Fire Modeling

NUREG-1934
EPRI 1023259
Final Report
Nuclear Power Plant

Fire Modeling Analysis
Guidelines
(NPP FIRE MAG)
Final Report
U.S. Nuclear Regulatory Commission

Office of Nuclear Regulatory Research
Washington, D.C. 20555-0001
Electric Power Research Institute

3412 Hillview Avenue
Palo Alto, CA 94303
AVAILABILITY OF REFERENCE MATERIALS

IN NRC PUBLICATIONS
NRC Reference Material
Non-NRC Reference Material
As of November 1999, you may electronically access

NUREG-series publications and other NRC records at
the NRCs Public Electronic Reading Room at
http://www.nrc.gov/reading-rm.html. Public records
include NUREG-series publications; Federal Register
notices; applicant, licensee, and vendor documents and
correspondence; NRC correspondence and internal
memoranda; bulletins and information notices;
inspection and investigative reports; licensee event
reports; and Commission papers and their attachments.
Documents available from public and special technical

libraries include all open literature items, such as
books, journal articles, and transactions, Federal
Register notices, Federal and State legislation, and
congressional reports. Such documents as theses,
dissertations, foreign reports and translations, and nonNRC conference proceedings may be purchased from
their sponsoring organization.
NRC publications in the NUREG series, NRC

regulations, and Title 10, Energy in the Code of Federal
Regulations may also be purchased from one of these
two sources:
1. The Superintendent of Documents
U.S. Government Printing Office
Mail Stop SSOP
Internet: bookstore.gpo.gov
Telephone: 202-512-1800
Fax: 202-512-2250
2. The National Technical Information Service
Springfield, VA 22161-0002
www.ntis.gov
1-800-553-6847 or, locally, 703-605-6000
A single copy of each NRC draft report for comment is
available free, to the extent of supply, upon written
request to the following:
Address: Office of the Chief Information Officer,
Reproduction and Distribution
Services Section
E-mail:
[email protected]
Facsimile: 301-415-2289
Some of the NUREG-series publications posted under
http://www.nrc.gov/reading-rm/doc-collections/nuregs
are updated periodically and may differ from the last
printed version. Although references to material found
on a website bear the date the material was accessed,
the material available on the date cited may
subsequently be removed from the site.
Copies of industry codes and standards used in a

substantive manner in the NRC regulatory process are
maintained at:
The NRC Technical Library
Two White Flint North
11545 Rockville Pike
Rockville, MD 20852-2738
These standards are available for reference by the
public. Codes and standards are usually copyrighted
and may be purchased from their originating
organization, or, if they are American National
Standards, from:
American National Standards Institute
11 West 42nd Street
New York, NY 10036-8002
www.ansi.org
212-642-4900
Legally binding regulatory requirements are stated only
in laws; NRC regulations; licenses, including technical
specifications; or orders, not in NUREG-series
publications. The views expressed in contractorprepared publications in this series are not necessarily
those of the NRC.
The NUREG series comprises (1) technical and
administrative reports and books prepared by the staff
(NUREG-XXXX) or agency contractors
(NUREG/CR-XXXX), (2) proceedings of conferences
(NUREG/CP-XXXX), (3) reports resulting from
international agreements (NUREG/IA-XXXX), (4)
brochures (NUREG/BR-XXXX), and (5) compilations of
legal decisions and orders of the Commission and
Atomic and Safety Licensing Boards and of Directors
decisions under Section 2.206 of the NRCs regulations
(NUREG-0750).
Nuclear Power Plant

Fire Modeling Analysis
Guidelines
(NPP FIRE MAG)
NUREG-1934
EPRI 1023259
Final Report
November 2012

Office of Nuclear Regulatory Research (RES)
Electric Power Research Institute (EPRI)

Palo Alto, CA 94304-1338
U.S. NRC-RES Project Manager

M.H. Salley
EPRI Project Manager

R. Wachowiak
DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES

THIS DOCUMENT WAS PREPARED BY THE ORGANIZATION(S) NAMED BELOW AS AN ACCOUNT
OF WORK SPONSORED OR COSPONSORED BY THE ELECTRIC POWER RESEARCH INSTITUTE,
INC. (EPRI). NEITHER EPRI, ANY MEMBER OF EPRI, ANY COSPONSOR, THE ORGANIZATION(S)
BELOW, NOR ANY PERSON ACTING ON BEHALF OF ANY OF THEM:
(A) MAKES ANY WARRANTY OR REPRESENTATION WHATSOEVER, EXPRESS OR IMPLIED,
(I) WITH RESPECT TO THE USE OF ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR
SIMILAR ITEM DISCLOSED IN THIS DOCUMENT, INCLUDING MERCHANTABILITY AND FITNESS
FOR A PARTICULAR PURPOSE, OR (II) THAT SUCH USE DOES NOT INFRINGE ON OR INTERFERE
WITH PRIVATELY OWNED RIGHTS, INCLUDING ANY PARTYS INTELLECTUAL PROPERTY, OR (III)
THAT THIS DOCUMENT IS SUITABLE TO ANY PARTICULAR USERS CIRCUMSTANCE; OR
(B) ASSUMES RESPONSIBILITY FOR ANY DAMAGES OR OTHER LIABILITY WHATSOEVER
(INCLUDING ANY CONSEQUENTIAL DAMAGES, EVEN IF EPRI OR ANY EPRI REPRESENTATIVE
HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES) RESULTING FROM YOUR
SELECTION OR USE OF THIS DOCUMENT OR ANY INFORMATION, APPARATUS, METHOD,
PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT.
REFERENCE HEREIN TO ANY SPECIFIC COMMERCIAL PRODUCT, PROCESS, OR SERVICE BY
ITS TRADE NAME, TRADEMARK, MANUFACTURER, OR OTHERWISE, DOES NOT NECESSARILY
CONSTITUTE OR IMPLY ITS ENDORSEMENT, RECOMMENDATION, OR FAVORING BY EPRI.
THE FOLLOWING ORGANIZATIONS PREPARED THIS REPORT:
U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research
Hughes Associates, Inc.
National Institute of Standards and Technology
California Polytechnic State University
Westinghouse Electric Company
University of Maryland
Science Applications International Corporation
ERIN Engineering
NOTE
For further information about EPRI, call the EPRI Customer Assistance Center at 800.313.3774 or e-mail
[email protected].
Electric Power Research Institute, EPRI, and TOGETHERSHAPING THE FUTURE OF ELECTRICITY
are registered service marks of the Electric Power Research Institute, Inc.
Copyright 2012 Electric Power Research Institute. All rights reserved.
ABSTRACT
There is a movement to introduce risk-informed and performance-based (RI/PB) analyses into
fire protection engineering practice, both domestically and worldwide. This movement exists in
both the general fire protection and the nuclear power plant (NPP) fire protection communities.
The U.S. Nuclear Regulatory Commission (NRC) has used risk-informed insights as a part of its
regulatory decision making since the 1990s.
In 2001, the National Fire Protection Association (NFPA) issued NFPA 805, Performance-Based
Standard for Fire Protection for Light-Water Reactor Electric Generating Plants, 2001 Edition.
In July 2004, the NRC amended its fire protection requirements in Title 10, Section 50.48 of the
Code of Federal Regulations (10 CFR 50.48) to permit existing reactor licensees to voluntarily
adopt fire protection requirements contained in NFPA 805 as an alternative to the existing
deterministic fire protection requirements. In addition, the NPP fire protection community has
been using RI/PB approaches and insights to support fire hazard analysis in general.
One key element in RI/PB fire protection is the availability of verified and validated (V&V) fire
models that can reliably estimate the effects of fires. The U.S. NRC, together with the Electric
Power Research Institute (EPRI) and the National Institute of Standards and Technology
(NIST), conducted a research project to verify and validate five fire models that have been used
for NPP applications. The results of this effort are documented in a seven-volume NUREG
report, NUREG-1824 (EPRI 1011999), Verification & Validation of Selected Fire Models for
Nuclear Power Plant Applications.
This report describes the implications of the V&V results for fire model users. The features and
limitations of the five fire models documented in NUREG-1824 (EPRI 1011999) are discussed
relative to NPP fire hazard analysis (FHA). Finally, the report provides information on the use of
fire models in support of various commercial NPP fire hazard analysis applications.
iii
CONTENTS
ABSTRACT ................................................................................................................................. iii
CONTENTS .................................................................................................................................. v
FIGURES ..................................................................................................................................... xi
TABLES ...................................................................................................................................... xv
REPORT SUMMARY ............................................................................................................... xvii
PREFACE .................................................................................................................................. xix
CITATIONS................................................................................................................................ xxi
ACKNOWLEDGEMENTS ....................................................................................................... xxiii
LIST OF ACRONYMS .............................................................................................................. xxv
NOMENCLATURE ................................................................................................................. xxvii
1 INTRODUCTION .................................................................................................................... 1-1
1.1 Background ..................................................................................................................... 1-1
1.2 Objective ......................................................................................................................... 1-2
1.3 Scope .............................................................................................................................. 1-2
1.3.1 User Capabilities ..................................................................................................... 1-2
1.3.2 Training Resources ................................................................................................. 1-3
1.4 Fire Modeling Theory ...................................................................................................... 1-4
1.5 Fire Modeling Tools ......................................................................................................... 1-7
1.5.1 Algebraic Models ..................................................................................................... 1-7
1.5.2 Zone Models ............................................................................................................ 1-7
1.5.3 CFD Models............................................................................................................. 1-9
1.5.4 Fire Model Verification and Validation (V&V) ........................................................ 1-11
1.6 Fire Modeling Applications ............................................................................................ 1-11
1.6.1 License Amendments and Exemptions ................................................................. 1-11
1.6.2 Fire Induced Circuit Failures .................................................................................. 1-11
1.6.2.1 Deterministic Application ............................................................................. 1-12
1.6.2.2 Risk-Informed, Performance-Based Application.......................................... 1-12
1.6.3 NFPA 805 Performance-Based Applications ......................................................... 1-12
1.6.3.1 Fire Modeling ............................................................................................... 1-13
1.6.3.2 Fire Risk Evaluations ................................................................................... 1-14
1.6.4 Fire Modeling in Support of Fire PRA .................................................................... 1-14
1.7 Organization of the Guide ............................................................................................. 1-15
2 THE FIRE MODELING PROCESS ......................................................................................... 2-1
2.1 Step 1: Define Fire Modeling Goals and Objectives........................................................ 2-2
2.2 Step 2: Characterize Fire Scenarios ............................................................................... 2-4
2.2.1 General Considerations ........................................................................................... 2-4
2.2.2 Enclosure Details..................................................................................................... 2-5
2.2.3 Fire Location ............................................................................................................ 2-6
2.2.4 Credited Fire Protection........................................................................................... 2-7
2.2.5 Ventilation Conditions .............................................................................................. 2-8
2.2.6 Target Locations ...................................................................................................... 2-8
2.2.7 Secondary Combustibles ......................................................................................... 2-9
2.2.8 Source Fire .............................................................................................................. 2-9
2.3 Step 3: Select Fire Models ............................................................................................ 2-10
2.3.1 Fire Dynamics Tools (FDTs) .................................................................................. 2-14
2.3.2 Fire Induced Vulnerability Evaluation (FIVE-Rev1) ............................................... 2-16
2.3.3 Consolidated Fire Growth and Smoke Transport (CFAST) Model ........................ 2-18
2.3.4 MAGIC ................................................................................................................... 2-19
2.3.5 Fire Dynamics Simulator (FDS) ............................................................................. 2-19
2.3.6 Verification and Validation (V&V) .......................................................................... 2-22
2.3.7 Fire Modeling Parameters Outside the Validation Range ..................................... 2-28
2.3.7.1 Sensitivity Analysis ...................................................................................... 2-29
2.3.7.2 Additional Validation Studies ....................................................................... 2-31
2.4 Step 4: Calculate Fire-Generated Conditions................................................................ 2-32
2.5 Step 5: Conduct Sensitivity and Uncertainty Analyses.................................................. 2-33
2.6 Step 6: Document the Analysis ..................................................................................... 2-34
2.7 Summary ....................................................................................................................... 2-35
3 GUIDANCE ON FIRE MODEL SELECTION AND IMPLEMENTATION................................ 3-1
3.1 Model Implementation of Fire Scenario Elements........................................................... 3-1
3.1.1 Heat Release Rate .................................................................................................. 3-1
3.1.2 Plant Area Configuration ......................................................................................... 3-2
3.1.3 Ventilation Effects .................................................................................................... 3-3
3.1.4 Targets .................................................................................................................... 3-4
3.1.5 Intervening Combustibles ........................................................................................ 3-5
3.2 Guidance on Model Selection and Analysis .................................................................... 3-5
3.2.1 Targets in the Flames or Plume .............................................................................. 3-8
3.2.1.1 General Objective .......................................................................................... 3-8
3.2.1.2 Modeling Strategy.......................................................................................... 3-8
3.2.1.3 Recommended Models.................................................................................. 3-9
3.2.1.4 Detailed Examples....................................................................................... 3-10
3.2.2 Scenario 2: Targets Inside or Outside the Hot Gas Layer ..................................... 3-11
3.2.2.1 General Objective ........................................................................................ 3-11
3.2.2.2 Modeling Strategy........................................................................................ 3-11
3.2.2.3 Recommended Modeling Tools ................................................................... 3-12
3.2.3 Scenario 3: Targets Located in Adjacent Rooms .................................................. 3-14
3.2.4 Scenario 4: Targets in Rooms with Complex Geometries ..................................... 3-16
3.2.5 Scenario 5: Main Control Room Abandonment ..................................................... 3-18
3.2.6 Scenario 6: Smoke Detection and Sprinkler Activation ......................................... 3-20
3.2.7 Scenario 7: Fire Impacting Structural Elements .................................................... 3-22
vi

4 UNCERTAINTY AND SENSITIVITY ...................................................................................... 4-1
4.1 Validation of Fire Models ................................................................................................. 4-2
4.2 Derivation of the Model Uncertainty Statistics ................................................................. 4-6
4.3 Calculation of Model Uncertainty..................................................................................... 4-8
4.3.1 Example 1: Target Temperature .............................................................................. 4-8
4.3.2 Example 2: Critical Heat Flux .................................................................................. 4-9
4.4 Parameter Uncertainty .................................................................................................. 4-10
4.4.1 Parameter Uncertainty Propagation ...................................................................... 4-10
4.4.2 Sensitivity Analysis ................................................................................................ 4-13
4.5 Chapter Summary ......................................................................................................... 4-17
5 REFERENCES ....................................................................................................................... 5-1
6 INTRODUCTION TO THE APPENDICES .............................................................................. 6-1
APPENDIX A CABINET FIRE IN MAIN CONTROL ROOM .................................................... A-1
A.1 Modeling Objective ........................................................................................................ A-1
A.2 Description of the Fire Scenario..................................................................................... A-1
A.3 Selection and Evaluation of Fire Models........................................................................ A-6
A.3.1 Temperature Criterion ............................................................................................ A-6
A.3.2 Heat Flux Criterion ................................................................................................. A-6
A.3.3 Visibility Criterion.................................................................................................... A-6
A.3.4 Validation ............................................................................................................... A-7
A.4 Estimation of Fire-Generated Conditions ..................................................................... A-10
A.4.1 Algebraic Models .................................................................................................. A-10
A.4.2 Zone Model .......................................................................................................... A-13
A.4.3 CFD Model ........................................................................................................... A-16
A.5 Evaluation of Results ................................................................................................... A-17
A.5.1 Temperature Criterion .......................................................................................... A-18
A.5.2 Heat Flux Criterion ............................................................................................... A-19
A.5.3 Visibility Criterion.................................................................................................. A-20
A.6 Conclusion ................................................................................................................... A-21
A.7 References................................................................................................................... A-22
A.8 Attachments (on CD) ................................................................................................... A-22
APPENDIX B CABINET FIRE IN SWITCHGEAR ROOM........................................................ B-1
B.1 Modeling Objective ........................................................................................................ B-1
B.2 Description of the Fire Scenario..................................................................................... B-1
B.3 Selection and Evaluation of Fire Models........................................................................ B-6
B.3.1 Temperature Criterion ............................................................................................ B-6
B.3.2 Heat Flux Criterion ................................................................................................. B-6
B.3.3 Validation ............................................................................................................... B-6
B.4 Estimation of Fire-Generated Conditions ....................................................................... B-9
B.4.1 Algebraic Models .................................................................................................... B-9
B.4.2 Zone Model .......................................................................................................... B-12
B.4.3 CFD Model ........................................................................................................... B-16
B.5 Evaluation of Results ................................................................................................... B-19
B.5.1 Cable Ignition and Damage .................................................................................. B-19
B.5.2 Cabinet Damage .................................................................................................. B-21
B.5.3 Parameter Uncertainty Propagation ..................................................................... B-22
vii
B.6 Conclusion ................................................................................................................... B-24

B.7 References................................................................................................................... B-24
B.8 Attachments (on CD) ................................................................................................... B-24
APPENDIX C LUBRICATING OIL FIRE IN PUMP COMPARTMENT ..................................... C-1
C.1 Modeling Objective ........................................................................................................ C-1
C.2 Description of the Fire Scenario .................................................................................... C-1
C.3 Selection and Evaluation of Models............................................................................... C-5
C.3.1 Fire Sustainability................................................................................................... C-5
C.3.2 Temperature Criterion ............................................................................................ C-5
C.3.3 Validation ............................................................................................................... C-5
C.4 Estimation of Fire-Generated Conditions....................................................................... C-7
C.4.1 Calculation of Oxygen Availability .......................................................................... C-7
C.4.2 Zone Model ............................................................................................................ C-8
C.4.3 CFD Model ........................................................................................................... C-14
C.5 Evaluation of Results ................................................................................................... C-16
C.5.1 The Fire................................................................................................................ C-16
C.5.2 Temperature Criterion .......................................................................................... C-17
C.6 Conclusion ................................................................................................................... C-23
C.7 References .................................................................................................................. C-23
C.8 Attachments (on CD) ................................................................................................... C-24
APPENDIX D MOTOR CONTROL CENTER FIRE IN A SWITCHGEAR ROOM .................... D-1
D.1 Modeling Objective ........................................................................................................ D-1
D.2 Description of the Fire Scenario .................................................................................... D-1
D.3 Selection and Evaluation of Fire Models ....................................................................... D-5
D.3.1 Temperature Criterion ............................................................................................ D-5
D.3.2 Heat Flux Criterion ................................................................................................. D-5
D.3.3 Validation ............................................................................................................... D-5
D.4 Estimation of Fire-Generated Conditions....................................................................... D-8
D.4.1 Algebraic Models ................................................................................................... D-8
D.4.2 Zone Models .......................................................................................................... D-8
D.4.3 CFD Model ........................................................................................................... D-12
D.5 Evaluation of Results ................................................................................................... D-14
D.5.1 Damage to Cabinet .............................................................................................. D-14
D.5.2 Cable Damage Based on Temperature Alone ..................................................... D-15
D.5.3 Cable Damage Based on Incident Heat Flux ....................................................... D-16
D.6 Conclusion ................................................................................................................... D-17
D.7 References .................................................................................................................. D-17
D.8 Attachments (on CD) ................................................................................................... D-18
APPENDIX E TRANSIENT FIRE IN CABLE SPREADING ROOM ......................................... E-1
E.1 Modeling Objective ........................................................................................................ E-1
E.2 Description of the Fire Scenario..................................................................................... E-1
E.3 Selection and Evaluation of Fire Models........................................................................ E-6
E.3.1 Temperature Criterion ............................................................................................ E-6
E.3.2 Heat Flux Criterion ................................................................................................. E-6
E.3.3 Validation ............................................................................................................... E-6
E.4 Estimation of Fire-Generated Conditions ....................................................................... E-9
E.4.1 Algebraic Models .................................................................................................... E-9
viii
E.4.2 Zone Model .......................................................................................................... E-11

E.4.3 CFD Model ........................................................................................................... E-17
E.5 Evaluation of Results ................................................................................................... E-19
E.5.1 Smoke Detection .................................................................................................. E-20
E.5.2 Temperature Criterion .......................................................................................... E-20
E.5.3 Heat Flux Criterion ............................................................................................... E-21
E.5.4 Parameter Uncertainty Propagation ..................................................................... E-23
E.6 Conclusion ................................................................................................................... E-24
E.7 References................................................................................................................... E-25
E.8 Attachments (on CD) ................................................................................................... E-25
APPENDIX F LUBRICATING OIL FIRE IN A TURBINE BUILDING ........................................F-1
F.1 Modeling Objective......................................................................................................... F-1
F.2 Description of the Fire Scenario ..................................................................................... F-1
F.3 Selection and Evaluation of Fire Models ........................................................................ F-6
F.3.1 Heat Flux and Target Temperature Algorithms ...................................................... F-6
F.3.2 Validation................................................................................................................ F-6
F.4 Estimation of Fire-Generated Conditions ....................................................................... F-9
F.4.1 Algebraic Models .................................................................................................... F-9
F.4.2 CFD Model ........................................................................................................... F-13
F.5 Evaluation of Results ................................................................................................... F-14
F.5.1 Column Heat Flux and Column Temperature ....................................................... F-14
F.6 Conclusion ................................................................................................................... F-17
F.7 References ................................................................................................................... F-17
F.8 Attachments (on CD).................................................................................................... F-18
APPENDIX G TRANSIENT FIRE IN A MULTI-COMPARTMENT CORRIDOR ....................... G-1
G.1 Modeling Objective ........................................................................................................ G-1
G.2 Description of the Fire Scenario .................................................................................... G-1
G.3 Selection and Evaluation of Models .............................................................................. G-5
G.3.1 Temperature Criterion............................................................................................ G-5
G.3.2 Validation ............................................................................................................... G-5
G.4 Estimation of Fire-Generated Conditions ...................................................................... G-7
G.4.1 Algebraic Models ................................................................................................... G-7
G.4.2 Zone Model ............................................................................................................ G-7
G.5 Evaluation of Results ................................................................................................... G-13
G.5.1 Temperature Criterion.......................................................................................... G-13
G.5.2 Smoke Detection ................................................................................................. G-15
G.6 Conclusion ................................................................................................................... G-16
G.7 References .................................................................................................................. G-17
G.8 Attachments (on CD) ................................................................................................... G-17
APPENDIX H CABLE TRAY FIRE IN ANNULUS.................................................................... H-1
H.1 Modeling Objective ........................................................................................................ H-1
H.2 Description of the Fire Scenario .................................................................................... H-1
H.3 Selection and Evaluation of Fire Models ....................................................................... H-5
H.3.1 Damage to Cables ................................................................................................. H-5
H.3.2 Sprinkler Activation ................................................................................................ H-6
H.3.3 Validation ............................................................................................................... H-6
H.4 Estimation of Fire-Generated Conditions....................................................................... H-7
ix
H.4.1 Algebraic Models ................................................................................................... H-7

H.4.2 CFD Model ............................................................................................................. H-8
H.5 Evaluation of Results ................................................................................................... H-11
H.5.1 Heat Flux and Temperature ................................................................................. H-11
H.5.2 Fire Protection Systems ....................................................................................... H-12
H.6 Conclusion ................................................................................................................... H-13
H.7 References .................................................................................................................. H-13
H.8 Attachments (on CD) ................................................................................................... H-13
INDEX ................................................................................................................................ Index-1
FIGURES
Figure 1-1.
Figure 1-2.
Figure 1-3.
Figure 2-1.
Figure 3-1.
Figure 3-2.
Figure 3-3.
Figure 3-4.
Figure 3-5.
Figure 3-6.
Figure 3-7.
Figure 3-8.
Figure 3-9.
Figure 4-1.
Figure 4-2.
Figure 4-3.
Figure 4-4.
Figure 4-5.
Figure 4-6.
Figure A-1.
Figure A-2.
Figure A-3.
Figure A-4.
Figure A-5.
Figure A-6.
Figure A-7.
Figure A-8.
Figure A-9.
Figure A-10.
Figure A-11.
Figure A-12.
Figure A-13.
Figure A-14.
Figure B-1.
Figure B-2.
Figure B-3.
Characteristics of compartment fires.................................................................... 1-5

