updated the hotwork severity factors to address finding FSS-C4-01.

Increased the assumed transient heat release rate to 317 kW to address finding FSS-D4-01.

Re-calculated the turbine-generator catastrophic fire frequency and incorporated the scenario into the model to address finding FSS-F3-01.

Updated the multi-compartment analysis to address findings FSS-G2-01, FSS-G2-02, FSS-G4-01, and FSS-G5-01.

Following the March 2011 peer review of the detailed fire modeling and scenario development in the 1C switchgear room, incorporated results from detailed fire Page 147 of 227

modeling in the 1D switchgear room which was completed after the peer review using the same methodology as the 1C switchgear room.

Refined scenarios in various PAUs utilizing same methodology that was subject to the final peer review.

The previously identified unfinished aspects at the time of the LAR are the subject of other RAIs (refer to PRA RAIs 01a and 01c) as noted above. The resolutions of these items will be incorporated into the RAI Response Fire PRA Model. Subsequent changes to the Fire PRA model made in response to other RAIs are described in the response to PRA RAI 27. The numerical effect of all these changes will be reflected in the base quantification of the RAI Response Fire PRA Model.

References:

[1] ERIN 0247-07-0005-03, Revision 1, Palisades Nuclear Plant Fire Probabilistic Risk Assessment Model, November 2012.

[2] EA-PSA-FPIE-FIRE-12-04, Palisades Full Power Internal Events and Fire PRA Model, Revision 0, November 2012.

NRC Request PRA RAI 03 The ASME/ANS PRA Standard and RG 1.200, Rev. 2, provide guidance for the technical adequacy, including supporting requirements and peer reviews. Section 2.4.3.3 of NFPA 805 states that the PSA approach, methods, and data shall be acceptable to the AHJ. RG 1.205, provides guidance for use in complying with the requirements promulgated for risk-informed, performance-based fire protection programs that meet the requirements of 10 CFR 50.48(c) and the referenced 2001 Edition of NFPA 805. RG 1.205 identifies NUREG/CR-6850 as documenting a methodology for conducting a FPRA and endorses, with exceptions and clarifications, NEI 04-02, Rev. 2, as providing methods acceptable to the NRC for adopting a fire protection program consistent with NFPA-805. The following additional information is requested in order for the staff to complete its review:

A number of F&O dispositions listed in LAR Attachments U and V are indicated as having an open status. For each open F&O disposition, discuss its significance to the NFPA 805 LAR (transition and post-transition).

ENO Response Attachment U of the LAR addresses the quality of the full power internal events (FPIE)

PRA model. Attachment V addresses the fire PRA (FPRA). The table in Attachment U is arranged such that Non-Flooding related findings are presented first, followed by findings applicable only to the flooding analysis. Suggestions are presented in the last portion of that table in the same order. As is described further in the response below, findings related to internal flooding have no significance to the NFPA 805 LAR for either Page 148 of 227

the transition or post-transition phases.

From Attachment U of the LAR, there are four non-flooding findings listed as open.

There is one non-flooding suggestion listed as open.

For internal flooding findings, Attachment U includes 14 open findings and three open suggestions.

From Attachment V, there are 13 open findings, and there are no open suggestions.

Each of the open findings and suggestions are revisited in the following sections, and the significance to the LAR from their dispositions noted.

In several instances, the responses for multiple findings/suggestions are the same or similar (e.g., for several findings related to human failure event dependency analyses).

In those instances, the findings/suggestions are listed together, with the response provided addressing each and all of the items in the grouping.

Non-flood related F&Os from Attachment U are addressed first, followed by flood-related F&Os. Finally, F&Os from Attachment V are listed.

Page 149 of 227

ATTACHMENT U Attachment U Open F&Os - Non-Flooding SR (NON- Topic Finding/Observation Significance to the NFPA 805 Related RAI(s)

FLOODING) (From LAR Attachment U) LAR HR-G7-01 (Finding) For multiple human actions in the same PNP has not completed their HFE The methodology used to PRA RAI 28c accident sequence or cut set, identified in Dependency Evaluation for their complete the FPRA dependency accordance with supporting requirement QU- updated HRA. This is specifically analysis reported as part of the C1, ASSESS the degree of dependence, and noted in Section 5.2 of PLP-HRA. LAR is the same as that calculate a joint human error probability that Failure to meet explicit employed for the FPIE PRA and reflects the dependence. ACCOUNT for the requirement of the standard. is well understood. No impact on influence of success or failure in preceding the LAR due to the dependency human actions and system performance on the After the HRA is complete, redo analysis approach is expected for human event under consideration including (a) and document the dependency either the transition or post-the time required to complete all actions in evaluation. transition phases.

relation to the time available to perform the actions (b) factors that could lead to dependence (e.g., common instrumentation, Transition Phase: NONE common procedures, increased stress, etc.) (c) Post-Transition Phase: NONE availability of resources (e.g., personnel)

QU-C1-01 (Finding) IDENTIFY cutsets with multiple HFEs that Conditional HEPs were The methodology used to PRA RAI 28c potentially impact significant accident developed by PNP for several complete the FPRA dependency sequences/ cutsets by requantifying the PRA HFEs and incorporated in the analysis reported as part of the model with HEP values set to values that are fault tree models. Some accident LAR is the same as that sufficiently high that the cutsets are not sequences revealed HFE employed for the FPIE PRA and truncated. The final quantification of these post- combinations for which is well understood. No impact on initiator HFEs may be done at the cutset level or dependency between the HFEs the LAR due to the dependency saved sequence level. has not been assessed and analysis approach is expected for documented. either the transition or post-While the PNP model has been transition phases.

quantified and cut sets for accident sequences have been Transition Phase: NONE identified, the review and update of those sequences with respect Post-Transition Phase: NONE to combinations of HFEs is not complete.

Complete review and update of accident sequence cut sets relating to combinations of HFEs.

QU-B2-01 (Finding) TRUNCATE accident sequences and PNP used a truncation level of The CDF is expected to be in the PRA RAI 28f associated system models at a sufficiently low 1E-09 for quantification and range of 1E-5 to 1E-6, indicating cutoff value that dependencies associated with conducted evaluation of that truncation to 1E-9 to 1E-10 significant cutsets or accident sequences are convergence of the results down may be adequate. Modifying the not eliminated. Note: Truncation should be to a truncation level of 1E-12. The truncation, and lowering the Page 150 of 227

SR (NON- Topic Finding/Observation Significance to the NFPA 805 Related RAI(s)

FLOODING) (From LAR Attachment U) LAR carefully assessed in cases where cutsets are truncation should be set to 1E¬11 convergence level accordingly, merged to create a solution (e.g., where system based on the PNP definition of will have no impact on the FPRA level cutsets are merged to create sequence significant accident sequences. model results as it is subject to its level cutsets). own truncation sensitivity study.

Transition Phase: NONE Post-Transition Phase: NONE QU-D1-01 (Finding) REVIEW a sample of the significant accident The final model review has not Reviews of the interim versions of PRA RAI 28g sequences/cutsets sufficient to determine that been completed and the FPIE PRA were conducted for the logic of the cutset or sequence is correct. documented.The final review of each version, and insights from accident sequence results has those reviews have been used to not been completed and improve each subsequent version documented so that the of the PRA. The PSAR3 model reasonableness of the results can cutset review of accident be verified. PNP indicated that sequence results are documented this review is required but not in Attachment N of EA-PSA-FPIE-complete. This finding is being FIRE-12-04 [1] for all sequences written against all of the QU-D from the base FPIE model that supporting requirements as well are applied in the fire analysis.

as some QU-F requirements. Upon completion of the next PNP needs to complete the update to the FPIE PRA, the formal review of accident cutsets will again be reviewed for sequence quantification results accuracy and reasonableness.

and make modifications as The findings of the review will be needed to address issues found used as input to a similar review in that review. The final results to be conducted on the next should then be documented in the version of the fire PRA, and the corresponding notebooks. contents of Attachment N of Reference [1] updated as necessary.

No impact to the LAR (transition or post-transition phase) is expected.

Transition Phase: NONE Post-Transition Phase: NONE QU-A1-01 INTEGRATE the accident sequences, system Figure 5-1 of the Quantification This suggestion refers to a NA (Suggestion) models, data, and HRA in the quantification Report provides a small flow chart documentation issue, and does process for each initiating event group, on the process of integrating the not impact the actual modeling or accounting for system dependencies, to arrive CAFTA models into the SAPHIRE quantification process utilized in at accident sequence frequencies. code and the additional APIs the FPIE and fire PRAs.

used to prepare the SAPHIRE Implementing this suggestion will model for quantification (including have no impact on the NFPA-805 the integration of CCF trees, HRA analysis.

rules, etc.) While this flow chart Transition Phase: NONE gives an upper level explanation Page 151 of 227

SR (NON- Topic Finding/Observation Significance to the NFPA 805 Related RAI(s)

FLOODING) (From LAR Attachment U) LAR of the process, a more detailed Post-Transition Phase: NONE flow chart would be useful in ensuring a consistent integration for personnel that do not perform this task frequently.

The integration process is fairly complex and involves multiple codes and tools. Missing any step in this process could impact quantification.

Develop a more detailed flow chart for those performing the quantification.

Page 152 of 227

Attachment U Open F&Os - Flooding Findings and suggestions associated with the internal flooding analysis are in fact specific only to that analysis, and have no bearing on the fire PRA and the NFPA-805 effort. Thus, although the findings may be relevant to the results of the flooding analysis, their disposition will not have significance to the transition or post-transition phases of the LAR.

Finding/Observation Significance to the NFPA 805 SR (FLOODING) Topic Related RAI(s)

(From LAR Attachment U) LAR IFEV-A5-01 DETERMINE the flood-initiating event Key-words are used to identify These findings (IFEV-A5-01, NA (Finding) frequency for each flood scenario group by the potential applicable LERs in IFEV-A5-02, IFEV-A5-03, and using the applicable requirements in 2-2.1. the INPO Database (5 found). IFEV-A5-04) are related to the However, as noted by the flood analysis only. There is no additional LERs identified in 5750 significance to the NFPA-805 (3 additional found) - key word LAR.

searches on the LER database The findings refer to the are not comprehensive (this development of flood initiating appears to be because some event frequencies. The initiating utilities did not provide any key event frequencies calculated for words on their LERs, or the key flooding events have no bearing words provided are not consistent on the fire PRA or its results.

or comprehensive).

Transition Phase: NONE As evidenced by the 5750 report, an LER search using key words is Post-Transition Phase: NONE not comprehensive or complete.

Since the 5750 report only covers a subset of the 16 years of LERs that are being considered for the generic prior data, it is probable that additional applicable LERs were missed.

Since the population is so small, missing even 1 LER has an impact on the internal-flood frequency.

The analysis performed could perform a review of all LERs in the INPO LER Database over the period of 1987-2002 to identify potentially missed Internal-flooding LERs (would be very time-intensive) to ensure completeness.

OR A number of utilities calculate the Page 153 of 227

Finding/Observation Significance to the NFPA 805 SR (FLOODING) Topic Related RAI(s)

(From LAR Attachment U) LAR internal-flood frequency based on the EPRI TR-101341report (note a newer report is to be issued imminently). This approach could be used for PNP.

IFEV-A5-02 DETERMINE the flood-initiating event When calculating the internal- These findings (IFEV-A5-01, NA (Finding) frequency for each flood scenario group by flooding generic prior, a capacity IFEV-A5-02, IFEV-A5-03, and using the applicable requirements in 2-2.1. factor of 75% was assumed. IFEV-A5-04) are related to the The 75% capacity factor was flood analysis only. There is no stated to be assumed based on significance to the NFPA-805 industry operating data from LAR.

1987- 1995 (as reported in The findings refer to the NUREG-CR/5750). Since this development of flood initiating only covers a subset of the years event frequencies. The initiating contained in the LER review, and event frequencies calculated for since the later years (where flooding events have no bearing capacity factors for industry were on the fire PRA or its results.

higher) are not being included, Transition Phase: NONE the capacity factor is not reflective of the actual operating history, Post-Transition Phase: NONE and appears to be under-estimated. Note: a quick calculation of the capacity factor based on Table A1.4-4 information calculated a capacity factor >80% for the years specified in NUREG-CR/5750).

Use the data available on the NRC website, and calculate the actual capacity factor for the years of interest.

IFEV-A5-03 DETERMINE the flood-initiating event The plant-specific data is based These findings (IFEV-A5-01, NA (Finding) frequency for each flood scenario group by on operating history for the "life" IFEV-A5-02, IFEV-A5-03, and using the applicable requirements in 2-2.1. of PNP. The generic priors is IFEV-A5-04) are related to the based on industry data from 1987 flood analysis only. There is no

- 2002 (including PNP data). significance to the NFPA-805 Need to provide a justification as LAR.

to why the "overlap" of data is The findings refer to the acceptable from a Bayesian development of flood initiating updating perspective. event frequencies. The initiating Bayesian updating principles event frequencies calculated for require the priors to be flooding events have no bearing "independent" of the update data. on the fire PRA or its results.

Remove the PNP data from the Transition Phase: NONE generic priors, or only use PNP Page 154 of 227

Finding/Observation Significance to the NFPA 805 SR (FLOODING) Topic Related RAI(s)

(From LAR Attachment U) LAR data since 2002. Post-Transition Phase: NONE IFEV-A5-04 DETERMINE the flood-initiating event LER Screening criteria A1.3.1g These findings (IFEV-A5-01, NA (Finding) frequency for each flood scenario group by appears to be non-conservative. IFEV-A5-02, IFEV-A5-03, and using the applicable requirements in 2-2.1. This assumption/screening IFEV-A5-04) are related to the criteria states: Leaks in the HPSI flood analysis only. There is no system or in the diesel generator significance to the NFPA-805 cooling systems were not LAR.

considered since these systems The findings refer to the would be operating only as a development of flood initiating result of another event. Since event frequencies. The initiating testing and maintenance of these event frequencies calculated for systems at power also require the flooding events have no bearing systems to be in operation, on the fire PRA or its results.

events associated with these systems should not be excluded Transition Phase: NONE (maintenance events could - and Post-Transition Phase: NONE probably are - the most likely source of potential flooding events associated with these systems).

Since the frequency of maintenance events is based on "back calculating" from the total frequency - screening out these events appears to be non-conservative.

Don't screen out the events associated with systems that can be / are tested / maintained at power.

IFEV-A7-01 INCLUDE consideration of human-induced PNP calculates the human- These findings (IFEV-A7-01 and NA (Finding) floods during maintenance through application induced floods during IFSO-A4-01) are also related to of generic data. maintenance by "back- the flood analysis only. There is calculating" the maintenance- no significance to the NFPA-805 induced floods by taking the LAR.

overall internal-flood frequency The findings refer to the and subtracting out "passive development of flood initiating failures" which contribute to the event frequencies. The initiating frequency. This number is then event frequencies calculated for further reduced by assuming that flooding events have no bearing only 30% of the maintenance- on the fire PRA or its results.

induced failures would required at power based on interviews and Transition Phase: NONE operating practices from the early Post-Transition Phase: NONE 1990's.

Page 155 of 227

Finding/Observation Significance to the NFPA 805 SR (FLOODING) Topic Related RAI(s)

(From LAR Attachment U) LAR The 70/30 split is based on discussions that occurred during the IPE days (early 1990's).

During that time frame, most sites maintenance practices included a majority of maintenance being performed during outages. However, this philosophy has changed for the industry, and maintenance being performed at power in order to shorten outage durations is much more common. Therefore, the 70/30 split may no longer be applicable.

Since this split is used to reduce the overall internal-flood frequency, it has a direct impact on the internal-flood frequencies for the various scenarios being induced.

IFSO-A4-01 For each potential source of flooding water, PNP did not explicitly identify and These findings (IFEV-A7-01 and NA (Finding) IDENTIFY the flooding mechanisms that would characterize human induced IFSO-A4-01) are also related to result in a fluid releaser. INCLUDE: (a) failure flooding events for each flood the flood analysis only. There is modes of components such as pipes, tanks, area. Instead, PNP chose to no significance to the NFPA-805 gaskets, expansion joints, fittings, seals, etc. (b) characterize the human-induced LAR.

human-induced mechanisms that could lead to flooding events by setting a The findings refer to the overfilling tanks, diversion of flow through generic element and then back- development of flood initiating openings created to perform maintenance; calculating a frequency without event frequencies. The initiating inadvertent actuation of fire suppression system actually delineating what the event frequencies calculated for (c) other events resulting in a release into the human induced event was. flooding events have no bearing flood area Without a reasonable on the fire PRA or its results.

characterization of the specific Transition Phase: NONE human induced flooding events it is difficult to understand their full Post-Transition Phase: NONE impact on the results or address them should they be found to be significant contributors.

PNP should either more fully characterize the human induced flooding events or they should be explicitly called out as assumptions so that they can be assessed for applications affecting internal flooding.

Page 156 of 227

Finding/Observation Significance to the NFPA 805 SR (FLOODING) Topic Related RAI(s)

(From LAR Attachment U) LAR IFQU-A3-01 SCREEN OUT a flood area if the product of the Because PNP used a truncation These findings (IFQU-A3-01, NA (Finding) sum of the frequencies of the flood scenarios for limit of 1E-09/yr, it is potential that IFQU-A7-01, and IFSN-A3-01) the area, and the bounding conditional core 1 of the 2 flood areas that are are related to the flood analysis damage probability (CCDP) is less than 10- reported as having a CDF < 1E- only. There is no significance to 9/reactor year. The bounding CCDP is the 9/yr (F01 - E sfgrd, and F06 - Aux the NFPA-805 LAR.

highest of the CCDP values for the flood Bldg) may be artificially Regardless of which areas are scenarios in an area. "screened" even though there is screened or not screened, the no positive evidence this criteria screening of areas in the flood was met for the zone. analysis has no impact on the fire Since the east safeguard room PRA.

has 6 scenarios associated with Therefore, there is no impact on it, it could exceed the 1E-9/yr the LAR from these findings.

CDF if each of the scenarios were in the 2E-10/yr range. Based on Transition Phase: NONE the CDF summary provided in Post-Transition Phase: NONE NB-PSA-IF Rev. 0, the 2 zones with a CDF <1E-9/yr are not considered as one of the eleven flood zones defined for the PNP, so it may have been inappropriately screened.

Lower the truncation limit used during quantification.

IFQU-A7-01 PERFORM internal flood sequence A truncation limit of 1E-09/yr was These findings (IFQU-A3-01, NA (Finding) quantification in accordance with the applicable used for the Internal Flooding IFQU-A7-01, and IFSN-A3-01) requirements described in 2-2.7 analysis. The acceptability of this are related to the flood analysis truncation limit was not provided. only. There is no significance to There is no evidence that this the NFPA-805 LAR.

truncation is sufficiently low to Regardless of which areas are meet the requirements of QU-B3 screened or not screened, the (demonstrates that the overall screening of areas in the flood internal flood model results analysis has no impact on the fire converge and no significant PRA.

accident sequences are Therefore, there is no impact on inadvertently eliminated.) the LAR from these findings.

Lower the truncation limit until Transition Phase: NONE convergence is obtained.

Post-Transition Phase: NONE IFSN-A3-01 For each defined flood area and each flood Because PNP used a truncation These findings (IFQU-A3-01, NA (Finding) source, IDENTIFY those automatic or operator limit of 1E-09/yr, it is potential that IFQU-A7-01, and IFSN-A3-01) responses that have the ability to terminate or 1 of the 2 flood areas that are are related to the flood analysis contain the flood propagation. reported as having a CDF < 1E- only. There is no significance to 9/yr (F01 - E safeguard, and F06 - the NFPA-805 LAR.

Aux Bldg) may be artificially Regardless of which areas are "screened" even though there is Page 157 of 227

Finding/Observation Significance to the NFPA 805 SR (FLOODING) Topic Related RAI(s)

(From LAR Attachment U) LAR no positive evidence this criteria screened or not screened, the was met for the zone. screening of areas in the flood Since the east safeguard room analysis has no impact on the fire has 6 scenarios associated with PRA.

it, it could exceed the 1E-9/yr Therefore, there is no impact on CDF if each of the scenarios were the LAR from these findings.

in the 2E-10/yr range. Based on Transition Phase: NONE the CDF summary provided in NB-PSA-IF Rev. 0, the 2 zones Post-Transition Phase: NONE with a CDF <1E-9/yr are not considered as one of the eleven flood zones defined for the PNP, so it may have been inappropriately screened.

Lower the truncation limit used during quantification.

IFQU-A6-01 For all human failure events in the internal flood In Section 7.3.2 of EA-PSA- This finding is related to the flood NA (Finding) scenarios, INCLUDE the following scenario- INTFLOOD 03(03), PNP analysis only. There is no specific impacts on PSFs for control room and states "Human errors developed significance to the NFPA-805 ex-control room actions as appropriate to the as a part of the internal events LAR.

