Text

Tuesday Presentations
	ML17284A214
Person / Time
Site:	Nuclear Energy Institute
Issue date:	10/03/2017
From:	; Nuclear Energy Institute
To:	; NRC/RES/DRA/HFRB
Shared Package
ML17284A210	List: ML17284A214;
References
	Download: ML17284A214 (118)
	v • d • e

WORKSHOP ON IMPROVING FIRE PRA REALISM - IMPACT ON RISK APPLICATIONS Tom Basso Director - Corporate Programs Exelon Generation October 3, 2017

FIRE PRA REALISM Improve Fire PRA realism to better reflect risk due to fire events

Improved PRA realism allows the industry to move forward using PRA information in risk informed applications

Implement a probabilistic, data driven approach to fire PRA issues using available and applicable data that requires no further testing

Incorporate operational experience to reflect the number of industry severe fire events

Any future testing should focus on actual plant design and operational configurations

Streamline the current approach for improving methods and data to 2 meet industry needs (2 year window)

FIRE PRA REALISM GOALS

NRC Research and EPRI should partner to implement the new EPRI road map to realistic fire models

Fire research activities within academia should be pursued to leverage and accelerate the resolution of issues

Expeditious resolution of impactful fire PRA drivers in the next 2 years is needed

Risk informed applications being sought rely on a realistic fire PRA platform

Rework of risk-informed programs will be required if resolution is not implemented in a timely manner

Improved fire models facilitate application of insights 3

Victoria Anderson, NEI Overview of Industry NRC Fire PRA Workshop October 3, 2017 Experience

Importance of Fire PRA Realism in Supporting Safe Operations

Transition to NFPA 805 means that Fire PRA outputs are driving more operational decisions

Future transitions to 50.69 and TSTF-505 will increase this impact

Critical that decisions are based on accurate information that comports with operating experience

Progress to Date

Nearly a decade of work improving on original methods outlined in NUREG/CR-6850

Substantial improvements to large-scale conservatisms (e.g. electrical cabinet heat release rates)

Multiple moderate conservatisms remain and, in aggregate, result in Fire PRAs that still do not comport with operating experience

Objectives for Workshop

Review specific conservatisms continuing to drive results

Evaluate relative impact of various conservatisms

Inform prioritization of work for research plans

Fire Spread Between Adjacent Electrical Enclosures Ashley Lindeman Senior Technical Leader Francisco Joglar JENSEN HUGHES Fire PRA Workshop: Improving Realism October 4, 2017

Background

- Fire PRAs include scenarios where a fire initiated in one electrical enclosure is assumed to propagate to an adjacent enclosure.

This is referred to as cabinet-to-cabinet fire spread in NUREG/CR-6850 A fire is assumed to spread from the electrical enclosure of fire origin (the exposing enclosure) to an adjacent enclosure (the exposed enclosure) after 10-15 minutes unless the partition(s) between the two enclosures meet one of two exclusionary configurations.

There are no distinctions based on factors that would be expected to influence this behavior. Such factors would include:

- Physical characteristics of both the exposing and exposed electrical enclosures, and

- Fire properties (e.g., peak heat release rate (HRR)) assigned to the exposing enclosure.

There is also no limit stated as to how many electrical enclosures might ultimately become involved so the extent of fire spread is

- Under some interpretations, the number of cabinets that may become involved is limited only based on the non-suppression probability as a function of time 2

Objectives Refine the current methodology for the analysis of multi-cabinet fire scenarios.

- Explicitly consider scenario likelihood and characteristics

- Provide guidance for avoiding double-counting of risk contributions when evaluating propagation between electrical cabinets.

Applicable to Bin 4 (main control board), Bin 10 (battery chargers), or Bin 15 (plant-wide electrical cabinets, non-HEAF) fire ignition sources.

3

Terminology Exposing enclosure: The electrical enclosure where the fire is assumed to originate (referred to as the exposing cabinet in NUREG/CR-6850).

Exposed enclosure: An electrical enclosure that is adjacent to the exposing enclosure and to which fire spread may be considered (referred to as the exposed cabinet in NUREG/CR-6850).

It should also be noted that the term enclosure refers to a vertical sections counted as part of electrical cabinets.

4

Fire Propagation Rules for Adjacent Enclosures The following rules are recommended to be used for determining the likelihood ignition of adjacent enclosures

- The Double Wall Rule: Do not postulate fire spread between adjacent electrical enclosures if both the exposing and exposed enclosures have solid steel panels on their adjacent sides (i.e., the double wall configuration).

- The Open Top/Vertically Oriented Cables Rule: Do not postulate fire spread between adjacent electrical enclosures if:

(1) the exposing or exposed enclosure (or both) has (have) an open top; and (2) there is an internal wall between vertical sections, possibly with some unsealed openings; and (3) there are no cables running in an upward direction (i.e., either vertically or diagonally) leading from the exposing enclosure into the exposed enclosure through the partition wall between sections. (Note: all three conditions must be met for the exclusion to apply.)

An open top electrical enclosure is defined as one whose top surface (i.e., the horizontal surface at the upper extreme of the side panel(s) is either entirely open or is largely open with obstructions (enclosure panels or framing but excluding objects external to the enclosure such as overhead cables or raceways) blocking no more than 50% of the total top area (e.g., a vented top surface with 50% or more venting).

- Vents positioned on (or in) the vertical side panels of an electrical enclosure, even if adjacent to the top surface, do not meet the top venting criteria.

Assessing the openings in a wall panel between sections will require judgment. As a general rule, openings representing up to 5% of the total surface area of the neighboring face would be acceptable under this rule.

5

Fire Propagation Rules for Adjacent Enclosures

- The Very Low Fuel Load Rule: Do not postulate fire spread to any adjacent electrical enclosure if the exposing enclosure has a very low fuel load per NUREG-2178.

- The Small Enclosure Rule: Do not postulate fire spread to any adjacent electrical enclosure if the exposing enclosure has been categorized as a small electrical enclosure per NUREG-2178.

6

Fire Propagation Rules for Adjacent Enclosures (cont.)

The following rules are recommended to be used for determining the likelihood of ignition of adjacent enclosures

- The Low Fuel/Steel Partition Rule: Do not postulate fire spread between adjacent electrical enclosures if there is a full steel panel partition between the exposing and exposed enclosures possibly with some unsealed/open penetrations (up to 5% of the total area of the panel partition) in combination with any one of the following conditions:

A low fuel load in the exposing enclosure, or A low fuel load in the exposed enclosure provided that the combustible fuels in the exposed enclosure do not come into contact with the separating steel partition, or A very low fuel load in the exposed enclosure provided that the combustible fuels in the exposed enclosure do not come into contact with the separating steel partition.