A two-zone enclosure fire with an HGL above and a cool lower layer
below. ................................................................................................................... 1-8
A Smokeview visualization of a CFD model of a compartment fire
experiment. ........................................................................................................ 1-10
Fire modeling process. ......................................................................................... 2-1
Pictorial representation of the fire scenario and corresponding technical
elements described in this section. ...................................................................... 3-6
Legend for fire modeling sketches presented in this chapter. .............................. 3-7
Pictorial representation of scenario 1. .................................................................. 3-8
Pictorial representation of scenario 2. ................................................................ 3-11
Sample set of results from NUREG-1824 (EPRI 1011999). ................................ 4-4
Two examples demonstrating how the validation data is tested for
normality............................................................................................................... 4-8
Normal distribution of the true value of the cable temperature in a
hypothetical fire. ................................................................................................... 4-9
Distribution of HRR for an electrical cabinet fire. ............................................... 4-12
Distribution of flame heights for the entire range of cabinet fires. ...................... 4-13
FDS predictions of HGL Temperature as a function of time due to a
1,000 kW fire (solid line) and a 1,150 kW fire (dashed). .................................... 4-15
Geometry of the main control room..................................................................... A-2
Main control room details. ................................................................................... A-3
Photograph of a typical open-grate ceiling. ......................................................... A-4
Photograph of a typical control cabinet. .............................................................. A-4
Time history of the HRR used by all models in the MCR scenario. .................... A-5
Schematic diagram of the FPA calculation for the MCR smoke purge
scenario............................................................................................................. A-11
Compartment geometry and surface material selection in CFAST for the
MCR Fire. .......................................................................................................... A-13
Specification of the fire in CFAST for the MCR Fire.......................................... A-14
Mechanical ventilation inputs in CFAST for the MCR fire. ................................ A-15
Smokeview rendering of the CFAST simulation of the MCR fire with
mechanical ventilation. ...................................................................................... A-15
FDS/Smokeview rendering of the MCR, as viewed from above. ...................... A-17
Hot Gas Layer Temperature and Height for the MCR scenario. ....................... A-19
Predicted heat flux at the location of the operator............................................. A-20
Optical density predictions for the MCR scenario. ............................................ A-21
Geometry of the switchgear room. ...................................................................... B-5
Schematic diagram of cabinet fire in switchgear room........................................ B-9
Plume temperatures at cable trays located above a cabinet fire. ..................... B-10
xi
Figure B-4.
Figure B-5.
Figure B-6.
Figure B-7.
Figure B-8.
Figure B-9.
Figure B-10.
Figure B-11.
Figure B-12.
Figure B-13.
Figure B-14.
Figure B-15.
Figure B-16.
Figure C-1.
Figure C-2.
Figure C-3.
Figure C-4.
Figure C-5.
Figure C-6.
Figure C-7.
Figure C-8.
Figure C-9.
Figure C-10.
Figure C-11.
Figure C-12.
Figure C-13.
Figure C-14.
Figure C-15.
Figure C-16.
Figure D-1.
Figure D-2.
Figure D-3.
Figure D-4.
Figure D-5.
Figure D-6.
Figure D-7.
Figure D-8.
Figure D-9.
Figure D-10.
Average HGL temperature (red line) from FPA correlation and HRR (blue
line) for SWGR cabinet fire scenario. ................................................................ B-12
Average CFAST/Smokeview rendering of SWGR. ........................................... B-12
CFAST inputs for compartment geometry in SWGR scenario. ......................... B-13
CFAST fire specification inputs for the SWGR scenario. .................................. B-14
CFAST mechanical ventilation inputs for the SWGR scenario. ........................ B-15
CFAST inputs for cabinet and cable targets for the SWGR scenario. .............. B-16
FDS/Smokeview rendering of the SWGR. ........................................................ B-18
FDS/Smokeview rendering of the SWGR fire showing localized ignition of
extinction of secondary cable fires resulting from initial cabinet fire. ................ B-18
Heat release rate inputs to CFAST and FDS for a SWGR cabinet fire
scenario............................................................................................................. B-20
Estimated temperatures for Cable Tray A directly above the fire source for
a SWGR cabinet fire scenario. .......................................................................... B-21
Estimated temperature and heat flux to a cabinet adjacent to the fire
source in a SWGR cabinet fire scenario. .......................................................... B-22
Distribution of HRR for an electrical cabinet fire. .............................................. B-23
Distribution of flame heights for the entire range of cabinet fires. ..................... B-23
Geometry of the pump room. .............................................................................. C-2
MAGIC view of the pump room. .......................................................................... C-9
Heat release rate curve for lubricating oil fire.................................................... C-10
MAGIC fire input screen for pump fire............................................................... C-11
Modeling multi-conductor cables in MAGIC. Source: NUREG-1824
(EPRI 1011999), Volume 6, Figure 3-3. ............................................................ C-12
Representation of the ERFBS protected cable raceway for MAGIC. ................ C-13
MAGIC input screen for ERFBS layers. ............................................................ C-13
MAGIC input screen for ventilation through door. ............................................. C-14
FDS/Smokeview rendering of the pump room scenario at the early stage of
the fire, before the compartment becomes under-ventilated............................. C-15
HRR predicted by algebraic methods, MAGIC, and FDS for the pump room
fire scenario....................................................................................................... C-17
HGL Temperature Predicted by MAGIC and FDS for the pump room fire
scenario............................................................................................................. C-17
Cable surface temperature predicted by MAGIC and FDS for the pump
room fire scenario. ............................................................................................ C-19
Cable surface temperature predicted by MAGIC for changes to insulation
wrapping............................................................................................................ C-20
Temperature predicted by MAGIC for increased door size. .............................. C-21
HRR for base case and HRR sensitivity case. .................................................. C-22
Temperature for base case and HRR sensitivity case. ..................................... C-22
Geometry of the MCC/SWGR in a BWR. ............................................................ D-3
Typical electrical cabinet in the lower part of the SWGR. ................................... D-4
View of the high ceiling space. ............................................................................ D-4
Smokeview rendering of the geometry of two-height ceiling SWGR, as
modeled in CFAST. ............................................................................................. D-9
CFAST inputs for compartment geometry for SWGR. ...................................... D-10
CFAST fire inputs for two-height ceiling SWGR scenario. ................................ D-10
CFAST target inputs for two-height ceiling SWGR scenario. ............................ D-11
CFAST inputs for vents connections in two-height ceiling SWGR scenario. .... D-12
FDS/Smokeview representation of the MCC/SWGR scenario. ......................... D-13
Heat flux and temperature predictions for the adjacent cabinet. ....................... D-15
xii
Figure D-11. Summary of the cable temperature and heat flux predictions for the
MCC/SWGR. ..................................................................................................... D-16
Figure E-1. Geometry of the CSR. ......................................................................................... E-4
Figure E-2. Photograph of the CSR used for analysis. Note that the cables are located
in the trays in the overhead. ................................................................................ E-5
Figure E-3. Geometric detail of the CSR. .............................................................................. E-5
Figure E-4. Schematic diagram of transient trash fire in CSR (not to scale). ........................ E-9
Figure E-5. Plume temperatures at cable trays located above a transient trash fire. .......... E-10
Figure E-6. Average HGL temperature (red line) from FPA correlation and heat release
rate (blue line) for the CSR trash fire scenario. ................................................. E-11
Figure E-7. CFAST/Smokeview rendering of the CSR scenario.......................................... E-12
Figure E-8. CFAST inputs for compartment geometry for the CSR scenario. ..................... E-13
Figure E-9. CFAST inputs for the fire in the CSR scenario.................................................. E-14
Figure E-10. CFAST inputs for cable targets above the fire source in the CSR scenario. .... E-15
Figure E-11. CFAST inputs for natural ventilation for the CSR scenario. Note the
opening of the doorway at 170 s, the time estimated for smoke detectors to
activate. ............................................................................................................. E-16
Figure E-12. CFAST inputs for smoke detectors in the CSR scenario. ................................. E-17
Figure E-13. FDS/Smokeview rendering of the CSR scenario. ............................................. E-17
Figure E-14. HRR and estimated HGL temperature for CSR scenario. ................................. E-19
Figure E-15. Estimated cable conditions for the CSR............................................................ E-22
Figure E-16. Distribution of HRR for a trash fire. ................................................................... E-23
Figure E-17. Distribution of plume temperatures at Trays 3 and 6, respectively. .................. E-24
Figure F-1. Geometry of the turbine building. .........................................................................F-2
Figure F-2. Main turbine lubricating oil tanks in the turbine building.......................................F-3
Figure F-3. Typical steel column in the turbine building. ........................................................F-3
Figure F-4. Schematic diagram of the fire impinging on the ceiling. .....................................F-10
Figure F-5. Detail from Figure F-1 with estimated flame extension beneath ceiling
superimposed.....................................................................................................F-11
Figure F-6. FDS geometry for the turbine building fire scenario. ..........................................F-13
Figure F-7. Temperatures of columns for Curb Location 1. ..................................................F-15
Figure F-8. Temperatures of columns for Curb Location 2. ..................................................F-16
Figure G-1. Geometry of the multi-compartment corridor. ..................................................... G-2
Figure G-2. Geometry details of the multi-compartment corridor. .......................................... G-3
Figure G-3. HRR for the corridor fire scenario. ...................................................................... G-4
Figure G-4. Effective corridor layout for implementation in zone models (not to scale). ........ G-8
Figure G-5. MAGIC rendering of the corridor scenario. ......................................................... G-9
Figure G-6. MAGIC screen capture of the fire in corridor scenario. ..................................... G-10
Figure G-7. MAGIC screen capture for supply vent specification. ....................................... G-11
Figure G-8. MAGIC screen capture of the smoke detector specification. ............................ G-12
Figure G-9. HGL temperature predictions by MAGIC for the corridor scenario. .................. G-14
Figure G-10. HGL temperature for reduced ceiling height by MAGIC. ................................... G-15
Figure G-11. Detector temperature prediction by MAGIC for fire corridor.............................. G-15
Figure G-12. Detector temperature for two radial distances predicted by MAGIC. ................ G-16
Figure H-1. Geometry of the annulus. .................................................................................... H-3
Figure H-2. Geometry details of redundant cable trays located in the annulus. .................... H-4
Figure H-3. HRR for a cable fire in the annulus. .................................................................... H-5
Figure H-4. FDS/Smokeview rendering of the annulus fire scenario. .................................. H-10
Figure H-5. Summary of simulation results for the annulus. ................................................ H-12
Figure H-6. Predicted sprinkler link temperature for the annulus fire scenario. ................... H-12
xiii
TABLES
Table 2-1.
Table 2-2.
Table 2-3.
Table 2-4.
Table 2-5.
Table 3-1.
Table 3-2.
Table 4-1.
Table 4-2.
Table 4-3.
Table 6-1.
Table A-1.
Table A-2.
Table A-3.
Table A-4.
Table B-1.
Table B-2.
Table B-3.
Table B-4.
Table B-5.
Table C-1.
Table C-2.
Table C-3.
Table C-4.
Table C-5.
Table C-6.
Table D-1.
Table D-2.
Table D-3.
Table E-1.
Table E-2.
Table E-3.
Table E-4.
Table E-5.
Table E-6.
Table F-1.
Table F-2.
Table F-3.
Summary of common fire model tools................................................................ 2-13

Routines included in the FDTs............................................................................ 2-14
Routines included in FIVE-Rev1. ....................................................................... 2-17
Fire modeling attributes included in NUREG 1824/EPRI 1011999 (2007). ........ 2-23
Summary of selected normalized parameters for application of the
validation results to NPP fire scenarios (NUREG-1824/EPRI 1011999,
2007). ................................................................................................................. 2-24
Material properties. .............................................................................................. 3-3
Listing of generic scenarios described in this chapter.......................................... 3-5
Results of the V&V study, NUREG-1824 (EPRI 1011999). ................................. 4-2
Experimental uncertainty of the experiments performed as part of the
validation study in NUREG-1824 (EPRI 1011999)............................................... 4-7
Sensitivity of model outputs from Volume 2 of NUREG-1824 (EPRI
1011999). ........................................................................................................... 4-14
Appendix fire scenario descriptions and cross-index. .......................................... 6-1
Data for MCR fire based on XPE/neoprene electrical cable. .............................. A-5
Normalized parameter calculations for the MCR fire scenario. See Table
2-5 for further details. .......................................................................................... A-9
Summary of input parameters for the FPA calculation of the MCR. ................. A-10
Summary of the model predictions of the MCR scenario. ................................. A-18
Products of combustion for switchgear room cabinet and cable fire. .................. B-4
Normalized parameters and their ranges of applicability to NUREG-1824
(EPRI 1011999). ................................................................................................. B-8
Summary of input parameters for FPA analysis of switchgear room
scenario............................................................................................................. B-11
Summary of the model predictions of the cabinet fire scenario. ....................... B-19
Estimated time to ignition of lowest cable tray by CFAST for the SWGR
cabinet fire......................................................................................................... B-20
Data for ERFBS and cable insulation.................................................................. C-3
Data for lubricating oil fire. .................................................................................. C-4
Normalized parameter calculations for the pump room fire scenario. ................. C-6
Calculated input for lubricating oil fire. ................................................................ C-8
Summary of the model predictions of the pump room scenario. ....................... C-16
Integrated thermal exposure of an object subjected to the ASTM E119
temperature curve. ............................................................................................ C-19
Products of combustion for the MCC fire. ........................................................... D-2
Normalized parameter calculations for the MCC fire scenario. ........................... D-7
Summary of the model predictions of the MCC fire scenario. ........................... D-14
Products of combustion for CSR fire. .................................................................. E-3
Normalized parameters and their ranges of applicability to NUREG-1824. ........ E-8
Summary of input parameters for FPA analysis of CSR scenario. ................... E-10
Cross-sectional area as a function of height used for CFAST calculation. ....... E-12
Summary of the model predictions for the CSR scenario. ................................ E-20
Smoke detector activation times, CSR.............................................................. E-20
Structural steel failure criteria (ASTM E119-10a).................................................F-4
Data for lubricating oil fire. ...................................................................................F-5
Normalized Parameter Calculations for the Turbine Building Fire Scenario. .......F-8
xv
Table F-4.
Table G-1.
Table G-2.
Table G-3.
Table G-4.
Table H-1.
Table H-2.
Table H-3.
Summary of results for the turbine building fire scenarios. ................................F-14

Products of combustion for a wood pallet fire. .................................................... G-4
Normalized parameter calculations for the multi-compartment corridor fire
scenario............................................................................................................... G-6
Compartment dimensions for corridor scenario. ................................................. G-8
Summary of the model predictions of the corridor scenario. ............................. G-13
Products of combustion for a PE/PVC cable fire................................................. H-2
Normalized parameter calculations for the annulus fire scenario. ...................... H-7
Summary of model predictions for the annulus fire scenario. ........................... H-11
xvi
REPORT SUMMARY
Background
Beginning in the 1990s, when the U.S. Nuclear Regulatory Commission (NRC) adopted the
policy of using risk-informed methods to make regulatory decisions whenever possible, the
nuclear power industry has been moving from prescriptive rules and practices toward the use of
risk information to supplement decision making. Several initiatives have furthered this transition
within the fire protection field, including risk-informed, performance-based fire protection
programs (FPPs) compliant with Title 10, Section 50.48(c) of the Code of Federal Regulations
(10 CFR 50.48(c)) and FPP change evaluation under the existing Title 10 Section 50.48 and
Regulatory Guide 1.189, Fire Protection for Nuclear Power Plants, October 2009. RI/PB fire
protection often relies on fire modeling to estimate the effects of fires.
Objectives
To provide guidance on the proper application of fire models to NPP fire scenarios
To fulfill the need as a teaching tool and support the NRC/EPRI Fire PRA training
Approach
There were five different fire models evaluated in NUREG-1824 (EPRI 1011999), Verification
and Validation of Selected Fire Models for Nuclear Power Plant Applications: (1) the NRCs Fire
Dynamics Tools (FDTs), (2) the Electric Power Research Institutes (EPRI) Fire-Induced
Vulnerability Evaluation Revision 1 (FIVE-Rev1), (3) the National Institute of Standards and
Technologys (NIST) Consolidated Model of Fire Growth and Smoke Transport (CFAST),
(4) Electricit de Frances (EdF) MAGIC, and (5) NISTs Fire Dynamics Simulator (FDS). To
support consistency, the same fire models were used in the development of this report. The
project team developed guidance on the selection and application of each model and treatment
of uncertainty and/or sensitivity as part of the fire modeling analysis. Based on this guidance,
the project team selected appropriate models and conducted fire modeling analyses of eight
different fire scenarios of interest in nuclear power plants (NPPs). The results of each analysis
were to be documented in a format appropriate for fire model calculation submittals.
Results
This report presents a step-by-step process for using fire modeling in NPP applications. The
recommended methodology consists of a six step process: (1) define fire modeling goals, (2)
characterize fire scenarios, (3) select fire models, (4) calculate fire-generated conditions, (5)
conduct sensitivity and uncertainty analyses, and (6) document the results.
This report is designed to assist fire model users in applying this technology in the NPP
environment. There are a number of unique construction and fire hazard attributes associated
with NPPs. It was the authors goal to explore and demonstrate the use of different models for
this application. In addition, a fifth module, Advanced Fire Modeling, has been added to the
annual NRC/EPRI Fire PRA training workshop in 2011. This report expands upon the
information provided in NUREG/CR-6850 (EPRI 1011989), EPRI/NRC-RES Fire PRA
Methodology for Nuclear Power Facilities, and will serve as the training material for the
Advanced Fire Modeling module.
xvii
The report contains a CD to assist the reader in reproducing the examples discussed in the
appendices. The CD contains the installation files for CFAST and FDS. It also contains the
latest versions (November 2012) of the NUREG-1805 spreadsheets. The CD also contains the
input files for each model used to analyze the eight fire scenarios.
EPRI Perspective
The use of fire models requires a good understanding of their limitations and predictive
capabilities, and also presents challenges that should be addressed if the fire protection
community is to realize the full benefit of fire modeling and performance-based fire protection.
EPRI, in partnership with NRC under a Memorandum of Understanding (MOU), will continue to
provide training to the fire protection community, using this document to promote fire modeling
and gain feedback on how the results of this work may affect known applications of fire
modeling. In the long term, model improvement and additional experiments should be
considered.
This report supersedes EPRI 10002981, Fire Modeling Guide for Nuclear Power Plant
Applications, August 2002, as guidance for fire modeling practitioners in NPPs. The report has
benefited from the insights gained since 2002 on the predictive capability of selected fire models
in improving confidence in the use of fire modeling in NPP applications.
Note: Due to EPRI limitations on distribution of software, the FDTs spreadsheets, CFAST, and
FDS are not included on the EPRI version of the CD. The FDTs can be obtained from the NRC
web site (www.nrc.gov), and CFAST and FDS can be obtained from the NIST web site
(www.nist.gov).
Keywords
Performance-Based
Fire Hazard Analysis (FHA)
Fire Protection
Probabilistic Risk Assessment (PRA)
Fire
Verification and Validation (V&V)
Risk-Informed Regulation
Fire Safety
Nuclear Power Plant (NPP)
Fire Modeling
xviii
PREFACE
This report describes research sponsored jointly by EPRI and U.S. Nuclear Regulatory
Commission, Office of Nuclear Regulatory Research (RES). The main purpose of this report is
to provide guidance on the application of fire models to nuclear power plant (NPP) fire scenarios
and to serve as a teaching tool and support the Advanced Fire Modeling module of the NRCRES/EPRI Fire PRA Course. The fire modeling analyses presented in this report represent the
combined efforts of individuals from RES and EPRI. Both organizations provided specialists in
the use of fire models/fire hazard analysis tools to support this work. These results are intended
to provide technical analysis of the predictive capabilities of five fire modeling calculation tools.
This report is the fifth in a series designed to assist those responsible for performing fire
modeling in NPP applications.
In August 2002, EPRI published EPRI 1002981, Fire Modeling Guide for Nuclear Power Plant
Applications. This report offered step-by-step guidance that analysts could follow when using
fire modeling to support nuclear power plant fire protection applications. It also included FIVERev1, an Excel-based library of fire models previously documented by EPRI, and additional
models from fire protection literature.
In December 2004, the NRC published NUREG-1805, Fire Dynamics Tools (FDTs) Quantitative
Fire Hazard Analysis Methods for the U.S. Nuclear Regulatory Commission Fire Protection
Inspection Program. This report provided an introduction to the principles of fire dynamics, and
included an Excel-based library of fire models comparable to EPRI FIVE-Rev1.
In a follow-up effort as a part of the NRC/RES-EPRI Memorandum of Understanding (MOU),
NRC/RES and EPRI jointly conducted a verification and validation of selected fire models for
use in nuclear power plant fire modeling to gain insight into the predictive capabilities of these
models. The results of this work were published in NUREG-1824 (EPRI 1011999), Verification
and Validation of Selected Fire Models for Nuclear Power Plant Applications, May 2007. Using,
in part, the findings of this work, the NRC conducted a Phenomena Identification and Ranking
Table (PIRT) study to evaluate the current state of knowledge for fire modeling for NPP
applications. The results of this work were published in NUREG/CR-6978, A Phenomena
Identification and Ranking Table (PIRT) Exercise for Nuclear Power Plant Fire Modeling
Applications, November 2008.
This document does not constitute regulatory requirements. RES participation in this
study does not constitute or imply regulatory approval of applications based upon this
methodology.
xix
CITATIONS
This report was prepared by:
U.S. Nuclear Regulatory Commission,
Office of Nuclear Regulatory Research (RES)
Principal Investigator:
D. Stroup
Electric Power Research Institute (EPRI)

Palo Alto, CA 94303
R. Wachowiak
National Institute of Standards and

Technology
Principal Investigators:
K. McGrattan
R. Peacock
Kleinsorg Group Risk Services

Division of Hughes Associates, Inc.
F. Joglar
S. LeStrange
B. Najafi
University of Maryland
J. Milke
California Polytechnic State University

F. Mowrer
Hughes Associates, Inc.
S. Hunt
Westinghouse Electric Company
C. Worrell
Science Applications International Corp (SAIC)
Principal Investigators:
D. Birk
ERIN Engineering
K. Zee
This report describes research sponsored jointly by the U.S. Nuclear Regulatory Commissions
(NRC) Office of Nuclear Regulatory Research (RES) and the Electric Power Research Institute
(EPRI).
The report is a corporate document that should be cited in the literature in the following manner:
Nuclear Power Plant Fire Modeling Analysis Guidelines (NPP FIRE MAG), U.S. Nuclear
Regulatory Commission, Office of Nuclear Regulatory Research (RES), Washington, D.C.,
2012, and Electric Power Research Institute (EPRI), Palo Alto, CA, NUREG-1934 and EPRI
1023259.
xxi
ACKNOWLEDGEMENTS
This report is the result of a multi-year effort by the authors, during which it has been
significantly rewritten in response to comments from various reviewers. The authors wish to
thank those reviewers for taking the time to provide their thoughts on this document. The
authors express appreciation to the members of the peer review panel who provided comments
on the original draft of this document in 2009: Professor Jose L. Torero and his students at BRE
Centre for Fire Safety Engineering, The University of Edinburgh, Scotland; Professor Frederick
Mowrer, formerly of the Department of Fire Protection Engineering, University of Maryland, and
currently with the Department of Fire Protection Engineering, California Polytechnic State
University; Mr. Patrick Finney, NRC Resident Inspector at Susquehanna Nuclear Plant;
Mr. Naeem Iqbal, Fire Protection Engineer, Office of Nuclear Reactor Regulation (NRR), NRC;
and Mr. Thomas Gorman, Project Manager, Pennsylvania Power and Light, Susquehanna
Nuclear Plant.
Drafts of this report were Noticed twice in the Federal Register for public comment: first on
December 29, 2009 (74 FR 68873), and second on August 2, 2011 (76 FR 46331). The authors
thank those members of the public who provided comments during the two public comment
periods, specifically Robert M. Brady, Schirmer Engineering Corporation; Patricia L. Campbell,
GE Hitachi Nuclear Energy; Jason E. Floyd, Hughes Associates, Inc.; Pablo Guardado, Entergy;
Michael D. Jesse, Excelon Nuclear; Daeil Kang, Korea Atomic Energy Research Institute;
Nancy McNabb, National Fire Protection Association; Daniel Orr, NRC, Region 1; Mark
Schairer, Engineering Planning and Management (EPM); and Robert Webster, AREVA, all of
whom provided insightful comments on the two drafts of the document.
This report was used as a textbook for pilot offerings of an advanced fire modeling course held
as part of the annual NRC/EPRI Fire PRA training in 2011 and 2012. The authors would like to
thank the students in those two classes for their many constructive comments that significantly
improved the final report.
Meetings were held before the Reliability and PRA Subcommittee of the Advisory Committee on
Reactor Safeguards (ACRS) and before the full ACRS committee on March 21, 2012 and
July 11, 2012, respectively. We would like to thank the members of the ACRS for their time and
efforts in supporting the meetings and reviewing and providing comments and suggestions for
improvements to the final report.
Finally, the authors express their thanks to Laurent Gay and Eric Wizenne, Electricit de France
(EdF), for reviewing the MAGIC calculations. The authors would also like to acknowledge Mr.
Bryan Klein of Thunderhead Engineering and Drs. Nathan Siu and Raymond Gallucci of the
NRC for their valuable contributions to this report. The authors also express appreciation to Ms.
Aixa Belen, Mr. Ken Canavan, Mr. Stuart Lewis, Mr. Nicholas Melly, Ms. Carolyn Siu, and Mr.
Robert Vettori for their reviews of and comments on various drafts of this document. Publication
of this document would not have been possible without assistance from the NRCs Office of
Administration. Specifically, the authors thank Tojuana Fortune-Grasty and Guy Beltz for their
efforts in publishing this report.
xxiii
LIST OF ACRONYMS
ACH
Air Changes per Hour
ACRS
Advisory Committee on Reactor Safeguards
AGA
American Gas Association
AHJ
Authority Having Jurisdiction
ANS
American Nuclear Society
ASET
Advanced Science and Engineering Technologies
ASME
American Society of Mechanical Engineers
ASTM
American Society for Testing and Materials
ATF
Bureau of Alcohol, Tobacco, and Firearms
BE
Benchmark Exercise
BFRL
Building and Fire Research Laboratory
BRE
Building Research Establishment
BWR
Boiling Water Reactor
CAROLFIRE Cable Response to Live Fire
CDF
Core Damage Frequency
CFAST
Consolidated Fire Growth and Smoke Transport Model
CFD
Computational Fluid Dynamics
CFR
Code of Federal Regulations
CHRISTIFIRE Cable Heat Release, Ignition, and Spread in Tray Installations during Fire
COR
Code of Record
CSR
Cable Spreading Room
DBD
Design Basis Document
ECCS
Emergency Core Cooling Systems
EdF
Electricit de France
EPRI
ERFBS
Electrical Raceway Fire Barrier System
FDS
Fire Dynamics Simulator
FDTs
Fire Dynamics Tools (NUREG-1805)
FHA
Fire Hazard Analysis
FIVE-Rev1
Fire-Induced Vulnerability Evaluation, Revision 1
FFT
Fast Fourier Transform
FLASH-CAT Flame Spread over Horizontal Cable Trays
FM/SNL
Factory Mutual & Sandia National Laboratories
FPA
Foote, Pagni, and Alvares
FPRA
Fire Probabilistic Risk Assessment
FRA
Fire Risk Analysis
GRS
Gesellschaft fr Anlagen- und Reaktorsicherheit (Germany)
HGL
Hot Gas Layer
HRR
Heat Release Rate
HRRPUA
Heat Release Rate Per Unit Area
HVAC
Heating, Ventilation, and Air Conditioning
IAFSS
International Association of Fire Safety Science
iBMB
Institut fr Baustoffe, Massivbau und Brandschutz
ICFMP
International Collaborative Fire Model Project
ID
Identification
IEEE
Institute of Electrical and Electronics Engineers
xxv
IPEEE
LERF
LES
LFS
LLNL
LOL
MCC
MCR
MEFS
MOVs
MQH
MOU
NBS
NEI
NFPA
NIST
NPP
NRC
NRR
PE
PMMA
PRA
PVC
PWR
RCP
RES
RG
RI/PB
RIS
RTE
RTI
SBDG
SDP
SFPE
SNL
SWGR
THIEF
TP
TS
UL
V&V
WTC
XPE
ZOI
Individual Plant Examination of External Events

Large Early Release Frequency
Large Eddy Simulation
Limiting Fire Scenario
Lawrence Livermore National Laboratory
Low Oxygen Limit
Motor Control Center
Main Control Room
Maximum Expected Fire Scenario
Motor-Operated Valves
McCaffrey, Quintiere, and Harkleroad
Memorandum of Understanding
National Bureau of Standards (now NIST)
Nuclear Energy Institute
National Fire Protection Association
National Institute of Standards and Technology
Nuclear Power Plant
Office of Nuclear Reactor Regulation (NRC)
Polyethylene
Polymethyl-Methacrylate
Probabilistic Risk Assessment
Polyvinyl Chloride
Pressurized Water Reactor
Reactor Coolant Pump
Office of Nuclear Regulatory Research (NRC)
Regulatory Guide
Risk-Informed, Performance-Based
Regulatory Issue Summary
Radiation Transport Equation
Response Time Index
Stand-By Diesel Generator
Significance Determination Process
Society of Fire Protection Engineers
Sandia National Laboratory
Switchgear Room
Thermally-Induced Electrical Failure
Thermoplastic
Thermoset
Underwriters Laboratory
Verification & Validation
World Trade Center
Cross-Linked Polyethylene and Neoprene
Zone of Influence
xxvi
NOMENCLATURE
area (m2)
floor area (m2)
opening area (m2)
surface area of enclosure boundary (m2)
specific heat, solid material (kJ/kg/K)
specific heat, gas, constant pressure (kJ/kg/K)
fire diameter (m) or optical density (m-1)
characteristic diameter
experimental measurement
acceleration of gravity (m/s2)
heat transfer coefficient (kW/m2/K)
ceiling height (m)
height of fire base above floor (m)
opening height (m)
thermal conductivity (kW/m/K)
constant
volumetric entrainment coefficient
light extinction coefficient (m-1)
mass-specific extinction coefficient (m2/kg)
compartment length (m)
flame height (m)
total mass per unit length of a single cable (kg/m)
combustible mass per unit area of tray (kg/m2)
mass loss or flow rate (kg/s)
mass loss rate per unit area (kg/s/m2)
model prediction
number of cables per tray
probability, wall perimeter (m)
heat flux (kW/m2)
critical heat flux (kW/m2)
heat release rate (kW)
fire Froude number
radial distance (m)
ceiling jet distance (m)
specific area (m2/g)
burnout time (s)
time (s)
temperature (C)
volume (m3)
volume flow rate (m3/s)
compartment width (m), width of tray (m)
cell dimension (m)
critical value
product yield (kg/kg)
mass fraction (kg/kg)
xxvii
smoke layer interface position above base of fire (m)