HRA methodology being used: (a) additional PRA have been left at their Because the flood-related human workload and stress (above that for similar existing failure probabilities. This failure events (HFEs) and their sequences not caused by internal floods) (b) would be reasonable given that human error probabilities (HEPs) cue availability (c) effect of flood on mitigation, plant response to transient and are not used in the fire PRA, this required response, timing, and recovery loss of offsite power related finding has no impact on the fire activities (e.g., accessibility restrictions, events should be similar PRA results. Therefore, there is possibility of physical harm) (d) flooding-specific regardless of the exact cause of no significance to the LAR from job aids and training (e.g., procedures, training the initiating event. However, with this finding.

exercises) the additional complication of a flood, performance shaping Transition Phase: NONE factors (PSFs) in the internal Post-Transition Phase: NONE events PRA may not be as appropriate." As part of their quantification, PNP did not change the HEPs from the internal events HFEs. Therefore, PNP did not address the flood specific impacts on the PSFs.

The flood specific impacts are such that the HEPs carried over are non-conservatively low.

Revise the quantification of the internal events HEPs to address the impact of the flood on the PSFs.

Page 158 of 227

Finding/Observation Significance to the NFPA 805 SR (FLOODING) Topic Related RAI(s)

(From LAR Attachment U) LAR IFQU-A9-01 INCLUDE, in the quantification, both the direct A specific discussion of jet This finding is related to the flood NA (Finding) effects of the flood (e.g., loss of cooling from a impingement and pipe whips was analysis only. There is no service water train due to an associated pipe not identified. significance to the NFPA-805 rupture) and indirect effects such as Consideration of jet impingement LAR.

submergence, jet impingement, and pipe whip, and pipe whips (as appropriate) Pipe whip and jet-impingement do as applicable. are a requirement of the standard not affect the fire scenario for this element. modeling or quantification results.

Provide a discussion of how jet There is no significance to the impingements and pipe whips LAR from this finding.

were considered and handled. Transition Phase: NONE The Internal Flooding Analysis Post-Transition Phase: NONE Report referenced walkdowns performed for the IPE. The scope of these walkdown was limited as a result of time constraints placed on the walkdown team by the authorized team escort. PNP indicated that they had performed a more recent complete walkdown, but that walkdown was not referenced in the Internal Flooding Analysis Report. The consideration of jet impingement and pipe whip is qualitatively and semi-quantitatively discussed in the walkdown notes for the more recent walkdown. If PNP wants to credit the more recent walkdown, they need to reference it in the Internal Flooding Analysis Report.

IFSN-A15-01 For each defined flood area and each flood The Heater Drain pump suction This finding is related to the flood NA (Finding) source, IDENTIFY the propagation path from tank T-5 has insufficient capacity analysis only. There is no the flood source area to its area of to flood the room. Tank T-60 NA significance to the NFPA-805 accumulation. Dirty Waste Drain Tank RI-In- LAR.

Service Inspection (ISI) does not This finding refers to a flood evaluate tanks, only pipes. propagation path from the turbine However, this tank has insufficient building or auxiliary building to the volume to flood area to any EDG room given a tank or pipe significant height. HBD-13-3 Misc rupture initiating event. Such West Drain Tank T89A/B From information has no applicability to Spool To Condensate Storage fire initiating events. Thus, there Tank Water 20. The is no impact on the LAR.

documentation states that it is assumed that there is insufficient Transition Phase: NONE volume to flood to level of EDG - Post-Transition Phase: NONE but the justification/basis for this Page 159 of 227

Finding/Observation Significance to the NFPA 805 SR (FLOODING) Topic Related RAI(s)

(From LAR Attachment U) LAR assumption is not provided.

Need to verify that the basis for the assumption is valid or an additional EDG failure mode could be missed.

Provide basis for determining or assuming insufficient volume.

IFSN-A17-01 CONDUCT a plant walkdown(s) to verify the The scope of the walkdown was This finding is related to the flood NA (Finding) accuracy of information obtained from plant limited to the identified areas as a analysis only. There is no information sources and to obtain or verify (a) result of time constraints placed significance to the NFPA-805 SSCs located within each defined flood area (b) on the walkdown team by the LAR.

flood/spray/other applicable mitigative features authorized team escort. His The finding describes a historical of the SSCs located within each defined flood limited availability resulted in the walkdown of some flood areas; area (e.g., drains, shields, etc.) (c) pathways walkdown team prioritizing the that walkdown was incomplete.

that could lead to transport to the flood area areas reviewed. This is a flooding risk assessment Because of the limited walkdown documentation issue that has no time, some rooms were not impact on the fire PRA. Thus, walked down, therefore the pipe there is no significance to the lengths for these rooms were not LAR.

identified. This results in the Transition Phase: NONE "frequency" for the rooms relying on RI-ISI data instead of being Post-Transition Phase: NONE able to use the best available data for pipe/component failure rates.

The Internal Flooding Analysis Report referenced walkdowns performed for the IPE. The scope of these walkdown was limited as a result of time constraints placed on the walkdown team by the authorized team escort. PNP indicated that they had performed a more recent complete walkdown, but that walkdown was not referenced in the Internal Flooding Analysis Report. If PNP wants to credit the more recent walkdown, they need to reference it in the Internal Flooding Analysis Report.

IFSN-A6-01 For the SSCs identified in IFSN-A5, IDENTIFY Spray effects from chilled water This finding is related to the flood NA (Finding) the susceptibility of each SSC in a flood area to systems pipe failures don't seem analysis only. There is no flood-induced failure mechanisms. INCLUDE to be addressed. The basis for significance to the NFPA-805 failure by submergence and spray in the elimination as a spray Page 160 of 227

Finding/Observation Significance to the NFPA 805 SR (FLOODING) Topic Related RAI(s)

(From LAR Attachment U) LAR identification process. EITHER: a) ASSESS consideration is only documented LAR.

qualitatively the impact of flood-induced for 1 room - not for all rooms This finding refers to the impact of mechanisms that are not formally addressed transgressed. a pipe rupture or spray initiating (e.g., using the mechanisms listed under This requirement states that the event from the chilled water Capability Category III of this requirement), by susceptibility of each SSC in a system, and is not applicable to using conservative assumptions; OR b) NOTE flood area to flood-induced fire initiating events. Thus, there that these mechanisms are not included in the failures mechanisms by either is no significance to the LAR.

scope of the evaluation. submergence or spray are Transition Phase: NONE included.

Post-Transition Phase: NONE Address the potential spray effects from chilled water system pipe failures in all zones transgressed.

IFSN-A12-01 For each defined flood area and each flood Although the screening of rooms This suggestion is related to the NA (Suggestion) source, IDENTIFY the propagation path from appears to be reasonable, it is not flood analysis only. There is no the flood source area to its area of clear what criteria from the significance to the NFPA-805 accumulation. Standard was used for the LAR.

various flood areas screened. This suggestion refers to the Because of the multiple screening need for better documentation of criteria available, specifying the the screening criteria applied to criteria applied would be determine which plant rooms are beneficial to ensure no zones susceptible to flooding affect, and were inappropriately screened. has no applicability to fire related This may result in additional initiating events. Thus, there is zones being able to be screened, no significance to the LAR.

and would ensure the zones Transition Phase: NONE already screened were done in accordance with the Standard's Post-Transition Phase: NONE requirements.

Specify the criteria from the Standard that was applied to screen the various zones from further consideration.

IFSN-A8-01 IDENTIFY inter-area propagation through the Inter-area propagation through These suggestions (IFSN-A8-01 NA (Suggestion) normal flow path from one area to another via the normal flow path from one and IFSO-A1-01) are related to drain lines; and areas connected via backflow area to another via drain lines the flood analysis only. There is through drain lines involving failed check valves, were addressed by the GOTHIC no significance to the NFPA-805 pipe and cable penetrations (including cable runs. Areas connected via LAR.

trays), doors, stairwells, hatchways, and HVAC backflow through drain lines The suggestions describe the ducts. INCLUDE potential for structural failure involving failed check valves, pipe need to evaluate the potential (e.g., of doors or walls) due to flooding loads. and cable penetrations (including flood propagation and flood level cable trays), doors, stairwells, due to a rupture in the chilled hatchways. There didn't appear to water system, and/or to verify that be any evaluation of the chilled the basis for eliminating the water system flow Page 161 of 227

Finding/Observation Significance to the NFPA 805 SR (FLOODING) Topic Related RAI(s)

(From LAR Attachment U) LAR rates/propagation from pumps chilled water system as a flood through piping through the room source remains valid. They have coolers. no applicability to fire initiating Add documentation on the chilled events. Thus, there is no water system and the flow significance to the LAR.

rates/propagation potential. Transition Phase: NONE Post-Transition Phase: NONE IFSO-A1-01 For each flood area, IDENTIFY the potential The chilled water system was These suggestions (IFSN-A8-01 NA (Suggestion) sources of flooding [Note (1)]. INCLUDE: (a) identified as being in the Bus 1D, and IFSO-A1-01) are related to equipment (e.g., piping, valves, pumps) located Cable Spreading and Electrical the flood analysis only. There is in the area that are connected to fluid systems Equipment rooms, but was no significance to the NFPA-805 (e.g., circulating water system, service water eliminated from consideration as LAR.

system, component cooling water system, a flood source because it has The suggestions describe the feedwater system, condensate and steam insufficient volume to flood these need to evaluate the potential systems) (b) plant internal sources of flooding areas (Ref. [4], Appendix A, Final flood propagation and flood level (e.g., tanks or pools) located in the flood area List Of Potential due to a rupture in the chilled (c) plant external sources of flooding (e.g., Hazards/Postulated Effects In water system, and/or to verify that reservoirs or rivers) that are connected to the The Bus 1D Room). Table A2.8- the basis for eliminating the area through some system or structure (d) in- 3c: Sources Not Considered for chilled water system as a flood leakage from other flood areas (e.g., back flow the Internal Flood analysis source remains valid. They have through drains, doorways, etc.) update, from the plant walkdown no applicability to fire initiating supporting the IPE (Ref. [6]). The events. Thus, there is no basis for the insufficient capacity significance to the LAR.

cannot be found.

Transition Phase: NONE Since the chilled water was eliminated in the original IPE, Post-Transition Phase: NONE need to verify and document that the elimination criteria used is still valid, especially since there may have been chillers installed in the plant that use Chilled Water since the IPE was performed.

Provide the criteria/basis for how it was determined that the Chilled Water system could be eliminated from flooding impacts, and ensure the basis is still valid.

Page 162 of 227

ATTACHMENT V Attachment V Open F&Os Attachment V SR Topic Finding/Observation Significance to the NFPA 805 Related RAI(s)

(From LAR Attachment V) LAR CS-A9-01 (Finding) INCLUDE consideration of proper polarity hot PLP has conducted updates to Although the data gathering is PRA RAI 1a shorts on ungrounded DC circuits; requiring up the original cable selection to complete, the data was not fully to and including two independent faults could ensure multiple hot short failures incorporated into the model used result in adverse consequences. are identified. It is not evident for the LAR.

that the supplemental analysis Cable data for the PLP FPRA was work specifically looked for proper obtained from two separate polarity hot shorts on ungrounded sources: the SAFE database and DC circuits NEXUS spreadsheets.

The supplemental analysis work considered proper polarity hot shorts on ungrounded DC circuits up to and including two independent faults. The completed cable analysis will be integrated into the RAI Response Fire PRA model. There is no longer an exclusion from the risk estimates.

Transition Phase: MINIMAL Post-Transition Phase: MINIMAL CS-C1-01 (Finding) DOCUMENT the cable selection and location The cable selection and location Although data verification is PRA RAI 1c methodology applied in the Fire PRA in a methodology is documented in complete, the results had not manner that facilitates Fire PRA applications, Section 4 the Model Development been fully implemented into the upgrades, and peer review. Report (0247-07-0005.03) and model used for the LAR.

associated appendices. The Section 4 of the Model methodology for completed work Development Report, Reference is documented in a manner [3], has been updated in a consistent with this supporting manner that ensures consistent requirement; however, the interpretation of Fire PRA methodology for the supplemental applications.

cable selection review (Attachment 1) is not formally Additionally, the verification of documented in a manner that Appendix R Non-Safe Shutdown ensures consistent interpretation Cable Routing to Support the Fire for Fire PRA applications and PRA has been separately upgrades. Additionally, the documented in PLP-RPT sample cable routing verification 0134, Reference [4].

check is not formally documented The results of the circuit analysis in the Fire PRA Report or any data verification are the updated Page 163 of 227

other plant document, and thus cable selection and circuit does not lend itself to consistent analysis data tables loaded into treatment for future Fire PRA the SAFE program. The applications and upgrades. verification is complete and will be implemented in the RAI Response Fire PRA Model. There is no longer an exclusion from the risk estimates.

Transition Phase: MINIMAL Post-Transition Phase: MINIMAL FQ-C1-01 (Finding) ADDRESS dependencies during the Fire PRA PRA document NB-PSA-HR-1, Although a dependency analysis PRA RAI 1h plant response model quantification in Rev 3 provides an HEP was completed prior to the PRA RAI 1ff accordance with HLR-QU-C and its SRs in Part dependency analysis and submittal of the LAR, an update 2 and DEVELOP a defined basis to support the develops adjustment factors to to the FPRA is being completed.

claim of nonapplicability of any of the apply to the cutsets. Multiple As part of that update, the requirements under HLR-QU-C in Part 2. HFE's are evaluated for dependency analysis will be dependencies using the EPRI updated as well. The majority of HRA calculator. Dependency the operator actions and adjustment factors are developed dependencies evaluated for the and applied in the cutsets. LAR continue to exist in the However, the "Q" model [which updated FPRA, and their relative was reviewed] does not impact on CDF is expected to incorporate this work. Therefore remain the same. Some the F&O and the not met conservatism will be removed in assessment. the update as the HEPs for some HFEs are explicitly developed (to replace the screening values used for the LAR).

There is expected to be some impact on the CDF due to the updates, but the impact is expected to be minimal.

(Note: the event identifiers specifically referenced in the Finding have been either renamed or removed from the FPRA models. Those that remain in the model (with different names) are documented in Reference [2]).

Transition Phase: MINIMAL Post-Transition Phase: MINIMAL HRA-D2-01 INCLUDE operator recovery actions that can Many of the operator recovery Although a dependency analysis PRA RAI 1h (Finding) restore the functions, systems, or components actions associated with fire was completed prior to the PRA RAI 1ff on an as-needed basis to provide a more response are still modeled with submittal of the LAR, an update realistic evaluation of significant accident screening values; i.e., not to the FPRA is being completed.

sequences (same as HRA-D1-01). accounting for all of the relevant As part of that update, the Page 164 of 227

PSFs. Dependency analysis has dependency analysis will be been performed for the current updated as well. The majority of set of fire scenarios and operator the operator actions and actions in the "T" model. The dependencies evaluated for the results generated from the "Q" LAR continue to exist in the model did not incorporate the updated FPRA, and their relative dependency analysis. The impact on CDF is expected to dependency analysis needs to be remain the same. Some re-analyzed before finalization of conservatism will be removed in the Fire PRA model. This task is the update as the HEPs for some not complete yet. Also, HRA HFEs are explicitly developed (to Calculator evaluation sheets replace the screening values cannot be located for PCP- used for the LAR).

PMOF-P-50X-LOC and EDG- There is expected to be some PMOE-PORT-PUMP, and AFW- impact on the CDF due to the AVOA-CV-2010-D, SWS-AVOA- updates, but the impact is CV-0823-26, and SWS-AVOB- expected to be minimal.

CV-082447M still need to be modified for fire related conditions (Note: the event identifiers specifically referenced in the Finding have been either renamed or removed from the FPRA models. Those that remain in the model (with different names) are documented in Reference [2]).

Transition Phase: MINIMAL Post-Transition Phase: MINIMAL FSS-B1-01 DEFINE and JUSTIFY the conditions that are The current Fire PRA does not Control room abandonment PRA RAI 1k (Finding) assumed to cause MCR abandonment and/or consider abandonment of the scenarios with respect to reliance on ex-control room operator actions main control room due to lack of environmental effects are including remote and/or alternate shutdown equipment/control due to fire addressed for the LAR. However, actions. damage the LAR model did not address abandonment due to equipment damage.

Main Control room abandonment scenarios have been postulated based on damage to equipment and controls. An abandonment analysis (Attachment 1 of Reference [5]) was performed to determine the response of the CR envelope given a range of possible fire events. The analysis considered three different operating states of the CR mechanical ventilation system and three different configurations of the CR Door.

Page 165 of 227

As a result of this analysis and to address other issues, the RAI Response Fire PRA Model will include additional scenarios that model control room abandonment due to equipment damage, with control being transferred to other locations, such as the alternate shutdown panel. There is expected to be minimal significance to the LAR, as there may be additional analyses and operator actions that increase in importance.

Transition Phase: MINIMAL Post-Transition Phase: NONE FSS-E3-01 PROVIDE a mean value of, and statistical A qualitative characterization of As a part of the current efforts to PRA RAI 1nn (Finding) representation of, the uncertainty intervals for the parameters used in the fire update the FPRA, a parametric the parameters used for modeling the significant modeling in significant fire uncertainty evaluation will be fire scenarios. scenarios have not been included as part of the RAI completed as the Fire PRA still Response Fire PRA model needs detailed analysis to reduce development.

the plant CDF. The qualitative There will be no significance to discussion required to meet the NFPA-805 analysis as the category 1 should be completed results are based on the point once key scenarios are identified. estimate values which approximate the mean values.

The issues identified in these F&Os do not have a significant impact on the mean value; and have no impact on the point estimate mean values used in the analysis.

Transition Phase: NONE Post-Transition Phase: NONE UNC-A2-01 INCLUDE the treatment of uncertainties, The uncertainty intervals As a part of the current efforts to PRA RAI 1nn (Finding) including their documentation, as called out in assigned to Fire IEs, Severity update the FPRA, a parametric SRs PRM-A4, FQ-F1, IGN-A10, IGN-B5, FSS- Factors and Non Suppression uncertainty evaluation will be E3, FSS-E4, FSS-H5, FSS-H9, and CF-A2 and Probabilities are not based on included as part of the RAI that required by performing Part 2 referenced acceptable systematic methods. Response Fire PRA model requirements throughout this Standard. 1) Uncertainty distributions for fire development.

IEs have been assigned the same There will be no significance to error factor of 10 rather than the NFPA-805 analysis as the using posterior distributions from results are based on the point Bayesian update estimate values which

2) SF distributions have been approximate the mean values.

assigned without an underlying The issues identified in these Page 166 of 227

basis. F&Os do not have a significant

3) NSP uncertainty distribution impact on the mean value; and has been derived on the basis of have no impact on the point NUREG/CR 1278. This provides estimate mean values used in the guidance on HEP uncertainty analysis.

assessment. However, NSP Transition Phase: NONE terms are an output of a Post-Transition Phase: NONE combination of fire growth and suppression modeling and guidance in NUREG/CR 1278 has therefore little relevance. A valid approach would be to address the uncertainties in damage times in combination with uncertainties in suppression probabilities based on specific contributing factors.

4) Uncertainties associated with spurious actuation probabilities have been characterized according to a set of rules defined for severity factors. In this case spurious actuation probabilities with a failure probability of > 0.25 are assigned an error factor of 1.0. In contrast NUREG/CR 6850 recommend use of a uniform distribution with the following limits Cables with 15 or less conductors: +20%

Cables with more than 15 conductors: +50%

Alternatively the values included in tables 10-1 to 10-5 NUREG/CR 6850 could be used where limits appear to be wider. The PNP analysis has not accounted for larger uncertainties associated with cables with > 15 conductors.

HRA-A4-01 TALK THROUGH (i.e., review in detail) with As the fire scenario refinement Procedures, modification detail, PRA RAI 1bb (Finding) plant operations and training personnel the continues, additional fire operations review, and detailed PRA RAI 1dd procedures and sequence of events to confirm response actions will be identified HRA model development had not that interpretation of the procedures relevant to and evaluated, which will require been completed prior to the actions identified in SRs HRA-A1, HRA-A2, and the performance of additional submittal of the LAR. These HRA-A3 is consistent with plant operational and operator interviews. As such, this items are being addressed as training practices. task is not fully completed yet. part of the FPRA efforts.

Also, operator interviews for those As screening values were used Page 167 of 227

fire response actions that are still for all HFEs in the LAR, it is using screening values (e.g., expected that use of explicitly ACP-DGOT-B5B-DG, ACP- developed HEPs for some HFEs, PMOE-383-11A, ACP-PMOE- scoping HEPs for others, and 383-12A, AFW-PMOA-P8B- screening values for the CRAB, etc.) may not have been remainder, may reduce the CDF completed. to some degree. Whatever the (Note: Specific HEP basic event impact on CDF is, that impact is identifiers cited by the peer review not expected to substantially team may have been impact the results of the LAR.

subsequently renamed or (Note: the event identifiers removed from the model as part specifically referenced in the of the F&O resolution process.) Finding have been either renamed or removed from the FPRA models. Those that remain in the model (with different names) are documented in Reference [2]).