- The Low Fuel Exposing/Very Low Exposed Rule: Do not postulate fire spread to an adjacent electrical enclosure if the exposing enclosure has a low fuel load and the exposed enclosure has a very low fuel load regardless of the nature of the separation/partitions - no separation, vented, or open partitions - between enclosures.

Note: If you have default fuel loading you must assume propagation if there is no separating partition.

- The MCC Rule: Do not postulate fire spread between adjacent motor control center (MCC) vertical sections.

- The Switchgear Rule: Do not postulate fire spread between adjacent switchgear vertical sections.

7

Fire Modeling of Adjacent Enclosures Fire spread will be limited to one adjacent enclosure The HRR profile for the two adjacent enclosures will follow as:

- The exposing enclosure will be assumed to have reached peak intensity after 12 minutes, and should be assumed to hold that peak intensity for 8 minutes at which time fire decay can be assumed to begin per the guidance provided in NUREG-CR-6850.

- Using the average decay stage times for the tests referenced in NUREG/CR-6850, the decay phase for the exposing enclosure will last 19 minutes. That is, the exposing enclosure will burn out 39 minutes after ignition.

- Assume the peak fire intensity for the exposing enclosure corresponds to the 98th percentile of the peak HRR distribution applicable to the exposing enclosure (i.e., based on size, function, and/or fuel loading conditions).

- Fire spread to an adjacent electrical enclosure should be assumed to occur 10 minutes after ignition of the exposing enclosure, per the most conservative guidance presented in NUREG/CR-6850. Therefore, the exposed enclosure will begin its growth stage concurrent with ignition at 10 minutes.

- The exposed enclosure will also be assumed to conservatively achieve the 98th percentile peak HRR based on the distribution that is applicable to the exposed enclosure. Note that the exposing and exposed enclosures may be characterized by different peak HRR distributions depending on the characteristics of each.

- Consistent with the treatment of the exposing enclosure, the exposed enclosure should be assumed to maintain its peak intensity for 8 minutes, after which it will begin a decay phase that will last no more than 19 minutes for a total fire duration in the exposed enclosure of 39 minutes (49 minutes from exposing enclosure ignition).

8

Fire Modeling of Adjacent Enclosures (cont.)

Total Combined HRR (kW) Exposing Enclosure HRR (kW) Exposed Enclosure HRR (kW)

HRR (kW) 0 10 20 30 40 50 Time (min) 9

Consequence for Adjacent Enclosures Fire Propagation When fire spread to an adjacent enclosure is postulated, a conditional probability with a value of 0.02, may be applied in the form of a modifier (severity factor) against the frequency of fires initiated in the exposing enclosure or, more correctly, in the form of a split fraction between single- and multi-enclosure fire scenarios.

- In instances where the rules outlined above suggest propagation in two directions, two options are available for the application of the 0.02 conditional probability:

A value of 0.01 (0.02/2) may be used and fire spread may be postulated to the enclosures on either side of the exposing enclosure (Note: When applying the probability in this manner the fire is probabilistically being spread to a single cabinet in either direction with a modifier of 0.01. Therefore, the HRR profile for such a scenario will only be that of the exposing and a single exposed enclosure), or A value of 0.02 may be used to determine the most risk significant scenario for fire spread among the enclosures in both directions. Once the most risk significant path is determined, fire spread will only be postulated to that single adjacent enclosure.

10

Pictorial Representation Propagation to Ignition Adjacent Cabinet Yes - 0.02 Source Freq No - 0.98 Exposing Exposed 11

TogetherShaping the Future of Electricity 12

NRC/INDUSTRY WORKSHOP ON IMPROVING REALISM IN FIRE PRAs REVISED GROWTH CURVE ELECTRICAL PANEL AND TRANSIENT FIRES Usama Farradj October 3, 2017

OUTLINE REVISED GROWTH CURVE Basis for Current Electrical Fire Growth Curve Applicability of Approach to Transient Fires Impact of Revised Fire Growth Curve on Non-Suppression Probability Application to Sensitivity Evaluation www.jensenhughes.com 2

BASIS FOR CURRENT ELECTRICAL CABINET FIRE GROWTH CURVES Electrical Cabinet Fire Growth Curve

NUREG/CR-6850, Section G.3.1, Table G-6 provides a list and brief description of electrical cabinet tests used to define the fire growth curve Time to Peak, 12 minutes, t2 growth curve 8 minutes at peak 19 minutes decay

Growth timeframe defined primarily by method of ignition (transient -

16/22, electrical - 3/22, gas - 3/22)

Growth curve defines applicable NSP Longer time to peak = longer time to damage = lower NSP www.jensenhughes.com 3

TRANSIENT FIRE GROWTH The same approach outlined for electrical cabinet fire growth may be applied to transient fires A proportional increase in the time to peak for a transient fire similar to that applied to an electrical panel fire will result in a similar proportional decrease in the corresponding transient fire NSP www.jensenhughes.com 4

IMPACT ON SCENARIO NSP OF LONGER TIME TO PEAK HEAT RELEASE RATE Using e-t curve

Doubling the time to peak (increase from 12 minutes to 24 minutes) results in a reduction of NSP, for a given time, to a value that is the square of the current NSP (NSP2), using the current NSP for electrical fires the reduction is approximately a factor of 3 (at 12 minutes)

Increasing the time to peak by a factor of 1.5 (18 minutes) results in a reduction of NSP for a given time to a value that is the current NSP1.5, using the current NSP for electrical fires the reduction is approximately a factor of 2 (at 12 minutes) www.jensenhughes.com 5

HEAT RELEASE RATE VERSUS TIME FOR NUREG-2178, CONFIG 4a, closed panel, 98th %ile 450 400 350 300 250 HRR, kW 24 min 200 18 min 12 min 150 100 50 0

0 10 20 30 40 50 60 Time, minutes www.jensenhughes.com 6

HRR VERSUS NSP FOR NUREG-2178, CONFIG 4a, closed panel, 98th %ile 450.0 400.0 350.0 300.0 HRR, kW 250.0 200.0 150.0 100.0 50.0 0.0 1.00 0.82 0.68 0.56 0.46 0.38 0.31 0.25 0.21 0.17 0.14 0.12 0.10 NSP 24 min 18 min 12 min www.jensenhughes.com 7

METHOD APPLIED FOR SENSITIVITY EVALUATION NSP for scenarios are decreased by a factor of 3 or 2, for peak at 24 min or 18 min, respectively Does not account for additional risk reduction due to the slower rate of increase of the HRR which would provide a longer time to damage using the NUREG/CR-6850, Appendix H damage accrual data Applicable to transient fires also The potential exists for additional reduction for transient fires where a shorter current analysis time to peak is typically used Approach is applied based on an increased time to peak heat release curve but is also conservatively representative of incorporation of a fire pre-growth phase during which the fire may be detected (when smoke detection is present) but the HRR is minimal www.jensenhughes.com 8

QUESTIONS?