Greek:
Subscripts:
,
fire growth coefficient (kW/s2)

model bias factor, difference operator
heat of combustion (kJ/kg)
pressure difference (Pa)
equivalence ratio
stoichiometric oxygen to fuel ratio
mean
density (kg/m3)
standard deviation
relative standard deviation, experiment
relative standard deviation, model
stoichiometric coefficient, residue yield (kg/kg)
ambient
average
distributed point source
effective
fuel or fire
floor
ith element in set
leakage
total number of elements in a set
opening
relating to oxygen
plastic
point source
required
soot or smoke
tenability limit
total
xxviii
INTRODUCTION
1.1 Background
In 2001, the National Fire Protection Association (NFPA) issued the first edition of NFPA 805,
Performance-Based Standard for Fire Protection for Light-Water Reactor Electric Generating
Plants, 2001 Edition1. Effective July 16, 2004, the U.S. Nuclear Regulatory Commission (NRC)
amended its fire protection requirements in Title 10, Section 50.48(c) of the Code of Federal
Regulations (10 CFR 50.48(c)) to permit existing reactor licensees to voluntarily adopt fire
protection requirements contained in NFPA 805 following a performance-based approach as an
alternative to the existing deterministic2 fire protection requirements. One important element in
a performance-based approach is the estimation of fire hazard using mathematical fire models.
Fire modeling is often used in constructing Fire PRAs to determine the effects of fire hazard so
that the associated risk can be quantified.
As part of its fire modeling requirements, NFPA 805 states that fire models shall be verified and
validated (section 2.4.1.2.3) and that only fire models that are acceptable to the authority
having jurisdiction (AHJ) shall be used in fire modeling calculations (section 2.4.1.2.1). This is
an important requirement because the verification and validation (V&V) of fire models is
intended to ensure the correctness, suitability, and overall quality of the method. Specifically,
verification is the process used to determine whether a model correctly represents the
developers conceptual description (i.e., whether it was built correctly), while validation is used
to determine whether a model is a suitable representation of the real world and is capable of
reproducing phenomena of interest (i.e., whether the correct model was built).
In 2007, the NRCs Office of Nuclear Regulatory Research (RES) and the Electric Power
Research Institute (EPRI) completed a collaborative project for the V&V of five select fire
modeling tools. The results of this study, which was performed under the NRC/RES-EPRI
Memorandum of Understanding (MOU), are documented in NUREG-1824 (EPRI 1011999),
Verification and Validation of Selected Fire Models for Nuclear Power Plant Applications. The
National Institute of Standards and Technology (NIST) was also an important partner in
developing this publication, providing extensive fire modeling and experimentation expertise.
The V&V effort is intended to support the use of fire modeling for various NPP fire hazard
analysis applications.
This report builds on the V&V research described earlier by incorporating the results into a set
of guidelines and recommendations for conducting fire modeling studies in support of
1
All references in this report to NFPA 805 are specific to the 2001 edition of the standard, which is the code of record (COR)
required by 10 CFR 50.48(c).
2
In nuclear fire protection, the term deterministic is typically used to refer to prescriptive requirements while deterministic is often
used as an adjective with fire model in the general fire protection field. Within this report, the usage should be clear from the
context in which the word is used.
1-1
INTRODUCTION
commercial nuclear industry applications. When the NRCs Advisory Committee on Reactor
Safeguards (ACRS) issued a letter to Luis Reyes, Executive Director for Operations,
recommending publication of NUREG-1824 (EPRI 1011999), they identified two major items to
be included in the users guide (Wallis, 2006). Specifically, the ACRS recommended that the
users guide include:
Estimates of the ranges of normalized parameters to be expected in nuclear plant

applications
Quantitative estimates of the uncertainties associated with each models predictions,
preferably in the form of probability distributions
The ACRS indicated that quantitative estimates of the intrinsic model uncertainty would be a
valuable input in risk-informed as well as non-risk-informed applications. Chapters 2 and 3
address the first ACRS recommendation. Chapter 4 specifically addresses the second ACRS
recommendation, that is, the development of V&V results into quantitative estimates of model
uncertainty. Finally, the appendices contain examples that illustrate the entire process for
several nuclear power plant (NPP) scenarios.
1.2 Objective
The objective of this guide is to describe the process of conducting a fire modeling analysis,
principally for commercial NPP applications. The process described in this guide addresses
most of the technical elements relevant to fire modeling analysis, such as the selection and
definition of fire scenarios and the determination and implementation of input values, sensitivity
analysis, uncertainty quantification, and documentation. In addition, requirements associated
with fire modeling analyses and analytical fire modeling tools are addressed through generic
guidance, recommended best practices, and example applications.
1.3 Scope
1.3.1 User Capabilities
This guide should be used as a complement to, not a substitute for, users manuals for specific
fire modeling tools, fire dynamics textbooks, technical references, education, and training. This
guide only compiles information and organizes it procedurally for NPP applications. Analysts
are encouraged to review the references identified throughout the guide for in-depth coverage of
the advantages and the range of applicability of specific models. Once a fire scenario has been
selected, this guide will help the fire model user define the necessary modeling parameters,
select an appropriate model, and properly interpret the fire modeling results. Since all models
are merely approximations of reality, this guide also provides useful insights for translating real
configurations into modeling scenarios. Due to the technical nature of this guide, users with the
following areas of expertise will benefit the most from it:
General knowledge of the behavior of compartment fires

General knowledge of basic engineering principles, specifically thermodynamics, heat
transfer, and fluid mechanics
1-2
INTRODUCTION
Ability to understanding the basis of mathematical models involving algebraic and

differential equations
This guide focuses on the capabilities of the models selected for V&V. However, some generic
guidance is also provided, and most of the discussion is applicable to any fire model of the
respective type (algebraic model, zone model, or computational fluid dynamics (CFD) model).
Five specific models are discussed in this guide:
(1) The NRCs Fire Dynamics Tools (FDTS), NUREG-1805 and Supplements
(2) EPRIs Fire-Induced Vulnerability Evaluation, Revision 1 (FIVE-Rev1)
(3) NISTs Consolidated Model of Fire Growth and Smoke Transport (CFAST) Version (6)
(4) Electricit de Frances (EdF) MAGIC code Version (4.1.1)
(5) NISTs Fire Dynamics Simulator (FDS) Version (5)
Finally, the user of this document would benefit from a familiarity with the use of fire modeling in
a fire protection performance-based approach and NUREG/CR-6850 (EPRI 1011989),
EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities.
1.3.2 Training Resources

For individuals seeking to enhance or update their expertise in the areas noted in section 1.3.1,
there are several resources available, including academic courses, short courses, and written
materials.
Information on academic institutions with degree programs or single classes in fire protection
engineering can be found at the Society of Fire Protection Engineers (SFPE) web site:
http://www.careersinfireprotectionengineering.com/career_types.htm
A background in engineering fundamentals is essential for fire modelers, especially in the areas
of fluid mechanics, thermodynamics, and heat transfer. These subjects are offered at virtually
any academic institution with programs in fire protection and/or mechanical, aerospace, civil,
and chemical engineering. While general courses provide basic background discussions,
courses involving fire applications are preferable, and would be provided by the institutions
offering courses or degree programs in fire protection engineering.
In addition to the academic programs, short courses in fire behavior and fire modeling are
available through professional and industry associations, such as the SFPE
(http://www.sfpe.org) and EPRI (http://www.epri.com).
Key written references on fire behavior and fire modeling include:
1-3
INTRODUCTION
ASTM E135505a (2005), Standard Guide for Evaluating the Predictive Capability of
Deterministic Fire Models, American Society for Testing and Materials, West
Conshohocken, PA, 2005.
Buchanan, A. H. (2001), Structural Design for Fire Safety, John Wiley and Sons, LTD,
Chichester, England, 2001.
Babrauskas, V., Ignition Handbook, Fire Science Publishers/Society of Fire Protection
Engineers, Issaquah WA (2003).
Drysdale, D., An Introduction to Fire Dynamics, 3rd Ed., John Wiley, 2011.
Karlsson, B. and Quintiere, J.G., Enclosure Fire Dynamics, CRC Press, 2000
Quintiere, J.G., Principles of Fire Behavior, Delmar Publishers, 1998.
Quintiere, J.G., Fundamentals of Fire Phenomena, John Wiley, 2006.
NFPA, Fire Protection Handbook, National Fire Protection Association, 20th Ed., A.E.
Cote, (Editor) 2008.
NUREG-1824 (EPRI 1011999), Verification and Validation of Selected Fire Models for
Nuclear Power Plant Applications, Nuclear Regulatory Commission, 2007.
SFPE, SFPE Engineering Guide to Assessing Flame Radiation to External Targets from
Pool Fires, SFPE Engineering Guide, Society of Fire Protection Engineers,
Bethesda, MD, March, 1999.
SFPE, SFPE Engineering Guide to Fire Exposures to Structural Elements, SFPE
Engineering Guide, Society of Fire Protection Engineers, Bethesda, MD,
November, 2005.
SFPE, SFPE Engineering Guide to Piloted Ignition of Solid Materials Under Radiant
Exposure, SFPE Engineering Guide, Society of Fire Protection Engineers,
Bethesda, MD, January, 2002.
SFPE, SFPE Engineering Guide to Predicting Room of Origin Fire Hazards, SFPE
Engineering Guide, Society of Fire Protection Engineers, Bethesda, MD,
November, 2007.
SFPE, SFPE Handbook of Fire Protection Engineering, 4th Ed., P. DiNenno (Editor),
National Fire Protection Association, 2008.
1.4 Fire Modeling Theory

Fire development in compartments is often divided into phases depending on the dominant
processes at any given stage of development. Ignition is dictated by the characteristics of the
fuel item being ignited (i.e., ignition temperature, geometry, orientation, and thermophysical
properties3) and the strength of the ignition source. Once the flames are sustained on a burning
fuel item, a smoke plume develops, transporting mass and heat vertically as a result of the
buoyancy of the smoke (see Figure 1-1). The plume will entrain air as it rises, thereby causing
the smoke to cool and become diluted; as a result, the quantity of smoke being transported will
increase with increasing elevation. After a smoke plume strikes the ceiling, the smoke travels
horizontally under the ceiling in a relatively thin layer, referred to as a ceiling jet. As the ceiling
jet travels, the smoke cools with increasing distance from the plume impingement point, in part
because of air entrainment into the ceiling jet as well as heat losses from the ceiling jet to the
solid ceiling boundary.
3
Thermophysical properties include thermal conductivity, specific heat, and density.
1-4
INTRODUCTION
In an ideal situation, once the ceiling jet reaches the enclosing walls, a Hot Gas Layer4 (HGL)
develops. As a result of the continuing supply of smoke mass and heat via the plume, the HGL
becomes deeper, and its temperature increases. Other properties of the smoke in the HGL also
increase (including concentration of gas species and solid particulates).
Radiant heat from the HGL to other combustibles not involved in the fire increases their
temperature. Similarly, the temperature of non-burning combustibles will also increase as a
result of receiving thermal radiation from the burning item(s). As the other combustibles reach
their respective ignition temperatures, they will also ignite. In some cases, the ignition of many
other combustibles in the space caused by heating from the HGL occurs within a very short time
span. This is commonly referred to as flashover.
Several aspects of fire behavior may be of interest when applying fire models, depending on the
purpose of the modeling application. Analysts may seek to determine the effects associated
with heating of targets submerged in smoke or receiving radiant heat from the flames, the
response of ceiling-mounted detectors or sprinklers to the fire environment, or other
phenomena.
Figure 1-1. Characteristics of compartment fires.
Hot Gas Layer or HGL is also called smoke layer or hot upper layer in other publications in fire protection engineering.
1-5
INTRODUCTION
The most common aspects of fire behavior that typically are of interest in such analyses include,
but are not limited to:
Rate of smoke production

Rate of smoke filling
Properties of the ceiling jet
Properties of the HGL
Target response to incident heat flux via either thermal radiation or convection
A detailed review of each of these aspects is provided in texts on fire dynamics. A brief review
of each is provided here.
Rate of smoke production
Smoke is defined as a combination of the gaseous and solid particulates resulting from the
combustion process, plus the air that is entrained into the flame and/or smoke plume.
Consequently, the rate of smoke production at a particular height in the plume is the
combination of the generation rate of combustion products and air entrainment rate into the
flame and/or smoke plume between the top of the fuel and the height of interest. In most cases,
the air entrainment rate greatly exceeds the generation rate of fuel volatiles. Thus, the
correlations used to estimate the rate of smoke production are usually taken from experimental
research on entrained air.
Rate of smoke filling
The rate of smoke filling is dependent on the rate of smoke production, the heat release rate
(HRR), floor area, height and configuration of the space, and time from ignition. For a fire with a
steady HRR, the rate of smoke filling in a compartment will decrease with time due to a
decrease in the smoke production rate, which decreases as the height available to entrain air
decreases when the HGL deepens.
Properties of the ceiling jet
The ceiling jet transports smoke and heat horizontally away from the plume after it impacts with
the ceiling. The response of ceiling-mounted fire detectors or sprinklers is governed primarily by
their interaction with a ceiling jet. The temperature and concentration of smoke in a ceiling jet is
principally dependent on the height and configuration of the space, distance to the ceiling
impact point of the smoke plume, and the HRR of the fire.
Properties of the HGL
As smoke and heat are transported to the HGL via the smoke and fire plumes, the properties of
the HGL will change. The principal properties of interest include the depth, temperature, and
gas concentrations in the HGL. The magnitude of the properties depends on the HRR of the
fire, geometry of the space, ventilation openings (permitting material from the HGL to leave the
space, providing air to the fire, and/or causing a stirring action), yields of combustion products,
1-6
INTRODUCTION
and the elapsed time after ignition. These changes can be tracked by considering the
conservation of energy, mass, and species relative to the HGL.
Target response to incident heat flux via either thermal radiation or convection
The targets temperature will increase as a result of receiving heat via either thermal radiation or
convection. Radiation heat transfer is dependent on the intensity of thermal radiation emitted by
a source, the size of the source, and the proximity of the target to the source. For this
application, the flame height, the portion of heat released from the fire as radiation, and the
distance separating the target from the flame are the dominant parameters. Convective heating
occurs whenever the target is submerged in the smoke plume or HGL.
1.5 Fire Modeling Tools

1.5.1 Algebraic Models
Algebraic models may be standalone equations found in literature or may be contained within
spreadsheets (such as the NRCs FDTs), and can help give a general understanding of one of
the fire environment phenomena. These equations are typically closed-form algebraic
expressions, many of which were developed as correlations from empirical data. In some
cases, they may take the form of a first-order ordinary differential equation, and, when used
properly, can provide an estimate of fire variables, such as; HGL temperature, heat flux from
flames or the HGL, smoke production rate, depth of the HGL, and the actuation time for
detectors.
Algebraic models are helpful because they require minimal computational time and a limited
number of input variables. When applying the results of the algebraic models, users need to be
aware that the development of most equations involved approximations to simplify the analysis.
Algebraic models are useful primarily as screening tools (i.e., to provide a rough approximation
for an analysis, perhaps as a check of an aspect of the results of the computer-based models),
and are also applicable when only one phenomenon can be treated in isolation: for instance,
plume or ceiling jet correlations are not applicable if there is a significant HGL unless they are
modified to account for this effect.
1.5.2 Zone Models

A zone model, such as the Consolidated Fire Growth and Smoke Transport Model (CFAST) or
MAGIC, calculates fire environment variables using control volumes, or zones, of a space. The
zones correspond to a cooler lower layer and a HGL, as depicted in Figure 1-2. The
fundamental idea behind a zone model is that each zone is well-mixed and that all fire
environment variables (temperature, smoke concentration, etc.) are therefore uniform
throughout the zone. Conditions in each zone are calculated by applying conservation
equations and the ideal gas law. The variables in each zone change as a function of time and
rely on the initial conditions specified by the user. There is a well-defined boundary separating
the two zones, though this boundary may move up or down throughout the simulation.
1-7
INTRODUCTION
HGL
Plume mass
flow
Mass
outflow
Mass inflow
Figure 1-2. A two-zone enclosure fire with an HGL above and a cool lower layer below.
Zone models are most applicable in situations involving simple geometries or where spatial
resolution within a compartment is not important. The preparation of input for a zone model, the
computation time, and the amount of output data generated are slightly more extensive than a
simple algebraic model; however, the overall computational time cost is still low.
Zone models can easily analyze conditions resulting from fires involving single compartments or
compartments with adjacent spaces, and are often used to compute the HGL temperature, HGL
composition, and target heat fluxes. They are also capable of modeling some effects of natural
and mechanical ventilation in both horizontal and vertical directions. Some zone models allow
the user to select a thermal plume model, which may assist in better characterization of a known
fire scenario, while others use an axisymmetric smoke plume. Other features of a zone model
may include a user-specified single zone or multiple fire plumes.
Simulations of spaces with complex ceilings or numerous compartments can be challenging
with a zone model. Because zone models specify uniform conditions in the HGL and lower
layer, results cannot be distinguished between locations at different distances from the fire. Due
to the zone approach, smoke transport time lags are not considered in the simulation, which is
an acceptable approximation in relatively small spaces but may lead to significant error in largevolume spaces or spaces with large aspect ratios.
Smoke production, fire plume dynamics, ceiling jet characteristics, heat transfer, and ventilation
flows are all algebraic models embedded within zone models. Other parameters that can be
calculated with a zone model include thermal behavior, detection response, and suppression
response. The output of a zone model is typically simple to understand and is generally
presented through an automatic user interface.
1-8
INTRODUCTION
Most zone models have default values that must be recognized and adjusted as necessary to
obtain an accurate solution. The model user must understand and justify the relevance of the
default values used in any application. Fire model users are expected to assess the
appropriateness of default values provided in the fire models and make changes or adjust
values as necessary. User manuals and technical references for each zone model outline such
values and may provide recommended ranges for the parameters.
1.5.3 CFD Models

A computational fluid dynamics (CFD) model is often useful when trying to determine fire
variables at a specific location or when there are geometric features that are expected to play a
significant role in the results beyond what can be calculated in a zone model approximation. A
typical CFD model consists of a preprocessor, a solver, and a postprocessor. CFD models can
provide a detailed analysis in both simple and complex geometries.
CFD models essentially apply a series of conservation and state equations across multiple cell
boundaries in a space. The number of cell boundaries depends on the mesh size, which breaks
the geometry into three-dimensional subvolumes called cells. Solutions to the conservation
equations of mass, momentum, and energy are updated as a function of time within each
numerical grid cell, with the solutions in all cells collectively describing the fire environment
within the geometry at the cell resolution.
The number of grid cells defines the type of mesh. A fine mesh is made up of numerous grid
cells. Since the equations are applied at each cells boundaries, a more detailed distribution of
fire parameters is characterized. A coarse mesh is made up of fewer grid cells and can result in
less accurate results. The type of mesh and number of grid cells should be based on the
geometry and the desired results. If a more detailed simulation is needed, then a finer mesh
should be used. Be aware that a finer mesh significantly increases the computational running
time of the model as well as the quantity of output data.
CFD models have much better spatial fidelity than zone models, being able to distinguish
conditions in one part of the space from another. Because of the appreciable amount of time
and effort required to apply CFD models as compared to zone models or algebraic models, CFD
models are generally applied when:
Spatial resolution is important, relative to either the locations of fuel packages or targets.
Large compartments relative to the fire size are involved.
Compartments have complex geometries, flow connections, or numerous obstructions in
the upper part of the compartment.
Large numbers of compartments are within the area of interest and the presence of each
compartment is expected to affect the fire environment in the area of interest.
An example of a CFD simulation of a fire experiment is shown in Figure 1-3. The purpose of the
calculation was to simulate an experiment that was part of the validation study described in
NUREG-1824 (EPRI 1011999). In the experiment, a pan fire was placed in a relatively small
1-9
INTRODUCTION
compartment, and temperatures and heat fluxes were measured at various locations. The CFD
simulation is able to describe the changing behavior of the fire as it interacts with its
surroundings.
Figure 1-3. A Smokeview visualization of a CFD model of a compartment fire experiment.
While CFD models provide a detailed analysis of a space, they are costly to create, simulate,
and maintain. The input files created in the preprocessing stage require a significant effort to
create. The user must understand the code syntax and the implications and approximations
embedded in the model. A firm understanding of fire dynamics is important in providing input
data that is relevant to the application. Most CFD models have default values that must be
recognized and adjusted as necessary to obtain an accurate solution. The model user must
understand and justify the relevance of the default values used in any application. Fire model
users are expected to assess the appropriateness of default values provided in the fire models
and make changes or adjust values as necessary. User manuals and technical references for
each CFD model outline such values and may provide recommended ranges for the
parameters.
Depending on the complexity of the scenario and the computers computational power, the
solver within the model can take anywhere from a few hours to weeks to complete all the
calculations. This time cost depends on the measured parameters, the size of the geometry,
and the mesh size of the calculations. Outputs of CFD models are visualized through a post-
1-10
INTRODUCTION
processing program. The CFD model developed at NIST, Fire Dynamics Simulator (FDS),
employs the program Smokeview to represent distributions of temperature, mass, heat flux,
burning rate, etc. throughout the geometry. These parameters can be described through point
locations, isocontours, or vector diagrams. Output data may also be stored in a commaseparated value file format that can be read by a standard spreadsheet program.
1.5.4 Fire Model Verification and Validation (V&V)

capabilities. For example, NFPA 805 states that fire models shall only be applied within the
limitations of that fire model (section 2.4.1.2.2). ASTM E 1355, Standard Guide for Evaluating
the Predictive Capability of Deterministic Fire Models, provides definitions of the terms model
verification and model validation.
Model Verification is the process of determining that the implementation of a calculation method
accurately represents the developers conceptual description of the calculation method and the
solution to the calculation method. The fundamental strategy of verification of computational
models is the identification and quantification of error in the computational model and its
solution.
Model Validation is the process of determining the degree to which a calculation method is an
accurate representation of the real world from the perspective of the intended uses of the
calculation method. The fundamental strategy of validation is the identification and
quantification of error and uncertainty in the conceptual and computational models with respect
to intended uses.
As noted in Section 1.1, NRC/RES and EPRI conducted a collaborative project for V&V of five
fire models. The results of this project were documented in NUREG-1824 (EPRI 1011999),
Verification and Validation of Selected Fire Models for Nuclear Power Plant Applications.
1.6 Fire Modeling Applications

Fire modeling is used in a variety of NPP applications. Examples include license amendments
or exemption requests, fire induced circuit failure analyses, NFPA 805 performance based
applications, and fire PRA support. This section provides a brief overview of common fire
modeling applications.
1.6.1 License Amendments and Exemptions

Fire modeling has been used to justify requests for changes from the deterministic requirements
contained in 10 CFR 50.48 and Appendix R of 10 CFR 50. License amendments and
exemption requests are evaluated by NRC staff on a case by case basis.
1.6.2 Fire Induced Circuit Failures

The disposition of certain fire-induced circuit failures, such as multiple spurious operations
(MSOs), is another accepted fire modeling application used in commercial NPPs. MSOs
typically involve one or more fire-induced component failures that include spurious operation
due to hot shorts, shorts to ground, or open circuits as a result of fire damage to electrical
cables.
1-11
INTRODUCTION
1.6.2.1 Deterministic Application

When one of the redundant safe-shutdown trains in a fire area is maintained free of fire damage
by one of the specified means in Regulatory Position 5.3.1.1 of Regulatory Guide (RG) 1.189,
Revision 2, then fire modeling may be used to demonstrate that components important to safe
shutdown, including systems, structures, and components (SSCs) that are not part of the
success path, are protected from fire damage. When fire modeling is used to demonstrate that
components important to safe shutdown are protected from fire damage, the analysis would
consider all in situ and transient fire sources in the area and all targets that involve components
important to safe shutdown. Variations in fire compartment configuration (e.g., doors open or
closed, ventilation system on or off, etc.) are to be included in this fire modeling analysis. The
analysis needs to clearly demonstrate that the largest expected fire will not damage
components important to safe shutdown. The phrase largest expected defines a fire that is
consistent with the characteristics of the ignition sources and other combustibles in the area
where the fire can propagate.
1.6.2.2 Risk-Informed, Performance-Based Application

RG 1.205 provides regulatory guidance on the evaluation of fire-induced circuit failures when
used in applications that support risk-informed, performance-based decision making. Within a
risk-informed, performance-based NFPA-805 license basis, fire-induced circuit failure5 scenarios
can be evaluated using a fire PRA. When applicable, the fire-induced circuit failure scenarios
are included in the plant response model. Chapter 5 of NUREG/CR-6850 provides information
on the definition of and methods for developing plant response models, such that the fireinduced circuit failure contribution is included in the quantification of core damage and large
early release frequencies. The model needs to address both the possible plant impact caused
by spurious operation(s) and the inability to restore equipment operability as a result of fire
damage. With the applicable fire-induced circuit failure scenarios modeled in the risk
assessment, quantitative risk calculations could be used to show that the scenarios present
acceptable risk contributions within the overall plant fire risk profile, as discussed in RG 1.205.
Fire modeling can be used within the framework of the fire PRA to characterize the risk
associated with fire scenarios. Determining the risk associated with fire scenarios may include
such evaluations as developing a range of fire sizes that may damage cables at varying
distances from an ignition source, with the corresponding times that it would take a fire to grow
to those damaging sizes. In this context, the fire modeling objective is to predict the fire
conditions that could induce cable damage and thus trigger fire-induced circuit failure scenarios,
so that the corresponding risk contribution is captured in the analysis. If the fire-induced circuit
failure is successfully resolved using fire modeling in a probabilistic approach, it must
demonstrate that the damage necessary to cause the fire-induced circuit failure would not occur
for the postulated fire scenario (e.g., a specific ignition source is screened from the analysis), or
that the risk associated with the fire-induced circuit failure condition is acceptable, given the
specific characteristics of the fire necessary to produce the postulated condition.
1.6.3 NFPA 805 Performance-Based Applications

NFPA 805 allows the use of either fire modeling or fire risk evaluations to demonstrate that
performance based requirements are satisfied.
The term fire-induced circuit failure is intended to address loss of function due to fire damage to cables and equipment as well as
spurious operation(s) of that equipment either singly or in multiples.
1-12
INTRODUCTION
1.6.3.1 Fire Modeling