Transition Phase: MINIMAL Post-Transition Phase: MINIMAL HRA-C1-01 For each selected fire scenario, QUANTIFY the Fire response HFEs modeled with Procedures, modification detail, PRA RAI 1bb (Finding) HEPs for all HFEs and ACCOUNT FOR screening values have not yet operations review, and detailed PRA RAI 1dd relevant fire-related effects using detailed been evaluated in a manner HRA model development had not analyses for significant HFEs and conservative accounting for relevant PSFs been completed prior to the estimates (e.g., screening values) for (e.g., ACP-DGOT-B5B-DG, FPS- submittal of the LAR. These nonsignificant HFEs, in accordance with the PMOE-START-L, ACP-PMOE- items are being addressed as SRs for HLR-HR-G in Part 2 set forth under at 383-11A, ACP-PMOE-383-12A, part of the FPRA efforts.

least Capability Category II, with the following etc.). Also, HRA Calculator As screening values were used clarification: evaluation sheet cannot be for all HFEs in the LAR, it is (a) Attention is to be given to how the fire located for PCP-PMOF-P-50X- expected that use of explicitly situation alters any previous assessments in LOC and EDG-PMOE-PORT- developed HEPs for some HFEs, nonfire analyses as to the influencing factors PUMP, and AFW-AVOA-CV- scoping HEPs for others, and and the timing considerations covered in SRs 2010-D, SWS-AVOA-CV-0823- screening values for the HR-G3, HR-G4, and HR-G5 in Part 2 And 26, and SWS-AVOB-CV- remainder, may reduce the CDF 082447M still need to be modified to some degree. Whatever the (b) DEVELOP a defined basis to support the for fire related conditions. This claim of nonapplicability of any of the impact on CDF is, that impact is task is not completed. not expected to substantially requirements under HLR-HRG in Part 2.

impact the results of the LAR.

(Note: the event identifiers specifically referenced in the Finding have been either renamed or removed from the FPRA models. Those that remain in the model (with different names) are documented in Reference [2]).

Transition Phase: MINIMAL Page 168 of 227

Post-Transition Phase: MINIMAL HRA-E1-01 DOCUMENT the Fire PRA HRA including Documentation for HFEs Procedures, modification detail, PRA RAI 1bb (Finding) (a) those fire-related influences that affect the associated with selected fire operations review, and detailed PRA RAI 1dd methods, processes, or assumptions used as response HFEs (e.g., FPS- HRA model development had not well as the identification and quantification of PMOE-START-L, ACP-PMOE- been completed prior to the the HFEs/HEPs in accordance with HLR-HR-I 383-11A, ACP-PMOE-383-12A, submittal of the LAR. These and its SRs in Part 2, and DEVELOP a defined etc.) in the risk significant fire items are being addressed as basis to support the claim of nonapplicability of scenarios need to be provided. part of the FPRA efforts.

any of the requirements under HLR-HR-I in Part Also, HRA Calculator evaluation As screening values were used 2, and (b) any defined bases to support the sheets cannot be located for for all HFEs in the LAR, it is claim of nonapplicability of any of the PCP-PMOF-P-50X-LOC, EDG- expected that use of explicitly referenced requirements in Part 2 beyond that PMOE-PORT-PUMP, and developed HEPs for some HFEs, already covered by the clarifications in this Part PULLFUSE; AFW-PMOT-P-8B- scoping HEPs for others, and LOC seems to have been screening values for the changed to AFW-PMOT-P-8B- remainder, may reduce the CDF SBO in HRA notebook (but not to some degree. Whatever the changed in Fire PRA model); and impact on CDF is, that impact is AFW-AVOA-CV-2010-D, SWS- not expected to substantially AVOA-CV-0823-26, and SWS- impact the results of the LAR.

AVOB-CV-082447M still need to be modified for fire related (Note: the event identifiers conditions. This task is not specifically referenced in the complete. Finding have been either renamed or removed from the FPRA models. Those that remain in the model (with different names) are documented in Reference [2]).

Transition Phase: MINIMAL Post-Transition Phase: MINIMAL PRM-B11-01 MODEL all operator actions and operator Complete work Procedures, modification detail, PRA RAI 1bb (Finding) influences in accordance with the HRA element operations review, and detailed PRA RAI 1dd of this Standard. HRA model development had not been completed prior to the submittal of the LAR. These items are being addressed as part of the FPRA efforts.

As screening values were used for all HFEs in the LAR, it is expected that use of explicitly developed HEPs for some HFEs, scoping HEPs for others, and screening values for the remainder, may reduce the CDF to some degree. Whatever the impact on CDF is, that impact is not expected to substantially impact the results of the LAR.

Page 169 of 227

(Note: the event identifiers specifically referenced in the Finding have been either renamed or removed from the FPRA models. Those that remain in the model (with different names) are documented in Reference [2]).

Transition Phase: MINIMAL Post-Transition Phase: MINIMAL HRA-B3-01 COMPLETE the definition of the HFEs identified The impact of loss of all A detailed review of possible PRA RAI 1cc (Finding) in SRs HRA-B1 and HRA-B2 by specifying the redundant/diverse operator responses to postulated following, taking into account the context instrumentation on HEPs has instrumentation failures was presented by the fire scenarios in the Fire PRA: been modeled by OR-ing the conducted. In this review, three to (a) accident sequence specific timing of cues, instrumentation logic with its four licensed plant operators and time window for successful completion associated HEP. Thus, in cases participated. The exercise was where total instrument failure (by conducted in the plant simulator b) accident sequence specific procedural hardware fault or fire) occurs area (the simulator was not guidance (e.g., AOPs, EOPs) (including the failure of the only utilized, but the panels, c) the availability of cues or other indications for instrument available), the HEP is instruments, and gauges of the detection and evaluation errors appropriately failed. However, simulated control room were d) the specific high-level tasks (e.g., train-level) the failure impact of partial energized).

required to achieve the goal of the response. instrumentation on an HEP has As a result of the knowledge not yet been implemented. There gained from the exercises with are cases in the model where plant operations personnel and multiple instruments provide cues the work done to identify multiple to the operators to perform indicators for various plant status actions. Operator actions based parameters, the impact on the on false indication have not been LAR results and conclusions due considered. In addition, HFEs to the effects of loss of modeled using screening values instrumentation on operator (for some of the fire response actions is expected to be small.

actions identified; e.g., ACP-DGOT-B5B-DG, FPS-PMOE- Transition Phase: MINIMAL START-L, ACP-PMOE-383-12A, Post-Transition Phase: MINIMAL ACP-PMOE-383-11A, etc.) and those fire response actions that will be identified as the fire scenario refinement continues have not yet accounted for the scenario context including timing, procedural guidance, instrumentation, task complexity, etc. Also, HRA Calculator evaluation sheets cannot be located for PCP-PMOF-P-50X-LOC and EDG-PMOE-PORT-PUMP, and AFW-AVOA-CV-2010-D, SWS-AVOA-CV-0823-Page 170 of 227

26, and SWS-AVOB-CV-082447M still need to be modified for fire related conditions.

SF-A1-01 (Finding) For those physical analysis units within the Fire The current seismic fire The Seismic-Fire Interaction PRA RAI 1ll PRA global analysis boundary, interactions analysis relies on the Report (Report 0247-07-0005.05)

(a) LOOK for fire ignition source scenarios that IPEEE study. The report needs to has been revised to address the might arise as the result of an earthquake that demonstrate that the scope of findings from the Fire PRA Peer would be unique from those postulated during that work meets the objectives of Review. The revision used the the general analysis of each physical analysis the Standard and that plant Fire PRA Standard, ASME/ANS unit, and (b) PROVIDE a qualitative assessment changes since the work was RASa2009 Part 4, as a of the potential risk significance of any unique performed do not compromise the guideline. The Seismic-Fire fire ignition source scenarios identified conclusions. Interaction Report now addresses each supporting requirement for HLR-SF-A, and the report itself fulfills the requirement of HLR-SF-B. Each of the supporting requirements is addressed individually in the report. No significant deficiencies were noted during the reviews and revision of the report.

Additionally, since the Standard only requires a qualitative analysis, there is no impact on the quantified results in the Fire PRA model.

Transition Phase: NONE Post-Transition Phase: NONE Page 171 of 227

References:

EA-PSA-FPIE-FIRE-12-04 Rev. 0, Palisades Full Power Internal Events and Fire PRA Model NB-PSA-HR, Vol. 1, Rev. 4, Human Reliability Analysis Notebook Volume 1 (Post Initiator Operator Actions)

Report 0247-07-0005.03 Rev. 1, Palisades Fire Probabilistic Risk Assessment Model Development Report PLP-RPT-12-00134 Rev. 0, Validation of Appendix R Non-Safe Shutdown Cable Routing to Support the Fire PRA Report 0247-07-0005.06 Rev. 1, Palisades Fire Probabilistic Risk Assessment Fire Scenario Development Report Report 0247-07-0005.05 Rev. 1, Palisades Fire Probabilistic Risk Assessment Seismic/Fire Interaction Report Report 0247-07-0005.01 Rev. 1, Palisades Fire Probabilistic Risk Assessment Fire Risk Quantification and Summary NRC Request PRA RAI 05 The ASME/ANS PRA Standard and RG 1.200, Rev. 2, provide guidance for the technical adequacy, including supporting requirements and peer reviews. Section 2.4.3.3 of NFPA 805 states that the PSA approach, methods, and data shall be acceptable to the AHJ. RG 1.205, provides guidance for use in complying with the requirements promulgated for risk-informed, performance-based fire protection programs that meet the requirements of 10 CFR 50.48(c) and the referenced 2001 Edition of NFPA 805. RG 1.205 identifies NUREG/CR-6850 as documenting a methodology for conducting a FPRA and endorses, with exceptions and clarifications, NEI 04-02, Rev. 2, as providing methods acceptable to the NRC for adopting a fire protection program consistent with NFPA-805. The following additional information is requested in order for the staff to complete its review:

Section 4.4.3 of the Fire Ignition Frequency Development Report states that the approach followed to apportion transient and cable fire frequencies associated with hot work deviates from NUREG/CR-6850 given that a separate hot work influence factor is developed. A review of LAR Attachment H, and the Fire Ignition Frequency Development Report indicates that FAQ 12-0064, Hot Work/Transient Fire Frequency Influence Factors (ADAMS Accession No. ML12346A488, closure memo) is not referenced. Clarify whether the guidance from NUREG/CR-6850 or FAQ 12-0064 was used, including:

a) Clarification that the methodology used to calculate hot work and transient fire frequencies applies influencing factors using NUREG/CR-6850 guidance or FAQ 12-0064 guidance.

b) Identification and description of administrative controls used to reduce transient fire frequency, e.g., using low (1) influencing factors per FAQ 12-0064, as well as justification of the reduction assumed for these controls.

Page 172 of 227

c) Additional justification for apportioning turbine building fire frequency bins for general transients and transient activities (e.g., hot work) to the diesel-generator-related PAUs (i.e., 5, 6, 7 and 8).

ENO Response a) The approach followed to apportion transient and cable fire frequencies associated with hot work was developed in accordance with FAQ 12-0064. As such, it deviates from NUREG/CR-6850 in that a separate hot work influence factor was developed. FAQ 12-0064 is not referenced in section 4.4.3 of the Scenario Development Report, as the FAQ was not closed at the time the report section was written.

b) The ranking factor numerical values were assigned to each location to reflect relative weighting values within each applicable frequency bin location set per FAQ 12-0064. Administrative controls were not used to reduce transient fire frequency, e.g., using low (1) influencing factors.

c) Table 4-9, in section 4.4.3 of the Fire Ignition Frequency Development Report, indicates that diesel-generator-related PAUs 05, 06, 07 and 08 are assigned to the turbine building transient fire standard generic location group. The statement referenced is in error. These PAUs are assigned to the plant wide standard generic location group and treated as such in the transient fire ignition frequency calculation. Accordingly, plant wide fire frequency bins for general transients and transient activities are apportioned to diesel-generator-related PAUs 05, 06, 07 and 08.

NRC Request PRA RAI 07 The ASME/ANS PRA Standard and RG 1.200, Rev. 2, provide guidance for the technical adequacy, including supporting requirements and peer reviews. Section 2.4.3.3 of NFPA 805 states that the PSA approach, methods, and data shall be acceptable to the AHJ. RG 1.205, provides guidance for use in complying with the requirements promulgated for risk-informed, performance-based fire protection programs that meet the requirements of 10 CFR 50.48(c) and the referenced 2001 Edition of NFPA 805. RG 1.205 identifies NUREG/CR-6850 as documenting a methodology for conducting a FPRA and endorses, with exceptions and clarifications, NEI 04-02, Rev. 2, as providing methods acceptable to the NRC for adopting a fire protection program consistent with NFPA-805. The following additional information is requested in order for the staff to complete its review:

Section 9.6 indicates that scenarios in PAU 02 (CSR) for both fixed and transient ignition sources are defined according to a grid system related to the arrangement of sprinkler heads within the PAU such that only those targets within the grid coordinate(s) associated with the physical location of the ignition source are modeled as failed given successful suppression of the fire by the wet-pipe suppression system. However, the Page 173 of 227

impacted grid coordinates associated with fixed ignition sources do not appear to appropriately bound the respective ZOIs of each source as determined by the generic fire modeling treatments. In addition, transient and hot work fire scenarios appear to only be postulated as affecting a single grid coordinate, resulting in the potential exclusion of locations (e.g., in between two or more grid coordinates) within the PAU where CCDPs are highest (i.e., pinch points). Provide further justification that the current method appropriately bounds fire risk within this PAU for both fixed and transient ignition sources, or provide the results of a sensitivity study (i.e., CDF, LERF, CDF, and LERF) that appropriately considers all pinch points associated with transient and hot work fire scenarios as well as those fixed sources for which the respective ZOI is not encompassed by the grid coordinates associated with the physical location of the ignition source alone.

ENO Response The method used to model transient and fixed ignition sources in the Cable Spreading Room (CSR) will be updated in the RAI Response Fire PRA Model to more appropriately bound the fire risk in this PAU. The updated approach is described in detail below.

Transient ignition source scenarios in the CSR are being updated to ensure transient and hot work fire scenarios consider multiple grid coordinate impacts. This is accomplished by postulating a transient at each intersection of the lines defining the grid coordinate system. The corresponding target damage set is comprised of the targets located within the grid coordinates that meet at that intersection. The ignition frequencies for general transients and transients due to hotwork are equally distributed to each scenario. For example, a scenario postulated at the intersection of the lines separating columns A, B and rows 1, 2 has a target set of grid coordinates A1, B1, A2, and B2. As a result of postulating a scenario at each intersection, the target sets for the scenarios overlap.

Fixed ignition source scenarios in the CSR are also being updated to ensure that the impacted grid coordinates appropriately bound the respective ZOIs of each source as determined by the Generic Fire Modeling Treatments. In addition to updating the impacted grids for the existing scenarios that correspond to the NUREG/CR-6850 98th percentile heat release rate (HRR), additional scenarios are being modeled for lower HRRs. These scenarios have fewer impacted grid coordinates due to the smaller ZOI dimensions. The severity factors for the various points selected from the HRR distribution curve are included in the calculation of the non-suppression probabilities (NSPs) in order to credit manual suppression preventing damage to a nearby target.

These updated scenarios will be in the base case results of the RAI Response Fire PRA Model. Therefore, a sensitivity study to determine the possible risk increase reflecting the changes described will not be necessary.

Page 174 of 227

NRC Request PRA RAI 08 The ASME/ANS PRA Standard and RG 1.200, Rev. 2, provide guidance for the technical adequacy, including supporting requirements and peer reviews. Section 2.4.3.3 of NFPA 805 states that the PSA approach, methods, and data shall be acceptable to the AHJ. RG 1.205, provides guidance for use in complying with the requirements promulgated for risk-informed, performance-based fire protection programs that meet the requirements of 10 CFR 50.48(c) and the referenced 2001 Edition of NFPA 805. RG 1.205 identifies NUREG/CR-6850 as documenting a methodology for conducting a FPRA and endorses, with exceptions and clarifications, NEI 04-02, Rev. 2, as providing methods acceptable to the NRC for adopting a fire protection program consistent with NFPA-805. The following additional information is requested in order for the staff to complete its review:

Describe the process used to identify and locate cable fires scenarios due to either self-ignition or hot work within a PAU, and discuss how frequencies are apportioned to such scenarios. In addition, clarify the special consideration applied to cable fire scenarios in the turbine building as noted in Section 7.2 of the Fire Scenario Development Report.

ENO Response In the fire PRA model that supported the LAR, scenarios for self-ignited cable fires and cable fires due to hot work were developed by locating scenarios at cable tray physical pinch points. In the case of the turbine building, room based transient scenarios were developed. This assured a bounding treatment for turbine building areas with numerous cable trays.

Subsequent to the LAR submittal, the NRC issued additional guidance for treatment of fire scenarios for self-ignited cable fires and cable fires due to hot work. This guidance is documented as FAQ 13-0005 (draft). The purpose of the additional guidance is to provide a more realistic approach for development of cable fire scenarios. The development of scenarios for self-ignited cable fires and cable fires due to hot work in the PNP fire PRA has been re-performed in accordance with FAQ 13-0005. The new scenarios have been integrated into the fire PRA model and included in the risk estimates. The numerical effect will be reflected in the base quantification of the of the RAI Response Fire PRA model.

NRC Request PRA RAI 09 The ASME/ANS PRA Standard and RG 1.200, Rev. 2, provide guidance for the technical adequacy, including supporting requirements and peer reviews. Section 2.4.3.3 of NFPA 805 states that the PSA approach, methods, and data shall be acceptable to the AHJ. RG 1.205, provides guidance for use in complying with the requirements promulgated for risk-informed, performance-based fire protection programs that meet the requirements of 10 CFR 50.48(c) and the referenced 2001 Page 175 of 227

Edition of NFPA 805. RG 1.205 identifies NUREG/CR-6850 as documenting a methodology for conducting a FPRA and endorses, with exceptions and clarifications, NEI 04-02, Rev. 2, as providing methods acceptable to the NRC for adopting a fire protection program consistent with NFPA-805. The following additional information is requested in order for the staff to complete its review:

Describe how heating ventilation and air conditioning (HVAC) modeling was performed to support the FPRA and whether HVAC cable tracing and fire modeling were performed to support this modeling. Confirm that additional operator actions are not needed for crediting HVAC. Heat load calculations performed for the IEPRA do not account for the additional heat load from fires. Confirm that heat loads from fires do not fail additional equipment in rooms including those that do not credit HVAC.

ENO Response Requirements for room cooling were evaluated in Attachment 8 of the Event Trees and Success Criteria Notebook [1]. The conclusions of these evaluations were applied to both the FPIE PRA and the fire PRA.

It was determined that the emergency diesel generator (EDG) rooms require active cooling systems to support equipment operation and the containment structure requires active cooling (containment spray or containment air coolers) to preserve containment integrity. These active cooling systems are included in the fire PRA logic model, and cable tracing was performed for required components. Additional operator actions are not required for crediting HVAC to support equipment operation or containment integrity.

For fires in the Control Room, the status of Control Room HVAC is evaluated when calculating abandonment times. Control Room HVAC was not explicitly treated in the fire PRA model that supported the LAR submittal. Subsequent to the LAR submittal, a fault tree supported by cable tracing has been developed for Control Room HVAC and will be evaluated for control room abandonment scenarios in the RAI Response Fire PRA Model. Additional operator actions are not required for crediting HVAC for control room abandonment scenarios, however, temporary cooling of the control room using portable fans is credited for non-abandonment scenarios for defense-in-depth.

A summary of the basis for excluding, or including, HVAC is provided below for physical analysis units (PAU) that could potentially have a room cooling dependency:

Containment (PAU 14)

PRA components within containment are designed to operate in a post high energy line break (HELB), loss of coolant accident (LOCA) or once through cooling (OTC) environment. Fire initiating event sequences, from fires within or external to containment, that lead to a LOCA would require heat removal from the containment spray or containment air cooler systems to ensure the integrity of the containment structure. The additional heat loads introduced by fires are negligible in comparison to the environmental heat loads and therefore will have no impact on HVAC requirements.

Page 176 of 227

East and West Engineered Safeguards Rooms (PAU 10 and PAU 28)

The analysis performed, as summarized in [1], for these spaces considered loss of ventilation, concurrent with a large break LOCA and no intervention by operators to align alternate cooling or restore installed ventilation. Under these conditions, it was shown that the most limiting component would not reach its environmentally qualified (EQ) rated temperature for at least 4.5 days. Reaching this limit does not imply immediate failure of the component, but a temperature above which the component has not been evaluated for continuous long term operation.

Short term heat addition to the space from a fire would be quickly dissipated by firefighting activities. Opening doors and spraying water by the fire brigade will provide additional ventilation and heat removal not credited in the room heat-up analysis.

Therefore, a fire in either the east or west safeguards rooms will not negatively impact components outside the fires zone of influence (ZOI) for the duration of the PRA mission time.

Auxiliary Feedwater (AFW) Pump Room (PAU 24)

The room heat-up analysis for this area, summarized in Reference [1], assumes the turbine driven AFW pump is in service, and a failed steam trap is leaking in the room. It does not credit operator action to restore primary ventilation or establish alternate means of cooling. The room is passively ventilated through a ceiling vent pipe that is always open.