Contact Usama Farradj

+1 925-943-7077 ufarradj@jensenhughes.com For More Information Visit www.jensenhughes.com www.jensenhughes.com 9

Conservatism in Non-Suppression Probability (NSP) Data FPRA Workshop October 3-5 2017 Presented by:

Mark Schairer Engineering Planning and Management, Inc.

NSP Data Conservatism There is a disconnect between the average durations of fire scenarios in Fire PRAs vs. fire event experience NUREG-2169 uses over 400 total fire events from 1981-2009 for FPRA applications.

The fire durations can have time lags or delays associated with reporting, which inflates the fire time.

Time to control the fire versus time to extinguish the fire Delays in declaring a fire event extinguished are associated with de-energizing equipment, offsite fire brigade, and water application In Fire PRAs, the source scenarios reach peak HRR in 12 minutes.

Using FLASH-CAT, multiple trays are involved well before 20 minutes. Peak ZOI and HGL occur before 20 minutes.

2

Path Forward From the NSP data, more than 25% of fire events have durations longer than 20 minutes.

Typically, on-site fire brigade arrival is expected to be within 5-10 minutes, and upon arrival, fire is under control within 5-10 minutes.

Additional review of the fire event data would be beneficial to re-examine when the fires were under control, rather than totally extinguished.

At the control point, the fire is no longer a threat to fire spread, hot gas layer formation, additional target damage.

3

Initial Assessment Additional fire event information can be obtained from NRC Event Reports, Licensee Event Reports, or through Plant contacts.

Some initial scoping of high-duration electrical cabinet fires from NRC Event Reports:

NSP Suppression Time EPRI Fire ID Event Date Power Mode Fire Severity Bin Designation Category (min) 1097 11/15/1986 Low-power operation Undetermined 26 Electrical 95 418 4/28/1984 Low-power operation Challenging 10 Electrical 60 642 11/4/1987 Power operation Challenging 15.1 Electrical 50 98 10/8/1998 RF PC 15 Electrical 46 30362 12/16/2001 RF PC 21 Electrical 45 50829 9/11/2004 PO PC 23 Electrical 45 175 11/22/2009 CD CH 15 Electrical 45 121 4/26/2003 PO U 21 Electrical 37 505 1/8/1986 Low-power operation Undetermined 21 Electrical 36 20302 7/25/1993 PO U 15 Electrical 35 238 1/24/1981 Power operation Challenging 21 Electrical 30 557 1/31/1987 Low-power operation Challenging 22 Electrical 30 656 12/17/1987 Power operation Challenging 22 Electrical 30 97 6/10/1998 PO PC 22 Electrical 29 10626 12/11/2002 PO PC 21 Electrical 27 235 12/30/1992 PO PC 26 Electrical 25 4

NSP Data from NUREG-2169 Sum of Total Number of Average Duration Fire Event Category Durations NSP at t = 20 min Events [min]

[min]

T/G Fires 30 1167 38.9 0.598 Control Room 12 37 3.1 0.002 PWR Containment 3 40 13.3 0.223 Containment (LPSD) 31 299 9.6 0.126 Outdoor transformers 24 928 38.7 0.596 Flammable gas 8 234 29.3 0.505 Oil fires 50 562 11.2 0.169 Cable fires 4 29 7.3 0.063 Electrical fires 177 1815 10.3 0.142 Welding fires 52 484 9.3 0.117 Transient fires 42 386 9.2 0.113 High energy arcing faults 8 602 75.3 0.767 All fires 442 6583 14.9 0.261 5

Estimate of Improvement: Electrical Fires Of the 177 total fire events, 25 had durations of over 20 minutes Simple exercise: Events over 20 minutes duration were reduced by half, but not less than 20 minutes; Events under 20 minutes were unaltered i.e., 50 min event was reduced to 25 min, while 30 min event to 20 min Average Total Number Sum of NSP at time =

NSP Curve Duration Events Durations 20

[min]

NUREG-2169 177 1815 10.3 0.142 Estimate of Improvement 177 1492 8.4 0.093 The average duration was reduced by ~20%

NSP at 20 minutes reduced by 35%

6

Summary NSP data still conservative with respect to fire durations More than 25% NSP after 20 minutes Time to Control versus time to extinguish a fire Time to control is a better data point for plant response and risk/damage assessment.

Future work - review NRC event reports, LERs, and other sources of event information to identify when a fire was controlled, rather than extinguished to refine NSP calculations.

7

Moving towards more Realistic Cabinet Damage Rob Cavedo

The Conservatism All trains and functions that can be affected by a fire are assumed to occur at the cabinet ignition frequency.

For most cabinets, this is a mildly conservative assumption (e.g. breaker goes open/breaker goes closed).

For cabinets that control multiple trains from different power supplies, this can be overly conservative.

Industry events of cabinets with multiple trains do not show the loss of all trains (e.g. single alarm card damaged others functional, single HS malfunction, etc.).

1 Moving towards more Realistic Cabinet Damage

Uninformative Prior Approach Assume each possibility has an equal likelihood of occurrence.

For simplification, the impact option should be grouped by train.

A train is defined as any equipment in the cabinet powered from the same ultimate external power supply to the cabinet.

If all equipment in the cabinet is powered from the same external source, then this method cannot be applied.

This would only apply to scenarios with NO EXTERNAL damage. If the fire is large enough to damage external equipment, then larger internal losses are expected.

2 Moving towards more Realistic Cabinet Damage

Example Flow controllers for Pump A and C are in Cabinet X.

Given a per panel ignition frequency of 1E-4, the conservative and more realistic results are:

Conservative full Cabinet Improved Cabinet Only Loss Modeling Benefit in Case Description Impact Frequency Frequency the Fire X1 Fire within Cabinet FCs A and C Lost 7.00E-05 2.33E-05 within a X2 Fire within Cabinet FC A lost; FC C functional N/A 2.33E-05 cabinet region X3 Fire within Cabinet FC C lost; FC A functional N/A 2.33E-05 X4 Fire Damages cabinet and Target FCs A, C and 1st Target 2.80E-05 2.80E-05 X5 Fire Damages whole room Whole Room 2.00E-06 2.00E-06 33.3% chance of each impact possibility 3 Moving towards more Realistic Cabinet Damage

Questions 4 Moving towards more Realistic Cabinet Damage

Reduction in the NSP Floor Rob Cavedo

The Conservatism The NSP Floor is at 1-in-1000 for an infinite duration fire.

This can be a significant contributor to control room abandonment likelihoods with lower heat release rate scenarios.