The NFPA 805 requirements associated with fire modeling are organized into two sections,
Section 2.4.1 and Section 4.2.4.1. Section 2.4.1 describes the requirements associated with the
fire modeling calculations and analysis. Section 4.2.4.1 describes the requirements for the
implementation of a performance-based fire modeling analysis.
Section 2.4.1.2 of NFPA 805 describes the requirements for the use of fire models, which
include:
The use of fire models acceptable to the authority having jurisdiction (AHJ), i.e., the U.S.
Nuclear Regulatory Commission
The application of fire models within their range and limitations
Fire models shall be verified and validated
In the context of this application, the specific analytical capabilities within the fire model need to
be verified and validated. Model capabilities not invoked in a specific calculation are outside the
scope of this requirement. NUREG-1824 (EPRI 1011999) is an example of a verification and
validation study for fire models specifically developed for NPP applications.
Section 4.2.4.1 is subdivided by process element as follows:
Identify Targets (NFPA 805 4.2.4.1.1): This subsection requires the description of the
targets (e.g., equipment or cables) and target locations (specific locations of raceways/conduits
containing the cables, electrical cabinets, or equipment) associated with them needed to
achieve the nuclear safety performance criteria.
Establish Damage Thresholds (NFPA 805 4.2.4.1.2): This subsection requires the
description of damage thresholds for the equipment and cables needed to achieve the nuclear
safety performance criteria. The damage threshold (i.e., target vulnerability) for cables exposed
to fire is expressed in most cases in the form of an incident heat flux on the cables or the cables
surface temperature shall be established in accordance with Section 2.5 of NFPA 805.
Determine Limiting Conditions (NFPA 805 4.2.4.1.3): This subsection requires the
description of the combination of equipment or required cables with the highest susceptibility to
any fire environment.
Establish Fire Scenarios (NFPA 805 4.2.4.1.4): This subsection requires the description of
a given areas fire conditions resulting from the identified and analyzed fire scenarios. It should
be noted that the scenario definition is consistent with the requirements listed under 2.4.1.3 of
NFPA 805. Appendix C of NFPA 805 provides two categories of fire scenarios used in the
standard as follows:
Maximum Expected Fire Scenario: The maximum expected fire scenario (MEFS) is
defined in NFPA 805 Section C.3.2, as the scenario that is used to determine by fire
modeling whether the performance criteria are met in the fire area being analyzed. The
input data for the fire modeling of the MEFS should be based on the following:
Existing in-situ combustibles in the fire area
1-13
INTRODUCTION
Types and amounts of transient combustibles that industry experience and

specific plant conditions indicate can reasonably be anticipated in the fire area
Heat release and fire growth rates for actual in-situ and transient combustibles
that are realistic and conservative based on available test data and applicable
fire experience
Ventilation within normal operating parameters with doors in the open or closed
position
Active and passive fire protection features operating as designed
Limiting Fire Scenario: The limiting fire scenario (LFS) is defined in NFPA 805 Section
C.3.3 as fire scenario(s) in which one or more of the inputs to the fire modeling
calculation (e.g., heat release rate, initiation location, or ventilation rate) are varied to the
point that the performance criterion is not met. The intent of this scenario(s) is to
determine that there is a reasonable margin between the expected fire scenario
conditions and the point of failure [T]he LFS can be based on a maximum possible,
though unlikely, value for one input variable or an unlikely combination of input variables.
The values used for LFS input should remain within the range of possibility but can
exceed that determined or judged to be likely or even possible.
For each fire scenario, the environmental conditions resulting from each MEFS are compared to
the damage thresholds for the targets in the fire area. If damage thresholds are not exceeded,
the targets in the fire area can be considered free of fire damage under the conditions of the
postulated MEFS.
By definition, the effects of the LFS include damage to the targets in the fire area under
consideration. Fire modeling parameters that have been varied to establish the LFS conditions
are identified and described.
1.6.3.2 Fire Risk Evaluations

A fire risk evaluation for the NFPA 805 performance-based approach uses fire PRA methods,
tools, and data to compare the risk associated with implementation of the proposed alternative.
Sections 2.4.3 and 4.2.4.2 of NFPA 805 contain specific requirements for the use of fire risk
evaluations. The fire PRA methods, tools, and data must be acceptable to the AHJ, and they
should be appropriate for the nature and scope of the evaluation, be based on the as-built and
as-operated and maintained plant, and reflect the operating experience at the plant. The
evaluation uses core damage frequency (CDF) and large early release frequency (LERF) as
measures for risk. The proposed alternative must ensure that the philosophies of defense in
depth and sufficient safety margin are maintained. Section 1.6.4 of this report provides
information on using fire modeling in support of fire PRA.
1.6.4 Fire Modeling in Support of Fire PRA

Fire PRA often applies fire modeling in the fire scenario development and analysis process. A
fire scenario in a Fire PRA is often modeled as a progression of damage states over time, which
is initiated by a postulated fire resulting from an ignition source. Each damage state is
characterized by a time and a set of targets damaged within that time. Fire modeling is
commonly used to determine the targets affected in each damage state and the associated time
at which this occurs. The first damage state usually consists of damage only to the ignition
source itself. Depending on the characteristics and configuration of the scenario, the last
damage state may consist of an HGL formation that leads to full room damage by exceeding
1-14
INTRODUCTION
predetermined damage criteria. Damage states between the first and final states capture target
sets compromised as the fire propagates through intervening combustibles.
Each scenario progression postulated in a Fire PRA is quantified to determine its contribution to
fire risk. The most common fire risk metrics are the reactor Core Damage Frequency (CDF)
and Large Early Release Frequency (LERF) of radioactive material. NUREG/CR-6850 provides
guidance on the quantification of fire risk.
The Fire PRA standard (ASME/ANS RA-Sa-2008 and Addenda RA-Sa-2009) lists requirements
for all the technical elements associated with a Fire PRA, includes specific requirements for the
use of fire models. The standard addresses: (1) the selection of appropriate fire modeling tools
for estimating fire growth and damage behavior, considering the physical behaviors relevant to
the selected fire scenarios, and (2) the application of fire models that are sufficiently capable of
modeling the conditions of interest within known limits of applicability. In the case of analytical
fire models, the standard requires the use of appropriate fire modeling tools with the ability to
model the conditions of interest within known limits of applicability.
1.7 Organization of the Guide

The guidance material provided in this document is divided into five chapters and a number of
appendices, as outlined below.
Chapter 2 presents a qualitative overview of the process for conducting fire modeling,
including the basic principles of fire simulation, advantages and limitations of the
technology, and brief descriptions of the five models
Chapter 3 provides specific guidance on selecting models to address typical scenarios in
commercial NPPs
Chapter 4 contains information on determining the sensitivity and uncertainty associated
with fire modeling calculations
Chapter 5 contains the list of references identified throughout this document
Chapter 6 contains general information on the eight scenarios documented in the
appendices
Appendices A through H provide detailed examples of fire modeling analyses of typical
NPP scenarios:
o Appendix A Cabinet Fire in Main Control Room
o Appendix B Cabinet Fire in Switchgear Room
o Appendix C Lubricating Oil Fire in Pump Compartment
o Appendix D Motor Control Center Fire in Switchgear Room
o Appendix E Trash Fire in Cable Spreading Room
o Appendix F Lubricating Oil Fire in Turbine Room
o Appendix G Transient Fire in Multi-Compartment Corridor
o Appendix H Cable Tray Fire in Annulus
CD containing the following files:
o Appendix A Files
Algebraic models
CFAST and FDS input files
o Appendix B Files
1-15
INTRODUCTION
o
o
o
o
o
Algebraic models
Appendix C Files
FDS and MAGIC input files
Appendix D Files
Appendix E Files
Algebraic models
Appendix F Files
Algebraic model files
FDS files
Appendix G Files
Algebraic model files
MAGIC input files
Appendix H Files
FDS input files
Fire Modeling Software
CFAST Software Guides
FDS Software Guides
NUREG-1805 Spreadsheets
Related Publications
1-16
THE FIRE MODELING PROCESS

This chapter provides a general step-by-step process for modeling fires in commercial nuclear
power plants (NPPs). The recommended methodology comprises six steps: (1) define fire
modeling goals and objectives, (2) characterize the fire scenarios, (3) select fire models,
(4) calculate fire-generated conditions, (5) conduct sensitivity and uncertainty analyses, and (6)
document the analysis. A simplified process involving the six steps is shown in Figure 2-1.
Start
Definefiremodelinggoals
andobjectives(Section2.1)
Characterizethefire
scenarios(Section2.2)
Selectthefiremodel(s)
(Section2.3)
Yes
Canthe
problem
be
redefined
?
Yes
No
Is
another
model
available
?
No
No
Isthe
model
suitable?
Yes
Analternateanalysismaybe
required
Calculatethefiregenerated
conditions(Section2.4)
End
Conductasensitivityand
uncertaintyanalysis(Section
2.5)
Arethe
goalsand
objective
satisfied?
Yes
Documenttheanalysis
(Section2.6)
End
Figure 2-1. Fire modeling process.
2-1
No
2.1 Step 1: Define Fire Modeling Goals and Objectives

The first step in a fire model analysis is to identify and state the fire modeling goals and
objectives. A fire modeling goal is a broad statement of what needs to be accomplished using
fire modeling (SFPE, 2007). The goal(s) should also identify whether the analysis results are
intended to help resolve a deterministic issue or are intended as input for a probabilistic risk
assessment (PRA).
The objectives are more specific and describe in engineering terms how the goal(s) will be met
(SFPE, 2007). Clearly defined objectives provide focus and are needed to correctly select the
fire scenarios that will be evaluated and the fire modeling tools that will be used. In order to
define the objectives, some understanding of the conditions by which success or failure are
measured (i.e., the performance criteria) is necessary at the analysis outset. Any fire modeling
objective may thus be viewed as a statement defining what needs to be accomplished, which
criteria will be used to define success or failure, and which analysis process will be followed.
The criteria should be stated in terms that can be achieved by the fire modeling analysis.
Some common situations in commercial NPPs where fire modeling may be used to determine
which goals and objectives would be developed include, but are not limited to:
Evaluating whether or when a fire could damage a single electrical cable or component
Evaluating whether or when a fire could damage multiple electrical cables or

components
Evaluating whether conditions are habitable in an enclosure
Evaluating the potential for fire propagation through or across a fire barrier
Evaluating detection or sprinkler actuation
Evaluating the potential for fire propagation between fire zones or fire areas, or to
secondary combustibles
Evaluating the structural response under fire conditions
The performance criteria are an important aspect for developing the fire modeling objectives
and will be specific to the fire modeling application. The performance criteria will often include,
but are not limited to one or more of the following:
Maximum acceptable surface temperature or jacket-insulation interface temperature for

a cable, component, secondary combustible, structural element, or fire-rated
construction
Maximum acceptable incident heat flux for a cable, component, structural element, or
secondary combustible
Maximum acceptable exposure temperature for a cable, component, structural element,

or secondary combustible
Maximum acceptable enclosure or key structural element temperatures
2-2
Maximum smoke concentration or minimum visibility
Maximum or minimum concentration of one or more gas constituents, such as carbon

monoxide, oxygen, hydrogen cyanide, etc.
The performance criteria may also involve sequences of events, such as detection or sprinkler
actuation before cable damage, which occurs when the surface temperature exceeds a
specified threshold value. NUREG/CR-6850 (EPRI 1011989) provides some performance
thresholds for common NPP targets (see Appendix H) as well as for habitability (see Section
11.5.2.11).
A few simple examples will illustrate the various ways in which a fire modeling goal may be
stated. In many NPP fire modeling applications, the motivation for a fire modeling analysis is a
need to know whether or not an electrical cable or a component remains free of damage from a
fire. This could be very specific (i.e., a particular exposure fire exposing a particular cable) or
general (i.e., the maximum distance from which a particular type of fire could damage cables).
In addition, it may only be necessary for a single fire to damage a single cable, or it may be
necessary to simultaneously damage two particular cables with a known separation. The
following are examples of goals that would lead to a fire modeling analysis:
Ensure that Panel X and Cable Y are not subject to failure given a fire in Room Z.
Tabulate the distance through which an electrical cabinet ignition source could
adversely affect electrical targets for use as a screening tool in support of a PRA.
Ensure that Cable X and Cable Y both do not fail in Room Z when exposed to a
single transient fuel package fire.
Quantify the potential benefit of the sprinkler system in Room X for preventing damage
to Cable Y.
Each of the goals explicitly states the purpose of the analysis and the means by which success
is determined in terms that can be achieved by a fire modeling analysis. The objectives are
more specific and may include specific steps that will be followed to satisfy the goal. Some
examples of objectives that support the previously cited example goals are as follows:
Perform a calculation to determine if a fire in Fire Area X involving Panel Y could

cause the surface temperature of Cable Z to exceed a specified threshold value.
Evaluate the maximum distance from any surface of an electrical cabinet at which a 98th
percentile heat release rate (HRR) fire in Fire Zone X could cause the surface
temperature of a cable to exceed a specified threshold value.
Perform a calculation to determine if a fire in Fire Area X involving a transient fuel

package could simultaneously cause the surface temperature of both Cables Y and Z
to exceed a specified threshold value.
Determine if any ignition sources in Fire Zone X could damage cables in Raceway Y
by causing the surface temperature of any cables in the raceway to exceed a specified
threshold value before the sprinkler system actuates.
2-3

Thus, the objectives define in specific terms the means by which a goal will be fulfilled and the
means by which success is determined.
2.2 Step 2: Characterize Fire Scenarios

The second step in the fire modeling process is to characterize the relevant fire scenarios that
capture those technical elements necessary to address the goals. A fire scenario is defined
within this guide as a set of elements needed to describe a fire event. These elements usually
include the following:
the enclosure details (i.e., compartment)
the fire location within the enclosure
the fire protection features that will be credited
the ventilation conditions
the target locations
the secondary combustibles
the fire, which is sometimes referred to as the ignition source
A number of the fire scenario elements may also be viewed as fire model input. This section
provides a broad perspective on the considerations that apply when formulating the appropriate
fire scenario, given a fire modeling goal. Chapter 3 provides additional guidance on specific fire
scenario elements as they apply to various fire modeling goals and objectives evaluated with a
particular fire model.
Note that when characterizing the fire scenarios, preliminary consideration should also be given
to how many scenarios are needed to address a particular goal and which specific fire event
characteristics each scenario should capture (i.e., which scenarios are needed). In general, at
least one fire scenario would be necessary to assess the effects for a single ignition sourcetarget set pair. The analyst should become familiar with the information necessary to develop
input files for the fire modeling tools. In practice, this information should be collected during the
process of selecting and describing fire scenarios to minimize the number of walkdowns and
document/drawing reviews.
2.2.1 General Considerations

Various documents provide guidance for describing fire scenarios from a technical and
regulatory perspective. Most of these documents are application-specific; for instance, NFPA
805 defines two general categories of fire scenarios, limiting fire scenarios (LFSs) and
maximum expected fire scenarios (MEFSs). According to NFPA 805, the input values
necessary to determine the MEFS are those that lead to the most challenging fire that could
reasonably be anticipated for the occupancy type and conditions in the space. For LFSs, one
or more of the inputs to the fire modeling calculation (e.g., heat release rate, initiation location,
or ventilation rate) are varied to the point that the performance criterion is not met. If there is a
2-4

sufficiently large" margin between the MEFS and the LFS, then the fire scenario can be
screened from the risk assessment (Gallucci, 2011).
In a Fire PRA, the goal is to quantify the risk contribution from individual scenarios and to
identify potential risk-contributing scenarios (e.g., fires impacting important targets in the
compartment). Although specific elements in the scenario selection process are standardized
for guidance and completeness purposes, a certain degree of fire protection engineering
judgment is also necessary. NUREG/CR-6850 (EPRI 1011989) contains information on fire
frequency, cable (target) selection, HRR, damage criteria, and other information that is useful in
developing fire scenarios.
Selected scenarios should represent a complete set of fire conditions that are important to the
fire modeling goal. For example, if the goal of the fire modeling analysis is to estimate whether
specific cable(s) will remain free of fire damage, the analyst should consider exposures that are
close to the cables as well as exposures that are farther away. A small, localized fire exposure
could be a greater challenge than a larger fire that is farther away, or vice versa. It may not
always be appropriate to select, or at times even possible to define, the worst case fire scenario
prior to conducting the analysis, due to the different exposure mechanisms associated with
various ignition sources. In large enclosures with a limited number of targets to protect, such as
a turbine building in an NPP when the protection of a safety-related circuit is the fire modeling
goal, it is easier to locate the targets of interest and then identify those fire sources capable of
affecting that target.
When attempting to characterize the fire scenario, plant walkdowns should be an essential
aspect of the scenario selection. Many key decisions relevant to fire modeling, including those
related to model selection and input parameters, are influenced by observations made during
walkdowns. The occupants, the access level to a particular area, and the fire brigade/fire
department access should be observed during the walkdowns, as applicable.
It should also be noted that not all the elements associated with a commercial NPP fire scenario
can be directly modeled using the tools within the scope of this guide (e.g., the effect of
suppression activities by the fire brigade or the conditions in a space after a sprinkler system
has actuated). It is important, however, not to limit the scenario selection and description to
those elements that can be modeled.
2.2.2 Enclosure Details

The enclosure details include the identity of the enclosures that belong in the fire model
analysis, the physical dimensions of the enclosures included in the fire model, and the boundary
materials of each enclosure. The enclosure(s) that belong in the fire model may depend on the
fire modeling goal, the complexity and connectivity of the spaces in the general area of interest,
the type of analysis conducted, and the type of fire model selected. It is possible that no
enclosure may be involved, as would be the case for an exterior transformer fire. As a
minimum, the space containing the fire would normally be included in the fire model, though
treatments involving algebraic plume temperature or flame height correlations would not model
the enclosure effects per se. Multiple enclosures might be necessary if there are flow
connections (natural or forced) to adjacent areas and if the conditions in both areas could affect
the analysis results or are of interest. Care should be taken to consider the potential effects of
fires in adjacent areas on the targets of interest. In some cases, a heating, ventilation, and air
conditioning (HVAC) recirculation system may involve areas that are fairly remote from the area
of interest. Depending on the type of analysis conducted, the conditions within either or both
areas may be of interest, and the fire model would thus include both spaces.
2-5

The physical dimensions of the enclosure and the boundary materials are model input and
should be determined once the fire model has been selected, since the level of detail varies
considerably among the fire models. One-zone models may only require a volume and
boundary area; two-zone models will typically require the length, width, and height; and
computational fluid dynamics (CFD)-type models will require details commensurate with the
model grid resolution. The determination of the correct physical dimensions and boundary
materials are described in Chapter 3 for various types of NPP fire scenarios.
2.2.3 Fire Location

The location of the fire will depend strongly on the fire modeling goal, the target location, and
the fire modeling tool selected. For example, when evaluating the performance of a fire barrier
system, fire scenarios challenging the barriers are of interest; when conducting a risk analysis,
fire scenarios impacting safety-related circuits may be of primary interest. The selected
scenarios for these two applications may not be the same.
When selecting the fire location, the fire scenario should challenge the conditions being
estimated. For example, if the goal is to evaluate flame radiation to a target, locating the ignition
source relatively far from the target may not provide the best representation of the fire hazards.
If the goal is to determine whether a fire can cause two circuits in different raceways to fail, it
may be appropriate to locate the fire between the two raceways. There will be situations in
which the target location is fixed within the plan area of a space, but there is some flexibility in
the vertical placement. A good example of this is an electrical cabinet fire. For a given
electrical cabinet, the floor position is fixed; however, the base of the fire is not. Depending on
the type of cabinet, it may be appropriate to locate the fire base at the cabinet floor (e.g., open
back and containing thermoplastic (TP) cables), at the top (open top and no side vents), or
somewhere in between. NUREG/CR-6850 Supplement 1 (EPRI 1019259), Fire Probabilistic
Risk Assessment Methods Enhancements, recommends a fire elevation equal to the top of the
upper vent or 0.3 m (1 ft) below the cabinet top for cabinets meeting certain physical
constraints.
In the case of transient fuel package fires or other types of fires that are not fixed, some
consideration of the effects of the wall or corner on the upper gas layer temperature is
necessary. If the primary exposure mechanism is the Hot Gas Layer (HGL), then locating the
fire in a corner or near a vertical boundary will produce higher HGL temperatures. However, the
type of analysis may dictate that multiple locations be postulated not necessarily yielding the
worst fire effects. Other features that affect the fire location could include the presence or slope
of a floor, particularly when a melting plastic or liquid hydrocarbon fuel is considered and
transient fuel packages may be staged on mezzanine levels, scaffolding, or platforms. The
October 19, 1989, turbine lubricating oil fire at the Vandellos NPP in Spain illustrated this
problem. Additional information can be found in NUREG/CR-6738, Risk Methods Insights
Gained from Fire Incidents.
The following general guidelines and considerations for locating the fire for different fire
exposure mechanisms may be followed as applicable:
Targets in the fire plume or ceiling jet. Locating a source on top of a cabinet ignition
source usually results in the most severe fire conditions, since the cabinet walls will not
affect fire-generated conditions. Furthermore, since the fire is located in the highest
possible position, flames are expected to be higher, and temperatures in the plume and
ceiling jet will also be high. The user should judge whether this is conservative based on
2-6

the goal of the analysis. For example, this would not necessarily be conservative if
detection of the fire was a critical aspect of the analysis.
Targets affected by flame radiation. Combustible materials that are not fixed, such as
transient fuel packages and unconfined liquid spills, should be located so that there is an
unobstructed (when no passive fire protection system is credited) view between the
source and the target. A horizontal path between flame and target provides the highest
heat flux to the target.
Targets engulfed in flames. Flame height calculations should be performed to determine

whether the selected location will result in targets engulfed in flames. Proper justification
should be provided as to the location of the fire to ensure that the target is out of the
flames. For example, consider the case where the analyst locates the fire on top of an
enclosed cabinet, resulting in a cable tray engulfed in flames. This would represent the
most severe exposure for the cable tray since the fire is expected to start somewhere
inside the cabinet. The analyst may choose to lower the fires position and ignore the
cabinet walls after a visual examination identifies the actual location of the combustibles.
Targets immersed in the HGL. The fires elevation may influence how far down the HGL
will develop as predicted by some fire models, although other important scenario
characteristics will also be influential.
2.2.4 Credited Fire Protection

The fire protection features that will be credited in a fire modeling analysis usually require a fire
protection engineering evaluation of the systems effectiveness in performing its design
objectives. This may include both an assessment of the system compliance with applicable
codes, including maintenance and inspection, and an assessment of the system performance
against the particular fire scenario considered. The evaluation should determine whether the
detection, suppression, and/or passive systems can protect the selected target from firegenerated conditions. Once the decision to credit a fire protection system is made, the analyst
should specify the type of system selected for the scenario.
There are several common fire protection features that may be present in a typical NPP area:
Fire detection systems. These include smoke detectors, heat detectors, or high
sensitivity detection systems
Fire suppression systems. These include automatic or manually activated fixed

systems, fire extinguishers, and fire brigades
Passive fire protection systems. These include structural fire barriers, fire doors,
Electrical Raceway Fire Barrier Systems (ERFBSs), radiant shields, and fire stops
Administrative controls. These include combustible or transient-free zones, combustible

fuel load limits, and hotwork procedures
When assessing the performance of a system against the postulated fire hazard, it is necessary
to consider the conditions under which the system is designed (fire size, fuel load, exposure
temperature, plant operation mode, etc.). For example, an ordinary hazard sprinkler system
may not have a sufficient water spray delivery to protect against a large hydrocarbon pool fire.
Another example would be passive fire protection systems that are rated against an ASTM
2-7

E119 (2008) fire exposure. Under some exposure conditions, such as prolonged flame
impingement or large hydrocarbon pool fire scenarios, such systems may not provide sufficient
fire resistance. There may be circumstances in which the thermal material properties are well
defined, such that the response of the passive systems can be analytically determined. In other
situations, it may be necessary to specify at the start of the fire modeling analysis that the
passive systems do not protect the target or prevent fire propagation because the exposure
exceeds the design capability of the passive fire protection feature.
When considering the effect of active fire protection features, a valid set of response
characteristics of the system is needed. For manually actuated fixed suppression systems or
manual intervention in the fire modeling analysis, additional information relating to the
occupants, the fire brigade, and the fire department are usually necessary. This may include
the means of access to the area considered, the presence of a fire watch, the potential for plant
personnel to be in the area, etc. Notice that the fire modeling tools within the scope of this
guide may not be capable of modeling the impact of some of the fire protection features that
may be credited in a given scenario. Credit for active fire suppression may thus be limited to
preventing further increases in the HRR or alteration of the HRR profile using suppression
curves such as those described in Appendix P of NUREG/CR-6850 (EPRI 1011989) and
Chapter 14 of NUREG/CR-6850 Supplement 1 (EPRI 1019259). Nevertheless, fire protection
features are designed to impact the outcome of a scenario, so their effects should be included in
the analysis.
2.2.5 Ventilation Conditions

Ventilation conditions collectively refer to the operation of the mechanical ventilation system
(e.g., the system will continue in normal operational mode, the system will transfer to smoke
purge mode, the system will transfer off with closed dampers, etc.) and the position of doors or
other openings during the fire event (e.g., doors closed, doors open, doors opening at fire
brigade arrival, etc.). Typically, both normal and off-normal ventilation conditions are
considered. Spaces in which doors are normally closed may have the doors propped open or
opened during the fire by plant personnel, or damaged during the fire. HVAC flows that are
normally present in a space may change during the fire due to dampers closing, activation of
purge modes, filter plugging, or fan damage by the hot gases. Characterization of the flow field
from mechanical devices may be important in some scenarios, especially if the inlet or outlet of
the mechanical system is in close proximity to the fire or target.
2.2.6 Target Locations

The target location refers to the physical dimensions of the target relative to the source fire or
the fire model coordinate system. These could include its horizontal and vertical distances from
the ignition source or source fire, or its spatial position within the room itself. It may be
necessary to further specify the location of a vulnerable portion of a target, such as the junction
box on a service water pump motor. The orientation of the target with respect to the exposure
fire may be of interest as well. An elevated target that is exposed only in the vertically upward
direction may be susceptible to thermal radiation from an HGL, but possibly shielded from
thermal radiation from the source fire itself. Note that in some types of analyses (e.g., a control
room abandonment calculation), occupants may be a target.
The fire exposure mechanisms should also be assessed when quantifying the target location.
Fire exposure mechanisms, such as flame impingement, fire plume, ceiling jets, HGLs, and
flame radiation, should be considered based on the relative location of the ignition source,
2-8

intervening combustibles, and the targets. The subsequent fire model analysis should quantify
relevant fire conditions and include a discussion of the proper disposition of those that are not
expected to affect the target.
2.2.7 Secondary Combustibles