Under these conditions, it was demonstrated that it would require in excess of 4 days to reach the design temperature of the rooms limiting component. Therefore, fires occurring in other areas of the plant, that result in failure of AFW room ventilation, will not adversely impact the operation of PRA components in this area for the duration of the PRA mission time and no additional operator actions are required. Fires occurring within the PAU are assumed to fail all equipment in the room [3]. The impacts of heat loads imparted to the area by fires are bounded by this treatment.

1-C Switchgear and 1-D Switchgear (PAU 4 and PAU 3)

The room heat-up analyses for these areas, summarized in Reference [1], assumes loss of ventilation and does not credit operator action to restore primary ventilation or establish alternate means of cooling. The analyses considered heat loads from room lighting as well as local panels and cables; in addition the 1-C room temperature profile assumed the EDGs in the adjacent spaces were in service. Under these conditions, the results demonstrate that neither switchgear room will reach the rooms HVAC design temperature prior to 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> (EQ temperature limits were not established for these rooms as they are not considered harsh areas). The door between the electrical equipment room and the 1-D switchgear is normally open, but would close automatically for a fire on either side of the door.

Reaching the HVAC design temperature does not imply failure of any room components, but a temperature above which the rooms HVAC system was designed to maintain under the most limiting outdoor ambient temperature. (HVAC design Page 177 of 227

temperatures are occasionally used as conservative acceptance criterion at PNP for room heat-up analysis if specific limiting component temperatures are unavailable for a given area.)

Short term heat addition to the space from a fire would be quickly dissipated by installed sprinkler systems and firefighting activities. Opening doors and spraying water by the fire brigade will provide additional ventilation and heat removal not credited in the room heat-up analysis.

Therefore, it is concluded that a fire in either the 1-C or 1-D switchgear rooms will not negatively impact components outside the fires zone of influence (ZOI) for the duration of the PRA mission time.

EDG Room 1-1 and EDG Room 1-2 (PAU 5 and PAU 6)

PRA components within each EDG room are assumed to fail on loss of ventilation if the generator within the space is required to operate. The EDG ventilation system is modeled in the fire PRA and cable tracing was performed. Fires occurring within the PAU are assumed to fail all equipment in the room [3]. The impact of heat loads imparted to the area by fires is bounded by this treatment.

Cable Spreading Room (PAU 2)

The room heat-up analyses for these areas, summarized in Reference [1], assumes loss of ventilation and does not credit operator action to restore primary ventilation or establish alternate means of cooling. The analyses established room heat loads with ventilation isolated based on test data collected during the Systematic Evaluation Program (SEP) evaluation. The resulting temperature profile demonstrated a peak temperature of 122 °F at 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br />.

A subsequent EEQ evaluation was performed for all components in the room that assumed the temperature was constant at 122 °F for the duration of the transient. This analysis concluded that all PRA components in the room would have reasonable assurance of operability under these conditions.

Fire scenarios in the cable spreading room are developed using a grid approach that is based on the location of automatic suppression. In scenarios where automatic suppression fails to initiate, all equipment in the room is assumed to be failed. When automatic suppression is successful, fire damage is confined to the affected grid(s). In this case, the additional heat loads imparted by the fire would be quickly dissipated by installed sprinkler systems and firefighting activities. Opening doors and spraying water by the fire brigade will provide additional ventilation and heat removal not credited in the room heat-up analysis.

Therefore, it is concluded that a fire in the Cable Spreading room will fail the PRA components within the fires zone of influence (ZOI), but will not negatively impact components outside the ZOI for the duration of the PRA mission time.

Page 178 of 227

Battery Room #1 and Battery Room #2 (PAU 12 and PAU 11)

The room heat-up analyses for these areas, summarized in Reference [1], assumes loss of ventilation and does not credit operator action to restore primary ventilation or establish alternate means of cooling. During the first four hours (design life of the batteries) the heat load is based on lighting and battery discharge, after which the heat load is based only on lighting. Under these conditions, the analysis concluded that the battery rooms reach the HVAC design temperature of 104°F approximately 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br /> into the transient. Reaching the HVAC design temperature does not imply failure of any room components, but a temperature above which the rooms HVAC system was designed to maintain under the most limiting outdoor ambient temperature.

For station blackout conditions the batteries have completely discharged by this time and loss of ventilation and room heat-up has no impact on battery operation.

For fire-induced control room abandonment scenarios (or other scenarios) in which the shunt trip pushbutton is actuated and the primary control station is transferred to the EC-150 panel, restoration of battery room ventilation is not required. When the shunt trip pushbutton is actuated, the battery discharge rate decreases (EC-150 panel is the only load) and the total heat rejected to the room simply occurs over a longer duration than the 4 hour4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> duration assumed in the analysis. This results in lower actual room temperatures at any point in time than those predicted in the calculation, and results in approximately the same final room temperature if the battery is completely discharged by the end of the analysis. Therefore, analysis final temperature at 75 hours8.680556e-4 days <br />0.0208 hours <br />1.240079e-4 weeks <br />2.85375e-5 months <br /> represents a conservative upper bound of battery room air temperature over the 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> PRA mission time.

Barriers between these spaces and adjacent rooms are considered two-hour fire barriers [2] which would provide enough thermal inertia to prevent any greater than a negligible impact on heating of adjacent spaces over the duration of the fire.

Fires occurring within these PAUs are assumed to fail all equipment in the room [3].The impact of heat loads imparted to the area by fires is bounded by this treatment.

Component Cooling Water Room (PAU 16)

The room heat-up analyses for the Component Cooling Water room elevations, summarized in Reference [1], assumes room heat loads based on a large break LOCA, loss of ventilation, and does not credit operator action to restore primary ventilation or establish alternate means of cooling. Under these conditions, the room temperature profiles are enveloped by the design EEQ profile, which demonstrates that PRA equipment in the PAU would function for 30 days; well beyond the PRA mission time.

Short term heat addition to the space from a fire would be quickly dissipated by firefighting activities. Opening doors and spraying water by the fire brigade will provide additional ventilation and heat removal not credited in the room heat-up analysis.

Therefore, it is concluded that a fire in the Component Cooling Water room will fail the PRA components within the fires zone of influence (ZOI), but will not negatively impact Page 179 of 227

components outside the ZOI or adjacent spaces for the duration of the PRA mission time.

Control Room (PAU 1)

Control room HVAC is not required for fires outside the control room as discussed in the response to PRA RAI 28h. The critical impact of heat loads imparted by fires within the control room is the effect on operator abandonment time. The control room abandonment analysis includes evaluation of these heat loads in the CFAST model and uses the CAFTA model and cable tracing to evaluate the availability of control room HVAC. In addition, short term heat addition to the space from a fire would be quickly dissipated by firefighting activities and opening doors.

Therefore, it is concluded that a fire in the control room will fail the PRA components within the fires zone of influence (ZOI), but will not negatively impact components outside the ZOI.

References:

Palisades Probabilistic Safety Assessment Notebook NB-PSA-ETSC Rev. 3, Event Trees and Success Criteria 0247-07-0005-07, Revision 1, Palisades Nuclear Plant Fire Probabilistic Risk Assessment Multi-Compartment Analysis, ERIN Document, November 2012 0247-07-0005-06 Revision 1, Fire Scenario Development Report, ERIN Document, November 2012 NRC Request PRA RAI 12 ASME/ANS-RA-Sa-2009, Standard for Level 1/Large Early Release Frequency Probabilistic Risk Assessment for Nuclear Power Plant Applications, as clarified in RG 1.200, Rev. 2, describes when changes to a PRA should be characterized as a PRA upgrade, e.g., new common cause failure (CCF) or HRA methods. Identify any such changes made to the IEPRA or FPRA subsequent to the most recent full-scope peer review. In addition, address the following:

a) If any changes are characterized as a PRA upgrade, indicate if a focused-scope peer review was performed for these changes consistent with the guidance in ASME/ANSRA-Sa-2009, and describe any findings and their resolution.

b) If a focused-scope peer review has not been performed for changes characterized as a PRA upgrade, describe what actions will be implemented to comply with the ASME/ANS standard.

Page 180 of 227

ENO Response Changes to the FPIE and FPRA models since the performance of the peer reviews are documented in the responses to PRA RAI 02 and PRA RAI 27. This includes changes that were made up to the time of the LAR submittal, and additional changes that are being made as part of the RAI Response Fire PRA model development activities. The three phases of the FPRA peer reviews covered all aspects of the FPRA model development. Besides the implementation of HEP screening values, subsequent changes to the FPRA model were performed consistent with the methods reviewed as part of the peer reviews or were performed using the approaches prescribed in the responses to other RAIs. As the use of the prior methods is not an upgrade, and the updates to the methodology in response to the RAIs are incremental in nature and are being reviewed as part of the PNP NFPA 805 LAR application, subsequent peer reviews are not warranted. Subsequent to the last Fire PRA peer review, the HRA methodology has been revised to utilize screening and scoping HEP values consistent with NUREG-1921 methods (refer to PRA RAI responses 01cc, 01dd, and 01ee). This screening and scoping method results in a conservative treatment and is also not considered an upgrade, but merely represents an update to the HEP analysis using established screening methods. The detailed HEP methodology was reviewed by the peer review and has not been changed. As such, a focused scope review of the HEP analysis is also not warranted.

NRC Request PRA RAI 13 Section 4.1 of the Plant Partitioning and Fire Ignition Frequency Development Report indicates that fire ignition frequencies were updated to reflect conclusions drawn in , Description of Treatment for Hot Work Fires to NEI Letter dated October 7, 2011, Recent Fire PRA Methods Review Panel Decision: Frequencies for Cable Fires Initiated by Welding and Cutting (ADAMS Accession Nos. ML113130465 and ML113130468). However, a review of Tables 4-1 and 4-4 indicates that the Bin 24 frequency was not altered. Additionally, the Bin 26 frequency is inconsistent with Supplement 1 to NUREG/CR-6850. Provide a re-analysis using values consistent with applicable guidance.

ENO Response A review was performed of Attachment 1, Description of Treatment for Hot Work Fires to NEI Letter dated October 7, 2011, Recent Fire PRA Methods Review Panel Decision: Frequencies for Cable Fires Initiated by Welding and Cutting and Supplement 1 to NUREG/CR-6850. The generic frequencies for bins 24 and 26 have been updated accordingly to match the respective sources as reflected in Table 1.

These bins have subsequently been Bayesian updated per a review of the plant specific fire events. The numerical effects of these updates will be reflected in the RAI Response Fire PRA Model.

Page 181 of 227

TABLE 1 GENERIC FIRE FREQUENCIES EPRI GENERIC FREQUENCY BIN# IGNITION SOURCE SOURCE PER REACTOR YEAR NEI Letter dated 24 Transient due to Cut & Weld 4.69E-03 Oct. 7 2011 Supplement 1 to 26 Ventilation Subsystems 6.12E-03 NUREG/CR-6850 NRC Request PRA RAI 16 The ASME/ANS PRA Standard and RG 1.200, Rev. 2, provide guidance for the technical adequacy, including supporting requirements and peer reviews. Section 2.4.3.3 of NFPA 805 states that the PSA approach, methods, and data shall be acceptable to the AHJ. RG 1.205, provides guidance for use in complying with the requirements promulgated for risk-informed, performance-based fire protection programs that meet the requirements of 10 CFR 50.48(c) and the referenced 2001 Edition of NFPA 805. RG 1.205 identifies NUREG/CR-6850 as documenting a methodology for conducting a FPRA and endorses, with exceptions and clarifications, NEI 04-02, Rev. 2, as providing methods acceptable to the NRC for adopting a fire protection program consistent with NFPA-805. The following additional information is requested in order for the staff to complete its review:

Section 2.1.1 and severity factors listed in Table 8-2 of the Fire Scenario Development Report appear to indicate that scenarios involving switchgear were assigned frequencies apportioned by the number of switchgear cubicles (i.e., horizontal divisions within a vertical section) in lieu of the number of vertical sections per NUREG/CR-6850 and FAQ 06-0016, Clarifying Guidance for Counting Electrical Panels and Cabinets, (ADAMS Accession No. ML072700475, closure memo). Describe the approach utilized to apportion frequencies to switchgear and other electrical cabinets. Provide a sensitivity analysis that aligns with guidance in NUREG/CR-6850 and FAQ 06-0016 if the approach does not.

ENO Response The approach utilized to apportion frequencies to switchgear and other electrical cabinets is consistent with the guidance in NUREG.CR-6850 as clarified in FAQ 06-0016. FAQ 06-0016 guidance for counting switchgears, load centers, unit substations and motor control centers is based on the number of vertical sections. The guidance for counting other electrical cabinets is based on the externally apparent vertical sections.

In the Fire Scenario Development Report [1], scenario naming convention examples Page 182 of 227

provided in Section 2.1.1 and severity factors in Table 8-2 are specific to switchgears.

In the case of the switchgears, switchgear breaker cubicles are vertical sections and are treated as such in accordance with the guidance in NUREG/CR-6850 as clarified in FAQ 06-0016.

Since the approach utilized to apportion frequencies to switchgear and other electrical cabinets is consistent with the guidance in NUREG.CR-6850 as clarified in FAQ 06-0016, a sensitivity analysis is not necessary.

Reference

[1] 0247-07-0005.06, Rev. 1, Fire Scenario Development Report, November 2012.

NRC Request PRA RAI 17 Not all elements of MCR fire modeling appear consistent with NUREG/CR-6850 guidance including treatment of electrical sub-enclosures. In particular, per NUREG/CR-6850 guidance the Main Control Board (MCB) is intended to be that subset of cabinets from which primary control and monitoring is performed. Clarify how MCR modeling was performed. Provide:

a) A sensitivity study that updates the MCB fire ignition frequency to treat the whole back panel of sub-enclosure 1 as electrical cabinets as opposed to part of the MCB.

b) Discuss the approach used to develop and apportion frequencies to MCB fire scenarios.

c) Clarify how panel fires within the MCR that are not associated with the MCB are treated by the FPRA.

d) Explain and justify how MCB or panel fire propagation in the MCR and MCR sub-enclosures was modeled, including how fire propagation was considered between panels with open backs opposite to each other, within close proximity, and connected by cable bundles.

e) Explain and justify the basis for transient fire placement including how locations next to open-back panels and inside sub-enclosures were considered.

f) Clarify any credit taken for ionization smoke detectors mentioned in Section 14.1.1 of the Fire Scenario Development report.

g) Describe how cable conduits mentioned in Section 14.1 of the Fire Scenario Development report are addressed in MCR fire scenarios.

h) Discuss how HVAC was treated in the MCR, including both fire-induced and random failures.

Page 183 of 227

ENO Response a) The complete back panel of sub-enclosure 1 includes panels C11-6, C11-3, C11-4, C11-5, C12-5, C12-6, C12-7, C12-8, C13-4, and C13-5. The ignition frequency calculation has been updated to count these 10 panels as electrical cabinets instead of main control board (MCB). The updated frequencies will be included in the base case results of the RAI Response Fire PRA model. Utilizing this approach for the base case which more closely matches the NUREG/CR-6850 guidance means that a specific sensitivity study is not needed.

b) Main Control Board (MCB) at PNP consists of three (3) distinctly separated sets of panels:

Section 1 - the bench-board-type cabinets (C01, C02, C03, and C08),

Section 2 - the main horseshoe which is made up of the front panels of sub-enclosure 1 (C11-1, C11-2, C12-1, C12-2, C12-3, C12-4, C13-1, C13-2, and C13-3),

Section 3 - the front panels of sub-enclosure 2 (C04-1 and C06-1).

The plant-wide MCB frequency is apportioned to each of the three MCB sections based on the length of the MCB sections.

NUREG/CR-6850 Appendix L guidance was used to analyze MCB fires.

Consistent with the guidance, the full MCB section frequency was applied to the postulated fire scenarios within the applicable section.

c) Non-abandonment scenarios involving electrical panel fires not associated with the MCB were analyzed using the Generic Fire Modeling Treatments in the same manner as electrical panels outside the control room. The possibility that fires in these electrical panels lead to control room abandonment was also considered using the results of the control room habitability calculation.

d) The treatment of MCB and electrical panel fire propagation is being updated to credit the ionization smoke detectors and to consider propagation to non-adjacent panels consistent with the guidance provided in NUREG/CR-6850 Appendix S.

There are no open back panels outside of the sub-enclosures. For non-abandonment scenarios, fire propagation to an adjacent electrical panel (i.e., non-MCB panels) within the sub-enclosure is considered to occur almost immediately as the location of the fire within the initiating panel is not known. However, given the presence of in-cabinet ionization smoke detectors, fire propagation was not postulated until after five (5) minutes (see part f of this RAI response). For non-adjacent panels within the sub-enclosure, the time to fire propagation was then calculated.

Fire propagation within the MCB for non-abandonment scenarios was modeled as described in the response to PRA RAI 17 b). Fire propagation in the MCB for abandonment scenarios was modeled consistent with the guidance provided in NUREG/CR-6850 Appendix S (i.e., fire spread was modeled at 15 min.).

Page 184 of 227

e) The placement of transient fires in the control room is being expanded to include open, wall, and corner transients leading to abandonment. Transient fires are also postulated in areas for which they would damage cable conduits. Transient fires are not postulated inside sub-enclosures due to the small space available inside the subenclosure. The physical configuration and sensitive nature of the MCB sub-enclosures prevents the realistic postulation of the co-existence of transient combustibles and transient ignition sources. Additionally, procedural guidance exists that prohibits the placement of transient combustibles within 3 feet of safety related cables [2].

f) The fire scenarios for the subenclosures are being updated to credit the in-cabinet ionization smoke detectors. To provide a more realistic treatment, consistent with the guidance in NUREG/CR-6850 Appendix P, the detectors are credited to provide an additional 5 minutes to the available suppression time for fires originating in these sub-enclosures.

g) Cable conduits in the control room are included in the target sets within the postulated ignition source zone of influence for the MCR fire scenarios.

h) As a further refinement in the RAI Response Fire PRA model, the treatment of control room HVAC will be updated to consider both random and fire induced failures when calculating the likelihood of control room abandonment. A logic model of the control room HVAC system is quantified for each fire scenario leading to abandonment with the fire induced failures of the ignition source applied. The resulting failure probability of the HVAC system is applied as a conditional probability that the non-suppression probabilities (NSPs) calculated for failure of the HVAC system for that ignition source should be applied. The complementary value is applied as a conditional probability that the HVAC system remains functional. The likelihood that the HVAC system is put into purge mode (which involves a different abandonment time) is modeled as a subsequent conditional probability. The three prorated HVAC NSPs are summed and applied to the abandonment fire scenario as a single NSP.

References:

[1] NUREG/CR 6850 (2005), EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities Volume 2 Detailed Methodology, EPRI 1011989 Final Report, NUREG/CR-6850, Nuclear Regulatory Commission, Rockville, MD, September, 2005.

[2] Palisades Procedure FPIP-7 Rev 21. Fire Prevention Activities.

NRC Request PRA RAI 18 The ASME/ANS PRA Standard and RG 1.200, Rev. 2, provide guidance for the technical adequacy, including supporting requirements and peer reviews. Section 2.4.3.3 of NFPA 805 states that the PSA approach, methods, and data shall be Page 185 of 227

acceptable to the AHJ. RG 1.205, provides guidance for use in complying with the requirements promulgated for risk-informed, performance-based fire protection programs that meet the requirements of 10 CFR 50.48(c) and the referenced 2001 Edition of NFPA 805. RG 1.205 identifies NUREG/CR-6850 as documenting a methodology for conducting a FPRA and endorses, with exceptions and clarifications, NEI 04-02, Rev. 2, as providing methods acceptable to the NRC for adopting a fire protection program consistent with NFPA-805. The following additional information is requested in order for the staff to complete its review:

Sections 15.1 and 15.2 of the Fire Scenario Development report indicate that secondary fires from HEAF events in PAUs 03 and 04 are assumed to be bounded by the detailed modeling performed to determine the fire impact associated with the Bin 8 HRR of an electrical cabinet with a single unqualified cable bundle obtained from Table E-4 of NUREG/CR-6850. Clarify whether or not this approach was applied to fire scenarios in other PAUs (e.g., PAU 02). Provide the risk results (i.e., CDF, LERF, CDF and LERF) from a sensitivity study that utilizes guidance provided in Appendix M of NUREG/CR-6850 to develop HEAF fire scenarios and damage sets that considers damage at time zero.

ENO Response The approach used for HEAF events in PAUs 03 and 04 were unique to the fire scenarios postulated for the 1D and 1C switchgear HEAF events. The fire scenarios postulated for HEAF events in 1D and 1C will be updated using calculations performed using FDS. HEAF events in the 1D and 1C switchgear were analyzed explicitly in the FDS models built for PAUs 03 and 04 using the guidance provided in Appendix M of NUREG/CR-6850.

These updated scenarios will be included in the base case results of the RAI Response Fire PRA model. Therefore, a sensitivity study to determine the possible risk increase reflecting the changes described is not necessary.

NRC Request PRA RAI 19 According to NUREG/CR-6850, where water from fire suppression efforts will likely enter a potentially vulnerable component (e.g., a panel with unsealed penetrations or an unshielded electrical motor), it is appropriate to include that component in the fire scenario damage set. Explain and justify the level of assessment of the impact of suppression system activation on component operation using guidance in Section 11.5.1.2 of NUREG/CR-6850. Include identification of walkdowns performed to identify unsealed penetrations or unshielded electrical equipment.