1 Reduction in the NSP Floor

Use a 1E-5 Floor The fire brigade composition and method of operation is similar to a reactor operational response crew.

There is a leader directing the actions of the fire brigade members. The leader will bring more and more resources to bear as the scenario progresses.

2 Reduction in the NSP Floor

Sample Improvements Control Room Transients Electrical Fires 0.324 0.111 0.098 NSP with NSP with NSP with NSP with NSP with NSP with Time Floor 1E-3 Floor 1E-5 Time Floor 1E-3 Floor 1E-5 Time Floor 1E-3 Floor 1E-5 0 1 1 0 1 1 0 1 1 5 0.198 0.198 5 0.57 0.57 5 0.61 0.61 10 0.039 0.039 10 0.33 0.33 10 0.38 0.38 15 0.0077 0.0077 15 0.19 0.19 15 0.23 0.23 20 1.52E-03 1.52E-03 20 0.11 0.11 20 0.14 0.14 25 1.00E-03 3.01E-04 25 0.06 0.06 25 0.09 0.09 30 1.00E-03 5.95E-05 30 0.04 0.04 30 0.05 0.05 35 1.00E-03 1.18E-05 35 0.02 0.02 35 0.03 0.03 40 1.00E-03 1.00E-05 40 0.01 0.01 40 0.02 0.02 50 1.00E-03 1.00E-05 50 3.81E-03 3.81E-03 50 7.63E-03 7.63E-03 60 1.00E-03 1.00E-05 60 1.25E-03 1.25E-03 60 2.88E-03 2.88E-03 70 1.00E-03 4.11E-04 70 1.08E-03 1.08E-03 80 1.00E-03 1.35E-04 80 1.00E-03 4.09E-04 90 1.00E-03 4.42E-05 90 1.00E-03 1.54E-04 100 1.00E-03 1.45E-05 100 1.00E-03 5.82E-05 110 1.00E-03 1.00E-05 110 1.00E-03 2.19E-05 120 1.00E-03 1.00E-05 120 1.00E-03 1.00E-05 3 Reduction in the NSP Floor

Benefits The largest benefit is a reduction in the control room abandonment likelihood. For most other types of fires, by the time the floor is used the room is already lost due to a damaging hot gas layer.

4 Reduction in the NSP Floor

Control Room Abandonment Improvement (w NUREG 2178)

HRR Time To NSP wo Contribution NSP with SF*NSP with Contribution Bin, i (kW) SF Abandonment Floor SF*NSP wo Floor wo Floor Floor Floor with Floor 1 34 0.161 33 2.27E-05 3.65E-06 0.65% 1.00E-03 1.61E-04 14.0%

2 130 0.554 25 3.04E-04 1.68E-04 30.05% 1.00E-03 5.54E-04 48.5%

3 221 0.205 22 8.02E-04 1.64E-04 29.38% 1.00E-03 2.05E-04 17.9%

4 310 0.059 19.1 2.05E-03 1.22E-04 21.71% 2.05E-03 1.22E-04 10.6%

5 400 1.61E-02 17.25 3.74E-03 6.00E-05 10.72% 3.74E-03 6.00E-05 5.3%

6 490 4.03E-03 15.67 0.0062 2.51E-05 4.49% 0.0062 2.51E-05 2.2%

7 579 9.72E-04 13.8 0.0114 1.11E-05 1.99% 0.0114 1.11E-05 1.0%

8 669 2.35E-04 12.59 0.0169 3.98E-06 0.71% 0.0169 3.98E-06 0.35%

9 759 5.47E-05 11.82 0.0217 1.19E-06 0.21% 0.0217 1.19E-06 0.10%

10 848 1.25E-05 11.3 0.0257 3.21E-07 0.06% 0.0257 3.21E-07 0.03%

11 938 2.90E-06 10.76 0.0306 8.88E-08 1.59E-04 0.0306 8.88E-08 7.77E-05 12 1,028 6.53E-07 10.46 0.0337 2.20E-08 3.94E-05 0.0337 2.20E-08 1.93E-05 13 1,118 1.46E-07 10.01 0.0390 5.71E-09 1.02E-05 0.0390 5.71E-09 4.99E-06 14 1,208 3.25E-08 9.68 0.0434 1.41E-09 2.53E-06 0.0434 1.41E-09 1.24E-06 15 1,462 9.24E-09 8.64 0.0608 5.62E-10 1.00E-06 0.0608 5.62E-10 4.92E-07 Pr(ab) 5.60E-04 Pr(ab) 1.14E-03 The reduction of the floor can provide a 20% to 50% reduction in the likelihood of control room abandonment. The amount of reduction is on the higher end for control room modeling that credits NUREG 2178. NUREG 2178 has a much higher likelihood of lower HRR fires.

5 Reduction in the NSP Floor

Questions 6 Reduction in the NSP Floor

HEAF Event Frequency Keith Vincent - NextEra Energy Fire PRA Technical Lead Fire PRA Workshop, October 3-5, 2017

HEAF Event Frequency Current Fire PRA Methodology

HEAF events modeled using Appendix M of NUREG/CR-6850

Assessment based on operating experience, San Onofre Unit 3 Event from 2001.

- Any vulnerable component with 3 horizontally suffers physical damage

- First overhead cable tray will be ignited if within 5 vertical distance and 1 foot vertical distance of the top of the cabinet

- Ensuing fire reaches peak HRR immediately (no t2 ramp up)

Does not differentiate between ignition sources 2

HEAF Event Frequency Operating Experience Review

Review of Operating Experience has shown that the classification of the electrical device plays a significant role in the potential for there to be a HEAF event.

Class 1E equipment has had fewer HEAF events compared with Non-Class 1E equipment

- Class 1E HEAF Events - 3

- Non-Class 1E HEAF Events - 11 3

HEAF Event Frequency Suggested Approach to Increase Realism

Develop a methodology that addresses the differences between Class 1E and Non-Class 1E electrical equipment.

- Class 1E equipment is subject to higher maintenance and inspection standards and as such common precursors to HEAF events are ameliorated prior to a HEAF event occurring.

Utilize operating experience to generate an overall split-fraction to differentiate the source of the HEAF events. Assign a smaller fraction of the overall HEAF ignition frequency to Class 1E equipment.