Secondary combustibles include any combustible materials that, if ignited, could affect the
exposure conditions of the target set considered. Intervening combustibles, which are typically
those combustibles that are located between the ignition source or source fire and the target,
are examples of secondary combustibles. However, secondary combustibles would also
include combustible materials that are not between the fire and the target but are exposed to the
fire effects. In this case, if the secondary combustibles were to ignite, the total HRR in the
enclosure (if applicable) would increase, resulting in a hotter gas layer, and the radiant energy
from the burning secondary combustible would augment the exposure from the initial source
fire, regardless of its location.
Secondary combustibles would include both fixed and transient materials. Typical fixed
combustibles include exposed cable jackets or cable insulation, combustible thermal insulation,
and combustible wall lining materials. Transient combustibles vary considerably from plant to
plant and plant area to plant area, but they may include trash containers, waste accumulations,
hoses, hand tools, cleaners and solvents, protective clothing, plastic containers, and so on. It is
essential to perform a visual survey of an area to obtain an understanding of the types of
combustibles present and the activities in the space, which can provide insight into the types of
combustibles that may be present from time to time. The combustible load calculations, fire
protection procedures, and fire hazards analysis could provide additional details on the nature of
fixed and transient fuel packages in a particular plant area.
Combustible materials in sealed or rated containers may be excluded from consideration if the
container is capable of resisting the effects of the fire. Some examples include cabinets
containing flammable liquids, solid bottom cable trays with fixed top covers, and bus ducts.
Secondary combustibles will take on the characteristics of a target prior to their ignition (see
Section 2.2.6). In this regard, the physical location and orientation of the secondary
combustibles with respect to both the ignition source and the target set are determined. The
performance criterion for a secondary combustible target is the ignition condition, which will
usually be a critical radiant heat flux or exposure temperature or an integrated heat flux. Unlike
a true target, once the performance criterion has been met, the secondary combustible is ignited
and then takes on the characteristics of a second source fire (see Section 2.2.8).
2.2.8 Source Fire

The source fire is the forcing function for the fire scenario. As all fire effects are directly related
to the characterization of the source fire, great care must be taken in characterizing it. A source
fire is often described as the ignition source, which introduces the concept of having both a
fuel package and a credible ignition mechanism. There are many ignition mechanisms in a
nuclear power plant; however, ignition sources are typically grouped by component or item such
as into electrical cabinets, transient fuel packages, self-ignited cable trays, hotwork-ignited cable
fires, and overheated motors.
2-9

The source fire is typically characterized by a HRR, though other important aspects include the
physical dimensions of the burning object, its composition, and its behavior when burning. The
HRR may be specified as a continuous function of time (e.g., proportional to the square of time
or a t2 fire), or it may be an array of HRR and time data. Algebraic models may only permit a
constant HRR. There may be situations in which the HRR is a function of the ventilation rather
than the object burning. Burning behaviors that may need consideration include whether the
material can melt and form a liquid pool, whether it can spread by dripping, and where a liquid
could pool. The HRR for many common ignition sources may be developed using the guidance
provided in both NUREG/CR-6850 (EPRI 1011989) and NUREG/CR-6850 Supplement 1 (EPRI
1019259). A summary of the information contained in these documents that may be used to
develop HRR profiles is as follows:
Appendix E of NUREG/CR-6850 (EPRI 1011989) provides distributions of peak HRR

values as a function of the types (qualified vs. unqualified) and amounts (single vs.
multiple bundles) of cables inside vertical electrical cabinets
Appendix G of NUREG/CR-6850 (EPRI 1011989) provides guidance on defining the

transient HRR profiles for electrical cabinets, transient combustible materials, and
flammable/combustible liquids
Appendix R of NUREG/CR-6850 (EPRI 1011989) provides guidance on calculating the

flame spread and HRR from cable fires
Appendix S of NUREG/CR-6850 (EPRI 1011989) provides guidance on estimating fire

propagation times between electrical cabinets, which would affect the total HRR profile
for a given area
Chapter 8 of NUREG/CR-6850 Supplement 1 (EPRI 1019259) provides further guidance

on fire propagation among groups of electrical cabinets
Chapter 11 of NUREG/CR-6850 Supplement 1 (EPRI 1019259) provides further

guidance on cable tray fires, including applicability of the Appendix R of NUREG/CR6850 cable fire parameters
Chapter 17 of NUREG/CR-6850 Supplement 1 (EPRI 1019259) provides further

guidance on the fire growth rate for transient combustible materials
When fire modeling is used to support a Fire PRA, the HRR for a source fire may be
represented as a conditional probability distribution. In this case, depending on the type of
analysis, a screening value may be selected (e.g., a 98th percentile peak HRR as recommended
in NUREG/CR-6850), or the effects may be represented using multiple points on the conditional
probability distribution.
As was the case with secondary combustibles, combustible materials in sealed or rated
containers may be excluded from consideration if the container is capable of resisting the
effects of the fire. In addition, self-ignited cable fires are generally postulated only for non-IEEE383 qualified power cables (NUREG/CR-6850/EPRI 1011989).
2.3 Step 3: Select Fire Models

A number of models are available for performing fire simulations. These models range from
algebraic models to sophisticated CFD computer codes that require days to set up a scenario
and perform the associated calculations. Given the availability of different models, the analyst is
2-10

responsible for understanding the advantages and limitations of a particular model in a specific
situation in order to achieve the established goals. In general, fire models can be classified into
three groups: (1) algebraic models, (2) zone models, and (3) CFD models. The level of effort
required to describe a scenario and the computational time consumed by each group increase
in the order in which they are listed. Table 2-1 provides a summary of the three groups of
models, their advantages and disadvantages, and typical applications.
In practical fire modeling applications, it is likely that a combination of all three types of models
would be useful for analyzing a specific problem. For example, algebraic models might be used
to estimate the radiative flux to a target for determination of a zone of influence (ZOI) or
minimum separation distance. A zone model would provide the temperature of the HGL and
height as a function of time for evaluating cable temperatures. CFD model calculations could be
used to provide more detailed information on fire-induced conditions in areas where the
algebraic models and zone models are not conclusive. Complex models can also be used as a
means of estimating the degree of conservatism in a simple model analysis.
The first step in selecting a model is to determine whether the scenario can be analyzed using
algebraic models, zone models, or CFD models. This guide focuses on the following five fire
models: Fire Dynamics Tools (FDTs) (NUREG-1805, 2004), Fire Induced Vulnerability
Evaluation (FIVE-Rev1) (EPRI 1002981, 2002), Consolidated Fire Growth and Smoke Transport
(CFAST) model (Jones et al., 2004), MAGIC (Gay et al., 2005), and Fire Dynamics Simulator
(FDS) (McGrattan et al., 2009). The FDTs and FIVE-Rev1 are a set of relatively simple
algebraic models codified in the form of electronic spreadsheets. CFAST and MAGIC represent
the class of fire models commonly referred to as zone models, which divide a compartment of
interest into two zones, an elevated temperature upper layer and a cool lower layer. FDS is an
example of a CFD model, which divides each compartment into thousands or millions of cells.
Temperatures and other quantities of interest are calculated for each cell.
Algebraic models can be performed by hand with relatively little computational effort. Karlsson
and Quintiere (2000) classify algebraic models into three categories: (1) those that deal with
combustion, (2) those that estimate resultant environmental conditions, and (3) those that
address heat transfer. Algebraic models related to the combustion process estimate fire
intensity based on the flammability characteristics of the fuel. Equations that estimate firegenerated conditions include plume, ceiling jet, and compartment temperatures. Heat transfer
equations deal with target temperatures and heat fluxes in the plume, ceiling jet, and lower and
upper layer regions.
Zone models are computer algorithms that solve conservation equations for energy and mass.
The fundamental approximation associated with zone models is that the enclosure is divided
into a limited number of distinct gas zones of uniform properties. In fire applications, the
enclosure is usually divided into two zones. The HGL is the volume of smoke generated by the
fire and accumulated below the ceiling of the enclosure. This layer is modeled as
homogeneous and will have uniform density and temperature. Its temperature and depth are
affected by the amount of mass and energy entering or leaving the volume in each time step
during the simulation. The lower layer, which can also experience a temperature increase, is
characterized by colder fresh air between the floor and the bottom of the HGL. This layer will
also have uniform density and temperature.
CFD models are sophisticated algorithms that solve a simplified version of the Navier-Stokes
equations. To run CFD codes, the enclosure must be divided into a large number of control
volumes (perhaps several million), and the equations solved for each control volume. CFD
models then provide a detailed estimate of temperature profiles because calculations are
2-11

performed for each control volume specified in the enclosure. CFD models also handle
turbulent gas flows. Another advantage of CFD models is their ability to simulate fire conditions
in geometries other than rectangular floor compartments with flat ceilings. Some CFD models
also attempt to estimate HRR values based on fuel flammability properties provided by the
analyst. The drawback of CFD models is the computational time and the level of effort required
to set up a scenario, as computational times are usually on the order of days. The time required
to set up a problem usually depends on the complexity of the geometry.
Another consideration when selecting a CFD-type model is that the amount of detail supplied to
the model is significantly greater than it is for the simpler empirical and zone models. Given the
large amount of information required for input, there is an intrinsically higher likelihood of errors
being introduced into the input, which is different from the model uncertainty and parameter
uncertainty described in Chapter 4. Furthermore, the features that may be described in the
input could include ductwork, cable trays, electrical cabinets, and other fixed contents that may
later be modified, relocated, or removed. New cabinets, cable trays, or other fixed contents that
would have been included in the fire model had they been present may be added to an area.
Although these changes may be minor, at the very least they would require an assessment by a
fire modeler as to whether the original analysis is still applicable or whether the model needs to
be adapted for the change. In some situations, such as the determination of a sprinkler
actuation time, such small modifications could have a significant effect on the model results.
2-12
Table 2-1. Summary of common fire model tools.

Fire Model
Class
Algebraic
models
Examples
Typical Applications
Advantages
FDTS
FIVERev1
Screening calculations;
ZOI; target damage by
thermal radiation, HGL,
or thermal plume acting
in isolation.
Simple to use; minimal

inputs; quick results;
ability to perform
multiple parameter
sensitivity studies.
Zone Model
CFAST
MAGIC
Detailed fire modeling

in simple geometries;
often used to compute
HGL temperatures and
target heat fluxes.
Simple to use; couples

HGL and localized
effects; quick results;
ability to perform
multiple parameter
sensitivity studies.
Computation
Fluid
Dynamics
Model
FDS
Detailed fire modeling

in complex geometries,
including computing
time to target damage
and habitability (main
control room (MCR)
abandonment or
manual action
feasibility).
Ability to simulate fire

conditions in complex
geometries and with
complex vent
conditions.
Disadvantages
Limited
application range;
treats phenomena
in isolation;
typically
applicable only to
steady state or
simply defined
transient fires
(e.g., proportional
to the square of
time or t fires).
Error increases
with increasing
deviation from a
rectangular
enclosure; large
horizontal flow
paths not well
treated.
Significant effort
to create input
files and postprocess the
results; long
simulation times;
difficult to model
curved geometry,
smoke detector
performance, and
conditions after
sprinkler
actuation.
An important consideration in the fire model selection process is the type of analysis performed.
Because of the large number of potential scenarios in a typical NPP, it is usually not practical to
default to the most sophisticated tool available. The analysis typically begins with a series of
screening analyses (NUREG/CR-6850/EPRI 1011989) performed to identify a subset of fire
scenarios and targets that require further analysis with greater resolution. The screening
process will typically use fairly simple fire modeling tools, such as algebraic models or generic
solutions. When such screening is conducted, it is important to remain within the model
limitations and the verification and validation (V&V) basis for the screening model. Section 2.3.6
and Chapter 4 provide additional guidance on the significance of the fire model V&V basis and
steps that the user should take to ensure that the fire model is used within acceptable limits.
2-13
2.3.1 Fire Dynamics Tools (FDTs)

Fire Dynamics Tools (FDTs) is a set of algebraic models preprogrammed into Microsoft Excel
spreadsheets. The FDTs library is documented in NUREG-1805, Fire Dynamics Tools (FDTs):
Quantitative Fire Hazard Analysis Methods for the U.S. Nuclear Regulatory Commission Fire
Protection Inspection Program (NUREG-1805, 2004) and Supplement 1 (NUREG-1805
Supplement 1, 2012). The primary objective of the FDTs library and the accompanying
documentation is to provide a methodology for U.S. Nuclear Regulatory Commission (NRC) fire
protection inspectors to use in assessing potential fire hazards in NRC-licensed NPPs. The
methodology uses simplified, quantitative fire hazard analysis techniques to evaluate the
potential hazard associated with credible fire scenarios.
The FDTs library includes a suite of spreadsheets that can be used to calculate various fire
parameters under varying conditions. Documentation of the theoretical bases underlying the
equations used in the FDTs spreadsheets helps to ensure that users understand the
significance of the inputs that each spreadsheet requires, and why a particular spreadsheet
should (or should not) be selected for a specific analysis. The governing equations for FDTs are
well established within the fire science community and are documented in handbooks and
scientific publications, such as the NFPA Fire Protection Handbook (NFPA Handbook, 2008),
the SFPE Handbook of Fire Protection Engineering (SFPE Handbook, 2008), and other fire
science literature.
The complete list of spreadsheets included in the FDTs library is shown in Table 2-2. A number
of the calculation methods included in the FDTs were part of the V&V study conducted by the
NRC, the Electric Power Research Institute (EPRI), and the National Institute of Standards and
Technology (NIST) (NUREG-1824 Volume 3, EPRI 1011999, 2007). These spreadsheets are
identified in Table 2-2. The NRC maintains a website at http://www.nrc.gov/reading-rm/doccollections/nuregs/staff/sr1805/final-report/index.html, where both new and updated
spreadsheets are posted.
Table 2-2. Routines included in the FDTs.
FDTS Spreadsheet
Function Name
02.1_Temperature_NV.xls
02.2_Temperature_FV.xls
02.3_Temperature_CC.xls
03_HRR_Flame_Height_
Burning_Duration_Calculation.
xls
NUREG-1805 Chapter and Function

Description
Chapter 2. Predicting Hot Gas Layer Temperature and
Smoke Layer Height in a Compartment Fire with Natural
Ventilation (Compartment with Thermally Thick/Thin
Boundaries): Method of McCaffrey, Quintiere, and
Harkleroad (MQH)
Chapter 2. Predicting Hot Gas Layer Temperature in a
Compartment Fire with Forced Ventilation
(Compartment with Thermally Thick/Thin Boundaries):
Method of Foote, Pagni, and Alvares (FPA) and
Method of Deal and Beyler
Chapter 2. Predicting Hot Gas Layer Temperature in a
Compartment Fire with Door Closed (Compartment has
Sufficient Leaks to Prevent Pressure Buildup; leakage is
Ignored): Method of Beyler
Chapter 3. Estimating Burning Characteristics of Liquid
Pool Fire, HRR, Burning Duration, and Flame Height
2-14
NUREG-1824
Verification
and Validation
Status
V&V provided
V&V provided
V&V provided
V&V provided for

flame height only

FDTS Spreadsheet
Function Name

Description
NUREG-1824
Verification
and Validation
Status
04_Flame_Height_Calculations
.xls
Chapter 4. Estimating Wall Fire Flame Height, Line

Fire Flame Height Against the Wall, and Corner Fire
Flame Height
V&V not provided
05.1_Heat_Flux_Calculations_
Wind_Free.xls
Chapter 5. Estimating Radiant Heat Flux from Fire to a

Target Fuel (Wind-Free Condition):
Point Source Radiation Model (Target at Ground Level);
Solid Flame Radiation Model (Target at Ground Level);
and Solid Flame Radiation Model (Target Above
Ground Level)
V&V provided for

the point source
model and the
solid flame
radiation model
(above ground)
only
05.2_Heat_Flux_Calculations_
Wind.xls
05.3_Thermal_Radiation_From
_Hydrocarbon_Fireballs.xls
06_Ignition_Time_Calculations.
xls
07_Cable_HRR_
Calculations.xls
08_Burning_Duration_Soild.xls
09_Plume_Temperature_
Calculations.xls
10_Detector_Activation_
Time.xls
13_Compartment_Flashover_
Calculations.xls
14_Compartment_Over_
Pressure_Calculations.xls
15_Explosion_Claculations.xls

Target Fuel (Presence of Wind): Solid Flame Radiation
Model (Target at Ground Level); and Solid Flame
Radiation Model (Target Above Ground Level)
Target Fuel: Estimating Thermal Radiation from
Hydrocarbon Fireballs
Chapter 6. Estimating the Ignition Time of a Target
Fuel Exposed to a Constant Radiative Heat Flux
Method of Estimating Piloted Ignition Time of Solid
Materials Under Radiant Exposures Method of:
(1) Mikkola and Wichman;
(2) Quintiere and Harkleroad;
(3) Janssens;
(4) Method of Toal, Silcock, and Shields; and
(5) Method of Tewarson
Chapter 7. Estimating Full-Scale Heat Release Rate of
a Cable Tray Fire
Chapter 8. Estimating Burning Duration of Solid
Combustibles
Chapter 9. Estimating Centerline Temperature of a
Buoyant Fire Plume
Estimating Detector Response Times:
Chapter 10. Estimating Sprinkler Response Time
Chapter 11. Estimating Smoke Detector Response
Time
Chapter 12. Estimating Heat Detector Response Time
Chapter 13. Predicting Compartment Flashover
Compartment Post-Flashover Temperature: Method of
Law
Minimum Heat Release Rate Required to Compartment
Flashover:
(1) Method of McCaffrey, Quintiere, and Harkleroad
(MQH);
(2) Method of Babrauskas; and
(3) Method of Thomas
Chapter 14. Estimating Pressure Rise Attributable to a
Fire in a Closed Compartment
Chapter 15. Estimating the Pressure Increase and
Explosive Energy Release Associated with Explosions
2-15
V&V not provided
V&V not provided
V&V not provided
V&V not provided

V&V not provided
V&V provided
V&V not provided
V&V not provided
V&V not provided

V&V not provided

FDTS Spreadsheet
Function Name
16_Battery_Compartment_
Flammable_Gas_Conc.xls
17.1_FR_Beams_Columns_
Substitution_Correlation.xls
Quasi_Steady_State_Spray_
Insulated.xls
Quasi_Steady_State_Board_
Insulated.xls
Quasi_Steady_State_
Uninsulated.xls
18_Visibility_Through_Smoke.
xls

Description
Chapter 16. Calculating the Rate of Hydrogen Gas
Generation in Battery Compartments:
Method of Estimating Hydrogen Gas Generation Rate in
Battery Compartments;
Method of Estimating Flammable Gas and Vapor
Concentration Buildup in Enclosed Spaces; and
Method of Estimating Flammable Gas and Vapor
Concentration Buildup Time in Enclosed Spaces
Chapter 17. Calculating the Fire Resistance of
Structural Steel Members (Algebraic Models):
Beam Substitution Correlation (Spray-Applied
Materials); and
Column Substitution Correlation (Spray-Applied
Materials)
Structural Steel Members (Algebraic Models): Heat
Transfer Analysis using Numerical Methods Protected;
and Steel Beams and Columns (Spray-Applied)
Structural Steel Members: Heat Transfer Analysis using
Numerical Methods Protected Steel Beams and
Columns (Board Materials)
Structural Steel Members: Heat Transfer Analysis using
Numerical Methods Unprotected Steel Beams and
Columns
Chapter 18. Estimating Visibility Through Smoke
NUREG-1824
Verification
and Validation
Status
V&V not provided
V&V not provided
V&V not provided
V&V not provided
V&V not provided

V&V not provided
2.3.2 Fire Induced Vulnerability Evaluation (FIVE-Rev1)6

In August 2002, EPRI published the Fire Modeling Guide for Nuclear Power Plant Applications
(EPRI 1002981, 2002) for the first time. Since then, it has provided fire protection engineers in
the commercial nuclear industry with a broad overview of fire modeling theory and applications,
including representative calculations performed with various state-of-the-art fire models. With
this guide, EPRI included a library of preprogrammed Microsoft Excel equations, which are
used to estimate some aspects of fire-induced conditions. This collection of algebraic models is
referred to as the Fire Induced Vulnerability Evaluation model (FIVE-Rev1). In general, the
equations in the library are closed-form analytical expressions that can be solved by hand. The
capabilities of the various equations in the library include predicting temperature and convective
heat fluxes in the fire plume or ceiling jet, radiation heat flux, upper-layer temperature, time to
detection, and target heating, among others. Some of the equations in FIVE-Rev1 were
included in the V&V study (NUREG-1824 Volume 4, EPRI 1011999, 2007). Like the FDTs,
several of the equations used in the examples have not been subject to V&V. Subsequent
efforts will be directed at the V&V of these equations and models. The calculations included in
the FIVE-Rev1 are summarized in Table 2-3.
FIVE-Rev1 is a proprietary software package available to EPRI members (www.epri.com).
2-16
Table 2-3. Routines included in FIVE-Rev1.

FIVE-Rev1
Function
Qf
Firr
FHeight
TpAlpert
TpMcCaffrey
TpHeskestad
Plcflux
VpAlpert
VpMcCaffrey
VpHeskestad
EpZukoski
EpThomas
EpHeskestad
PdHeskestad
TcjAlpert
TcjDelichatsios
Cjcflux
VcjAlpert
MQHTemperature
MQHFlashover
FiveTemp
Detact
Aset
CThrr
Function Description
Heat release rate profile considering t growth and four stages.
Estimates flame irradiation at distance r from the fire source.
Point source approximation for REMOTE targets.
Flame height based on Heskestad flame height correlation.
Plume temperature at a specific height based on Alpert plume
temperature correlation.
Plume temperature at a specific height based on McCaffrey plume
Plume temperature at a specific height based on Heskestad
plume temperature correlation.
Estimates convective heat flux in the fire plume.
Plume velocity at a specific height based on Alpert's plume
Plume velocity at a specific height based on McCaffrey plume
Plume velocity at a specific height based on Heskestad plume
Air entrainment into plume based on Zukoski plume entrainment
correlation.
Air entrainment into plume based on Thomas plume entrainment
correlation.
Air entrainment into plume based on Heskestad plume
entrainment correlation.
Estimates plume diameter based on Heskestad plume correlation.
Unconfined ceiling jet temperature based on Alpert ceiling jet
correlation.
Confined ceiling jet temperature based on Delichatsios ceiling jet
correlation.
Estimates convective heat flux in the ceiling jet.
Unconfined ceiling jet velocity based on Alpert ceiling jet
correlation.
Compartment temperature after a specified time, given a steady
HRR based on MQH approach.
Heat release rate required for flashover after a specified time
based on MQH approach.
Estimates compartment temperature based on FIVE-Rev1.
Activation time of heat detection devices based on heat release
rate profiles.
Time required by Hot Gas Layer to reach a specific height based
on heat release rate profiles and openings at the bottom of the
enclosure.
Estimates heat release rate from cable trays. The correlation is
based on 14 experiments with a stack of 12 horizontal cable trays
and 2 experiments with a combination of 12 horizontal cable trays
and 3 vertical trays.
2-17
NUREG-1824
Verification
and Validation
Status
V&V not provided
V&V provided
V&V provided
V&V not provided
V&V not provided
V&V not provided
V&V not provided
V&V not provided
V&V provided
V&V provided
V&V not provided
V&V not provided
V&V not provided
V&V not provided
V&V provided
V&V not provided
V&V not provided
V&V not provided
V&V provided
V&V not provided
V&V provided
V&V not provided
V&V not provided
V&V not provided

FIVE-Rev1
Function
Visib
Ttar
Ttdam
Function Description
Estimates the length of a visible path in a smoke environment.
The correlation applies to light-reflecting signs.
Estimates target temperature under constant heat flux.
Time to target damage under constant heat flux.
NUREG-1824
Verification
and Validation
Status
V&V not provided
V&V not provided
V&V not provided
2.3.3 Consolidated Fire Growth and Smoke Transport (CFAST) Model

The Consolidated Fire Growth and Smoke Transport (CFAST) model is a two-zone computer
fire model. For a given fire scenario, the model subdivides a compartment into two control
volumes, which include a relatively hot upper layer (i.e., the HGL) and a relatively cool lower
layer. Mass and energy are transported between the layers via the fire plume and mixing at the
vents. Combustion products accumulate via the plume in the HGL. Each layer has its own
energy and mass balances. The most important approximation for the model is that each zone
has uniform properties, that is, that the temperature and gas concentrations are constant
throughout the zone, only changing as a function of time. The CFAST model describes the
conditions in each zone by solving equations for conservation of mass, species, and energy,
along with the ideal gas law. The Technical Reference Guide for CFAST (Jones et al., 2004)
provides a detailed discussion concerning the specific derivation of these conservation laws.
Documentation for CFAST also includes a Users Guide (Peacock et al., 2008b), which details
the use of the model, and a Model Development and Evaluation Guide (Peacock et al., 2008a),
which presents the latest model V&V results.
For some applications, including long hallways or tall shafts, the two-zone approximation may
not be appropriate. To address this, CFAST includes empirical algorithms to simulate smoke
flow and filling in long corridors and for a single well-mixed volume in tall shafts. CFAST also
includes several correlations (as sub-models), based on experimental data that are used to
calculate various physical processes during a fire scenario: smoke production, fire plume
dynamics, heat transfer by radiation, convection, conduction, natural flows through openings
(vertical and horizontal), forced or natural ventilation, thermal behavior of targets, heat
detectors, and water spray from sprinklers.
CFAST models horizontal flow through vertical vents (doors, windows, wall vents, etc.), vertical
flow through horizontal vents (ceiling holes, hatches, roof vents, etc.), and mechanical
ventilation through fans and ductwork. Natural flow is determined by the pressure difference
across a vent, using Bernoullis law for horizontal vent flow, and by algebraic models for vertical
vent flow. Mechanical ventilation is based on an analogy to electrical current flow in series and
parallel paths where flow is split in parallel paths proportional to the flow resistance in each path
and resistance to flow is additive for paths in series.
CFAST includes algorithms to account for radiation, convection, and conduction within a
modeled structure. Radiative transfer occurs among the fire(s), gas layers, and compartment
surfaces (ceiling, walls, and floor). It is a function of the temperature differences and emissivity
of the gas layers, as well as the compartment surfaces. Convective heat transfer between gas
layers and compartment or target surfaces is based on typical correlations available in the
literature. The V&V results for CFAST are documented in Volume 5 of NUREG-1824 (EPRI
1011999). Additional validation results, particularly for plume temperature predictions that were
not included in the NUREG-1824 (EPRI 1011999) Volume 5 results, are included in the CFAST
Model Development and Evaluation Guide (Peacock et al., 2008a).
2-18
2.3.4 MAGIC7
MAGIC is another two-zone computer fire model, developed and maintained by Electricit de
France (EdF) specifically for use in NPP analysis. MAGIC is supported by three EdF
publications, including: (1) the technical manual, which provides a mathematical description of
the model (Gay et al., 2005b); (2) the users manual, which details how to use the graphical
interface (Gay et al., 2005a); and (3) the validation studies, which compare MAGICs results to
experimental measurements (Gay et al., 2005c). These three proprietary publications and the
MAGIC software are available through EPRI to EPRI members. Additional V&V results for
MAGIC are documented in Volume 6 of NUREG-1824 (EPRI 1011999).
MAGIC is fundamentally the same type of model as CFAST and thus solves the same basic set
of differential equations. The combustion model and vent flow models are similar as well.
Despite this, MAGIC differs from CFAST in that it does not have the corridor or shaft submodels, and the ceiling jet and wall jet treatments are different. The user should consult the
technical manual for a complete description of the MAGIC sub-models (Gay et al., 2005b).
Once a given simulation is completed, MAGIC generates an output file with all of the solution
variables. Through a post-processor interface, the user selects the relevant output variables
for the analysis. Typical outputs include the temperatures of hot and cold zones, concentrations
of oxygen and unburned gases, smoke migration into each compartment, the mass flow rates of
air and smoke through the openings and vents, the pressures at the floor level of each
compartment, the temperatures at the surfaces of the walls, and the thermal fluxes (radiative
and total) exchanged by the targets placed by the user.
2.3.5 Fire Dynamics Simulator (FDS)