ENO Response The guidance in NUREG/CR-6850 Section 11.5.1.2 identifies three items that should be considered to determine the additional impact fire suppression activities may have on the fire scenario definitions.

Page 186 of 227

Potentially vulnerable components to fire suppression water spray (e.g. a panel with unsealed penetrations or unshielded electrical motors)

Inadequate drainage of an area that could lead to flooding of components Unsealed floor penetrations that could allow water from firefighting activities to migrate These items were evaluated, with the support of plant walk downs, in the Fire Suppression Activities Effect on Nuclear Safety Performance Criteria [1] report prepared to support the transition to NFPA-805. This assessment was reviewed to identify the impacts of suppression activities on components in the fire PRA and the applicable fire PRA scenarios will be updated in the RAI Response Fire PRA model to include component failures because of fire suppression activities consistent with NUREG/CR-6850 guidance.

References:

PLP-RPT-12-00100 Rev. 1, Fire Suppression Activities Effect on Nuclear Safety Performance Criteria NRC Request PRA RAI 20 Section 4.1 of the Plant Partitioning and Fire Ignition Frequency Development report discusses that the FPRA was quantified using fire ignition frequencies obtained from Supplement 1 to NUREG/CR-6850. Explain whether the LAR Attachment W risk results are based on these frequencies, and provide the results (i.e., CDF, LERF, CDF and LERF) of a sensitivity study that utilizes NUREG/CR-6850-based ignition frequencies as indicated in Footnote 10 of Supplement 1 to NUREG/CR-6850. If the use of NUREG/CR-6850-based ignition frequencies produces risk results that exceed RG 1.174, Rev. 2, risk acceptance guidelines, discuss the fire protection, or related, measures that can be taken to provide additional DID to offset this risk.

ENO Response The LAR Attachment W risk results are based on fire ignition frequencies obtained from Supplement 1 to NUREG/CR-6850. Without consideration of the risk reduction afforded by the NFPA 805 modifications in other hazard groups, the calculated risk increase due to the use of NUREG/CR-6850-based ignition frequencies indicates the potential to exceed RG 1.174 risk acceptance guidelines.

Therefore, applicable compensatory measures have been implemented in accordance with the PNP Fire Protection Program.

The RAI Response Fire PRA model results will be based on fire ignition frequencies obtained from Supplement 1 to NUREG/CR-6850.

Page 187 of 227

Based on the RAI Response Fire PRA model, the use of NUREG/CR-6850-based ignition frequencies are not expected to exceed RG 1.174.

NRC Request PRA RAI 23 The ASME/ANS PRA Standard and RG 1.200, Rev. 2, provide guidance for the technical adequacy, including supporting requirements and peer reviews. Section 2.4.3.3 of NFPA 805 states that the PSA approach, methods, and data shall be acceptable to the AHJ. RG 1.205, provides guidance for use in complying with the requirements promulgated for risk-informed, performance-based fire protection programs that meet the requirements of 10 CFR 50.48(c) and the referenced 2001 Edition of NFPA 805. RG 1.205 identifies NUREG/CR-6850 as documenting a methodology for conducting a FPRA and endorses, with exceptions and clarifications, NEI 04-02, Rev. 2, as providing methods acceptable to the NRC for adopting a fire protection program consistent with NFPA-805. The following additional information is requested in order for the staff to complete its review:

Attachment W of the LAR provides the CDF and LERF for the VFDRs and the additional risk of RAs for each of the fire areas in which the LAR describes (but not in detail) how CDF and LERF or the additional risk of RAs were calculated. Describe the method(s) used to determine the changes in risk reported in the LAR Attachment W, Table W-2. The description should include:

a) A detailed definition of both the post-transition and compliant plants used to calculate the reported changes in risk and additional risk of RAs.

b) A description of how the reported changes in risk and the additional risk of RAs were calculated. Include in this description a discussion of PRA modeling mechanisms used to determine the reported changes in risk (e.g. altering the probabilities of basic events and modeled recovery actions). Clarify whether FAQ 08-0054, Demonstrating Compliance with Chapter 4 of NFPA 805, (ADAMS Accession No. ML110140183, closure memo) guidance was used.

c) A discussion of any exceptions to normal modeling mechanisms discussed in b.

(above), including those cases for which the PRA model lacks sufficient resolution to model the VFDR or those that utilize surrogate basic events or HFEs to estimate/bound the change in risk in lieu of manipulating components or actions directly associated with the VFDR.

d) A statement whether all PRA manipulations performed effectively bound CDF and LERF.

e) A description of any modeling manipulations that use data or methods not included in the FPRA peer review.

f) A separate description specific to how the CDF and LERF and additional risk of RAs were calculated for the MCR (PAU 01). Include in the description how this Page 188 of 227

calculation was performed for loss-of-control scenarios and for control room abandonment (CRA) scenarios (i.e., alternate shutdown).

g) An explanation of the following anomalies identified in LAR Attachment W, Table W-2:

i. Values for CDF and LERF exceed CDF and LERF values for some PAUs (e.g., 02, 03, 15, 16, 21, 23 and 32).

ii. Values for LERF exceed CDF values for some PAUs (e.g., 23, 24, 40 and 41).

iii. Values for additional risk of RAs appear to be too low for several fire areas (e.g., Fire Areas 01, 03, 13, 16 and 23), particularly in light of the respective CDF and LERF values reported, the use of HEP screening values and the number of VFDRs with RAs for some of the impacted PAUs.

iv. CDF and LERF values reported for PAU 15 in LAR Attachment W, Table W-2, differ from those reported in LAR Attachment C, Table B-3.

v. CDF and LERF values for Fire Area 56 are reported as epsilon.

h) A discussion of how RAs were quantified and treated in the fire risk evaluations (FREs) for fire scenarios inside and outside the MCR. Describe whether there are any previously approved RAs.

ENO Response LAR Attachment W provides for each fire area the CDF, LERF, CDF and LERF for VFDRs, and the additional risk of recovery actions. Additional detail describing how CDF, LERF, and the additional risk of RAs were calculated in the LAR is provided below, including a description of the methods used to determine the reported changes in risk. Also, proposed changes to the methods for determining (1) CDF and LERF for VFDRs and (2) the additional risk of recovery actions for the RAI response fire PRA model quantifications are discussed.

The PNP Fire PRA Model is currently being revised to incorporate RAI refinements. The aggregate numerical effect of the model refinements will be reflected in the base quantification of the RAI Response Fire PRA model. The RAI Response Fire PRA Model will represent the collection of changes identified in individual RAI responses that could impact the CDF, LERF and delta calculations that support the LAR submittal. The RAI Response Fire PRA Model will be quantified to provide updated Attachment W values as required for the LAR submittal once all issues that impact quantification have been identified and resolved via the RAI process. The final results are expected to provide reasonable assurance that the total risk remains within Region II of RG 1.174, Rev 2.

Given the two evaluation models, the LAR Fire PRA Model and the RAI Response Fire PRA model, question responses are provided under two headings as appropriate:

Page 189 of 227

Additional Detail Regarding LAR Attachment W Methods and Results Responses under this heading pertain to the LAR Fire PRA Model and the results as provided in Attachment W of LAR.

Changes Proposed for RAI Response Fire PRA Model Responses under this heading pertain to the RAI Response Fire PRA model as proposed in this and possibly future RAI responses.

This presentation is not included when differentiation in the response is not necessary.

a) The definitions of the plant models used to calculate the reported changes in risk and the additional risk of recovery actions in LAR Attachment W may differ from the commonly understood definitions. To indicate this, the term altered is added as a prefix to the plant model identifiers. This provides clarification and distinction between what was done for the LAR (i.e., altered) and what is proposed for the RAI Response Fire PRA Model.

Additional Detail Regarding LAR Attachment W Methods and Results The altered post-transition plant, altered compliant plant and altered recovered plant used to calculate the reported changes in risk (CDF and LERF for VFDRs) and the additional risk of recovery actions are defined below. Since it is used in the calculation of the reported additional risk of recovery actions, the altered recovered plant is also defined below. Additionally, the post-transition plant (used in the calculation of the additional risk of recovery actions in the LAR) is defined in section Changes Proposed for RAI Response Fire PRA Model to avoid duplication.

The difference between the altered post-transition plant and the altered compliant plant is that the altered compliant plant protects unresolved VFDR components from fire damage. These two plant models are used to identify the delta risks due to protection of unresolved VFDRs - in a context that credits both classes of modifications (those required for compliance and modifications beyond compliance) and does not credit recovery actions. The results demonstrate the risks due to unresolved VFDRs are acceptable.

The difference between the post-transition plant and the altered recovered plant is that the altered recovered plant credits fully reliable recovery actions. These two plant models are used to identify the additional risks of recovery actions - in a context that credits both classes of modifications and does not protect unresolved VFDR components from fire damage.

The altered post-transition plant model credits modifications required for compliance and modifications beyond compliance, feasible primary control station operator actions (including control room operator actions), feasible ex-primary control station operator actions not associated with recovering VFDRs, and does not credit protection of individual VFDR components beyond those protected by Page 190 of 227

modifications required for compliance. See column 2 in Table 23.1 below.

The altered compliant plant model credits modifications required for compliance and modifications beyond compliance, feasible primary control station operator actions (including control room operator actions), feasible ex-primary control station operator actions not associated with recovering VFDRs, and does credit protection of individual VFDR components beyond those protected by modifications required for compliance. See column 3 in Table 23.1 below.

The altered recovered plant model credits modifications required for compliance and modifications beyond compliance, feasible primary control station operator actions (including control room operator actions), feasible ex-primary control station operator actions not associated with recovering VFDRs, feasible ex-primary control station operator actions that are associated with recovering VFDRs (i.e., recovery actions) at 100% reliability, and does not credit protection of individual VFDR components beyond those protected by modifications required for compliance. See column 4 in Table 23.1 below.

These plant models are summarized in Table 23.1 below.

Table PRA RAI 23.1: LAR Fire PRA Model Delta Risk Case Definitions Altered Altered Altered Credit Taken in the LAR Fire PRA Model Post-Compliant Recovered For: Transition Plant Plant Plant Modifications Required for Compliance Modifications Beyond Compliance Feasible Primary Control Station Operator Actions (Including Control Room Operator Actions)

Feasible Ex-Primary Control Station Operator Actions not Associated with Recovering VFDRS (not RAs)

Feasible Ex-Primary Control Stations Operator Actions that are Associated with Recovering 1 VFDRS (RAs)

Protection of individual VFDR Components Beyond those Protected by Modifications Required for Compliance (Unresolved VFDRs) 1 Local operator actions credited as 100% reliable.

Changes Proposed for RAI Response Fire PRA Model The plant definitions in the LAR were chosen to separate out the delta risk due to unresolved VFDRs from the impact of the modifications beyond compliance. This also had the effect of skewing the CDF and LERF for unresolved VFDRs.

Page 191 of 227

To provide changes in risk (CDF and LERF for VFDRs) and additional risk of recovery actions in a format potentially more amenable for review, the post-transition plant, compliant plant and recovered plant used to calculate the reported changes in risk (CDF and LERF for VFDRs) and the additional risk of recovery actions for the RAI response fire PRA model will be defined as follows below.

The difference between the post-transition plant and the compliant plant is that the compliant plant protects VFDR components from fire damage but does not credit modifications beyond compliance. These two plant models are used to identify the delta risks due to protection of unresolved VFDRs - in a context that credits only modifications required for compliance (and not modifications beyond compliance).

Note that the act of protecting VFDRs that otherwise have associated recovery actions changes the context of these actions, since fire-induced VFDR failures are eliminated and recovery is then from random failures only.

The difference between the post-transition plant and the recovered plant is that the recovered plant protects VFDR components from fire damage. These two plant models are used to identify the additional risks of recovery actions - in a context that credits both classes of modifications and protection of unresolved VFDR components from fire damage.

The post-transition plant model will credit modifications required for compliance and modifications beyond compliance, feasible primary control station operator actions (including control room operator actions), feasible ex-primary control station operator actions not associated with recovering VFDRs, feasible ex-primary control station operator actions that are associated with recovering VFDRs (i.e., recovery actions), and will not credit protection of individual VFDR components beyond those protected by modifications required for compliance.

The compliant plant model will credit modifications required for compliance, feasible primary control station operator actions (including control room operator actions), feasible ex-primary control station operator actions not associated with recovering VFDRs, feasible ex-primary control station operator actions associated with recovering VFDRs (i.e., not actually recovery actions since fire-induced failures are eliminated and recovery is from random failure only), and will credit protection of individual VFDR components beyond those protected by modifications required for compliance.

The recovered plant model will credit modifications required for compliance, feasible primary control station operator actions (including control room operator actions), feasible ex-primary control station operator actions not associated with recovering VFDRs, feasible ex-primary control station operator actions that are associated with recovering VFDRs (i.e., recovery actions), and does credit protection of individual VFDR components beyond those protected by modifications required for compliance.

These plant models are summarized in Table 23.2 below.

Page 192 of 227

Table PRA RAI 23.2: RAI Response Fire PRA Model Delta Risk Case Definitions Post-Credit Taken in the RAI Response Fire PRA Compliant Recovered Transition Model For: Plant Plant Plant Modifications Required for Compliance Modifications Beyond Compliance Feasible Primary Control Station Operator Actions (Including Control Room Operator Actions)

Feasible Ex-Primary Control Station Operator Actions not Associated with Recovering VFDRS (not RAs)

Feasible Ex-Primary Control Stations Operator Actions that are Associated with Recovering 1 1 VFDRS (RAs)

Protection of individual VFDR Components Beyond those Protected by Modifications Required for Compliance (Unresolved VFDRs) 1 Since fire-induced VFDR failures are eliminated; recovery is from random failures only.

b) Additional Detail Regarding LAR Attachment W Methods and Results The changes in risk (CDF and LERF for VFDRs) for each fire area were calculated as the altered post-transition plant minus the altered compliant plant risk. The additional risks of recovery actions were calculated as the post-transition plant minus the altered recovered plant risk, for each fire area that required recovery actions.

Fire areas that required recovery actions were determined as follows. Credit for recovery actions was collectively removed from the post-transition plant (i.e.,

recovery action HEPs were set to 1.0). Fire areas with an increase in CDF of >1E-8 or LERF of >1E-9 were considered fire areas that require recovery actions (i.e.,

the recovery actions are important enough to be required). For these areas, the additional risk of recovery actions was computed by assuming the recovery actions are completely reliable (HEPs set to 0).

PRA modeling mechanisms used to determine the reported changes in delta risks are described as follows. When a component is assumed to fail as a result of the fire, the failure probability for the PRA basic events chosen to represent the affected component function are increased to 1.0 (or logically true) (i.e., the component fails with 100% certainty).

Page 193 of 227

Credit for protection of individual VFDR components beyond those protected by modifications required for compliance means protecting from fire damage only.

That is, the VFDR component and its cables are protected from the effects of fire, but allowed to fail in the model at the baseline random failure probability.

The compliant case, as defined in Regulatory Guide 1.205 Section 2.24, postulates hypothetical modifications such as moving or protecting cables associated with VFDRs such that the VFDRs no longer exist in the fire PRA model. A compliant case for a given fire PRA scenario is one in which those fire-affected components and associated cables within the scenario, that if protected would result in at least one success path required to meet NFPA 805 nuclear safety performance criteria (NSPC), remain at the random failure probability, i.e.,

the components are unaffected by the fire.

FAQ 08-0054 guidance was used as described above. Note that the FAQ submittal number was 08-0054 but the NRC closure memo for the FAQ was listed as 07-0054. Therefore, the FAQ is referred to in the LAR as 07-0054 to be consistent with the closure memo.

Changes Proposed for RAI Response Fire PRA Model The changes in risk (CDF and LERF for VFDRs) for each fire area will be calculated as the post-transition plant minus the compliant plant risk, as defined above. The additional risks of recovery actions for each fire area will be calculated directly (not using a surrogate approach, see RAI PRA 25) as the post-transition plant minus the recovered plant risk.

Modifications beyond compliance (MBC) will be defined as the subset of PRA modifications (LAR Attachment S, Table S-2, Items S2-1 through S2-15) considered as not required for compliance, as indicated below. The modifications beyond compliance help ensure RG 1.174 is met.

Table PRA RAI 23.3: PRA Modifications Beyond Compliance (MBC)

Mod Mod Title S2-1 Additional High Head Auxiliary Feedwater Pump S2-8 Insulate Emergency Diesel Generator Exhaust S2-12 Manual Bypass of the ADV Solenoids S2-14 Replace cables associated with CV-0910, CV-0911 and CV-0940 With the changes identified in part a) above for the recovered plant, the additional risks of recovery actions are determined as indicated in the FAQ 08-0054 guidance. The post-transition plant models the recovery actions explicitly in the fire PRA, with appropriate human error probabilities, to calculate the CDF and LERF.

The recovered plant is obtained by assuming that the plant was modified to remove the VFDRs. Subtracting these risks gives the CDF and LERF Page 194 of 227

associated with performing the action compared to maintaining the otherwise fire-failed equipment free of fire damage.

c) Exceptions include treatments of systems and components not modeled in the fire PRA, including various HVAC systems for room cooling and other non-consequential components, and systems and components that are modeled in the fire PRA but not in the same way as deterministically assumed.

For example, condensate and heater drain pump trips. VFDRs exist involving the loss of the ability to trip condensate pumps and heater drain pumps. Condensate pumps are important for maintaining secondary side cooling via main feed and low pressure feed (i.e., without main feedwater pumps). Failure of condensate pumps can also drain the condensate storage tank to the hotwell, given additional condensate makeup/reject valve failures. Heater drain pumps are not credited in the fire PRA since heater drain pumps are not able to provide or enhance secondary side cooling via main feed or low pressure feed. Condensate and heater drain pump trip functions are not modeled in the fire PRA. The pumps are not credibly capable of steam generator overfill at pressure due to pump head capacity and suction source limitations.

The NFPA 805 function to trip condensate and heater drain pumps is related to preserving controlled decay heat removal. The fire PRA model does not consider failure to trip either condensate or heater drain pumps as impacting decay heat removal. The NFPA 805 function to trip condensate and heater drain pumps for feedwater isolation to provide controlled decay heat removal has no impact in the fire PRA.

A modification to provide a passive, direct AFW suction source from demineralized water storage tank T-939 has been credited in the fire PRA post-transition plant.

The modification reduces the delta risks with respect to T-2 draindown to epsilon.

Exceptions to the modeling mechanisms, including those cases for which the PRA model lacks sufficient resolution to model the VFDR or those that utilize surrogate basic events or HFEs to estimate/bound the change in risk in lieu of manipulating components or actions directly associated with the VFDR, are discussed further in EA-PSA-805-FRE-10-03.

d) PRA manipulations performed for determining CDF and LERF for VFDRs and the additional risk of recovery actions effectively bound CDF and LERF.

e) The use of NUREG-1921 methods for screening, scoping and detailed HEP values constitutes data and methods not included in the fire PRA peer review. However, these data and methods are considered acceptable for use.

f) Additional Detail Regarding LAR Attachment W Methods and Results The CDF and LERF and additional risk of RAs calculated for the control room (PAU 01) were performed in the same way as for non-control room PAUs. No loss-of-control scenarios that lead to control room abandonment were quantified for the Page 195 of 227

LAR. Loss-of-control scenarios that do not lead to control room abandonment had CDF and LERF and additional risk of RAs calculated in the same way as for non-control room PAUs. Control room abandonment scenarios are based on habitability (smoke opacity, ambient temperature) and were quantified directly via CCDP calculation and not estimated. These control room abandonment scenarios were not treated differently than non-abandonment scenarios for CDF and LERF and additional risk of RAs.

Changes Proposed for RAI Response Fire PRA Model The CDF and LERF and additional risk of RAs for control room fire scenarios that do not lead to control room abandonment will be performed in the same way as for non-control room PAUs. Control room abandonment will be based on habitability (smoke opacity, ambient temperature) and scenarios will be treated as follows. Actions to operate equipment (e.g., P-8B) from the alternate shutdown panel will not be treated as recovery actions since the primary control has been transferred to the alternate shutdown panel. In addition, VFDRs representing failures that result in transfer and operation of equipment from the alternate shutdown panel will not be protected in the recovered plant model.

g)

i. Additional Detail Regarding LAR Attachment W Methods and Results The apparently counter-intuitive results of CDF exceeding CDF and/or LERF exceeding LERF is an artifact of the method used to separate out the delta risk due to unresolved VFDRs from the impact of the modifications beyond compliance. The method utilized the plant definitions described above. The results can be understood given these plant definitions.

The CDF and LERF values reported in the LAR are based on the post-transition plant as defined in Table 23.2. The CDF and LERF values reported in the LAR are based on the altered post-transition plant minus the altered compliant plant as defined in Table 23.1.