- Class 1E Split Fraction - 0.21

- Non-Class 1E Split Fraction - 0.79 4

FAQ 14-0007 TRANSIENT FIRE FREQUENCY ENHANCEMENT Kiang Zee Gregory Zucal Fire PRA Workshop, October 3-5, 2017

BACKGROUND FLOOR BASED TRANSIENT FREQUENCY ALLOCATION NUREG/CR-6850

Methodology to distribute transient frequency to Physical Analysis Units (PAUs)

Rankings for transient influence factors assigned to each PAU

Generic location frequencies distributed to PAUs

PAU frequency apportioned to transient scenarios using floor area ratio FAQ 12-0004

Provided enhancements to account for administrative controls

New ranking levels of Very Low and Extremely Low available if criteria is met Limitations of current approach

PAUs frequently consist of different types of areas Transient free zones Storage areas High traffic areas

No clear guidance on how to account for variability of ranking values within a PAU www.jensenhughes.com 2

SUBDIVIDING A PHYSICAL ANALYSIS UNIT Follow existing guidance to distribute frequency to PAUs Identify PAUs for which sub-division is warranted to account for varying transient influence factor rankings

Define Transient Ignition Source Regions (TISRs)

Determine applicable floor area Influence factor rankings assigned following guidance in FAQ 12-0064 PAU D PAU Bin 7 (GT) Bin 6 (WC) Total D 1.53E-03 1.67E-03 3.20E-03 Other Area Type ID Floor Area nM nO nS nH Storage PAU D 2000 10 3 10 3 Area TISR D_TFZ 200 1 3 1 1 TFZ TISR D_Storage 400 1 3 10 1 TISR D_Other 1400 10 3 3 3 www.jensenhughes.com 3

CALCULATING TISR FREQUENCIES Transient Ignition Source Region Factor (TISRF)

Used to apportion the frequency within a PAU

Considers TISR influence factor rankings and sizes of TISRs (available floor areas)

TISR Frequency

Product of PAU frequency and TISRF

Used to apportion frequency to scenarios instead of PAU frequency Equations

Similar to the PAU apportionment equations in FAQ 12-0064 with addition of floor area term

nA,k,J is the available floor area for TISR k within PAU J

( n M , k , J + nO , k , J + n S , k , J ) n A, k , J nH , k , J n A, k , J TISRFGT ,k , J = TISRFWC ,k , J =

[(n k

M ,k , J + nO , k , J + n S , k , J ) n A, k , J ] n k

H ,k , J n A, k , J p ,k = p , J

TISRFp ,k , J www.jensenhughes.com 4

CALCULATING TISR FREQUENCIES Transient Ignition Source Region Factor (TISRF)

Used to apportion the frequency within a PAU

Considers TISR influence factor rankings and sizes of TISRs (available floor areas)

TISR Frequency

Product of PAU frequency and TISRF

Used to apportion frequency to scenarios instead of PAU frequency Equations

Similar to the PAU apportionment equations in FAQ 12-0064 with addition of floor area term

nA,k,J is the available floor area for TISR k within PAU J TISR Floor Area nM nO nS nH FA*Sum(n) FA*nH TISRFGT TISRFWC Bin 7 (GT) Bin 6 (WC) Total FA Ratio Freq Dist D_TFZ 200 1 3 1 1 1000 200 0.03 0.04 5.28E-05 6.94E-05 1.22E-04 10.0% 3.8%

D_Storage 400 1 3 10 1 5600 400 0.19 0.08 2.96E-04 1.39E-04 4.35E-04 20.0% 13.6%

D_Other 1400 10 3 3 3 22400 4200 0.77 0.88 1.18E-03 1.46E-03 2.64E-03 70.0% 82.6%

Total 2000 - - - - 29000 4800 1.00 1.00 1.53E-03 1.67E-03 3.20E-03 100% 100%

www.jensenhughes.com 5

IMPACT ON SCENARIO FREQUENCIES Scenarios frequencies calculated using floor area ratio of TISR

If transient scenario spans TISRs, frequency calculated per TISR and then combined The TISR methodology allows risk informed administrative controls to be reflected in future PRA model updates Frequency may increase in areas for which administrative controls were not applied PAU D Floor Area Ratio Freq Ratio Freq %

D_T3 Scenario

[ft2] (PAU) (PAU) (TISR) (TISR) Change D_T1 100 0.050 1.60E-04 0.500 6.11E-05 -62%

Other D_T2 150 0.075 2.40E-04 0.375 1.63E-04 -32%

D_T4 Storage D_T3 200 0.100 3.20E-04 0.143 3.77E-04 18%

Area D_T4 50 0.025 7.99E-05 0.125 5.43E-05 -32%

(Storage)

TFZ D_T2 D_T4 50 0.025 7.99E-05 0.036 9.43E-05 18%

D_T1 (Other)

D_T4 100 0.050 1.60E-04 N/A 1.49E-04 -7%

(Combined) www.jensenhughes.com 6

QUESTIONS?

Contact Gregory Zucal

+1 610-431-8260 gzucal@jensenhughes.com For More Information Visit www.jensenhughes.com www.jensenhughes.com 7

High Energy Arcing Fault (HEAF) Non-Suppression Probability (NSP)

Fire PRA FAQ 17-0013 FPRA Workshop October 3-5 2017 Presented by:

Mark Schairer Engineering Planning and Management, Inc.

Introduction The non-suppression probabilities (NSP) for high energy arcing fault (HEAF) fires provided in NUREG/CR-6850 Supplement 1 (FAQ 08-0050) and NUREG 2169 are considered conservative.

Fire event durations used for NSP extend past the control point in the fire event.

As a result, the risk associated with HEAFs in critical fire areas may be artificially high.

2

Approach A review of LERs was conducted to assess whether HEAF non-suppression probability (NSP) guidance per NUREGs 6850 and 2169 is overly conservative.

Mean BIN 16 HEAF Total AVG

Events Suppression Analysis Duration time/event Rate (/min)

NUREG/CR-3 239 79.67 0.013 6850 NUREG/CR-6850 3 276 92.00 0.011 Supplement 1 NUREG 2169 8 602 75.25 0.013 HEAF fire durations were reviewed to reduce the time to when the fire is considered to be controlled (i.e., when fire spread has been arrested, the ZOI has reached it's peak and when the threat of further damage, or hot gas layer is minimal).

3

Revision to Fire Event Times Common factors in the reported events that contributed to extended durations past the control point in the fires.

There is a delay between when the fire is under control in the field and when it is reported to the control room as extinguished.

There is a delay in reporting when the fire is declared extinguished due to the need to de-energize high energy equipment.

The time to control the fire is more relevant to potential damage impacts than full extinguishment.

Five (5) fire event durations have been revised based on the above.

4

Revision to Fire Event Times NUREG/CR- NUREG-2169 Revised Duration EPRI Event ID 6850 Supp. 1 Duration Justification (min)

Duration (min) (min)

Plant personnel suppressed and controlled the fire at 46 minutes with CO2 extinguishers to the origin. The 947 - (OC 19890103) 59 59 46 additional 13 minutes per NUREG-2169 accounted for water-based extinguishment.