Fire Dynamics Simulator (FDS) (McGrattan et al., 2007) is a CFD model of fire-driven fluid flow.
The model numerically solves a form of the Navier-Stokes equations appropriate for low-speed,
thermally driven flow, with an emphasis on smoke and heat transport from fires. The partial
derivatives of the equations for conservation of mass, momentum, and energy are approximated
as finite differences, and the solution is updated in time on a three-dimensional, rectilinear grid.
Thermal radiation is computed using a finite volume technique on the same grid as the flow
solver. Lagrangian particles are used to simulate smoke movement and sprinkler discharge.
FDS computes the temperature, density, pressure, velocity, and chemical composition within
each numerical grid cell at each discrete time step. There are typically hundreds of thousands
to several million grid cells, and thousands to hundreds of thousands of time steps. In addition,
FDS computes the temperature, heat flux, mass loss rate, and various other quantities at solid
surfaces.
Time histories of various quantities at a single point in space, or global quantities, such as the
fires HRR, are saved in simple, comma-delimited text files that can be plotted in a spreadsheet
program. Some output quantities typically used to assess the conditions in a space or to
compare against measured data are not inherently calculated in a CFD-type model, such as the
HGL temperature and interface height. FDS does provide an estimate of these types of
parameters using the continuous vertical temperature distribution and an integrated averaging
scheme, as described by McGrattan et al. (2007).
Most field or surface data are visualized with a program called Smokeview, a tool specifically
designed to help analyze results generated by FDS. FDS and Smokeview are used in concert
7
MAGIC is a software package available to EPRI members (www.epri.com).
2-19

to model and visualize fire phenomena. Smokeview does this by presenting animated tracer
particle flows, animated contour slices of computed gas variables, and animated surface data,
and also presents contours and vector plots of static data anywhere within a scene at a fixed
time. The FDS Users Guide (McGrattan et al., 2007) provides a complete list of FDS output
quantities and formats, while the Smokeview Users Guide (Forney, 2008) explains how to
visualize the results of an FDS simulation. Volume 7 of NUREG-1824 (EPRI 1011999) contains
the results of V&V efforts for FDS. Additional V&V results for FDS are contained in the FDS
documentation series (McGrattan et al., 2007).
FDS solves conservation equations of mass, momentum, and energy for an expandable mixture
of ideal gases in the low Mach number limit. This means that the equations do not permit
acoustic waves, the result of which is that the time step for the numerical solution is bounded by
the flow speed, rather than the sound speed. Situations in which this limitation may be
encountered include jet fires, deflagrations, and detonations. This limitation also reduces the
number of unknowns by one, as density and temperature can be related to a known background
pressure. Flow turbulence is treated by large eddy simulation.
For most simulations, FDS uses a mixture fraction combustion model. The mixture fraction is a
conserved scalar that represents, at a given point, the mass fraction of gases originating in the
fuel stream. In short, the combustion is controlled by the rate at which fuel and oxygen mix.
Unlike versions of FDS prior to version 5, the reaction of fuel and oxygen is not necessarily
instantaneous and complete, and there are several optional schemes that are designed to
estimate the extent of combustion in under-ventilated spaces. The mass fractions of all of the
major reactants and products can be derived from the mixture fraction by means of state
relations, expressions achieved through a combination of simplified analysis and measurement.
The combustion model used by FDS is an area of active development. Consequently, FDS
users should consult the latest code documentation for a description of new features or submodels.
Numerical parameters play a very important role in a CFD model like FDS. A numerical
parameter is any input value that is needed for the mathematical solution of the equations, but
has little or no physical meaning. For example, the time step with which the numerical solution
of the HGL temperature is computed does have units of seconds, but it is not a value that has
meaning outside of that particular algorithm; nevertheless, these numerical parameters can
affect the solution, and their sensitivity should be assessed in some way. For the spreadsheet
and zone models, this procedure is relatively straightforward because the calculations run in
less than a minute. One simply varies the value and ensures that the solution does not change
appreciably. Specifically, one should simply demonstrate that the solution converges towards a
particular value as the parameter is varied; for instance, using a smaller and smaller time step
ought to lead to convergence of any evolution equation.
In FDS, the numerical parameter with the greatest importance is cell size. CFD models solve
an approximate form of the conservation equations of mass, momentum, and energy on a
numerical grid. The error associated with the discretization of the partial derivatives is a function
of the size of the grid cells and the type of differencing used. FDS uses second-order accurate
approximations of both the temporal and spatial derivatives of the Navier-Stokes equations,
meaning that the discretization error is proportional to the square of the time step or cell size. In
theory, reducing the grid cell size by a factor of two reduces the discretization error by a factor
of four; however, it also increases the computing time by a factor of at least sixteen (a factor of
two for the temporal and each spatial dimension). Clearly, there is a point in diminishing returns
2-20

as one refines the numerical mesh. Determining which size grid cell to use in any given
calculation is known as a grid sensitivity study.
Determining an optimal grid size in FDS is usually a matter of assessing the size of the fire. The
physical diameter of the fire is not always a well-defined property; a compartment fire does not
have a well-defined diameter, whereas a circular pan filled with a burning liquid fuel has an
obvious diameter. Regardless, it is not the physical diameter of the fire that matters when
assessing the size of the fire, but rather its characteristic diameter, :
/
(2-1)
where is the fire HRR (kW), is the ambient density of air (kg/m3),
is the specific heat of
air (kJ/kg/K), is the ambient air temperature (K), and is the acceleration of gravity (m/s2).
In many instances,
is comparable to the physical diameter of the fire. FDS employs a
numerical technique known as large eddy simulation (LES) to model the unresolvable or subgrid motion of the hot gases. The effectiveness of the technique is largely a function of the ratio of
the fires characteristic diameter, , to the size of a grid cell, . In short, the greater the ratio
/ , the more the fire dynamics are resolved directly, and the more accurate the simulation.
Past experience has shown that a ratio of 5 to 10 usually produces favorable results at a
moderate computational cost for problems where gross smoke movement is of interest.
As an example, suppose the HRR of the fire were 700 kW. The characteristic diameter may
then be calculated as follows:
/
700 kW
1.2 kg/m
1.012 kJ/kg/K
293 K 9.81 m/s
0.83 m
(2-2)
To perform a grid sensitivity analysis, a good place to start might be a cell resolution of 15 cm
5.5. Then choose a grid of 10 cm (4 in), and then 5 cm (2 in).
(6 in), which means that /
At this point, the calculation time will have increased by a factor of roughly three hundred,
making it potentially impractical to compute; however, if it can be shown that there is little
difference between the 5 cm (2 in) and 10 cm (4 in) grids, then the objective has been achieved.
The meaning of little difference can be interpreted in several ways. Given that NUREG-1824
(EPRI 1011999), the fire model V&V study, lists the relative error expected of the various
models for the various quantities, it is reasonable to interpret the difference in results on
different grids in light of what is expected of the model accuracy.
Although the fire size and dimensions often determine the optimum grid resolution, there are
other factors that can influence the selection of the grid resolution. These include the number of
cells used to resolve a flow path dimension, the number of cells used to describe the fire
dimension, and the number of cells used to resolve the conditions in a partially isolated volume.
These considerations are related in that it is generally advisable to include at least three cells
across any flow path, such as a door or a window, and fire dimension, regardless of the
minimum number of cells computed using the fire characteristic diameter. In some cases,
partially isolated volumes are created by various obstructions; if the temperature and flow
conditions are of interest in these areas, a minimum of three cells across any dimension should
be provided. Another consideration that could influence the grid resolution is the dimension of
2-21

the obstructions that are expected to influence the result. For example, if it is necessary to
quantitatively assess the effect that various conduits and light fixtures may have on the
actuation time of a nearby sprinkler, the maximum grid resolution would be comparable to the
dimensions of the smallest distinct obstruction included in the model.
FDS input files are frequently created with the assistance of preprocessing software, which may
include commercial software packages that can create input files for FDS or spreadsheet tools
created by users to insert obstructions or create stair-step approximations to curved geometries.
This type of software can reduce the tediousness of creating the geometric representation of a
space, but is not part of the FDS model. Any input files created by such software should be
carefully checked by the user to ensure that the geometry or boundary data are exactly as
intended.
2.3.6 Verification and Validation (V&V)

capabilities. NFPA 805, for example, states that fire models shall only be applied within the
limitations of the given model and shall be verified and validated. Previously, the NRCs Office
of Nuclear Regulatory Research (RES) and EPRI conducted a collaborative project for the V&V
of the five selected fire models described in Sections 2.3.1 through 2.3.5. NIST was also an
important partner in this project. The results of this project were documented in the seven
volumes of NUREG-1824 (EPRI 1011999), Verification and Validation of Selected Fire Models
for Nuclear Power Plant Applications.
The parameters for which NUREG-1824 (EPRI 1011999) provides V&V information are shown
in Table 2-4. Not all output parameters are available in all models. The information in Table 2-4
may be a useful element to consider when selecting the appropriate fire model tool; for
example, it is clear that the libraries of algebraic models (FDTs, FIVE-Rev1) have limited
capabilities when compared to the zone and CFD models. These libraries do not have
appropriate methods for estimating many of the fire scenario attributes evaluated in this study.
The correlations that the libraries do contain are typically empirically deduced from a broad
database of experiments; they are based on fundamental conservation laws, and have gained a
considerable degree of acceptance in the fire protection engineering community. However,
because of their empirical nature, they are subject to many limiting approximations. The user
must be cautious when using these tools.
CFD model predictions can be more accurate in complex scenarios; however, the time it takes
to obtain and understand a prediction may also be an important consideration in the decision to
use a particular model for a specific scenario. FDS is computationally expensive in all respects
(preprocessing, simulation, and post-processing), and, while the two-zone models produce
answers in seconds to minutes, FDS provides comparable answers in days to weeks. In
general, FDS is better suited to estimate fire environments within more complex configurations.
The fire experiments selected for inclusion in the V&V were limited to high-quality, real-scale
experiments with direct applicability to NPP applications. As it was not possible to consider all
possible NPP applications, a method for determining the applicability of validation results to
other specific NPP fire scenarios has been described in NUREG-1824 Volume 1 (EPRI
1011999). The applicability of the validation results is determined using normalized parameters
traditionally used in fire modeling applications. Normalized parameters allow users to compare
results from scenarios of different scales by normalizing the physical characteristics of the
scenarios.
2-22

Table 2-4. Fire modeling attributes included in NUREG 1824/EPRI 1011999 (2007).
Fire Modeling Attributes
Fire Model
FDTs
FIVE-Rev1
CFAST
MAGIC
FDS
Hot Gas Layer (HGL) Temperature
YES
YES
YES
YES
YES
Hot Gas Layer (HGL) Height
NO
NO
YES
YES
YES
Ceiling Jet Temperature
NO
YES
YES
YES
YES
Plume Temperature
YES
YES
NO
YES
YES
Flame Height
YES
YES
YES
YES
YES
Radiated Heat Flux to Targets
YES
YES
YES
YES
YES
Total Heat Flux to Targets
NO
NO
YES
YES
YES
Total Heat Flux to Walls
NO
NO
YES
YES
YES
Wall Temperature
NO
NO
YES
YES
YES
Target Temperature
NO
NO
YES
YES
YES
Smoke Concentration
NO
NO
YES
YES
YES
Oxygen Concentration
NO
NO
YES
YES
YES
Room Pressure
NO
NO
YES
YES
YES
Table 2-5 identifies normalized parameters that may be used to compare NPP fire scenarios
with validation experiments. A full description of each parameter follows Table 2-5. The
validation range for the normalized parameters shown in Table 2-5 were derived from NUREG1824 (EPRI 1011999), Table 2-4, and are intended to provide guidance on which groups of
validation experiments to consider when evaluating a certain attribute based on the validation
results. These parameters may not be the only ones appropriate for evaluating the applicability
of a specific experiment; Table 2-5 of NUREG-1824 (EPRI 1011999) Volume 1 lists the ranges
of values for different physical characteristics and normalized parameters based on the
experiments considered in the validation study.
It is seen in Table 2-5 that the fire diameter plays a role in three of the normalized parameters:
the fire Froude Number, the flame length ratio, and the radial distance ratio. The fire diameter
may be estimated from the total plan area of the burning fuel using the following equation
(SFPE, 2011):
4
(2-3)
where is the effective fire diameter (m) and is the plan area of the burning fuel (m) equal to
the fuel package length dimension multiplied by the width dimension. Equation 2-3 produces a
diameter of a circle having an equivalent area to that of the burning fuel and is therefore
2-23

applicable to fires with a plan area that is nearly circular. When the aspect ratio of a source fire
with a rectangular plan or footprint (i.e., the ratio of the long dimension to the short dimension)
becomes large, the flame length and thermal plume take on characteristics of a line type fire
(Grove et al., 2002). There is no clearly defined aspect ratio limit; however, a value of up to
about five is shown to be reasonable for using the effective diameter (Grove et al., 2002). Note
that the form of the Froude Number for a line type source fire is nearly the same as the
axisymmetric source, with the exception that the source area is used in place of the diameter
(Yuan et al., 1996). The fire tests used to V&V the various fire models in NUREG-1824 all had
aspect ratios below two.
Table 2-5. Summary of selected normalized parameters for application of the validation results to
NPP fire scenarios (NUREG-1824/EPRI 1011999, 2007).
Quantity
Normalized Parameter
Fire Froude
Number
Flame Length
Ratio
3.7
1.02
cj
Ceiling Jet
Distance Ratio
Equivalence
Ratio
O
Compartment
Aspect Ratio
General Guidance
0.23
1
2
0.23
(Natural)
Mechanical
or
Radial Distance
Ratio
2-24
Ratio of characteristic
velocities. A typical accidental
fire has a Froude number of
order 1. Momentum-driven fire
plumes, like jet flares, have
relatively high values.
Buoyancy-driven fire plumes
have relatively low values.
A convenient parameter for
expressing the size of the fire
relative to the height of the
compartment. A value of 1
means that the flames reach
the ceiling.
Ceiling jet temperature and
velocity correlations use this
ratio to express the horizontal
distance from target to plume.
The equivalence ratio relates
the energy release rate of the
fire to the energy release that
can be supported by the mass
flow rate of oxygen into the
compartment, O . The fire is
considered over- or underventilated based on whether
is less than or greater than 1,
respectively.
This parameter indicates the
general shape of the
compartment.
This ratio is the relative
distance from a target to the
fire. It is important when
calculating the radiative heat
flux.
NUREG1824
Validation
Range
0.4 2.4
0.2 1.0
1.2 1.7
0.04 0.6
0.6 5.7
2.2 5.7

Froude Number
The Froude Number is a measure of the buoyant strength of the fire plume. A large Froude
Number indicates a strong source, and, given a sufficiently high Froude Number, the fire plume
may take on characteristics of a jet fire. The Froude Number is determined through the
following equation:
(2-4)
is the Fire Froude number (non-dimensional), is the fire heat release rate (kW),
where
is the heat capacity at constant pressure for ambient air
is the density of ambient air (kg/m),
(kJ/kg-C), is the ambient temperature (K), is the fire diameter or the effective fire diameter
(m), and is the acceleration of gravity (9.81 m/s). The heat release rate that is used to
generate the validation range is the peak heat release rate; thus, when comparing the model
application range to these values, the peak heat release rate should be used. The effective fire
diameter may be computed using Equation 2-3. The ambient density and temperature are
typically coupled through the following equation:
352
(2-5)
where all terms have been defined. The heat capacity for air varies slightly with temperature,
and, at 298 K, it is equal to about 1.012 kJ/kg-C.
Flame Length Ratio
The flame length ratio is a measure of the flame height relative to the upper horizontal boundary
(ceiling). The NUREG-1824 validation tests all involved fires with a flame height that was at or
below the ceiling. In situations where the flame height is greater than the ceiling, the flames will
extend radially outward from the impingement point. This configuration is not well characterized
by simple algebraic models, or even by zone models that use plume and flame height
correlations.
Flame Length Ratio
(2-6)
where
is the base height of the fire (m),
is the flame height (m), and
is the enclosure
height (m). Note that the flame length ratio does not apply to fires modeled outside an
enclosure. The flame height is computed using the following equation:
3.7
1.02
(2-7)
where is the fire diameter or the effective fire diameter (m) and
is the fire Froude Number
calculated using Equation 2-4 (non-dimensional). The effective fire diameter may be computed
using Equation 2-3.
2-25

Ceiling Jet Distance Ratio
The ceiling jet distance ratio is a measure of the ceiling jet position at which data is sought
relative to the enclosure height and is applicable primarily when the temperature and velocity of
the ceiling jet are quantities of interest. A low ceiling jet distance ratio indicates that the position
is within the impingement zone and that the conditions would be dominated by the thermal
plume. A high ceiling jet position ratio suggests that the position is approaching the edge of the
ceiling jet, at least as idealized by algebraic correlations. A high ceiling jet ratio also suggests
that a considerable portion of the ceiling would need to be free of obstructions in order to
conform to the underlying approximations of the ceiling jet models. The ceiling jet ratio is given
by the following equation:
Ceiling Jet Ratio
(2-8)
is the
where cj is the horizontal distance within the ceiling jet from the fire centerline (m),
is the base height of the fire (m). The ceiling jet ratio is applicable
enclosure height (m), and
primarily when sprinkler or heat detector actuation is calculated. It is not applicable when the
fire is modeled outside an enclosure.
Equivalence Ratio
This quantity is the ratio of the generation rate of fuel to the supply rate of oxygen. When the
equivalence ratio is equal to one, the exact amount of oxygen required for complete combustion
is available. When the ratio is greater than one, the environment is fuel rich and the fire is
considered to be under-ventilated. The reverse is true when the ratio is less than one. The
upper limit for the equivalence ratio in enclosure fires is about three (Gottuk et al., 2008).
Modeling under-ventilated fires is challenging in part because the fuel mass loss rate is typically
specified by the user and is not adjusted by the fire models to the conditions actually present.
The equivalence ratio may be estimated for natural and forced ventilation by the expression:
(2-10)
O
where is the equivalence ratio (non-dimensional), is the heat release rate of the fire (kW),
is the heat of combustion for oxygen (kJ/kg), and O is the mass flow rate of oxygen
into the enclosure (kg/s). The heat of combustion for oxygen is the energy released per unit
mass of oxygen and varies by less than five percent among nearly all carbon-based fuels
(Janssens, SFPE Handbook, 4th edition). The value for
is typically taken to be 13,100
kJ/kg, an average value over a wide range of common fuels. A notable exception to this
convention is hydrogen, whose heat of combustion based on oxygen consumption is
approximately 18,000 kJ/kg. For other pure fuels, it is possible to calculate a more accurate
value than 13,100 kJ/kg, in which case Eq. (2-10) can be modified accordingly.
The mass flow rate of oxygen into the enclosure is given by the following equation:
0.23
0.23
1
2
(Natural)
Mechanical
2-26
(2-11)

where
is the effective area of the openings (m),
is the effective height of the openings
(m), is the density of ambient air given by Equation 2-5 (kg/m), and is the volumetric flow
rate of air into the enclosure (m/s). In the case of natural ventilation, Equation 2-11 is
applicable only to enclosures with one or more vertical openings. Note that the oxygen mass
flow parameter is the oxygen flow into an enclosure; if the enclosure has forced exhaust only,
the mass flow rate during the fire will likely change as the temperature of the enclosure
increases. The mass flow that corresponds to the heated environment should be used when
computing the equivalence ratio. An alternate means of estimating the mass flow rate of
oxygen into the enclosure is necessary if there are horizontal openings, which could include
using the model-predicted vent flow rates. This approach may also be used for enclosures in
which there is only forced exhaust. If there are multiple vertical openings, the effective opening
height ( ) is given by the following equation (Buchanan, 2001):
(2-12)
is the area of the ith opening (m),

is the height of the ith opening, and is the total
where
is given by the following equation
number of openings. The effective opening area (
(Buchanan, 2001):
(2-13)
where
is the area of the ith opening (m).
In many compartments, both natural and forced ventilation is present. The recommended
procedure to follow in this case is to select the dominant ventilation mode (i.e., the mode that
produces the highest mass flow of oxygen). If the two modes are comparable, the oxygen
masses may be added. As was the case for the ceiling jet ratio and the flame length ratio, the
equivalence ratio applies only to fires that are modeled within an enclosure.
Compartment Aspect Ratio
The compartment aspect ratio is a measure of the deviation of the enclosure dimensions from a
cube. When at least one of the compartment aspect ratios is large, the enclosure takes on the
characteristics of a corridor. In such cases, the transport time of the combustion products and a
non-uniform layer can both become significant parameters that require consideration. When at
least one of the compartment aspect ratios is low, the enclosure takes on the characteristics of
a shaft. In these cases, stratification of the combustion products, the interaction of the fire
plume and the enclosure boundaries, or choked flow could become parameters that influence
the results. These situations may lie outside the development basis for algebraic and zone fire
models. The compartment aspect ratio is computed using the following equation:
Compartment Aspect Ratio
2-27
(2-14)

is the
where is the compartment length (m),
is the compartment width (m), and
compartment height (m). The compartment aspect ratio is applicable to fires evaluated within
an enclosure.
Radial Distance Ratio
The radial distance ratio is a measure of the distance from the center of the source fire at which
a heat flux quantity is predicted. It is applicable only when the heat flux parameter is computed.
A low radial distance ratio indicates that the target location is close to the fire and that near-field
thermal radiation effects could be significant. Large radial distance ratios indicate that the target
is located far from the fire. The radiant heat flux at high radial distance ratios will approach that
predicted from a point heat source, barring other sources of external thermal radiation or
geometric factors (flame deflections under a ceiling or boundary re-radiation).
The radial distance ratio is given by the following equation:
Radial Distance Ratio
where is the actual distance between the target and the center of the fire base (m) and
the fire diameter as computed using Equation 2-3.
(2-15)
is
For a given set of experiments and NPP fire scenarios, the user can calculate the relevant
normalized parameters. If the fire scenario parameters fall within the ranges evaluated in the
study, then the results of the study offer appropriate validation for the scenario. If they fall
outside the range, then a validation determination cannot be made based on the results from
the study. For any given fire scenario, more than one normalized parameter may be necessary
for determining the applicability of the validation results.
The V&V study provides valuable insight into the predictive capability of the five fire models.
This insight is ultimately characterized in terms of a bias and a standard deviation for a number
of output parameters. The closer the bias is to unity, the more accurate the fire model tends to
be in predicting the given parameter; the smaller the standard deviation, the smaller the
expected scatter around the mean bias. Chapter 4 of this guide describes how the V&V
uncertainty information can be used to assign a probability function to the output data.
NUREG-1824 (EPRI 1011999) provides V&V documentation for specific versions of fire models.
Because the fire models considered are under active development, new releases are expected
to and do occur. The user has the option of using the model version that has been verified and
validated in NUREG-1824 (EPRI 1011999) or re-evaluating cases in NUREG-1824 (EPRI
1011999) to demonstrate that the predictive capability of the model has not decreased for the
application at hand. It is expected that NUREG-1824 (EPRI 1011999) will be updated from time
to time, as the need arises.
2.3.7 Fire Modeling Parameters Outside the Validation Range

The development of the sample problems documented in the appendices to this report suggests
that many commercial NPP fire modeling applications can fall outside the range of applicability
of the validation study documented in NUREG-1824 (EPRI 1011999). The primary reason for
this is that the range of applicability, as defined by the dimensionless parameters, is governed
by the experiments selected for the validation study. The selected experiments are
representative of various types of spaces in commercial NPPs, but do not encompass all
possible geometries or applications. Accordingly, the analyst will encounter many areas or
2-28

applications that will fall outside this application range. The predictive capabilities of the fire
models in specific scenarios can extend beyond the range of applicability defined in NUREG1824 (EPRI 1011999). Additional analysis and justification is required by the analyst to address
situations where some or all of the analysis parameters fall outside the range of applicability
defined in NUREG-1824 (EPRI 1011999). The additional analysis and justification should
address the applicability of fire modeling results generated by input parameters outside the
validation range to support the conclusions of the study. This section describes the
recommended strategies for addressing this situation.
2.3.7.1 Sensitivity Analysis