The changes made to create the altered post-transition plant were to eliminate credit for operator actions associated with recovering VFDRs (recovery actions). This increases the risk results for the altered post-transition plant to varying degrees depending on the importance of the recovery actions. Note importance in this sense is the risk increase when the action is assumed failed, which is not necessarily of similar magnitude as the risk decrease when the action is assumed 100% reliable (i.e., RAW vs.

RRW).

The altered compliant plant is different from the altered post-transition plant only in that all components associated with VFDRs are protected from fire effects. This lowers the risk of the altered compliant plant relative to the altered post-transition plant, hence all delta risks are positive. Protecting additional components is beneficial or neutral and not negative, as expected.

Page 196 of 227

The counter-intuitive results of CDF exceeding CDF and/or LERF exceeding LERF can occur for several reasons. First, there is not a 1 to 1 relationship in all cases between operator actions considered associated with resolution of VFDRs and the specific variant VFDR components (for example, the action representing restoration from loadshed). In these cases the recovery of the VFDR component occurs at a higher functional level in the fire PRA, resulting in the recovery (or failure when not credited) of a greater number of components or functions in the fire PRA than in deterministic evaluations. This can result in a greater risk increase due to failure of a VFDR recovery action than a risk decrease due to the protection of the associated VFDR components, since a greater number of component/functions may be impacted by the recovery action than by the associated VFDR component. Failure of the recovery action associated with the VFDR component in the altered post-transition case increases risk above the post-transition case. Compliant case protection of VFDR components may reduce compliant case risk significantly enough to result in a delta (altered post-transition - altered compliant) that is greater than the post-transition case. To avoid potential masking of insights, delta risk was not limited to the value of the post-transition plant risk.

Second, protection of the VFDR components helps only in scenarios that fire-impact these components, but failure of the operator action occurs on a PAU basis and impacts all scenarios for the PAU. Even though the operator action will be more important to scenarios in which the associated components are fire-failed, the actions can have appreciable importance all other scenarios due to recovery from random failures. The summation of the impact of failing the operator action in all scenarios can exceed the impact of protecting the component in the subset of scenarios that fire-impact the component. This can again result in a greater risk increase due to failure of a VFDR recovery action than a risk decrease due to the protection of the associated VFDR components for a given PAU.

Changes Proposed for RAI Response Fire PRA Model The method to separate out the delta risk due to unresolved VFDRs from the impact of the modifications beyond compliance will not be used. The method described in part b) above will utilize the plant definitions in part a) above.

This treatment is expected to reduce or eliminate the counter-intuitive results.

Any remaining counter-intuitive results will be explained in detail.

ii. If the benefit of protecting a VFDR or set of VFDRs preferentially impacts the LERF results, delta-LERF can exceed delta-CDF.

Given:

CDF1 - CDF2 = CDF LERF1 - LERF2 = LERF Page 197 of 227

The constraints:

LERF1 CDF1 LERF2 CDF2 CDF2 CDF1 LERF2 LERF1 Do not imply:

LERF CDF For a numerical example:

CDF = 1E-5, CDF2 = 8E-6 CDF = 2E-6 LERF1 = 5E-6, LERF2 = 2E-6 LERF = 3E-6 But:

LERF > CDF For a physical example with respect to FA-23 (turbine building), VFDRs exist for the main steam isolation valve bypass valves MO-0501 (VFDR-0293) and MO-0510 (VFDR-0294). MO-0501 is required to isolate the steam generator E-50B. MO-0510 is required to isolate the steam generator E-50A.

Due to plant asymmetries, only E-50A can supply turbine-driven AFW pump P-8B. Both valves provide containment isolation if steam generator tube rupture occurs following core damage. However, only MO-0510 provides SG isolation in support of P-8B operation. The VFDRs exist against both MO-0501 and MO-0510 isolation to support the deterministic decay heat removal NSPC for NFPA 805.

Protecting these valves from fire damage results in a moderate improvement in CDF. The CDF improvement is moderate (and not larger) since there are many fire-induced failures that result in loss of P-8B (absent recovery actions) for important scenarios in the turbine building. Only a moderate fraction of these involve failures (spurious open) of MO-0510. None involve spurious open of MO-0501 since this valve does not impact P-8B operation.

Protecting these valves from fire damage results in a greater improvement in LERF. In contrast to CDF, LERF results in FA-23 are dominated by many scenarios involving fire-induced spurious open failures of either or both MO-0501 and MO-0510. These containment isolation valves are important in LERF sequences that involve thermally induced or pressure induced consequential steam generator tube rupture. Many of these sequences involve MO-0501 and MO-0510 for LERF only, since other fire-induced Page 198 of 227

failures lead to loss of P-8B and core damage and since MO-0501 does not impact P-8B. This fact results in a proportionally greater impact for protecting these valves in LERF than in CDF. This is reflected in the counter-intuitive delta risk results.

iii. Additional Detail Regarding LAR Attachment W Methods and Results The additional risks of recovery actions were calculated as the post-transition plant minus the altered recovered plant risk, for each fire area that required recovery actions, as described in part b) above.

The post-transition plant is different from the altered recovered plant only in that recovery actions are set to 100% reliable HEPs (HEP set to 0 in the altered recovered plant). This lowers the risk of the recovered plant relative to the post-transition plant, hence all delta risks are positive. Crediting recovery actions as 100% reliable is beneficial or neutral and not negative, as expected. Note the risk increase when the action is assumed failed, which is not necessarily of similar importance as the risk decrease when the action is assumed 100% reliable (i.e., RAW vs. RRW).

The counter-intuitive results of low additional risk of recovery actions given relatively large CDF and/or LERF occurs due to the skewing of delta risks that results from part g), i) above.

Changes Proposed for RAI Response Fire PRA Model The method to separate out the delta risk due to unresolved VFDRs from the impact of the modifications beyond compliance will not be used. The method described in part b) above will utilize the plant definitions in part a) above.

iv. Table B-3 of LAR Attachment C reported PAU 15 CDF / LERF as 2.5E-10

/ 2.5E-11. These values are incorrect (and were transposed from PAU 14).

LAR Attachment W reports the correct results for PAU 15 as 1.8E-06 / 1.1E-06.

v. Fire area FA-56 (Diesel Fire Pump Fuel Oil Day Tank Room) is located at the 590 ft elevation and contains the two diesel fire pump fuel oil day tanks, associated fuel oil level gauges and control switches. Diesel fire pump P-9B day tank T-24 and diesel fire pump P-41 day tank T-40 are elevated about 8 ft off the floor to provide gravity feed of fuel oil to the associated diesel fire pump.

The room is a concrete block structure located against the south wall of the intake structure. The day tank room structure is on its own foundation and consists of three block walls with steel reinforced alternate concrete filled core blocks. The fourth wall is the exterior wall of the intake structure which is 24 inches of reinforced concrete. Penetrations through the intake structure wall are rated at 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> or more. Access to the tank room is through a locked exterior door on the south side of the structure. The foundation Page 199 of 227

structure is reinforced concrete and is designed to contain the contents of the two fuel oil tanks. A small sump area is formed within the concrete base. A sealed penetration in the intake structure wall provides passage of fuel oil lines and level circuitry.

The total core damage frequency (CDF) and large early release frequency (LERF) associated with the post-transition plant configuration for this fire area are not quantified. Fires in this area do not result in a reactor trip and no components modeled in the fire PRA exist in this fire area. Therefore, risk and delta risk results are epsilon.

h) Additional Detail Regarding LAR Attachment W Methods and Results By definition, recovery actions are actions taken to recover a deterministic nuclear safety performance criteria and recovery actions occur outside the primary control station location. For fire areas in which control room abandonment is not the deterministic strategy (i.e., for fire areas other than FA-01 (control room) and FA-02 (cable spreading room)), all ex-control room actions that recover NSPCs are treated as recovery actions in the fire risk evaluations.

For scenarios that require control room abandonment, actions taken in the control room prior to abandonment are treated as primary control station actions and not recovery actions, actions taken to transfer control to the primary control station are not treated as recovery actions, and actions taken outside the re-located primary control station (alternate shutdown panel) and recover NSPCs are treated as recovery actions.

The quantification of recovery actions was done per the PNP screening method.

These recovery actions were originally part of the Appendix R licensing basis.

Changes Proposed for RAI response Fire PRA Model The quantification of recovery actions will be done per NUREG-1921.

NRC Request PRA RAI 24 LAR Attachment W, Section W.2.2 states that the total CDF is reasonably estimated to be below 1E-4/year; however, when summing the CDFs for all hazards reported in this section and LAR Attachment W, Table W-2, a CDF that exceeds 1E-4/year is obtained.

Discussion in LAR Attachment W states that a better estimate for the fire CDF, which would result in the total CDF falling below 1E-4/year, is estimated as a factor of 5 to 10 lower than the fire CDF reported in LAR Attachment W, Table W-2 based on review of uncertainties associated with FPRA tasks; however, there appears to be no basis for the aforementioned factor (e.g., F&Os against UNCA1 and UNC-A2). Provide additional information to demonstrate that the total CDF is below 1E-4/year.

Page 200 of 227

ENO Response The updated post-transition plant fire CDF will be reflected in the base quantification of the RAI Response Fire PRA model.

The LAR risk estimates for internal events, internal flood and external events do not include the beneficial impact of the proposed NFPA 805 modifications.

Many NFPA 805 modifications were intentionally designed to provide significant risk reduction with respect to other hazards. For example, the additional auxiliary feedwater pump (LAR modification S2-01) is designed to provide significant risk reductions for internal events, internal flood and external events, in addition to fires. Other examples include the condensate storage tank passive cross-tie (LAR modification S2-10) and the manual bypass of ADV solenoids (LAR modification S2-12).

Based on current working models for internal events and internal flood, the risk benefits of crediting NFPA 805 modifications are expected to result in the following:

Internal events ~1E-5/yr Internal flood ~1E-5/yr These estimates together with the updated post-tranisiton plant fire CDF based on the RAI Response Fire PRA Model are expected to demonstrate total CDF less than 1E-4/yr.

NRC Request PRA RAI 25 The ASME/ANS PRA Standard and RG 1.200, Rev. 2, provide guidance for the technical adequacy, including supporting requirements and peer reviews. Section 2.4.3.3 of NFPA 805 states that the PSA approach, methods, and data shall be acceptable to the AHJ. RG 1.205, provides guidance for use in complying with the requirements promulgated for risk-informed, performance-based fire protection programs that meet the requirements of 10 CFR 50.48(c) and the referenced 2001 Edition of NFPA 805. RG 1.205 identifies NUREG/CR-6850 as documenting a methodology for conducting a FPRA and endorses, with exceptions and clarifications, NEI 04-02, Rev. 2, as providing methods acceptable to the NRC for adopting a fire protection program consistent with NFPA-805. The following additional information is requested in order for the staff to complete its review:

LAR Attachment W, Table W-2, provides the CDF and LERF for each fire area, where the beyond compliance modifications are credited for both the compliant and transition plants. LAR Attachment W, Table W-2 also presents a total risk offset attributable to beyond compliance modifications indicating a significant reduction in risk.

This risk offset is based on a surrogate approach described in Section 6.8 of the Fire Risk Evaluation report (EA-PSA-805- FRE-10-03) and is presented to show that the NFPA 805 transition meets RG 1.174 risk acceptance guidelines. The surrogate Page 201 of 227

approach utilizes a ratio of CCDPs (or conditional large early release probabilities (CLERPs)) with and without selected beyond compliance modifications (i.e., additional feedwater pump and atmospheric steam dump valve nitrogen bypass) obtained from a better than compliant plant in which no components are failed by fire. It is not clear that this surrogate approach produces the same risk reduction effect as calculating the CDF and LERF directly. Provide the CDF and LERF for each fire area that removes credit for beyond compliance modifications directly for the compliant plant (i.e.,

does not use the surrogate approach).

ENO Response The surrogate approach does not produce the same risk reduction effect as calculating the delta risks directly. However, the surrogate approach produces bounding delta risks as compared to calculating delta risks directly. That is, the surrogate approach produces a smaller risk offset attributable to modifications beyond compliance and therefore higher (more positive or less negative) delta risks.

The surrogate approach was used to facilitate the license amendment request process.

The approach supports the conclusion that the post-transition plant has lower risk than the compliant plant and the overall delta risks of transition are negative.

The numerical effect will be reflected in the base quantification of the RAI Response Fire PRA model, which would include CDF and LERF for each fire area that removes credit for beyond compliance modifications directly for the compliant plant (i.e., does not use the surrogate approach).

NRC Request PRA RAI 26 Address the following regarding modifications presented in LAR Attachment S, Tables S-1 and S-2:

a) Some modifications (e.g., S1-1, S2-21 and S2-23) are cited as being modeled in the FPRA but are assigned a risk ranking of N/A. Justify the dismissal of any risk impact of these modifications.

b) Several modifications (e.g., S1-2, S2-26, S2-29, S2-30, S2-31 and S2-34 through S2-42) are cited as not being modeled in the FPRA but appear to result in improvements to existing configurations, indicating that either a less robust configuration is retained in the FPRA or that the subject of the modification is not modeled. Describe how these modifications are treated by the FPRA, discussing whether or not they are either implicitly or explicitly modeled.

c) Some modifications (e.g., S2-2, S2-3, S2-5, S2-6, S2-7 and S2-9 through S2-15) either indicate or suggest the routing of new cabling or the movement of existing cabling. Discuss whether or not the exact locations of new or relocated cabling are known and modeled accordingly. Explain what approach (e.g., exclusion) was used.

Page 202 of 227

d) Modification S1-3 is indicated as not being evaluated in the FPRA. Discuss the implications of fuse failure, and further justify its exclusion from the model.

e) Describe and justify the PRA modeling of Modification S2-1. If a screening value is utilized in lieu of detailed system and HRA modeling, provide a basis that this value appropriately reflects or bounds the equipment- and human-related failures associated with the modification.

ENO Response a) Modification S1-1 provides additional fuses in the control circuits of three pumps.

The modification addresses potential simultaneous faults on power and control cables. The 2400 VAC breaker control power fusing was modified for service water pumps P-7B (breaker 152-103) and P-7C (breaker 152-205) and component cooling water pump P-52B (breaker 152-208).

Modification S2-21 MO-2160, Loss of Remote Shutdown Capability resolves an Information Notice 92-18 issue for motor-operated valve MO-2160. The modification reduces the potential for fire induced motor-operated valve failure from causing non-recoverable loss of charging pump suction from SIRWT. The modification modifies each valves motor-operator circuitry such that torque switch is not disabled by a fire that could also cause the MOV to spuriously operate.

Modification S2-23 resolves coordination issues identified during power supply evaluations. The modification will resolve the identified coordination issues to establish alignment with the fire PRA assumption that circuit protection features (breakers, fuses) are electrically coordinated. The modification will replace breakers52-345 and 52-325 in MCC-3; replaces or supplements overcurrent protection in DC panels 11-1, 11-2, 11A, 21-1, 21-2, and 21A; replaces fuses in panels served by the Instrument AC Bus EY-01; and adjusts breaker/relay settings in breakers 152-201, 152-115 and 152-108.

The risk impact was not dismissed. The risk benefit of these modifications is reflected in the fire PRA risk results. These modifications were listed as risk ranking N/A in the modification discussion based on the determination that the modifications are required to support the transition to NFPA-805 without regard to their risk ranking. The table provides risk ranking for other modifications to support ongoing discussions regarding the benefit of the modifications where cost/benefit could be an issue. Consequently a detailed risk analysis of each the identified modifications was not pursued to support the importance of the modifications. Modification S1-1 is complete. Modifications S2-21 and S2-23 are in process.

b) The modifications cited as examples with the exception of modification S2-26 (discussed separately) are related to compliance issues. These issues involve a range of conditions regarding barriers, seals, penetrations, dampers, etc. which do not meet code requirements. The modifications are intended to bring the components identified into compliance with code requirements. While they are improvements, they restore the components in question to the condition implicitly Page 203 of 227

assumed in the Fire PRA model (compliance). Not being in compliance does not automatically result in the component being considered to be in a failed state.

Establishing compliance for these components reduces concerns regarding their ability to perform their functions and supports the PRA assumptions for the Fire PRA model. None of the compliance issues were explicitly or implicitly treated in the Fire PRA model.

Modification S2-26 will be incorporated into the RAI Response Fire PRA Model.

This modification was not initially included in the Fire PRA model quantified to support the license amendment request (LAR). However, it directly relates to a PRA modeled system. The impact of the modification on the PRA is expected to be minimal as the realignment of power to the alternate charger is from the same source as the primary charger which will introduce a common power failure to the primary and alternate charger. The alternate charger will retain the ability to be powered from the opposite division of ac power. Hence no significant change in risk is expected from the inclusion of the additional source to the alternate charger.

The risk impact of this modification will be included in the base case results of the RAI Response Fire PRA Model.

c) As described below, the PRA model treatment of new cable routing for modifications that require new cable installation is by exclusion.

Modifications S2-5 and S2-11 will install alternate pump controls for primary coolant pumps and charging pumps in the control room. Proposed routing paths for cables to support these modifications have been identified during the design reviews and walkdowns to support the modifications. Currently the PRA model treatment for these modifications is by exclusion. This treatment is based on the current conceptual design specification that would require routing that maintains adequate fire scenario separation to support the intended risk reduction.

Modifications S2-2 and S2-3 considered the addition of new breakers at the power supplies (safeguards bus and startup transformer) to the 2400V buses. Design reviews determined that the location of the breakers and necessary interconnections with existing protective circuits at the safeguards bus and startup transformer would introduce a single point vulnerability into the current design of the 2400V ac system. In addition, subsequent to the LAR submittal it was determined there is no risk benefit to the modifications as currently designed.

Consequently these modifications are considered not feasible so that consideration of new cable routing is not currently an issue. Therefore, Modifications S2-2 and S2-3 are no longer being pursued at PNP.

Modification S2-6 will add control switches at the control room cabinet near the existing pump switches. The only wiring impact will be internal to the cabinet between switches. Therefore there is no new cable routing impact associated with this modification.

Modification S2-9 is the only case in which re-routing of existing cables was considered. During the design review an alternate route for the cables was Page 204 of 227

identified. However, an assessment of the modification benefit by assuming the cables were not in the affected fire area determined that there was not adequate risk benefit to the modification without expanding the modification scope.

Consequently determination of any change to risk in the proposed alternate route is not being analyzed. Therefore, Modification S2-9 is no longer being pursued at PNP.

Modifications S2-14 and S2-15 will replace existing cables with fire rated cables.

The new cables will use the same routing as existing cables. For these modifications the PRA model treatment was based on exclusion.

Modifications S2-7, S2-10, S2-12 and S2-13 do not require additional cables based on the current conceptual design. These modifications employ mechanical changes to component operation that would bypass fire induced cable failures that result in significant risk impacts.

Modification S2-7 conceptual design will provide the ability to manually align a nitrogen supply to the affected valve operators such that alteration to the valve control circuits is not required.

Modification S2-10 installs a passive piping connection between the demineralized water storage tank (T-939) and the condensate storage tank (T-2). The only cable impact is to local indication of tank level.

Modification S2-12 will install a manually controlled piping bypass of the current configuration of solenoid valves controlling the air supply to each atmospheric steam dump valve (ASDV). There are no new cables associated with this modification. The only cabling impact is the wiring to a new emergency lighting unit to support the operator action.

Modification S2-13 will be implemented by changing the operating configuration of the affected valves from air-to-open to air-to-close. There are no wiring changes associated with this modification.

d) Modification S1-3 required installation of fuses in fire pump control circuits to assure that certain cable faults did not result in loss of both control room and local control of the pumps. Prior to the modification local control is lost until operators take local action via operating the breaker at the local panel. Once the breaker is open local control is restored. Detailed circuit analysis revealed that in order to credit multiple fire pumps in the Fire PRA, the control circuitry needed to be modified to prevent a fire from disabling a fire pump by grounding/shorting out the wires between the local panel and the Control Room. Fuses were installed in the electrical circuits fed from the screen house to the Control Room for all three fire pumps. The modification isolates damaged cables from the Control Room to the local fire pump controls and allows fire pumps to start automatically upon suppression system demand.

Modification S1-3 installed fuses to protect automatic control circuits for the fire pumps (P-9A, P-9B and P-41) when cables associated with remote manual Page 205 of 227

controls or indications are damaged by fire. Prior to mod installation, faults on cables from the pumps to switches or indication in the control room could disable the auto-control circuitry which is located local to the pumps. The modification was not evaluated in the fire PRA since it was determined that the modification was necessary to better align with NFPA code requirements without consideration of the risk significance of the issue. Therefore the risk importance of the modification was not a necessary condition to support proceeding with the modification. The modification has been completed and incorporated into the fire PRA by adding the fuses to the Fire PRA model.

e) Modification S2-1 is to provide an additional high head auxiliary feedwater pump that is redundant to the existing auxiliary feedwater pumps. The Fire PRA model includes basic modeling of assumed components necessary for pump operation consistent with the conceptual design. The modeling includes events for pump failure to start or run, failure of valves to align the pump discharge into existing headers and connect the pump to a water supply, failure of the water supply (tank) and the operator action to operate the pump. The design specification is for a diesel driven pump therefore there are no existing external power supply requirements for pump operation post fire. The design also includes a local instrument for pump discharge flow to the steam generators. A scoping human error probability was developed based on the methods of NUREG-1921. A screening value was not used.