The local fire department applied water to the insulation 74 - (WF 19950610) 76 136 80 above the bus duct at ~80 minutes, the fire would be considered sufficiently controlled at that point.

LER states that the fire brigade extinguished the fire after 35 minutes, well before offsite assistance, which later 100 - (DC 20000515) N/A 78 35 cleared the room of smoke and declared the fire extinguished after 78 minutes.

The fire was suppressed and controlled after 30 minutes according to plant personnel (flames no longer visible).

106 - (SG 20010203) 141 154 31 Additional time was due to complete extinguishment concerns over de-energization and resistance to using water.

The fire brigade reported the fire was under control after 127 - (VY 20040618) N/A 71 37 37 minutes, but it not declared extinguished until 71 minutes.

5

Additional Fire Events During the review, two fire events were identified that were binned as electrical fires for NSP in NUREG-2169, but are HEAFs for ignition frequency.

Event #922 - bus bar fire connecting to Main Aux Transformer from 6160 volt busses caused by fault to ground, it exhibited characteristics of a typical HEAF fire (secondary fires, de-energizing equipment, resistance to water use).

Fire duration of 5 minutes per the LER Event #792 - occurred in the A isolated-phase bus duct due to damaged ground straps and weathering. The bus ducts were required to be de-energized prior to suppression.

Fire duration of 3 minutes per the LER An additional event included in EPRIs most recent FEDB, Fire event

162 (8/5/2009) is a HEAF fire with a 46 minute duration.

6

Revised NSP Using the revised (5) HEAF fire durations, the other three (3) HEAF events in NUREG 2169, and event 162 Compare with and without the 2 bus duct events Mean BIN 16 HEAF

Events Total Duration AVG time/event Suppression Rate Analysis

(/min)

NUREG/CR-6850 3 239 79.67 0.013 NUREG/CR-6850 Supplement 1 3 276 92 0.011 NUREG 2169 8 602 75.25 0.013 FAQ 17-0013 -

(including Bus 11 385 35.0 0.029 Ducts)

FAQ 17-0013 excluding Bus 9 377 41.89 0.024 Ducts) 7

Comparison with International Events The Organization for Economic Co-operation and Development (OECD) report: Fire Project Topical Report No. 1 Analysis of High Energy Arching Fault (HEAF) Fire Events Incorporates data from HEAF events in 10 countries (excluding USA)

The OECD average duration of HEAF events outside the US was 31.3 minutes The OECD average duration of HEAF events, both US and International is 32.7 minutes FAQ 17-0013 proposes an average duration of 35 min (41.89 w/o Bus Ducts), which is conservative WRT international data.

8

Summary/Conclusions Number of Total Average Suppression Curve Events in Duration Duration Mean Curve (minutes) (minutes)

T/G fires 30 1167 38.8 0.026 Control room 12 37 3.1 0.324 PWR containment (AP) 3 40 13.3 0.075 Containment (LPSD) 31 299 9.6 0.104 Outdoor transformers 24 928 38.7 0.026 Flammable gas 8 234 29.3 0.034 Oil fires 50 562 11.2 0.089 Cable fires 4 29 7.3 0.138 Electrical fires 175 1807 10.3 0.097 Welding fires 52 484 9.3 0.107 Transient fires 43 386 9.2 0.111 HEAFs 11 385 35.0 0.029 All fires 443 6358 14.35 0.070 9

Alternative Approach For the HEAF events in the database, the new average time to suppression is 35.0 (41.89) minutes.

It can be argued that the fire is controlled early on, and propagation to trays had either burned out or this portion of the fire had been extinguished successfully earlier on.

Thus, in PRAs, the longer duration HEAF scenarios, which may lead to HGLs, etc. are much more likely in line with an electrical cabinet, or cable fire NSP curve.

Therefore, the HEAF fires that are evaluated at time periods well beyond the HEAF itself (i.e., 20 minutes) use the electrical cabinet or cable tray NSP and lambda mean value.

10

Summary/Conclusions NUREG-2169 HEAF data includes durations that are conservative It is proposed that the mean suppression rate be increased by approximately a factor of two (from 0.013 to 0.024 (0.029)) to reflect the revised average fire duration for HEAFs originating in high energy equipment in the US.

Preliminary set of comments received from NRC EPRI/NRC/NEI Task Force to jointly review and discuss the comments.

11

FAQ 16-011 Bulk Cable Tray Ignition Rob Cavedo

The Conservatism 1 FAQ 16-011 Bulk Cable Tray Ignition

Modeling with Conservatism For FLASH-CAT, the recommended value of HRR per unit area is:

150 kW/m^2 for thermoset cables and

250 kW/m^2 for thermoplastics.

For a medium cabinet 6 tall with 3 trays 1 above the cabinet, this is a 1.2 MW fire in 20 minutes 3.7% of the time.

Cabinet Scenario Frequencies given 205 C Ignition This credits 2178 HRR distribution and Cabinet Only 4.40E-05 obstructed plume.

Cabinet with Tray Stack Lost 1.77E-05 Whole Room Lost 2.35E-06 This is for a single cabinet. There are hundreds of cabinets that in the aggregate contribute to risk.

2 FAQ 16-011 Bulk Cable Tray Ignition

A Few Wise Sites - Flame Ignition of Trays Only A few sites modeled that tray ignition/spread only occurs when the cables are exposed to flame. This was explicitly approved in a few Safety Evaluation Reports (SERs). This was implicitly approved in a few other SERs.

This was the original inspiration for this FAQ. This FAQ translates the flame ignition requirements to the equivalent fire model thresholds.

Flame Tip Temperature Ta C Tp(centerline) - Ta C 474.9 454.9 20.000 500.4 470.4 30.000 526.0 486.0 40.000 551.5 501.5 50.000 577.0 517.0 60.000 Possible flame tip temperatures in a nuclear power plant using the NUREG 1805 spreadsheets.

3 FAQ 16-011 Bulk Cable Tray Ignition

Revised Cable Impact Table for Bulk Ignition/Spread This is based on:

NUREG/CR-7010 Results

NUREG/CR-5384 Insights

Only flame causes cable ignition (already approved in several SERs)

Limited contribution of arcing failures to overall spread/HRR contribution

Flux equivalent to 500 C from THIEF Equation 2.6 (NUREG/CR 6931 Vol 3) 4 FAQ 16-011 Bulk Cable Tray Ignition

Modeling with Realistic Tray Ignition For a medium cabinet 6 tall with 3 trays 1 above the cabinet, this is a 1.2 MW fire in 20 minutes 1.3% of the time.