In the context of applicability of validation results, sensitivity analysis refers to varying selected
input parameters in the conservative direction so that they fall within the applicability range. If
the fire modeling conclusions are not affected by the variations in the parameters, the analyst
may use the sensitivity analysis results to further justify the conclusions. Based on the
dimensionless terms listed above, the following sensitivities could be evaluated:
Froude number: The two parameters that can be practically varied are the fire diameter
and the HRR. For fire sizes (i.e., HRR) that are small for the postulated diameter, the
resulting Froude number can fall under the low end of the applicability range. Similarly,
for fires that are relatively large for the postulated diameter, the Froude number can fall
above the applicability range. In the former situation, the analysts may consider
reducing the fire diameter and keeping the HRR profile unchanged. In most fire
modeling tools, the fire diameter is simply used to determine HRRs or to calculate the
fire plume conditions, such as the flame height or plume temperature. Considering that
the HRR is fixed in this sensitivity study, the fire diameter may not be a relevant
parameter in the analysis, with the important exception of scenarios where the fire plume
conditions are relevant. A similar approach could be used for the latter situation.
Increasing the fire diameter can force the dimensionless term into range. It should be
stressed that fire diameter is often a parameter that influences predicted flame height
and fire plume conditions, and that the effects of diameter variations should be explicitly
addressed in the analysis. This includes other dimensionless terms where the fire
diameter is a key input (e.g., target distance to diameter (r/D), etc.).
Flame length relative to ceiling height: This is a convenient parameter for expressing the
size of the fire relative to the height of the compartment. A value of 1 means that the
flames reach the ceiling. The validation range extends up to a value of 1.0, which
should cover most of the scenarios of interest in commercial NPPs. Scenarios that are
expected to fall out of the range are:
o
Those associated with relatively short flames. Typical ceiling heights in NPP
scenarios range from about 3 to 6.1 m (10 to 20 ft), excluding the containment and
turbine buildings, which have relatively large openings between elevations.
Consequently, flame lengths shorter than 0.6 to 1.2 m (2 to 4 ft) will be considered
outside of validation range. A sensitivity analysis increasing the HRR values should
provide a conservative estimate of fire conditions within the validation range. In
cases where the conclusion of the analysis does not change given the increased fire
intensity (e.g., no damage within the flame length of fire plume), the suggested
sensitivity analysis can be used as the justification for the evaluation of a
compartment that falls outside the validation range.
2-29

o
Flame extensions under ceilings. In this particular case, not only are such flame
lengths out of the range of validation, but also the models for predicting this
phenomenon have not been verified or validated with a process similar to the one
documented in NUREG-1824 (EPRI 1011999).
Ceiling jet radial distance relative to the ceiling height: Ceiling jet temperature and
velocity correlations use this ratio to express the horizontal distance from target to
plume. Ceiling jet applications in commercial NPPs should be carefully evaluated due to
the numerous obstructions near the ceiling (e.g., cable trays, HVAC ducts, piping, etc.).
Most of its applications include determination of time to detection and sprinkler
activation, in which the ceiling jet velocity is a sub-model in the analysis. An alternative
option is a sensitivity analysis consisting of moving the fire location to distances that
would fall within the validation range; it is recognized, however, that in many situations
the fire location cannot be altered, particularly in the case of fixed ignition sources or
transient fires postulated near areas where redundant targets are in close proximity
(pinch-points). In general, longer horizontal distances will result in longer activation time
results; by contrast, shorter horizontal distances would result in conservative time-todamage results. In situations where the ceiling jet geometry deviates significantly from
the idealized flat horizontal surface, as may be the case when there are large numbers
of obstructions or bays, a CFD model may be the better choice for calculating detector
response times.
Equivalence ratio as an indicator of the ventilation rate: The validation available is for
well-ventilated fires: that is, no model validation information is available for underventilated compartment fires, including fire extinction due to lack of oxygen. In general,
fires that are considered well ventilated in the enclosure should result in bounding
conditions as long as the HRR profile is appropriate. Conditions in the enclosure are not
expected to be worse in a fire where the combustion process is affected by lack of
oxygen than they would be under fire conditions where the combustion process is
unaffected. However, under-ventilated fire conditions should be considered carefully as
sudden air inflows into compartments with under-ventilated fire conditions could produce
relatively severe fire conditions.
Compartment aspect ratio: It is expected that some compartments in commercial NPPs

would have geometric characteristics outside the validation range (e.g., relatively long,
narrow corridors with high ceilings, etc.). These parameters are important in fire
scenarios involving HGL calculations, as the size and configuration of the compartment
are important input parameters. Clearly, these parameters should not be applicable in
scenarios where the enclosure conditions are not considered, such as flame radiation
calculations using the point source model and plume temperature calculations using
algebraic models where it has been determined that enclosure conditions are not a
factor. As part of the sensitivity analysis, the analyst may consider shortening the
length, width, or height of the compartment to values that fall within the validation range,
with the expectation that this will result in an elevated level of hazardous fire-generated
conditions, as predicted by the model (i.e., a conservative calculation). In cases where
the conclusion of the analysis does not change given the smaller compartment (e.g.,
the HGL temperature does not exceed damage threshold of cables in either case), the
2-30

suggested sensitivity analysis can be used as the justification for the evaluation of a
compartment that falls outside the validation range.
Radial distance relative to the fire diameter: This ratio is the relative distance from a
target to the fire, and is important when calculating the radiative heat flux. Note that the
validation range starts at a distance approximately twice the fire diameter. In practice,
targets at a very close distance to the fire (approximately two fire diameters or less)
should be expected to fail, given the relatively low damage threshold levels for cables.
An alternative option is a sensitivity analysis, which consists of moving the fire location to
distances that would fall within the validation range; it is recognized, however, that in
many situations the fire location cannot be altered, particularly in the case of fixed
ignition sources or transients fires postulated near pinch-points. In general, shorter
horizontal distances will result in higher heat flux levels.
2.3.7.2 Additional Validation Studies

There are other fire model validation studies besides NUREG-1824 (EPRI 1011999) that can
serve as a basis for establishing the applicability of fire modeling results. In developing the
examples documented in the appendices of this report, the research team identified relevant
validation studies outside of NUREG-1824 (EPRI 1011999), as summarized below:
Scenarios involving targets within the fire plumes: A useful discussion of fire plumes is
contained in Gunnar Heskestads chapter in the SFPE Handbook of Fire Protection
Engineering, 4th ed., Fire Plumes, Flame Height, and Air Entrainment. The plume
correlations used in the empirical and zone models are described, as well as their range
of applicability. NUREG-1824 (EPRI 1011999) contains experimental measurements of
fire plumes, but the range is somewhat limited. The plume correlations used by the
models have a much wider range of applicability than that exercised in NUREG-1824
(EPRI 1011999).
Scenarios involving targets within the ceiling jet: Similarly, Ronald Alperts chapter
Ceiling Jet Flows in the SFPE Handbook contains a description of the various
correlations used to estimate the temperature and gas velocity of ceiling jets. There are
extensive references to the original experimental test reports from which the correlations
were derived.
Scenarios involving targets exposed to flame radiation: A useful collection of techniques

and validation data for thermal radiation calculations is found in the SFPE Engineering
Guide for Assessing Flame Radiation to External Targets from Pool Fires, written by the
SFPE Task Group on Engineering Practices, 1999.
Scenarios involving flashover/post-flashover conditions: A series of experiments was

conducted at NIST as part of an investigation of the collapse of the World Trade Center
towers. Validation calculations with FDS are described in the report NIST NCSTAR 15F, Federal Building and Fire Safety Investigation of the World Trade Center Disaster:
Computer Simulation of the Fires in the WTC Towers, September 2005.
Scenarios involving electrical failure of cables: The Cable Response to Live FIRE
(CAROLFIRE) program led to the development and validation of the Thermally-Induced
2-31

Electrical Failure (THIEF) model (NUREG/CR-6931, Volume 3). This model can be
used to estimate the temperature within an electrical cable that is exposed to an
elevated temperature or heat flux.
Scenarios involving cable burning: The Cable Heat Release, Ignition, and Spread in
Tray Installations in Fire (CHRISTIFIRE) program led to the development and validation
of the Flame Spread in Horizontal Cable Trays (FLASH-CAT) model (NUREG/CR-7010,
Volume 1). This model addresses the growth and spread of fire within vertical stacks of
horizontal, open-back cable trays.
In addition to NUREG-1824 (EPRI 1011999) and the various documents cited above, the
individual model developers typically maintain a collection of validation cases that are included
as part of the model documentation. The algebraic spreadsheet models, FDTs and FIVE-Rev1,
are based directly on experimental correlations. Validation of these models is typically not part
of the model documentation; rather, there are references to source material like the SFPE
Handbook or the original test reports. Validation studies by the CFAST and FDS developers are
contained within:
NIST Special Publication 1086, CFAST Consolidated Model of Fire Growth and Smoke
Transport, Software Development and Model Evaluation Guide, 2008.
NIST Special Publication 1018, Fire Dynamics Simulator, Technical Reference Guide,
Volume 3, Validation, 2007.
In summary, the purpose of the sensitivity analysis is to re-shape the scenario with parameters
that fall within the V&V range and result in more severe fire generated conditions (e.g., higher
HGL or plume temperature, higher incident heat flux, etc.). Depending on the application, one
or more parameters may need to be varied affecting multiple dimensionless parameters. It is
recommended that the results from the sensitivity calculations always be compared to those
resulting from the base case to ensure that the input parameter manipulation produces more
severe fire generated conditions.
2.4 Step 4: Calculate Fire-Generated Conditions

This step involves running the model(s) and interpreting the results. When running a computer
model, the following general steps are recommended:
1. Determine the output parameters of interest. If the goal of the simulation is to estimate
wall temperatures, for example, the analyst should be interested in internal and external
wall temperatures. The analyst should ensure that the model will provide the output of
interest, or at least the fire conditions that can help achieve the objectives of the
analysis. The output file should be labeled with a distinctive file name.
2. Prepare the input file. In this step, the analyst enters the input parameters into the
model. The best way to enter input parameters is to follow the same guidelines
described in the scenario description section. Each model has a users manual with
instructions on creating the respective input file. These files are created either through
user-friendly menus and screens or through a text editor. If a text editor is used, it is
strongly recommended that the analyst start with an example case prepared by code
developers, and make appropriate changes to that file.
2-32

3. Run the computer model. The running time for zone models is on the order of minutes,
depending on the complexity of the scenario and the speed of the computer.
Calculations using a CFD model may take up to days or weeks in complex scenarios,
including multiple compartments, multiple fires, and mechanical ventilation systems.
4. Interpret the model results. Check that the results are intuitively consistent with the input
and expectations. Check that the output results accurately reflect the desired input;
common verifications would include the fire size and location, the location and status of
any doors or boundary openings, and the forced ventilation flow rate and location. The
model output should be checked for indications of a solution error. For example, the
pressure and the HGL should fall within the ranges observed in test data; the HGL
temperature should be greater than the lower gas layer temperature; and there should
not be anomalous areas of flow acceleration or temperature change. Determine whether
or not the fire scenario resulted in conditions that exceed the performance criteria, as
applicable.
5. Organize the results in a form that answers the question(s) and allows the calculations to
be reproduced. If the results are used in a PRA screening analysis, this may take the
form of a ZOI dimension or a maximum HGL temperature. If the results are part of a
deterministic analysis, the output form may be a conclusion with regard to the
performance of some component and an associated safety margin if the component is
predicted to be free of damage.
For the FDTs and FIVE-Rev1, the input data is entered directly into a spreadsheet, and the
results are presented in the spreadsheet. Some of the FDTs spreadsheets include graphical
and tabular results. FIVE-Rev1 typically provides a single result for a given set of input data;
however, many of the calculations in FIVE-Rev1 are implemented as Microsoft Excel functions.
These functions can be called from any cell in the spreadsheet, and can be used to easily
calculate the plume temperature at a specific location above the fire as a function of time for a
fire with a time-varying HRR.
CFAST, MAGIC, and FDS can handle user-specified transient heat release rates, as they
calculate the results for each zone or cell at each time step. The time step required to maintain
stable calculations is typically determined by the model. The interval at which results are
presented is a user-specified value. CFAST, MAGIC, and FDS can output results as text files,
which can be read or plotted using commercially available spreadsheet programs; CFAST and
FDS can also output their results in a form appropriate for SMOKEVIEW (Forney, 2010).
SMOKEVIEW is a software tool that visualizes smoke and other attributes of the fire using
traditional scientific methods, such as displaying tracer particle flow, two- or three-dimensional
shaded contours of gas flow data (e.g., temperature), and flow vectors showing flow direction
and magnitude. MAGIC includes its own post-processor for visually analyzing the results of a
simulation. Post-processing may also be performed using other graphical or graphical
animation software. If using a software package that is not designed for viewing the particular
fire model results, the user should check that the output parameters are interpreted and
displayed as intended.
2.5 Step 5: Conduct Sensitivity and Uncertainty Analyses

This document recommends a comprehensive treatment of uncertainty and/or sensitivity
analysis as part of a fire modeling analysis for the following reasons:
2-33
Models are developed based on idealizations of the physical phenomena and simplifying
approximations, which unavoidably introduces the concept of model uncertainty (i.e.,
model error) into the analysis.
A number of input parameters are based on available or generic data or on fire

protection engineering judgment, which introduces the concept of parameter uncertainty
into the analysis.
The concepts of model and parameter uncertainty have traditionally been addressed in fire
modeling using uncertainty and/or sensitivity analysis. The uncertainty in a variable represents
the lack of knowledge about the variable, and is often represented with probability distributions.
Its objective is to assess the variability in the model output, that is, how uncertain the output is
given the uncertainties related to the inputs and structure of the model. By contrast, the
sensitivity of a variable in a model is defined as the rate of change in the model output with
respect to changes in the variable. A model may be insensitive to an uncertain variable.
Conversely, a parameter to which a model is very sensitive may not be uncertain.
Details of the uncertainty and sensitivity analysis are included in Chapter 4.
2.6 Step 6: Document the Analysis

The amount of information required and generated by a fire modeling analysis can vary widely.
Simple algebraic models may not require a large number of inputs, and the complete analysis,
including output results, can be documented on a single piece of paper. On the other hand,
some fire modeling exercises may require use of multiple computer models, where outputs from
one are inputs to others. These cases, for the most part, will require a significant number of
input parameters and will produce outputs requiring documentation. Regardless of the amount
of information required or generated by the analysis, proper documentation is vital to identifying
the important findings of the exercise and providing clear, focused conclusions.
Documentation of the fire scenario selection and description process should include enough
information so that the final report is useful in current and future applications. This is particularly
relevant in the commercial nuclear industry, where compartment and equipment layouts or
processes do not change much over time. It is likely that fire scenarios analyzed for one
application may be useful for other applications as well; the key, however, is to develop and
maintain good documentation of the selected fire scenarios, including all the technical elements
discussed in this section. The SFPE Engineering Guide to Substantiating a Fire Model for a
Given Application (SFPE, 2010) provides general guidance on information to be included in fire
modeling analyses.
It is likely that the information necessary for documenting the fire scenario selection will be
gathered from a combination of observations made during engineering walkdowns and a review
of existing plant documents and/or drawings. The documentation process then involves
compiling the information from different sources into a well-organized package that can be used
in future applications and for NRC regional inspections. The documentation package may
consist of:
Marked-up plant drawings. Plant layout, detection, suppression, cable tray, HVAC, and
conduit drawings are often marked to highlight the location of the compartment, the
ignition sources, the targets, the ventilation flow paths, and the fire protection features.
The drawings also serve as sources of fire model input values, such as compartment
2-34

dimensions, ventilation flow rates, and relative locations of fire protection systems or
targets.
Design basis documents (DBDs). Design basis documents (DBDs) provide in-depth
assessments of plant features in various operation modes, such as the HVAC system.
Sketches. Sketches are perhaps one of the most useful ways of documenting a fire
scenario. A sketch typically consists of a drawing illustrating the ignition source,
intervening combustibles, targets, and fire protection features. A first draft of the sketch
is usually prepared during walkdowns. The analyst should take the opportunity to
include details such as raceways and conduit identifications (IDs), and other information
relevant to the fire modeling analysis. Pictures often supplement sketches.
Write-ups and input tables. Write-ups and input tables are used to compile the
information collected from drawings and walkdowns in an organized way. The write-up
should include a brief scenario description and detailed documentation supporting
quantitative inputs to the fire modeling analysis, as well as any relevant sketches or
pictures associated with each scenario.
Software versions, descriptions, and input files. The documentation package should
include the version numbers of any software, brief descriptions of the software, and
copies of the input files.
The examples presented in Appendices A through H of this guide illustrate techniques for the
proper documentation of fire modeling calculations using the format described. In conclusion, a
properly documented analysis should enable someone else to reproduce the results from the
information contained within the documentation.
2.7 Summary
This chapter described a recommended process for conducting and documenting a fire
modeling analysis. Chapter 3 provides guidance on selecting the appropriate fire modeling tool
and input parameters for typical commercial NPP applications. Fire model uncertainty is
addressed in Chapter 4 of this document. Specific fire modeling examples evaluated using the
process described in this Chapter are provided in Appendices A through H.
2-35
2-36
GUIDANCE ON FIRE MODEL SELECTION AND

IMPLEMENTATION
This chapter contains a catalogue of typical nuclear power plant (NPP) fire scenarios, including
relevant physical phenomena, model selection criteria, and suggested modeling strategies. In
particular, the chapter provides recommendations as to the appropriate use of empirical
correlations, zone models, and CFD calculations.
3.1 Model Implementation of Fire Scenario Elements

This section provides a description of fire modeling elements typically present in commercial
NPP scenarios. The following fire modeling elements are described:
Heat Release Rate
Plant Area Configuration
Ventilation Parameters
Targets
Intervening Combustibles
3.1.1 Heat Release Rate

For most fire model applications, the heat release rate (HRR) is the most important parameter to
specify. All enclosure fire models solve some form of the energy conservation equation (i.e., the
energy from the fire increases the temperature and drives hot air out and cold air into the
enclosure). The models essentially redistribute the energy from the fire throughout the
enclosure. For most NPP applications, the HRR is specified by the analyst in the form of a time
history. This time versus HRR curve typically has four stages: incipient, growth, steady burning
at peak intensity, and decay.
During the incipient stage, the fire burns at a low intensity (i.e., smoldering insulation or a small
trash can fire). The duration of this stage may vary from seconds to hours, and the energy
release is relatively low. Because of the uncertainty in the intensity of the fire during this stage,
and the exact time that the fire will transition to a significant fire, the incipient stage is often not
considered in the analysis.
Depending on the combustible and its arrangement, the growth to a fully developed fire will vary
from seconds to minutes. Unless experimental data is available, the HRR is usually specified to
increase following a so-called t2 growth profile (Karlsson and Quintiere, 2000). The basic idea
behind this approximation is that the burning surface area of a growing fire increases as the
square of the time from the beginning of the growth period.
The steady burning phase occurs when the fuel reaches its maximum burning rate. In most
cases, the peak HRR is obtained from experimental data. Alternatively, it can be estimated by
multiplying the heat of combustion by the maximum burning rate of the fuel, if known. The peak
3-1
GUIDANCE ON FIRE MODEL SELECTION AND IMPLEMENTATION

HRR may be limited by the available air supply. Estimates of the maximum HRR inside a
compartment with a given ventilation rate are available in fire protection engineering handbooks.
Following the steady burning phase, the HRR is usually specified to decrease to zero linearly.
The duration of this phase depends on the amount of fuel available. In fact, the integral of the
entire HRR time profile (in units of kJ) divided by the heat of combustion of the fuel (in units of
kJ/kg) should equal to the mass of combustibles (in units of kg).
In addition to the HRR time history, the following parameters may be important depending on
the model or the scenario.
Fire elevation: The elevation of the base of the fire, measured from the floor. It is important
in scenarios involving targets in the fire plume where the relative distance between the fire
and the target strongly influences the exposing temperature. It is also important because
the height of the fire relative to the hot gas layer (HGL) influences the air entrainment into
the plume, the position of the HGL, and, potentially, the actual HRR (since air entrained from
the HGL is oxygen-depleted).
Fire location: In scenarios where the fire is located near a wall or corner, the plume is
expected to entrain less air, resulting in higher plume temperatures (Karlsson and Quintiere,
2000, p. 72).
Fuel mass: This parameter is an important factor in determining the burning duration
Soot and product yields: The yield of a combustion product is the mass of the product
generated per unit mass of fuel consumed. In particular, the soot yield is an important factor
in radiative heat transfer (e.g., targets immersed in the HGL), visibility calculations, and
smoke detector response estimates. The yields of toxic gases can also be important in
habitability calculations
Radiative fraction: The fraction of energy emitted in the form of thermal radiation. For most
materials, the radiative fraction is approximately one third. That is, one third of the total
HRR radiates in all directions and two thirds is convected upwards into the smoke plume
3.1.2 Plant Area Configuration

The plant area configuration refers to the geometrical layout and construction of the enclosure.
Each of these elements is described in detail next.
Compartment Geometry
Compartment geometry refers to the physical layout of the volume in which the fire is
postulated. The length, width, and height of the room are the typical inputs required by the
model. The size of a compartment is an important factor in the volume, and is used to solve the
fundamental conservation equations. Algebraic and zone models employ considerable
simplifications of the geometry, while CFD models attempt to replicate as much of the geometry
as possible.
Compartment Boundary Materials
Boundary (e.g., wall or ceiling) materials are characterized with thermophysical properties,
which include the density, specific heat, and thermal conductivity of the material. In the majority
of commercial nuclear power plant (NPP) applications, the wall material is concrete. Other
materials may include steel, gypsum board, etc. Properties for these materials are often
3-2

available in drop down menus in the fire models or in fire protection engineering handbooks.
Table 3-1 provides typical properties for materials commonly found in NPPs. These properties
were used in the examples described later in this Guide. It should be noted that these
properties, as input parameters to the models, may also be available from sources other than
the ones listed in Table 3-1. Regardless of the source of the information, users should always
check that the material property values are appropriate for their specific application, as the
resulting fire conditions may be sensitive to these parameters.
Table 3-1. Material properties.
Material
Brick
Concrete
Copper
Gypsum
Plywood
PVC
Steel
XLP
Thermal
Conductivity
(W/m/K)
0.8
1.6
386
0.17
0.12
0.192
54
0.235
Density
(kg/m3)
Specific Heat
(kJ/kg/K)
Source
2600
2400
8954
960
540
1380
7850
1375
0.8
0.75
0.38
1.1
2.5
1.289
0.465
1.390
NUREG-1805, Table 2-3

SFPE Handbook, Table B.6
NUREG/CR-6850, Appendix R
NUREG/CR-6850, Appendix R
3.1.3 Ventilation Effects

Ventilation effects include natural ventilation through vertical or horizontal openings, the effects
of leakage paths, and/or the effects of mechanical ventilation. Each of these elements is
described next.
Vertical Openings
Vertical openings are usually doors, but they can also be other wall openings, such as windows
or passive ventilation ducts. In some cases, a compartment will have more vertical openings
than the number that can be specified in a simplified model. For example, the McCaffrey,
Quintiere, and Harkleroad (MQH) correlation for calculating HGL temperature uses only one
opening when calculating the ventilation factor which, is defined as the product of the opening
area and the square root of the opening height
(Karlsson et al., 2000; Drysdale, 2011).
Buchanan (2001) suggests the following method for calculating an effective height and area:
,
,
(3-1)
Here Ao,i and Ho,i are the individual door areas and heights. The effective width of multiple
vertical openings can be estimated by the ratio Ao/Ho.
Regarding doors (and other operable openings), consideration should be given to the doors
being opened (or closed) during a fire. For example, when the fire brigade arrives, they will
open the doors to the fire area to gain access, which will affect the ventilation and possibly
result in smoke spread.
3-3

Leakage Paths
The doors of most compartments in commercial NPPs are normally closed, but are not perfectly
sealed. Consequently, the resulting pressure and the rate of pressure increase are often kept
very small by gas leaks through openings in the walls and cracks around doors, or leakage
paths. Leakage paths must be specified in compartments with closed doors during the fire
event unless the analysis considers a completely sealed enclosure where pressure rise is an
important variable. By contrast, compartments with at least one open door or window can
maintain pressure close to ambient during the fire event. Leakage paths therefore do not need
to be specified, since the leakage opening area is negligible when compared with the opening
areas of doors and windows.
Horizontal Openings
Horizontal openings consist of hatches or stairwells. For modeling purposes, the areas of
horizontal openings can simply be added. Any zone model should provide similar answers with
single or multiple horizontal openings as long as the total opening area is the same. Note that
the Consolidated Fire Growth and Smoke Transport (CFAST) model only allows for a single
connection between any pair of compartments included in a simulation. For a CFD model, no
special provisions are necessary to describe a horizontal opening.
Mechanical Ventilation
Mechanical ventilation refers to any air injected into or extracted from a compartment by
mechanical means. This has a number of practical applications, such as extracting smoke from
the HGL (e.g., a smoke purge system). The ventilation rate and the vent position are the two
most important mechanical ventilation parameters. For some applications, the velocity of the
airflow may also be important. These mechanically induced flows have the potential to alter the
fire-induced flows. Mechanical ventilation often consists of a supply and an exhaust system that
are maintained to achieve a certain pressure level.
3.1.4 Targets
A target is an object of interest that can be affected by the fire-generated conditions and
typically consists of cables in conduits, cables in raceways, or plant equipment. Targets are
characterized by their location, damage criteria, and thermophysical properties.
A targets location simply refers to its location relative to the fire. The location is represented by
three-dimensional coordinates within the volume of the room in which the fire conditions are
simulated. Where the target faces in a particular direction, an orientation vector to indicate that
direction needs to be entered.
The damage criteria refer primarily to a damage/response threshold. In general, the damage
criteria for scenarios involving cable damage is expressed in terms of damage temperature or
incident heat flux.
The models within the scope of this Guide require specification of the targets thermophysical
properties, primarily the density, specific heat, and thermal conductivity, for the analysis. These
parameters are used to estimate heat conducted into the targets. The predicted time for the gas
temperature surrounding a target to reach a specific limit is usually less than the time it takes
the target to reach the same limit because the heat conduction inside the target will delay the
temperature rise at the surface during the heating process.
3-4

Information on target damage thresholds and target thermophysical properties can be found in
documents such as NUREG-1805, NUREG/CR-6931, and NUREG/CR-6850 (EPRI 1011989).
3.1.5 Intervening Combustibles

In many cases, commercial NPP fire scenarios do not require burning targets to be modeled
because it is sufficient to determine only when the target is damaged. This is clearly not the
case with intervening combustibles, whose flammability characteristics need to be incorporated
into the model so that the fire progression is considered. Therefore, the intervening
combustibles should be described not only in terms of their proximity to the fire and the targets,
but also in terms of their relevant thermophysical and flammability properties.
In many cases, intervening combustibles consist of cables in ladder back trays. Representing
intervening combustibles, like cables, in fire models presents technical challenges that the
analyst should consider, including (1) obtaining the necessary geometric and thermophysical
properties representing the intervening combustible and (2) the ability of the computer tools to
model the fire phenomena (e.g., fire propagation). Because of these challenges, simplified
models for determining the contribution to the HRR due to flame spread and fire propagation
through cable trays have been developed. Appendix R of NUREG/CR-6850 (EPRI 1011989)
provides guidance on the calculation of fire spread and HRRs for cable trays. Additionally,
research is underway to develop improved methods for predicting the HRR and flame spread of
electrical cables. A simple model, Flame Spread over Horizontal Cable Trays (FLASH-CAT)
which predicts flame spread over cables, has been developed as part of the Cable Heat
Release, Ignition, and Spread in Tray Installations during Fire (CHRISTIFIRE) project
(NUREG/CR-7010, vol. 1, 2012), sponsored by the U.S. Nuclear Regulatory Commission (NRC)
and conducted by the National Institute of Standards and Technology (NIST).
3.2 Guidance on Model Selection and Analysis

This section provides guidance on model selection and analysis of specific fire scenarios. Each
subsection is devoted to a specific fire scenario, as listed in Table 3-2. In addition, Figure 3-1
provides a pictorial representation of each of these scenarios. The circled numbers are
intended to direct the reader to the section in which the scenario is described. Please note:
these pictorial representations are not drawn to scale and are used for illustrative purposes only.
Table 3-2. Listing of generic scenarios described in this chapter.
Number
Chapter Section
3.2.1
3.2.2
3.2.3
3.2.4
3.2.5
6
7
3.2.6
3.2.7
Scenario Description
Scenarios consisting of determining time to damage of cables above the ignition
source located inside the flames or the fire plume.
Scenarios consisting of determining time to damage of cables located inside or
outside the HGL. This scenario also includes a secondary fuel source (i.e.,
propagation to cable trays).
Scenarios consisting of determining time to damage of cables located in a room
adjacent to the room of fire origin.
Scenarios consisting of determining time to damage of cables located inside or
outside the HGL in rooms with complex geometries.
Scenarios consisting of determining time to loss of habitability of the main control
room.
Scenarios consisting of determining time to smoke or heat detector activation.
Scenarios consisting of determining temperature of structural elements.
3-5

Each of the sections listed above is organized as follows:
A sketch capturing most of the technical elements relevant to the analysis. A legend
summarizing the different elements presented in the sketches is provided in Figure 3-2.
A scenario objective stating the purpose of the modeling exercise in engineering terms.
A description of the relevant technical fire scenario elements, such as mechanical

ventilation, room geometry, etc. Recall that fire scenario elements refer to the different
characteristics of the fire scenario that are relevant to the analysis, and should be properly
represented in the model.
A modeling strategy section summarizing the recommended steps for performing the
calculation.
A section listing fire model recommendations for the analysis.
A section referencing relevant detailed fire modeling examples documented in the Appendix
section of this guide.
3
7
4
1
6
Figure 3-1. Pictorial representation of the fire scenario and corresponding technical
elements described in this section.
3-6
Cable
Tray
1. Exposed
2. Target
1.
Open
Door
2.
Junction
Box and
Conduit
Closed
Door
Electrical
Cabinet
Pump
Fire
Plant
Employee
Sprinkler
Control
Board
HVAC
Smoke
Detector
1. Supply
2. Exhaust
Structural
Element
1.
2.
Obstruction
Figure 3-2. Legend for fire modeling sketches presented in this chapter.
3-7
3.2.1 Targets in the Flames or Plume
Figure 3-3. Pictorial representation of scenario 1.
3.2.1.1 General Objective