A revised Attachment S will be provided when the numerical effect from the base case results of the RAI Response Fire PRA Model is provided.

NRC Request PRA RAI 27 The ASME/ANS PRA Standard and RG 1.200, Rev. 2, provide guidance for the technical adequacy, including supporting requirements and peer reviews. Section 2.4.3.3 of NFPA 805 states that the PSA approach, methods, and data shall be acceptable to the AHJ. RG 1.205, provides guidance for use in complying with the requirements promulgated for risk-informed, performance-based fire protection programs that meet the requirements of 10 CFR 50.48(c) and the referenced 2001 Edition of NFPA 805. RG 1.205 identifies NUREG/CR-6850 as documenting a methodology for conducting a FPRA and endorses, with exceptions and clarifications, NEI 04-02, Rev. 2, as providing methods acceptable to the NRC for adopting a fire protection program consistent with NFPA-805. The following additional information is requested in order for the staff to complete its review:

Describe the revisions of the IEPRA and FPRA models seen by the respective peer review teams, and outline any subsequent revisions that the models have undergone.

Additionally, clarify any differences between the revision of the IEPRA model reviewed by the peer review teams and that used to support the FPRA model upon which the LAR is based. Lastly, identify the revision of the FPRA model that is being used for transition to NFPA-805, and clarify whether or not a different revision of the model is expected to be used following transition to NFPA-805.

Page 206 of 227

ENO Response A timeline of the IEPRA and FPRA models seen by the respective peer review teams is provided in Figure 27-1 below. A summary description of the major changes associated with each of the models is then provided. Additionally, a description of the subsequent changes to the models from the time of the LAR submittal up until the development of the RAI Response Fire PRA model is also provided.

FPIE Model Development PSAR2C Model of Record PSAR3, Release 2b (October 2009 FPIE Peer Review) ->

PSAR3.3.0 Finalization (December 2012 LAR Submittal) ->

PSAR3.3.1 Development (RAI Response) ->

PSAR3.3.1+ Finalization (NFPA-805 Transition) ->

Fire PRA Model Development PSAR3 Phase 1 (January 2010 Phase 1 FPRA Peer Review) ->

PSAR3 Phase 2 (August 2010 Phase 2 FPRA Peer Review) ->

PSAR3 Phase 3 (March 2011 Final FPRA Peer Review) ->

PSAR3.1.0 Finalization (December 2012 LAR Submittal) ->

PSAR3.1.1 Development (RAI Response) ->

PSAR3.1.1+ Finalization (NFPA-805 Transition) ->

Mar-06 Jun-06 Sep-06 Dec-06 Mar-07 Jun-07 Sep-07 Dec-07 Mar-08 Jun-08 Sep-08 Dec-08 Mar-09 Jun-09 Sep-09 Dec-09 Mar-10 Jun-10 Sep-10 Dec-10 Mar-11 Jun-11 Sep-11 Dec-11 Mar-12 Jun-12 Sep-12 Dec-12 Mar-13 Jun-13 Sep-13 Dec-13 Mar-14 Jun-14 Sep-14 Dec-14 PSAR3.1.0 PSAR3.3.0 3 is the major revision id 3 is the major revision id 1 refers to the fire model 3 refers to the full power internal events model 0 is the minor revision id 0 is the minor revision id Figure 27-1: Timeline of FPIE and FPRA Model Development An example of the changes from the PSAR2C model of record until the interim PSAR3 model reviewed by the FPIE peer review team is provided below. This covered the time frame from June 2006 until October 2009.

Updated component random failure data.

Updated common cause data.

Created a LERF model employing the simplified PWROG methodology.

Incorporated NFPA 805 logic changes to address spurious failures (MSOs).

Incorporated plant modification addressing GSI-191changing ESF pump seal cooling to closed-loop, increasing dependence on CCW/SWS.

Updated ATWS and Charging fault tree and event tree logic.

Eliminated conservatisms regarding isolation of steam generators on blowdown.

Added potential for battery depletion during non-SBO sequences.

Added consequential LOCA modeling for all transient initiators (new transfer event trees).

Modified assumptions regarding SG overfill during SGTR leading to stuck open SRV.

Refined Safeguards bus and non-safety related diesel breaker modeling.

Addressed additional flow diversion paths.

Re-evaluated the ISLOCA modeling.

Incorporated new IE data.

Incorporated new HRA data.

Page 207 of 227

The RG-1.200 FPIE peer review was then performed in October of 2009.

The initial fire model development was also ongoing by that time. The base fire model incorporates the same CAFTA fault trees as the FPIE and Flood models, but only utilizes the Transient with Main Condenser Available sequences. The activities that were completed prior to the Phase 1 fire PRA peer review are listed below. This encompasses the time frame up until January 2010.

Completed initial version of the Fire PRA model based on the PSAR3, Release 2b PRA model.

Completed initial Fire PRA model development and documentation.

Plant Partitioning and Ignition Frequency Development Report.

Model Development Report.

MSO Report.

Scenario Development Report.

Seismic/Fire Interactions Report.

Quantification and Summary Report.

Performed miscellaneous updates to address FPIE F&Os.

System notebook documentation Data notebook documentation Event tree and success criteria documentation The RG-1.200 Phase 1 Fire PRA peer review was then performed in January 2010.

Following the January 2010 peer review, activities ensued to address both FPIE and FPRA F&Os. Examples of activities that were completed prior to the Phase 2 fire PRA peer review are listed below. This encompasses the time frame from February 2010 until August 2010.

Updated initiating events documentation.

Updated systems analysis documentation.

Updated data notebook documentation.

Created safe and stable notebook (newly developed).

Updated event tree and success criteria documentation.

Updated pre-initiator HRA methodology.

Incorporated multi-compartment analysis results.

Refined fire scenario development to incorporate fire growth and propagation to address miscellaneous FSS F&Os.

Updated plant partitioning and ignition frequency document to address PP and IGN F&Os.

The RG-1.200 Phase 2 Fire PRA peer review was then performed in August 2010.

Following the August 2010 peer review, activities again ensued to address both FPIE and FPRA F&Os. Examples of activities that were completed prior to the final fire PRA Page 208 of 227

peer review are listed below. This encompasses the time frame from September 2010 until March 2011.

Updated initiating events documentation.

Updated systems analysis documentation.

Updated room heat up analyses to support success criteria.

Updated various modeling logic and data to address MSOs per findings HRA-A2-01, HRA-A3-01.

Completed the exposed structural steel analysis.

Updated the multi-compartment analysis.

Further refined fire scenario development to incorporate fire growth and propagation to address miscellaneous FSS F&Os.

Incorporated scenario development based on detailed fire modeling of the 1C switchgear room.

The RG-1.200 Final Fire PRA peer review was then performed in March 2011.

The model changes made between the final Fire PRA peer review and the freeze date of September 2012 that was implemented for the LAR submittal are provided in the response to PRA RAI 02, and are not repeated here. This covers the time frame between April 2011 and September 2012.

Additional changes to the FPIE model have also been ongoing that will not impact the FPRA model. A summary of those activities is provided below. This encompasses the time frame up until the current time (4th quarter of 2013).

New internal flood IEs (based on latest EPRI data)

New LOCA IEs (NUREG-1829)

New VSLOCA Event Tree headings New MSLB IEs (full power internal event modeling only, which has no impact on the fire PRA).

Maintenance and spurious sprinkler operation induced flood IEs (in support of internal flooding analyses which has no impact on the fire PRA).

Associated Documentation In addition, other changes to the FPIE model have also been ongoing that will impact the FPRA model. Examples of those activities are provided below. This encompasses the time frame from October 2012 up until the current time (4th quarter of 2013).

Revised the PCP seal LOCA event tree headings and event tree linkage rules to eliminate illogical cutsets that were indicating a coincident loss of offsite power, subsequent failure of emergency power sources, and operator fail to trip the primary coolant pumps could lead to a PCP seal LOCA.

Flow diversion logic in the LPSI and Service Water fault trees was corrected to eliminate illogical cutsets.

Added a fault tree to model availability of the non-safety related containment air cooler VHX-4 for purposes of preventing a containment high pressure signal in Page 209 of 227

very small break LOCA sequences Revised fault tree logic for inter and intra cable faults based on the latest interim guidance Added a control room HVAC fault tree to address system availability for each control room abandonment fire scenario.

Finally, additional changes have also been made to the Fire PRA model to address F&Os that were not fully closed at the time of the LAR submittal and to address various RAIs that resulted from the NRC audit that occurred in June of 2013. A summary of those activities is provided below. This encompasses the time frame from October 2012 up until the current time (4th quarter of 2013).

Integrate results of supplemental cable analysis work to address PRA RAIs 01a, 01c, and 01d.

Incorporate revised HFE values to address PRA RAI 01dd and 01ee.

Revise treatment of MCR fire scenarios and abandonment scenarios to address PRA RAIs 01l, and 17 as well as FM RAIs 01c, 01d, and 01o.

Update the treatment of sensitive electronics in adjacent cabinets to address PRA RAI 01o.

Update the treatment of manual detection and improved the basis for the times applied in scenario development to address PRA RAI 01r.

Update the postulated elevation of a transient fire in select areas to address PRA RAI 01s.

Revise the catastrophic turbine-generator frequency to address PRA RAI 01w.

Update the multi-compartment analysis to address PRA RAI 01x, 01y, and 01z.

Update the distribution of miscellaneous hydrogen fires to address PRA RAI 01aa.

Update the plant partitioning and ignition frequency development to address PRA RAI 01hh, and 13.

Update the treatment of transient fire placement at pinch points to address PRA RAI 06.

Update the treatment of the cable spreading room fire modeling to address PRA RAI 07.

Update the treatment of cable tray scenarios to address PRA RAI 08.

Revise the HEAF secondary fire development based on revised detailed fire modeling results to address PRA RAI 18.

Update selected scenarios to include component failures because of fire suppression activities to address PRA RAI 19.

Revise treatment of cable damage time in the MathCad sheets used for non-suppression probability development to address FM RAI 01p and 02b as described in PRA 01g.

Update the scenario development to account for the modification to the critical heat flux for targets immersed in a thermal plume to address FM RAI 01e.

Update the scenario development to account for the increased HRR and ZOI of fires that involve multiple burning items to address FM RAI 01f, 01g, 01k, and 01l.

Page 210 of 227

The resolutions of these items will be incorporated into the RAI Response Fire PRA Model (PSAR 3.1.1). The numerical effect of these changes will be reflected in the base quantification of the RAI Response Fire PRA model. Given that PRA models are living analyses, updated versions of the full power internal event (PSAR3.3.1) and fire PRA (PSAR 3.1.1) models may be used for transition to NFPA-805 and following transition to NFPA-805.

References:

[1] ERIN 0247-07-0005-03, Revision 1, Palisades Nuclear Plant Fire Probabilistic Risk Assessment Model, November 2012.

[2] EA-PSA-FPIE-FIRE-12-04, Palisades Full Power Internal Events and Fire PRA Model, Revision 0, November 2012.

NRC Request PRA RAI 28 Contrary to RG 1.200, Rev. 2, it is not clear that all the peer review F&Os were resolved to bring the findings into alignment with CC-II (Met) or justified why a lesser CC was acceptable. Clarify the following dispositions to IEPRA F&Os identified in LAR Attachment U that have the potential to impact the FPRA results:

a) F&O HR-A2-01 against HR-A2:

The pre-initiator methodology documented in NB-PSA-HR (Vol. 2) utilizes scoping values lower than those recommended by NUREG-1792, Good Practices for Implementing Human Reliability Analysis (HRA), for both independent and joint HFEs. Further justify the basis of chosen values. Assess the impact of using values consistent with those recommended.

b) F&O HR-E3-01 against HR-E3:

Although the disposition to this F&O notes that a review of HFEs credited in the IEPRA model was performed with operations and training personnel, the completeness of this review remains unclear given discussion in Appendix F of NB-PSA-HR (Vol. 1), which states that several reviews are currently in progress or under revision. Elaborate on the completeness of reviews performed for the IEPRA HFEs, clarifying the impact, if any, on HFEs credited in the FPRA.

c) F&O HR-G7-01 against HR-G7 and F&O QU-C1-01 against QU-C1:

The disposition to this F&O indicates that a dependency analysis has not been performed for the IE HRA. Estimate the impact on risk metrics (e.g., internal events CDF and LERF) reported in LAR Attachment W by performing a sensitivity study (e.g., setting any subsequent HEP beyond the maximum in a cutset to 1.0). Also, discuss the conclusion in the disposition that there is no impact to the NFPA-805 analysis because fire-specific HEPs are used in the FPRA. Recognize that there may be fire-affected cutsets carried from the IE Page 211 of 227

model where the internal HFEs are retained at their original values, (i.e., not fire-affected, such that failure to properly account for dependency may underestimate the cutset risk contribution to fire).

d) F&O LE-G5-01 against LE-G5:

Discuss which of the two source term models is used to support the NFPA-805 LAR, and identify limitations in the LERF analysis that would impact the NFPA-805 LAR.

e) F&O QU-A3-01 against QU-A3:

The RG 1.200 clarification of QU-A3 requires that the SOKC be taken into account for both CDF and LERF regardless of significance. Discuss the extent to which the correlation of basic event data has been addressed to meet CC-II.

f) F&O QU-B2-01 against QU-B2:

The disposition to this F&O indicates a truncation study has not been performed for the IEPRA. Estimate the impact on risk metrics (e.g., IE CDF and LERF) reported in Attachment W of the LAR.

g) F&O QU-D1-01 against QU-D1 through D7, QU-F1, and QU-F2:

Describe the reasonableness review performed on quantification results, including a discussion of the results, and provide additional justification for the conclusion of no impact on the NFPA 805 analysis.

h) F&O SY-B12-01 against SY-B12-01 and F&O IE-C6-01 against IE-C6:

Clarify the basis for excluding control room HVAC initiating events and dependencies from the IE model. Confirm that the FPRA results are conservative by excluding the potential mitigating effects from MCR HVAC in the event of an accident.

i) F&O DA-A8-01 against DA-A8, F&O HR-C2-01 against HR-C2, F&O IE-A9-01 against IE-A9, F&O IE-C2-01 against IE-C2, and F&O SY-A20-01 against SY-A20-01:

The following observations related to the gathering of plant-specific data are noted:

i. A review of NB-PSA-DA indicates that plant-specific data related to equipment failures, unavailability, equipment demands and run time, etc.

were collected and documented over a three-year period from January 1, 2005 to January 23, 2008.

ii. Section 2.4 of NB-PSA-HR (Vol. 2) indicates that plant-specific operating experience was reviewed from 2005 to 2009 to identify additional modes of Page 212 of 227

unavailability for pre-initiators.

iii. Section 2.2.6 and 2.2.8 of NB-PSA-IE indicates that license event reports from 1996 to 2003 and maintenance rule and work order databases from 2005 until 2008 were reviewed to identify potential precursor events.

iv. The disposition to IE-C2-01 indicates that the update of generic initiating event frequencies excluded plant-specific data prior to January 2003.

v. The disposition to SY-A20-01 indicates that plant operating experience over a three-year period was used to screen certain cases of coincident unavailability as non-repetitive activity.

Provide a basis for each of the above data collection windows. In doing so, justify the exclusion of past data, and discuss whether or not other time periods were considered.

ENO Response a) The basis for the values used is as stated in the reference HRA notebook.

NUREG 1792 recommends values of 1.0E-2 for an independent train, function or channel (TFC) and a high dependency factor of 0.5 for multiple trains or channels irrespective of the level of redundancy. These values were judged to be overly conservative when considered in light of plant experience.

Conceptually the PNP arguments are consistent with the best practices guidance.

The guidance recognizes that the risk contribution from individual trains, functions or channels is not likely to be significant. The guidance notes that the most significant contributors are those conditions in which the possibility of multiple trains, functions or channels are unavailable from improper performance of an action that affects multiple components or improper performance of a particular action on a large group of components. This is consistent with the implementation of the revised process for the PRA model as discussed below.

Once the initial scoping values were assigned and an updated core damage frequency was quantified, risk significant pre-initiator contributors were identified and detailed human error probabilities (HEPs) were developed for the events. For the most part these were events impacting two or more trains, functions or channels or redundant or multiple diverse equipment.

The high failure rate and dependency factor recommended in the guidance document represent bounding values that account for worst case influence factors that impact the HEP development. These include work performed by the same crew, under similar work environments over short periods of time. However, current work practices would require that work and testing activities on redundant trains, functions or channels be performed on a staggered basis. The activities would be performed during assigned work weeks for the train, function or channel of a system.

Page 213 of 227

These work practices provide reduction in the influencing factors by increasing time separation between activities on redundant equipment such that it is likely that work is performed by a different crew or the same crew with different individual influencing factors. These practices are consistent with NUMARC 93-01 (Maintenance Rule) guidance and elements of detailed human error probability development. Additionally, operating experience reviews of site work activities indicates that routine calibration activities increasingly indicate that components are found to be within the acceptance range such that a calibration activity is not required. The impact is that the occurrence of required calibrations is less than the scheduled frequency. This translates to a reduction in the possible occurrence of mis-calibration events.

The revised process employed a combined methodology of scoping and detailed HEP development. The values identified in the human reliability analysis (HRA) notebook were used as a basis for the development of an initial risk ranking of pre-initiator operator actions. The initial assignment of scoping values was used to develop an initial importance ranking of the pre-initiator human failure events (HFEs). Given the importance ranking, risk significant HFEs were identified.

Detailed HEPs were developed for the risk significant HFEs.

Risk significance was determined based a criterion of Fussel-Vesely (F-V) importance contribution greater than 0.005. The PRA model database was updated with the detailed HEPs in place of the scoping values for the risk significant pre-initiator operator actions. The HEPs for the HFEs that did not meet the risk significance criterion were left at the assigned scoping value.

The aggregate impact of employing NUREG 1792 scoping values for the HFEs that are not risk significant is an increase in core damage frequency by a factor of 1.8. Eighteen events in the model were identified as the primary contributors to the increase.

b) Screening, scoping and detailed HEP values will be utilized in the RAI Response Fire PRA model for both fire response actions and actions that were carried over from the internal events PRA. Actions that were carried over have been renamed and assessed as post-fire actions. Since HFEs in the fire PRA (both fire response and internal events carry-overs) are treated as post-fire HFEs, the completeness of the reviews for internal events only HFEs does not impact the fire PRA.

Detailed reviews with plant operations and training personnel have been completed for 31 operator actions having the highest risk impact and 54 additional operator actions utilizing screening and scoping values. The reviews confirm interpretation of current and planned procedures relevant to modeled actions is consistent with plant operational and training practices. Recovery action documentation will be completed as part of NFPA 805 implementation (see LAR Attachment S, Table S-3, Items 1 and 3). A detailed description of the Post-Initiator Operator Action Questionnaire (P-IOAQ), Recovery Action Feasibility Evaluation, and recovery action Validation form content, development and review process is provided in RAI 01cc.

Page 214 of 227

c) The FPIE PRA results referenced in Attachment W of the LAR are for the model of record that existed at the time of the submittal. The impacts of human error dependencies are included explicitly in the model of record and in FPIE results shown in Attachment W. Thus, no sensitivity study is required.

The methodology for evaluating human error dependency was developed as described in HRA Notebook NB-PSA-HR Volume 1 (Reference [1]). This notebook also contains the results of the dependency analysis performed on the FPIE PRA version maintained by the PNP PRA group just prior to the LAR submittal.

The methodology used to complete the FPRA dependency analysis reported as part of the LAR is the same as that employed for the FPIE PRA. However, the HFEs for FPIE and fire events are treated separately. To identify them as specific to fire scenarios, all HFEs in the fire PRA have different basic event identifiers from their equivalent actions in the FPIE models (the identifiers for the FPRA are appended with the suffix -FR).

The HEPs for the FPRA HFEs are set independently from the HEPs for the FPIE PRA events. For the LAR, screening values of 1, 0.1, and 0.01 were applied for all of the FPRA events, even if the FPIE PRA contains similar human failure events with different HEPs. In general, the HEPs used in the FPRA are higher than the HEPs used for the equivalent HFEs in the FPIE PRA. Any detailed HEPs developed for the FPRA HFEs will be completed specifically for conditions associated with the fire scenarios (e.g., timing, cues, accessibility, etc.), and thus will be unaffected by any changes that might occur to HEPs for FPIE PRA human failure events during the completion of the dependency analysis for the updated FPIE PRA models.

Combinations of HFEs are generated using the cut sets specific to the model being assessed (FPIE, or fire); although many combinations of actions are the same, the combinations for the fire PRA are explicitly generated without regard to combinations that may have been identified for the FPIE PRA. Similarly, the level of dependence between actions within FPRA combinations are set based on parameters specific to the fire scenarios, and are once again independent of the FPIE PRA dependencies. Thus, no impact on the FPRA dependency analysis can occur due to the FPIE PRA dependency analysis update for the next version of the FPIE models.

There are no fire-affected cutsets carried from the FPIE model where the internal HFEs are retained at their original values, therefore, there was no failure to properly account for dependency and the FPRA dependency analysis does not underestimate the cutset risk contribution to fire.