Cabinet Scenario Cabinet Scenario Frequencies given Frequencies given 205 C Ignition 500 C Ignition Cabinet Only 4.40E-05 4.40E-05 Cabinet with Tray Stack Lost 1.77E-05 1.92E-05 Whole Room Lost 2.35E-06 8.25E-07 As the cable damage criteria remains the same, the benefit is a factor of 3 reduction in whole room loss scenarios.

5 FAQ 16-011 Bulk Cable Tray Ignition

Day-to-Day Impact

Current Cabinet Install Screening Approach All trays must be more than 8 above a medium closed cabinet. If not, then multiple fire scenarios must be analyzed to determine growth issues. This is very complicated. This involves walkdown of conduits and trays near the closest tray to the cabinet.

Post FAQ Cabinet Install Screening Approach As long as the closest tray is more than 4 away. Propagation need not be considered. The evaluation would simply be the cabinet itself and a scenario with the cabinet and those targets in the plume. This is a much less complicated evaluation.

6 FAQ 16-011 Bulk Cable Tray Ignition

Questions 7 FAQ 16-011 Bulk Cable Tray Ignition

NRC/INDUSTRY WORKSHOP ON IMPROVING REALISM IN FIRE PRAs Cable Fire Spread Usama Farradj October 3, 2017

OUTLINE CABLE FIRE SPREAD Fire Scenarios with Cable Tray Intervening Combustibles NSP Lambda Values NSP for Electrical Fire versus NSP for Cable Fire Suggested Approach www.jensenhughes.com 2

FIRE SCENARIOS WITH CABLE TRAY INTERVENING COMBUSTIBLES Fires associated with electrical panel, transient and other ignition sources which impact electrical cable trays may result in ignition of cables in the cable trays and fire spread along the cable trays Fire scenarios involving both electrical cabinet fires and cable tray fire spread transition from electrical cabinet fires to cable tray fires at the time when the electrical cabinet fire growth rate begins to decay (at 20 minutes per current fire growth rate assumptions, NUREG/CR-6850, Section G.3)

Benefit is primarily for electrical panels since transient fires do not have a timeframe in which their HRR decreases from the peak value www.jensenhughes.com 3

LAMBDA VALUE (MEAN SUPPRESSION RATE) COMPARISON www.jensenhughes.com 4

COMPARISON OF NSP VALUE OVER TIME NSP Comparison 1

0.1 NSP 0.01 0.001 0.0001 0 10 20 30 40 50 60 70 Time (minutes)

Cable NSP Elect NSP Combined NSP at 20 min www.jensenhughes.com 5

SUGGESTED APPROACH TO INCREASE REALISM Transition NSP curve from Electrical to Cable NSP curve at 20 minutes Incorporation of this approach to fire in large volumes where HGL is reached after 20 minutes would allow reduction of risk based on application of a more appropriate NSP value for the period during which the cable fire is the primary source of HRR Current fire data supporting the NSP curve for cable fires is limited but can be supplemented with future fire events data www.jensenhughes.com 6

QUESTIONS?

Contact Usama Farradj

+1 925 943-7077 ufarradj@jensenhughes.com For More Information Visit www.jensenhughes.com www.jensenhughes.com 7

OBSTRUCTED RADIATION ZOI Jason Floyd, PhD Fire PRA Workshop, October 3-5, 2017

OBSTRUCTED RADIATION POINT SOURCE SOLID FLAME www.jensenhughes.com 2

1D APPROXIMATION 4 4

, + , =

4 4

, + ,

Painted surface: = =0.95 Thin metal wall: , ,

, , = 900 ; = 25

= 50 /K (Forced due to flame)

= 10 /K (Natural)

= 719 NRL/MR/6180-03-8711 (2003)

(heptane spray fire against steel bulkhead) www.jensenhughes.com 3

MODELING APPROACHES Open Door 1 Grid Cell Wide Louvers No accounting for radiation Some radiation shielding shielding If < ~5 grid cells across each Flow dynamics incorrect slot, will not resolve pressure drop through louvers www.jensenhughes.com 4

SELECTED MODELING APPROACH Localized Leakage Define multiple &HVAC leakage paths over the height of the cabinet.

Assign each path the actual open area of the louver(s)

No direct flame radiation from inside the cabinet www.jensenhughes.com 5

FDS GEOMETRY MODELS Open Face Closed (Louvered) Face www.jensenhughes.com 6

ZOI FOR STEADY STATE FIRES 3 3 N TP N TS 2.5 E/W TP 2.5 E/W TS S TP S TS 2 2 FDS ZOI (m) FDS ZOI (m) 1.5 1.5 1 1 0.5 0.5 0 0 0 1 2 3 0 1 2 3 FDT ZOI (m) FDT ZOI (m)

FDS predicted ZOI of zero means no ZOI outside the cabinet Variables: cabinet size, cabinet shape, fire size, fire shape+location www.jensenhughes.com 7

TIME DEPENDENT HEAT RELEASE RATE At threshold exposure it takes 19 minutes for damage to occur (NUREG/CR-6850 Tables H-7,H-8)

NUREG/CR-6850 Table G-2 heat release profile 1.2 12 min. growth HRR Wall Temp 8 min. steady-state 1 19 min. decay Normalized Profile 0.8 0.6 0.4 19 min 0.2 0

0 500 1000 1500 2000 2500 Time (s) www.jensenhughes.com 8

TARGET DAMAGE Simple threshold exposure Peak exposure reaches NUREG/CR-6850 App H Conservative as discussed on prior slide Time dependent exposure Assume damage occurs in a linear fashion with a rate given by the corresponding entry in Appendix H At 6 kW/m2 thermoplastic cable is damaged at a rate of 1/19 min.

Below threshold exposure, adjust rate assuming the same total energy deposition is required (e.g. 3 kw/m2 would be 1/38 min)

Integrate the damage rate over time Damage occurs when the integral exceeds 1 www.jensenhughes.com 9

METHOD 1 - TARGET SCREENING A

B B= Obs_fac

A www.jensenhughes.com 10

METHOD 2 - REDUCED SEVERITY FACTOR EC www.jensenhughes.com 11

METHOD 2 - REDUCED SEVERITY FACTOR www.jensenhughes.com 12

QUESTIONS?