References:

[1] NB-PSA-HR, Vol. 1, Rev. 4, Human Reliability Analysis Notebook Volume 1 (Post Initiator Operator Actions)

Page 215 of 227

d) Of the two available PNP Level 2 models (PWROG-L2 and PAL-L2) as described in the PNP PRA model summary notebook [1], the Level 2 LERF (PWROG Methodology) is used in the Fire PRA model. It is based on WCAP-16341-P (Simplified Level 2 Modeling Guidelines) and is referred to by the designator PWROG-L2.

The overall approach to the development of the PWROG-L2 model followed WCAP-16341-P which many plants are currently using as a basis for updated Level 2 analyses. This WCAP provides a common, standardized method for PWRs with large dry containments to produce an analysis that meets capability category II of the ASME PRA standard. The guidance particularly addresses the latest understanding for induced steam generator tube ruptures, direct containment heating, and other important Level 2 phenomena. The WCAP provides an event tree structure for both station-blackout-related scenarios and non-station-blackout scenarios to determine the likelihood of different accident progression scenarios. Each event tree question must be answered on a plant-specific basis, but the WCAP provides straightforward guidance for the process.

As such, there are no specific limitations in the LERF analysis that would impact the NFPA-805 LAR.

References:

[1] Palisades Probabilistic Safety Assessment Notebook NB-PSA-SM, Rev. 2, PSA Model Summary

[2] Palisades Probabilistic Safety Assessment Notebook, Level 2 Notebook, Rev. 3, November 2012.

[3] Westinghouse, Simplified Level 2 Modeling Guidelines - WOG Project: PA-RMSC-0088, WCAP-16341-P, Rev. 0, November 2005.

e) In the PNP FPIE, SAPHIRE is used for uncertainty. The fire analysis uses the UNCERT code [1].

The SAPHIRE code can estimate the variability (due to the uncertainties in the basic event probabilities) of both a fault tree top event probability or an event tree sequence frequency using Monte Carlo (MC) simulation or Latin Hypercube Sampling (LHS).

The approach being used at PNP is summarized below. This approach allows Capability Category III to be met for QU-A3.

Correlate basic events by collecting them into groups based on component type and failure mode. When the same component type, failure mode, failure probability, distribution and error factor exist among different basic events, a correlation class number is assigned to those basic events. Assignment of a correlation class ensures the events are treated appropriately by SAPHIRE to address uncertainty in basic event probabilities.

Page 216 of 227

Quantify accident sequences and propagate uncertainties taking into account the SOKC on the final CDF and large early release frequency (LERF) cutsets.

Perform sensitivity analyses to identify the basic events that drive the core damage and LERF results.

Perform sensitivity studies for basic event types that could not be correlated.

Tables of all PNP PRA basic events, correlated and un-correlated, are found in Attachment 14 of NB-PSA-DA [2].

PNP will update the results of the FPIE PRA uncertainty analysis, per the approach described above, based on the RAI Response Fire PRA Model now under development. As previously noted, the propagation of the uncertainty distributions ensures that the state-of-knowledge correlation between event probabilities is taken into account allowing Capability Category III for QU-A3 to be met. Furthermore, for basic events that are common to the FPRA model, the basic event correlation and uncertainty parameters are transferred for use in the fire PRA. This is not expected to impact the FPRA calculations which will utilize the point estimate CDF and LERF results, but will allow the propagation of uncertainties to also occur for the base case RAI Response Fire PRA model (refer to the response to PRA RAI 01nn).

References:

Calculation 0247-07-0005-01, Palisades Nuclear Plant Fire Probabilistic Risk Assessment Fire Risk Quantification and Summary, 11-27-12.

Palisades Probabilistic Safety Assessment Notebook NB-PSA-DA Rev. 6, Data Analysis 11-10-2012.

f) The FPIE PRA results referenced in Attachment W of the LAR are for the model of record that existed at the time of the submittal. A truncation study meeting supporting requirements of QU-B2 and QU-B3 was performed for this model. The impact on risk metrics was determined to be an 8% increase.

However, the QU-B2-01 finding is still considered open as a convergence study has not yet been performed for the current working FPIE model. An analysis to determine the future FPIE model of record truncation limit will be documented in Section 6.6 of PSA notebook NB-PSA-QU, after issuance of the complete internal events model report and full cutset review of all event trees.

As described in the response to PRA RAI 24, the risk benefits of crediting NFPA 805 modifications is expected to result in an internal events CDF in the range of 1E-5. Additionally, the internal events PRA model truncation level has no impact the FPRA model results reported in Attachment W of the LAR. Moreover, it is expected that the future FPIE model of record truncation level will have no impact on the RAI Response Fire PRA Model results.

Page 217 of 227

g) Reviews of the interim versions of the FPIE PRA, including the version used as the basis of the results submitted with the LAR, were conducted by PNP PRA staff, PRA consultants, and/or other parties, for each version, and insights from those reviews have been used to improve each subsequent version of the PRA.

Attachment O of EA-PSA-FPIE-FIRE-12-04 [1] documents the review history. For example:

The Combustion Engineering Owners Group (CEOG) conducted an industry peer review of the PNP PRA in 2000 [2]. All level A and B findings have been addressed.

Subsequent to the 2000 peer review, a gap analysis was performed in 2004

[3]. In general, recommended actions identified by this evaluation dealt with issues of documentation and/or justification for technical analyses in the PRA, although some recommendations potentially resulted in changes to the PRA and impacts to the CDF or LERF. These issues have all been subsequently addressed in this update, or earlier model updates.

Per Reference [4], the FPIE PRA also underwent a peer review against the requirements of RG 1.200 Rev. 2. Findings and observations from that review are included in Attachment U of the LAR submittal.

As a result of these and other (e.g., internal) reviews of the FPIE models and results, there is high confidence the results are accurate and reasonable. The PSAR3 model cutset review of accident sequence results as referenced in the LAR submittal are documented in Attachment N of Reference [1] for all sequences from the base FPIE model that are applied in the fire analysis. The review was conducted per Entergy engineering guideline EN-NE-G-014, Probabilistic Safety Assessment Level 1 Quantification Guideline [5]. The review individually evaluated the logic of the top 100 cutsets (based on CDF), 5 cutsets in each sequence (where cutsets were generated), and a sampling of non-dominant cutsets. When potential changes to logic models were identified as part of the cutsets review, they were classified per EN-NE-G-014 step 5.4.1. The cutset review process was (and is) iterative. For example, if any changes were identified that were classified as Immediate Change, the necessary change would be made and the model review repeated. As a result of the review documented in Attachment N of Reference [1], some potential changes to the logic models were identified, primarily in areas where additional credit for recovery of equipment may be possible (if such credit were included, the CDF might decrease). For example, several cutsets were identified in which off-site power recovery could be credited as successful - but such credit was not included. The possible changes to the CDF and LERF were not judged to be significant, and therefore these changes were classified as Change to be deferred; they were included in a Model Issue Database for possible inclusion at a later date.

Upon completion of the next update to the FPIE PRA, the cutsets will again be reviewed for accuracy and reasonableness. Through the use of deterministic and probabilistic analyses, operating experience, and interactions with plant staff, the PRA has continuously evolved and improved; its models and results are very representative of the plants actual and expected behavior. Thus, there is high Page 218 of 227

confidence that the review of the update, building upon the multiple iterations of review that have already been completed, will conclude that the model and its results continue to be accurate and complete.

The review documented in Attachment N of Reference [1] concludes that (T)he PSAR3 model generates sequence cutsets that are considered reasonable and logical. The findings of the review will be used as input to a similar review to be conducted on the RAI Response Fire PRA Model, and the contents of Attachment N of Reference [1] updated as necessary.

References:

EA-PSA-FPIE-FIRE-12-04 Rev. 0, Palisades Full Power Internal Events and Fire PRA Model.

Combustion Engineering Owners Group (CEOG), Industry Peer Review the Probabilistic Safety Analysis (PSA) against the Combustion Engineering Owners Group (CEOG) PSA checklists, RIE 2000-02, CE-NPSD-1194-P Task 1037.

ERIN Engineering and Research Inc., PALISADES GAP ANALYSIS REVIEW AND UPDATE, P0495060007-2711-061215, October 2004.

From David Finnicum to Bradford Grimmel, RG 1.200 PRA Peer Review Against the ASME PRA Standard Requirements For The Palisades Nuclear Power Plant Probabilistic Risk Assessment, LTR-RAM-II-10-015, March 12, 2010.

EN-NE-G-014 Rev. 0, Probabilistic Safety Assessment Level 1 Quantification Guideline.

h) Internal Events Initiating Events Considerations Control Room Heat-up Control room heat-up was evaluated considering a 72-hour period following a loss of ventilation. Offsite power was considered available to maximize control room heat load. The temperature reached the limit for habitability of 110°F at about 3.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> and exceeded the technical specification limit of 120°F in greater than 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br />.

Reactor Protection System (RPS) Considerations The RPS thermal margin monitor (TMM) is the system weak link. This unit feeds the variable high power trip and the thermal margin/low pressure trip, which provides protection for DNB and over-power events.

The design temperature of the RPS is 140°F, except for the TMM which as a design temperature of 131°F. Since the TMM is the limiting control room RPS component with respect to temperature rating and since it could reach its rated temperature when control room ambient temperature is 106°F, a control room administrative limit of 90°F was imposed. However, the thermal margin monitors are only required during normal operation and are not needed under the Page 219 of 227

emergency operating procedures after a reactor trip.

As noted in the FSAR, the Thermal Margin Monitor (TMM) was originally qualified to 131°F. However, the location of the TMM in the panel is such that cooling is required. Analysis shows that, with forced air cooling, 131°F is reached by the TMM when the control room ambient temperature is 106°F. Other portions of the Reactor Protective System located in the control room were designed to operate up to 135°F and 90% relative humidity. Individual components and modules of the RPS have been factory tested at design temperature and humidity conditions. With the exception of the TMM, the RPS cabinet (including all portions of the system located in the control room) has been tested for operation as a system at temperatures in excess of 135°F.

Loss of HVAC Impact Due to the diversity of the RPS signal inputs, the electrical reliability of the RPS system is dominated by the common cause terms in the voting logic. In addition, ATWS system provides an independent and diverse reactor trip.

While there are design basis transients that rely specifically on the TM/LP and/or VHP trips, all actual scenarios have at least several diverse reactor trips available.

The loss of the TM/LP and VHP trips would not result in a significant increase in RPS failure probability.

Given that the reliability of the RPS system is not significantly affected by the loss of the TMM, the remaining potential issue is control room abandonment. to SOP-24 contains guidance for aligning alternate ventilation for the control room. The alternate ventilation involves opening doors and aligning portable fans. Given an event followed by reactor trip, it is reasonable to expect operators to recognize the increased control room temperatures and open doors and bring in fans. Based on various control room heatup analyses, ample time is available to perform the actions.

However, the non-inclusion of control room HVAC in the internal events PRA model is not dependent solely on the operator action to open doors and bringing in portable fans. As described below.

PRA Review A review of the Westinghouse Owners Group PRA Model Methods and Result Comparison listed 14 plants that evaluated the impact of a Loss of HVAC. The average contribution to the internal events core damage frequency was about 1.6E-07 per year. The average associated loss-of-HVAC initiating event frequency was about 1E-02. Assuming PNP has a similar plant configuration; applying these results to the current PNP non-subsumed analysis-of-record CDF would result in less than a 1% increase in CDF, if a loss of HVAC were considered.

Procedural Controls Page 220 of 227

AOP-41, Alternate Safe Shutdown Procedure This procedure addresses the inability to maintain control of the plant from the Control Room. This procedure would be implemented on a loss of HVAC only if environmental conditions outside the control room prohibited alignment of alternate ventilation per SOP-24.

Implementation of this procedure may require Control Room evacuation (due to fires). Back to non-fire events, with the exception of the TMM, the operators would have access to the RPS instrumentation and controls for the 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> mission time of the PSA non control room abandonment scenarios. The aforementioned components have been proven to operate intermittently up to 140F. The expected peak room temperature in 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> would be less than 130F.

SHO-1, Operators Shift Items Modes 1, 2, 3, and 4 This procedure addresses operator surveillances that are required each shift. To protect the technical specifications limit, a shiftly check, that control room temperature is less than 88°F is performed. However, since the station blackout analysis assumes an initial control room temperature of 75°F, an additional procedural requirement exists to initiate an action to evaluate and correct the cause of the elevated control room temperature if the temperature exceeds 75°F for more than a shift.

Fire Initiating Events Considerations Control Room Fire Events Some fires that occur within the control room, if unsuppressed, have the potential to require operators to abandon the control room due to the presence of smoke or hot gases. The abandonment time is a function of fire severity, room boundary conditions (e.g. open or closed doors) and control room HVAC considerations.

These considerations include availability of the HVAC system as well as its operating mode. In the fire PRA, the room boundary conditions and HVAC operating mode (including failure of HVAC) are treated by evaluating each potential condition and selecting the bounding (shortest) abandonment time. The abandonment time is then used to develop the non-suppression probability factor for the scenario.

In the fire PRA that supports the LAR submittal, control room HVAC was not explicitly evaluated in the logic model. Its availability for control room fires was estimated by considering the system failed for fires that originate in cabinets where HVAC controls are located and successful where no controls are located.

Subsequent to the LAR submittal, a fault tree has been developed for control room HVAC and is used to determine system availability for each control room fire scenario. This refined treatment will be reflected in the results of the RAI Response Fire PRA model.

Page 221 of 227

Treatment of control room fires that do not result in control room abandonment is consistent with the treatment in the internal events PRA.

Non-Control Room Fire Events Treatment of non-control room fires abandonment is consistent with the treatment in the internal events PRA.

Control HVAC cooling in the PNP internal events PRA is not modeled based on the following:

the high design temperature limits of the major control room components, the philosophy of the operators with respect to remaining in the control room as long as possible, the TMMs are not credited in the EOPs, and the relative unimportance of HVAC failure on a variety of plant PRA studies.

Therefore it is considered unnecessary to model either loss of HVAC as an initiator or as a support system for the internal events model. Control room HVAC is evaluated for fire PRA control room abandonment scenarios to capture the impact on non-suppression probability factors. In the fire PRA submitted with the LAR, control room HVAC availability for control room fire scenarios was estimated, based on the location of HVAC controls. In the RAI Response Fire PRA Model, the treatment will be refined to use a fault tree to determine HVAC function for control room fire scenarios in assessing HVAC mitigating effects.

i) Part i: For Bayesian updates, consideration is given to the time frame of the generic data collection period. For data, the NUREG/CR-6928 generic data used as prior evidence covered the time frame through the end of 2002. Since the generic data already includes evidence from PNP, it is appropriate to start the plant-specific Bayesian updates after the generic data collection period. For data, the analysis utilized the available information starting on January 1, 2005 through the freeze date of the data analysis. This 3+ year time frame was chosen since a thorough review of plant experience was performed to identify component failures. In addition to reviewing Maintenance Rule (MR) data, the PRA team reviewed 7,693 Maintenance Work Orders (MWO). This review resulted in identification of an additional 17 component failures; 38 were identified through the MR Program. The additional failures involved components outside the scope of the MR and component failures that did not result in a functional failure. The review also concluded that all functional failures identified through the MWO process were identified in the MR program. Given the resources associated with the MWO review, the 3+ year time frame was deemed appropriate for the data collection period and judged representative of expected future performance. As such, a review of additional data periods was not warranted.

Part ii: Section 2.1 and 2.2 of the HRA Notebook (Vol. 2) described the detailed process used to identify potential pre-initiator HEPs for incorporation in the PRA Page 222 of 227

model. This process led to the identification of over 150 pre-initiators to be considered for risk significance and further evaluation. The pre-initiators identified were determined based on evaluation of each system and supporting system credited in the PRA. Each train, channel or function of each system was then examined to determine if pre-initiator conditions were credible (e.g. not self revealing during normal operation). This resulted in the more than 150 pre-initiators included in the model.

The plant operating history review was then performed as described in Section 2.4 of the HRA notebook (Vol. 2) to see if actual pre-initiator type events had occurred at PNP that involved systems or components not identified by the original scoping evaluation. For the operating experience review, condition Reports (CRs) for years 2005-2009 (inclusive) were gathered for a plant-specific evaluation of plant operating history to see if pre-initiator type events had occurred. The records were searched to determine if the events recorded in the CRs may have been due to mis-calibration or mis-positioning, or similar types of errors that could have prevented the affected system from performing its function in response to a postulated initiating event. The results of the operating history review were not used to eliminate any of the 150 pre-initiator events identified by the scoping process.

A screening process allowed the initial review of over 12,000 CRs to be reduced to 45 CRs for further evaluation. As a result of the review of the remaining 45 CRs, it was determined that no changes to the current PRA modeling were required. Each of the CRs in question was dismissed as either not applicable to the model, or already modeled with an appropriate pre-initiator HEP. As the scoping process utilized to identify potential risk-significant pre-initiator HEPs already included the identification of over 150 potential pre-initiator events, and the plant operating experience review for five years of data did not identify any new potential initiators, five years of data was judged sufficient for the evaluation, and an examination of additional past data was not warranted.

Part iii: The original LER review was performed for the initiating event analysis that covered the time frame from 1996 - 2003. The subsequent initiating event analysis update (see response to Part v below) covered the time frame from January 2003 to March 2009. Note that the PNP model includes a comprehensive set of initiating events typical for most PWR models, such that additional precursor reviews would not likely identify additional initiating events not previously considered. For the internal events initiating event analysis considered in the peer review, a review of additional LERs was not performed. However, potential initiating event precursors were also considered during the PRA teams review of the Maintenance Rule (MR) database and Maintenance Work Orders (MWO) in support of the data effort. The team reviewed the MR database and all 7683 plant MWOs collected between the period of January 1, 2005 to January 23, 2008. The MR collects functional failures based on a review of Corrective Action Reports and Condition Reports. No new initiating events were identified through this review.

Since the 1996 - 2003 LER review did not identify any additional initiating events, and since the 2005 - 2008 data review did not identify any new initiating events, there is high confidence that all internal events initiators are captured, and an examination of additional data periods was not warranted.

Page 223 of 227

Part iv: For Bayesian updates, consideration is given to the time frame of the generic data collection period. For initiating events, the NUREG/CR-6928 generic data used as prior evidence covered the time frame through the end of 2002. Since the generic data already includes evidence from PNP, it is appropriate to start the plant-specific Bayesian updates after the generic data collection period. For initiating events, this included all available information from January 1, 2003 through the freeze data of the initiating events analysis (March 2009).

Part v: Supporting Requirement (SR) SY-A20 refers to SR DA-C14 which states the following [4]:

EXAMINE coincident unavailability due to maintenance for redundant equipment (both intrasystem and intersystem) that is a result of a planned, repetitive activity based on actual plant experience.

CALCULATE coincident maintenance unavailabilities that are a result of a planned, repetitive activity that reflect actual plant experience. Such coincident maintenance unavailability can arise, for example, for plant systems that have installed spares (i.e., plant systems that have more redundancy than is addressed by tech specs). For example (intrasystem case), the charging system in some plants has a third train that may be out of service for extended periods of time coincident with one of the other trains and yet is in compliance with tech specs. Examples of intersystem unavailability include plants that routinely take out multiple components on a train schedule (such as AFW train A and HPI train A at a PWR, or RHR train A and LPCS train A at a BWR).

Three years of data was judged sufficient to determine the potential for repetitive coincident unavailability. This time period encompasses two full refueling cycles and therefore would include any activities that occurred on a repetitive basis at PNP where the maintenance schedule is based on a cycle schedule with the majority of activities repeating every 13 weeks and most all within one refueling cycle. Therefore, a review of two refueling cycles gives high confidence that routine periods of coincident unavailability would be identified, and an examination of additional data periods was not warranted.

References:

[1] Palisades Probabilistic Safety Assessment Notebook NB-PSA-DA, Rev. 6, Data Analysis Notebook

[2] Palisades Probabilistic Safety Assessment Notebook NB-PSA-HR, Vol. 2, Rev. 3, Human Reliability Analysis Notebook Volume 2 (Pre Initiator Operator Actions)

[3] Palisades Probabilistic Safety Assessment Notebook NB-PSA-IE, Rev. 4, Initiating Event Notebook

[4] ASME/ANS RA-S- 2009, Standard for Level 1/Large Early Release Frequency Probabilistic Risk Assessment for Nuclear Power Plant Applications, the American Society of Mechanical Engineers and the American Nuclear Society, Page 224 of 227

approved by the American National Standards Institute on February 2, 2009.

UPDATE - ENO RESPONSE TO PRA RAI 06 On October 24, 2013, ENO submitted the 90-DAY RAI response per PNP Letter 2013-079. In this letter, the RAI response for PRA RAI 06 was provided. It was subsequently determined that PRA RAI 06 needs to be updated to remove the reference to PRA RAI

23. The specific sentence that references PRA RAI 23 should be revised to read as follows:

Transient fire scenarios are being developed for these locations to demonstrate no pinch points were omitted and will be included in the base case results of the RAI Response Fire PRA model.

Page 225 of 227