Contact Name

+1 410-737-8677 jfloyd@jensenhughes.com For More Information Visit www.jensenhughes.com www.jensenhughes.com 13

Integrating OE into NUREG-2180 (FAQ 17-012)

Harold Stiles - Lead Engineer PSA

VEWFDS Performing Beyond Its Design Basis August 2013 Timeline OE Has Been Used In NUREG-2180 OE was used to meet certain project objectives for NUREG-2180:

Item C: Human Reliability Analysis Item D: System Availability and Reliability Item E: System response to common products of combustion applicable to NPP While not a specific project objective, OE was also used in NUREG-2180 to estimate the:

Duration of the incipient stage Frequency of a potentially challenging fires having an incipient stage of sufficient duration

Event Tree for Quantifying the Risk Benefit of VEWFDS Focus on Poor reporting of Split fraction Abbreviated Prevention and Suppression fails Prevention does not External Targets Incipient Phase conflicts with HRA Incipient Duration Suppression entire cabinet result in No Fire No OE for fire in cabinet with VEWFDS

NRC feedback on FAQ 17-012 NRC identified several serious faults with the methodology:

Crediting limited, possibly irrelevant, data for human intervention rather than HRA methods Using non-suppression methodology to credit prevention Directly using Appendix L, rather than a method like Appendix L, beyond its intended application Inadequate technical justification for:

Changing the incipient stage threshold without collecting data (Appendix G of NUREG-2180)

Assuming the ALARM occurs prior to the start of the fire Establishing the incipient stage duration and the time available Using the MCR non-suppression probability for area-wide applications NRC also provided detailed review comments:

Provide a different event tree Separate area-wide applications from in-cabinet applications Address potential for low-energy events never progressing to a potentially challenging fire Industry initiative to collect data (Appendix G) could provide a basis for crediting VEWFDS Clarify roles for prevention/suppression and related dependencies Provide basis for suppression agent not failing entire cabinet NRC also provided additional comments during a face-to-face exchange

Proposed Event Tree for Quantifying the Risk Benefit of VEWFDS Initiating Event is not Targeted Suppression Successful prevention based on Bin 15 frequency does not fail entire cabinet results in No Fire

Expected Results with Cloud Chamber ASD Reasonable results with modest HEPs

Expected Results with Light Scattering ASD Limitations of less effective equipment

Considerations for Parameter Estimation Incipient Event Frequency Section 6.5.1 of NUREG/CR-6850 provides guidance for treatment of unique ignition source types that are not reflected in the generic frequency model ignition source list.

Incipient events would only be considered when attributed to a monitored component and terminated either by actual fire or by intervention prior to fire.

When an incipient event is terminated by intervention prior to fire, the characteristics (e.g., low-energy, temperature relative to ignition threshold) of the specific component should be considered to assign a partial count reflecting the likelihood of the incipient event progressing to an actual fire.

When an incipient event is terminated by fire or by intervention (if partially counted), the characteristics of the surroundings (e.g., contents of affected cabinet) should be considered to classify the event into a fire severity group (e.g., challenging, potentially challenging)

Duration of incipient stage Given the lack of available information, NUREG-2180 acknowledged expert elicitation as one approach to develop a consensus or community opinion.

The expectation is that a consensus or community opinion would result in a significantly longer incipient stage duration.

For application to a particular ignition source, the most realistic duration of the incipient stage would be a composite distribution based on the number, types, and failure modes of associated components.

Going Forward Industry initiative to collect VEWFDS data comparable to Appendix G:

Application Number and type of ignition sources (e.g., BIN 15 or BIN 4) for which VEWFDS is credited Number and type of components comprising the ignition source Capable of being de-energized at the panel or component level Administrative control in place to support a response Level of detail in procedures Roles of responsible personnel (e.g., decision maker)

Pre-stage equipment Operating experience Defining t(end) based on t(de-energize) is misleading Analyze and Report Incipient Results (EPRI document)

Incipient Event Frequency Incipient Stage Duration Establish a Process for Maintaining and Periodically Updating Incipient Results

UNOBSTRUCTED RADIATION ZOI Jason Floyd, PhD Fire PRA Workshop, October 3-5, 2017

FDT APPROACHES Point Source

" =

D/2+L 4 2 + 2 L

LF Solid Flame

" =

L D LF E=emissive power H F=view factor www.jensenhughes.com D 2

POINT SOURCE Underlying assumption is that the targets view L factor, F, of the fire is near zero.

LF Shokri+Beyler: D/2 + L > 2.5 Max(D,LF)

Contemp. Health Phys: D/2 + L > 3 Max(D,LF)

D Flame Distance Ratio Size Diameter Source Height 11 6 1.7 (kW) (m)

(m) (kW/m2) (kW/m2) (kW/m2)

Small EC 45 0.34 0.73 0.4 0.6 1.1 Medium EC 325 0.69 1.7 0.5 0.7 1.3 Large EC 1,000 1.0 2.7 0.6 0.7 1.4 100 gal spill 670,000 26 24 1.6 2.2 4.1 www.jensenhughes.com 3

SOLID FLAME S

Solid Flame L " =

LF From Shokri-Beyler (large hydrocarbon fires)

H = 58 x 100.0823 D , ,

=

,

Limit as L0 =

=

www.jensenhughes.com 4

SOLID FLAME Pump, motor, Large oil fires (e.g. TB) transient, EC, &

small oil spill fires www.jensenhughes.com 5

PROPOSED SOLID FLAME CHANGES Adjust Emissive Power:

0.0823

= Min 58 x 10 ,

Or Adjust Target Flux:

58 x 100.0823

= Min 1, www.jensenhughes.com 6

VALIDATION (FLUERY) 35 All Tests 35 1:1 Aspect Ratio 30 30 25 25 Predicted Heat Flux (kW/m2) Predicted Heat Flux (kW/m2) 20 20 15 15 10 10 5 Solid Flame 5 Solid Flame Solid Flame, Adjusted E Solid Flame, Adjusted E 0 0 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 Measured Heat Flux (kW/m2) Measured Heat Flux (kW/m2)

New approach has lower error and bias (bias is still positive) www.jensenhughes.com 7

VALIDATION (NIST/NRC) 40 All Tests 15 Outlier Removed Burner location underneath 30 gauge, solid flame Predicted Heat Flux (kW/m2) Predicted Heat Flux (kW/m2) 10 not appropriate 20 5

10 Solid Flame Solid Flame Solid Flame, Adjusted E Solid Flame, Adjusted E 0 0 0 10 20 30 40 0 5 10 15 Measured Heat Flux (kW/m2) Measured Heat Flux (kW/m2)

Most are larger fires (> 1 MW), correction is minor New approach has lower error and bias (bias is still positive) www.jensenhughes.com 8

VALIDATION (WTC) 70 60 50 Predicted Heat Flux (kW/m2) 40 30 20 10 Solid Flame Solid Flame, Adjusted E 0

0 10 20 30 40 50 60 70 Measured Heat Flux (kW/m2)

Larger fires (> 1 MW), correction is minor New approach has lower error and bias (bias is still positive)

High scatter due to test objectives www.jensenhughes.com 9

QUESTIONS?

Contact Name

+1 410-737-8677 jfloyd@jensenhughes.com For More Information Visit www.jensenhughes.com www.jensenhughes.com 10

ML17284A214

Text

Background

=

Navigation menu

Search