The objective of this scenario is to calculate the time to damage for a target immediately above
a fire, as indicated by the dashed circle in Figure 3-3. For this scenario, the target is an
electrical raceway and the fire source is an electrical cabinet. This scenario is often
encountered as the initial part of a progression of target damage (i.e., the first item ignited, aside
from the ignition source). Consequently, the characterization of the first item ignited after the
ignition source is important in subsequent estimations of fire propagation through other
intermediate combustibles.
3.2.1.2 Modeling Strategy

The recommended modeling strategy is summarized in the following steps:
1. Determine whether the target cable, which is directly above the fire, is within the flame zone
or within the fire plume. The target should be considered inside the flame zone if it is
located directly above the base of the fire and its distance from the base of the fire is less
than the flame height. If the target is above the fire but is not within the flame zone, then it is
considered to be within the fire plume. It should be noted that unobstructed fire plumes will
increase in diameter as a function of height. Consequently, a target does not need to be
directly above the ignition source to be immersed in the fire plume.
2. Calculate the time to damage by finding the minimum of either:
a. The time at which the flame reaches the target. This is achieved by calculating the
flame height as a function of time using the HRR profile (HRR vs. time) and fire
diameter as input.
b. The time it takes the fire plume temperature to exceed the target damage
temperature. This is achieved by calculating the plume temperature at the specified
height as a function of time, using the HRR profile as an input. This approach can be
3-8

considered conservative, as it predicts cable damage occurs when the gas
temperature surrounding the target reaches the damage temperature (i.e., heating of
the cable is ignored). As an alternative, a potentially more accurate time to damage
can be obtained by calculating the surface temperature of the cable as a function of
time, given a heat flux profile generated by the flame or plume.
If non-target raceways are located between the ignition source and the target, the contributions
of intervening combustibles need to be considered in the analysis. For example, consider a
cabinet fire that ignites the first of a stack of trays overhead. The fire involving the combination
of the cabinet and first tray may then ignite the second tray in the stack, and the fire may
progress to damaging the target raceway. Considerations of the intervening combustibles in the
analysis include the HRR contribution and the corresponding effects on the target heating time.
Section 3.2.2 (Scenario 2) provides guidance on treatment of intervening combustibles.
In addition to the guidance provided above, the analyst should determine whether HGL effects
are relevant to the scenario. The portion of the fire plume immersed in the HGL entrains air at
higher temperatures (i.e., the HGL temperature) and is expected to have increased
temperatures when compared with portions of the fire plume outside the HGL. In scenarios
consisting of targets located relatively close to the ignition source (which is the case for the
scenario discussed in this section), the HGL effects on the plume temperature are generally not
considered, as the time to target damage is expected to be relatively short. For scenarios
involving targets in the fire plume, located relatively far from the ignition source, the HGL effects
on target heating should be considered. In the latter case, the rooms geometry and ventilation
(both natural and mechanical) conditions should be captured by the analysis.
3.2.1.3 Recommended Models

Algebraic Models
Both the FDTs and FIVE-Rev1 have models that can be useful for this scenario, provided that
the configuration is within the correlation basis and that there are no significant HGL effects.
Heskestads flame height correlation is an alternative for determining flame height. Similarly,
Heskestads fire plume temperature correlation is an alternative for determining plume
temperature and diameter (Heskestad, 2002).
The correlations listed above are particularly applicable for scenarios consisting of targets
relatively close to the ignition source, where HGL effects are not considered in the analysis.
As noted above, the time to damage is the minimum of either the time at which the flame
reaches the target, or the time it takes the fire plume temperature to exceed the target damage
temperature. This is simply the time at which the HRR reaches the value required for either of
the failure criteria. In both cases, the correlations are solved for a specific HRR at a specific
time up to when the fire conditions suggest target damage (e.g., flames reach the location of the
raceway or the plume temperature exceeds the damage temperature of the cables).
Another option, for scenarios where the flames do not reach the cables, would be to use the
Thermally-Induced Electrical Failure (THIEF) model (NUREG/CR-6931, volume 3) to determine
the surface temperature at the target. The THIEF model is included in NUREG-1805,
Supplement 1 (2012).
3-9

Zone Models
Zone models can be used for this scenario. To do so, set up the necessary input file, which
should include a target in the location of the electrical cable of interest, along with the
corresponding thermophysical properties, so that the surface temperature of the cable can be
tracked. Zone models have the ability to include HGL effects in their calculation of plume
temperature, and are thus particularly appropriate for scenarios where the HGL temperature
interacts with the fire plume at the location of the target.
Again, the time to damage is the minimum of either the time at which the flame reaches the
target or the time it takes the fire plume temperature to exceed the target damage temperature.
The zone models routinely calculate and report these values.
CFD Model
Although a CFD model could be used to analyze this scenario, the level of detail and resolution
offered by a CFD calculation is generally not necessary. On the other hand, the CFD model
would be particularly applicable if the scenario involved obstructions between the fire and the
target inside the fire plume or if HGL effects are significant. The effects of these obstructions on
the exposure conditions are not captured by algebraic models or zone models.
3.2.1.4 Detailed Examples

Readers are referred to Appendix B, which describes the analysis of an electrical cabinet fire in
the switchgear room, and Appendix E, which describes the analysis of a transient fire in a cable
spreading room.
3-10
3.2.2 Scenario 2: Targets Inside or Outside the Hot Gas Layer

The objective of this scenario is to calculate the time to damage for a target inside or outside the
HGL produced by a fire. The time to ignition of a secondary fuel source and the resulting
contribution to the total HRR can also be determined. For the case shown in Figure 3-4, the
target is a cable in an electrical raceway and the fire source is an electrical cabinet.

Two levels of analysis can be employed: (1) algebraic models for the average room temperature
as an indicator of the gas temperature surrounding the target, or (2) detailed heat transfer
analysis for determining the target temperature.
The first strategy consists of determining the overall room temperature using an algebraic model
(e.g., the MQH room temperature model) (McCaffrey et al., 1981). Such a calculation will
indicate whether the target may be subjected to damaging temperatures and the time at which
such temperatures may be observed. It should be noted that the room needs to be represented
as a rectangular parallelepiped and the area of all the surfaces in the room must be conserved.
In addition, if the target cable tray is relatively close, the target may be damaged by radiant
heating. This can be assessed with simple point source estimates that only require the HRR of
the fire, the separation distance between the fire and the target, and the damage criteria (i.e.,
critical heat flux for damage). The point source model may lead to over-predictions when used
to predict heat flux to targets in very close proximity to the fire. In many applications, the overpredictions will clearly suggest target damage or ignition.
The second strategy is best addressed with a model capable of including detailed heat transfer
analysis for determining the targets temperature. A raceway outside the fire plume may be
exposed to HGL conditions if the smoke accumulating in the upper part of the room descends to
3-11

the location of the raceway. Consequently, targets outside the fire plume are initially exposed to
lower layer conditions. As the smoke continues to accumulate, the target is immersed in HGL
conditions. As heat transfer conditions will be different for each case, a model with the ability to
track the relevant/applicable heat transfer interaction and calculations as a function of time,
such as a zone model or a CFD model, should be selected to handle this scenario at the
desired level of resolution.
With regard to the secondary fuel source, three distinct additional analyses must be made to
determine:
time at which the secondary fuel source ignites,
HRR of the secondary fuel source, and
combined HRR of the primary and secondary fires.
The more detailed models, such as FDS, can handle the ignition and contribution of multiple
fires, provided that the ignition criteria and source HRR characteristics are provided as input.
Other models, especially the algebraic models, only accept the total HRR as a function of time,
which is found by summing up the individual HRRs.
In the present example, consider a cable tray directly above the fire. The time to ignition of the
cable tray can be determined via algebraic models that estimate the flame height and plume
temperature as a function of time for the initial cabinet fire (see scenario previously discussed).
Once the flames from the cabinet reach the cable tray, the cable ignites. The same is true when
the plume temperature at the elevation of the cable tray reaches the ignition temperature of the
cables. Both calculations should be completed, and the shorter time used as the ignition time.
The HRR from the cable tray can be added to the HRR of the cabinet to determine a combined
HRR as a function of time. The resulting HRR profile can take into consideration both the fuel
consumption and propagation to additional intervening combustibles as a function of time. In
cases where fuel consumption is not considered, the resulting HRR profile is expected to
overestimate the fire intensity. This total rate can then be used in the various models as an
approximation of the HRR profile.
Appendix R of NUREG/CR-6850 (EPRI 1011989) addresses cable fires, including methods for
calculating the HRR for a variety of cable configurations.
It should be noted that the simple summation of the two HRRs is a simplification of a complex
phenomenon and only provides an approximation of the conditions created by the two separate
fires.
3.2.2.3 Recommended Modeling Tools

Algebraic Models
Select the appropriate HGL (or room temperature) model and then collect the required inputs,
including room size, opening sizes, boundary material properties, forced ventilation, and the
HRR profiles for the initial and secondary fuel packages. For screening purposes, the use of
algebraic models is recommended as long as the contributions of the first item ignited and
intervening combustibles are considered. As was mentioned earlier, this approach will provide
an approximation of the room temperature in which the target may be immersed. The methods
used by algebraic models to address the secondary fire source are discussed above. Target
damage due to radiant heating can be estimated using algebraic models; all that is required is
3-12

the HRR of the fire (as a function of time), the separation distance between the fire and the
target, and damage criteria (i.e., critical flux for damage).
Zone Models
Zone models provide a good alternative for modeling this scenario, as they provide the incident
heat flux profile, the surface temperature, and the internal temperature of the target in one
simulation. Set up the necessary input file with the required inputs, including room size,
opening sizes, boundary material properties, HRR, fire diameter, and a target and fire location
so that the cables surface temperature can be predicted.
Zone models also have the benefit of being able to handle secondary fire sources as separate
entities. Secondary fires can be ignited at a prescribed time, temperature, or heat flux.
However, zone models have limited capabilities for handling obstructions.
Target damage due to radiant heating from the fire is easily handled by zone models, as long as
there are no obstructions between the fire and the target that block radiant heat transfer. Zone
models can also account for radiant heating of targets by the HGL.
CFD Model
The use of CFD models for this scenario is recommended for complex geometries capable of
affecting the location of the HGL and the incident heat flux to the targets, or when greater
accuracy of the ignition and contribution of secondary fires is warranted. For instance,
obstructions between the ignition source and the target affect the heat balance at the surface of
the target. The CFD model will require inputs similar to the ones collected for the zone models;
however, the compartment geometry will need to be specified in greater detail.
Due to their detailed calculations, CFD models are best able to model secondary fire sources,
including their ignition and subsequent contribution to the HRR within the enclosure.
Like zone models, CFD models can handle targets damaged by radiant heating from the fire
and the HGL. CFD models can also include the effects of obstructions between the fire and the
target.

Appendix C describes the analysis of a relatively large lubricating oil fire affecting a raceway in a
pump room, and Appendix E describes the analysis of a transient fire in a cable spreading
room.
3-13
3.2.3 Scenario 3: Targets Located in Adjacent Rooms

The objective of this scenario is to calculate the time to damage for a target in the HGL in a
room adjacent to the room of fire origin. An opening connecting the room of origin to the
adjacent room allows combustion products to enter the adjacent room. For the case shown in
Figure 3-5, the target is a cable in an electrical raceway, the fire source is an electrical cabinet,
and there is an unprotected opening in the wall.

The recommended strategy for determining the temperature of targets located in a room
adjacent to the room of fire origin consists of four basic steps:
1. Determine the following characteristics for the HGL in the room of fire origin and the
adjacent compartment:
a. Temperature as a function of time
b. Depth as a function of time
2. Determine the incident heat flux surrounding the target cable.
3. Determine the surface and internal temperature of the target cable.
4. Compare the surface or internal temperature of the target with its damage temperature.
One way to approach this (using algebraic models or zone models) is to first model the HGL
temperature in the room of origin and the resulting effect on remote targets in the room. If this
approach indicates that target damage/ignition is unlikely in the room of origin, there would
probably be little benefit in evaluating similar targets in adjacent spaces. However, if target
damage is possible in the room of origin, the next step would be to model the room of origin and
the adjacent room and determine whether the resulting HGL is capable of causing
damage/ignition of the target(s).
3-14

Algebraic Models
Generally, algebraic models are not suitable for this calculation, as a model capable of tracking
fire conditions in adjacent rooms is necessary. Zone and CFD models can provide this
capability. As a screening tool, algebraic models could be used to model the HGL temperatures
in the room of fire origin. If the estimated fire conditions in the room of origin are determined not
to generate damage or ignition of targets, it can be concluded that targets in adjacent rooms are
also not expected to be damaged.
Zone Models
The zone model is an appropriate tool for addressing this scenario. Zone models are efficient
tools for scenarios involving relatively simple geometries (i.e., geometries and openings that can
be easily represented in rectangular parallelepipeds without compromising the technical
elements in the analysis). Consequently, the room geometry should be represented as
accurately as possible. One of the primary outputs of zone models is the height and
temperature of the HGL versus time in each of the rooms specified in the computational domain.
Zone models are also capable of determining target temperature (not just the temperature of the
gases surrounding the target), given the boundary conditions generated by the fire and the
thermophysical properties of the target.
CFD Model
A CFD model would be particularly appropriate for addressing targets located in adjacent rooms
in scenarios with complex geometries (i.e., geometries that cant be easily represented as
rectangular parallelepipeds). CFD models can describe the geometry of the compartment in
detail, including the opening(s) providing smoke migration paths to the adjacent room.

Readers are referred to Appendix G, which describes the analysis of targets in rooms remote
from the fire room.
3-15
3.2.4 Scenario 4: Targets in Rooms with Complex Geometries

The objective of this scenario is to calculate the time to damage for a target in the HGL in a
room with a complex geometry. For the case shown in Figure 3-6, the target is a cable in an
electrical raceway and the fire source is an electrical cabinet.

The first strategy involves using an algebraic or single compartment zone model to estimate the
HGL temperature. This approach requires that the complex geometry be reduced to a single
equivalent volume. In the case of two different ceiling heights, an empirical correlation or a
single compartment zone model may underestimate the temperature of the smaller volume and
overestimate the temperature of the larger. The fires energy is conserved, but it is not
expected to be uniformly distributed. For this reason, it is suggested that the model only be
applied to the portion of the compartment where the fire is postulated. This reduction in room
volume will result in a higher predicted HGL temperature. This strategy should not be applied if
the purpose of the calculation is to predict the activation of a sprinkler or smoke detector
because an overestimate of the HGL temperature will lead to an underestimate of the activation
time.
The second strategy is to use a model capable of describing the complex geometry. A zone
model can model the entire compartment as a collection of connected volumes. A CFD model
can block off portions of the numerical grid to account for geometric obstructions. In complex
geometries, HGL development can be significantly impacted by mixing associated with spilling
and ventilation, and these can only be modeled by zone and CFD models.
3-16

Algebraic Models
Detailed analyses of complex geometries typically cannot be easily accomplished with algebraic
models. However, for screening purposes, it is possible to use algebraic models. As mentioned
earlier, this approach can provide an approximation of the HGL temperature in which the target
may be immersed. To utilize this approach, first select the appropriate HGL (or room
temperature) model and then collect the required inputs, including room size, opening sizes,
boundary material properties, and HRR. Next, the complex geometry must be reduced to a
single equivalent volume while maintaining total surface area (due to the importance of energy
losses through the bounding surfaces) and ceiling height. It should be noted that the more
complex the space, the less ideal the equivalent volume/area approximation becomes. Based
on the estimates derived using the algebraic models, more detailed modeling may be indicated.
Zone Models
Zone models should also be used with caution when modeling this scenario. If the entire space
is modeled, the interface between lower and upper compartments is treated as a big door. The
entrainment correlations used by the zone model to handle vertical vents were not designed for
such large, open doors.
CFD Model
CFD models may be required when detailed analyses of complex geometries capable of
affecting fire development and the location of the HGL and the incident heat flux to the target
are desired. CFD models are expected to better estimate the overall compartment
temperatures, both upper and lower, because the basic methodology can handle non-uniform
ceilings.

Readers are referred to Appendix D, which consists of a switchgear fire in a room with a
complex geometry, and Appendix H, which consists of a fire inside the containment annulus.
3-17
3.2.5 Scenario 5: Main Control Room Abandonment

The objective of this scenario is to determine when control room operators will need to abandon
the control room due to fire-generated conditions inside the room. This scenario consists of a
fire, such as an electrical cabinet fire within the main control. A schematic diagram of this
scenario is shown in Figure 3-7. Notice the presence of a suspended ceiling in the room.

Main control room (MCR) abandonment is specified as being solely dependent on habitability
conditions. As mentioned in the previous sections, control room operators are considered
targets in this scenario, so it is necessary to establish the fire conditions that would force
operators out of the control room. This can be considered the abandonment criteria; for
example, visibility, temperature, heat flux, and toxicity are often the habitability indicators in
these scenarios. Tracking these conditions can provide the time at which the operator may
need to abandon the control room. Once the criteria have been established (see Chapter 11 of
NUREG/CR-6850 (EPRI 1011989) for details on habitability conditions), the fire-generated
conditions in the room can be calculated so that the abandonment time can be determined.
For MCR analyses, two ventilation conditions should be taken into consideration: (1) the
ventilation system is turned off, causing hot gases and smoke to accumulate inside the control
room and (2) the ventilation system is operating in smoke-purge mode.

Algebraic models
Algebraic models can be used to address individual abandonment criteria, but not all
simultaneously. Algebraic models are typically based on empirical correlations, not the basic
conservation laws of mass, momentum, and energy. For this reason, it is difficult to perform a
3-18

sensitivity analysis for a collection of correlations because they may not capture the
interdependent relationships of the predicted quantities.
Zone Models
Unlike algebraic models, zone models are capable of simultaneously tracking a number of
relevant output variables (e.g., habitability conditions) in this scenario. They are also capable of
modeling the impact of the various ventilation configurations required for modeling MCR
abandonment. Zone models are a good tool for modeling fires in the MCR as long as
compartment or zonal average conditions within a given enclosure provide the necessary
insights to support the conclusions. Localized radiation conditions can also be estimated using
zone models. Questions, such as average temperature, visibility conditions in the control room,
and smoke management, are well handled by zone models. In contrast, questions associated
with temperature, heat flux levels and visibility at specific locations within the control room (e.g.,
near a panel) are best handled by CFD models.
CFD Model
CFD models are also a good alternative to address this scenario, particularly if complex
geometries are involved or localized fire conditions are needed. CFD models have the added
advantage of handling rooms with complex geometries, intervening combustibles and
obstructions, and varying ventilation conditions.

Readers are referred to Appendix A, which describes the analysis of a fire in an MCR.
3-19
3.2.6 Scenario 6: Smoke Detection and Sprinkler Activation

The objective of this scenario is to calculate the response time of a smoke or heat detector. The
flow of heat and smoke from the fire to the detector may be obstructed by ceiling beams,
ventilation ducts, etc. Failure of a detector to actuate in response to a fire can delay the
response of either the fire brigade or an automatic suppression system. Typical scenarios are
shown in Figure 3-8.

For scenarios involving unobstructed smoke detector devices:
1. Determine the location of the detection device relative to the fire.
2. Select the detector response (activation) criteria. Chapter 11 of NUREG/CR-1805 contains
guidance on estimating smoke detector response times.
3. Calculate the detection time using the appropriate model.
3-20

For scenarios involving obstructed smoke detector devices:
1. Determine the following characteristics of the HGL using all the necessary inputs for an HGL
calculation, as described earlier in this chapter.
a. Temperature as a function of time.
b. Depth as a function of time. The smoke detector is expected to activate shortly after
the HGL reaches the bottom of the obstruction and spills into the location of the
device.
2. Select the detector response (activation) criteria. Chapter 11 of NUREG/CR-1805 contains
guidance on estimating smoke detector response times.
3. Calculate the response time of the given smoke detector once the combustion products
reach the detector.
For scenarios involving thermal devices (e.g., sprinklers, fusible links, or heat detectors), the
process is similar. The only difference is that the thermal device needs to be characterized with
relevant parameters, typically an activation temperature and the response time index (RTI). In
addition, the selected model should account for the heating process of thermally thin elements
(i.e., the heat detector device).

Algebraic models
Algebraic models can be used to determine time to heat or smoke detection when the fireinduced flows are not obstructed before reaching the detection device. By contrast, algebraic
models are typically not suitable when fire-induced flows, such as fire plumes or ceiling jets, will
be obstructed before reaching the detection device. In some cases, algebraic models that
estimate the HGL temperature as a function of time may be used for rough estimates of
activation times.
Zone Models
Zone models can address the different scenario conditions presented above; for instance,
CFAST and MAGIC are capable of determining time to smoke or heat detection when no
obstructions are present, and can simultaneously calculate smoke accumulation so that the time
for smoke detection activation can be estimated. This would provide an approximation, as zone
models do not directly account for complex geometries, including obstructions. These models
are not recommended for determining time to heat detection in obstructed geometries, since the
velocity of the gases impacting the heat detector is not available in zone model calculations. As
mentioned above, in some cases the HGL temperature alone may be used as a rough indicator
of smoke and heat activation times.
CFD Model
CFD models are good tools for estimating time to fire detection in complex geometries, including
obstructions, as they can describe the compartments complex geometries and mechanical
ventilation conditions in detail.

Readers are referred to Appendices B and E, which discuss the calculation of the time to smoke
and heat activation.
3-21
3.2.7 Scenario 7: Fire Impacting Structural Elements

The objective of this scenario is to characterize the temperature of structural elements exposed
to a nearby fire source. For the case shown in Figure 3-9, the exposure fire is an electrical
cabinet.

The fire modeling tools within the scope of this Guide should indicate whether the exposed
structural element will reach damaging temperatures. However, this information is often not
enough to determine whether the structural integrity of the compartment will be compromised by
the exposing fire conditions. A more detailed structural analysis (i.e., one that involves complex
temperature-dependent load-bearing calculations) may be needed if such a determination is
necessary.
Considering the limitations listed above, the following general guidance is provided:
1. Determine whether the structural element is directly above the fire, within the ceiling jet,
exposed to radiant heating, or within the HGL. The results of this determination will suggest
which model or combination of models should be used.
2. Calculate the temperature of the structural element based on the fire conditions affecting it.
This will require an initial estimate of the fire-generating conditions surrounding the structural
element, and, subsequently, the temperature of the element itself.
3-22

Algebraic models
Provided that the fire conditions affecting the structural element are appropriately identified
(e.g., a fire plume, a ceiling jet, flame radiation, or an HGL exposure without significant
contributions from any of the other three exposure mechanisms), algebraic models may be
capable of determining whether the structural element will be exposed to damaging conditions.
For example, plume temperature correlations can be used to determine the gas temperature
surrounding an element inside the fire plume. MQH calculations can indicate whether
compartment temperatures are near the critical temperature of structural elements; however,
these may provide overly conservative estimates, as the algebraic models do not account for
the heating of those structural elements that typically have large masses. Point source radiation
calculations can be used to estimate the heat flux to structural elements that are not directly in
the plume but close enough to the fire to become significantly heated.
Zone Models
Zone models are an appropriate tool to address this scenario, as the input file can be developed
to capture the relative location of the fire and the structural element(s) as long as compartment
or zonal average conditions within a given enclosure are appropriate to support the analysis.
Localized conditions, such as flame radiation or radiation from surfaces, can also be handled by
zone models. Structural elements can be represented as a target, and the incident fire
conditions can be tracked during the fire. Zone models are also capable of performing
conduction heat transfer calculations for the structural element, resulting in a prediction of the
temperature of the element itself.
CFD Model
CFD models are good tools for estimating temperatures in structural elements of complex
geometries, including obstructions, as they can handle the compartments complex geometries,
fire development, and mechanical ventilation conditions in detail, as well as the localized
heating of the structural elements.

Readers are referred to Appendix F, which describes the analysis of a lubricating oil fires effect
on structural elements.
3-23
3-24
UNCERTAINTY AND SENSITIVITY

The fire models discussed in this report are classified as deterministic to distinguish them from
stochastic models. In essence, this means that each model takes as input a set of values,
known as input parameters, that describe a specific fire scenario, and the models algorithms
then calculate the fire conditions within the compartment. The output of the models usually
takes the form of time histories of the various predicted quantities of interest, such as
temperature, heat flux or smoke concentration. In a sense, the model calculation is a virtual
experiment because the design of a model simulation often involves the same thought process
as the design of a physical experiment. The results of the calculation are likewise expressed in
terms similar to those of an experiment, including an estimate of the uncertainty. The sources
of uncertainty in a model prediction are different than those in an experimental measurement.
According to NUREG-1855, Volume 1, Guidance on the Treatment of Uncertainties Associated
with PRAs in Risk-Informed Decision Making (2009), there are three types of uncertainty
associated with a model prediction:
Parameter Uncertainty: Input parameters are often chosen from statistical distributions or
estimated from generic reference data. In either case, the uncertainty of these input parameters
is propagated through the calculation, and the resulting uncertainty in the model prediction is
known as the parameter uncertainty. The process of determining the extent to which the
individual input parameters affect the results of the calculation is known as a sensitivity analysis.
Model Uncertainty: Idealizations of physical phenomena lead to simplifying approximations in
the formulation of the model equations. In addition, the numerical solution of equations that
have no analytical solution can lead to inexact results. Model uncertainty is estimated via the
processes of verification and validation (V&V). The first seeks to quantify the error associated
with the mathematical solution of the governing equations, typically through numerical analysis,
while the second seeks to quantify the error associated with the simplifying physical
approximations, typically through comparison of model predictions and full-scale experiments.
Completeness Uncertainty: This refers to the fact that a model may not be a complete
description of the phenomena it is designed to predict. Some consider this a form of model
uncertainty because most fire models neglect certain physical phenomena that are not
considered important for a given application. For example, a model of sprinkler activation might
neglect water condensation.
The purpose of this chapter is to provide relatively simple methods to assess model and
parameter uncertainty. Completeness uncertainty is addressed, indirectly, by the same process
used to address the model uncertainty. Model uncertainty is based primarily on comparisons of
model predictions with experimental measurements as documented in NUREG-1824 (EPRI
1011999) and other model validation studies. Parameter uncertainty is addressed using simple
techniques to propagate input parameter uncertainty and to conduct sensitivity analyses.
4-1
4.1 Validation of Fire Models

capabilities. For example, NFPA 805 (NFPA, 2001) states that fire models shall only be applied
within the limitations of the given model and shall be verified and validated. The NRC Office of
Nuclear Regulatory Research (RES) and the Electric Power Research Institute (EPRI)
conducted a collaborative project for the V&V of the five selected fire models described in
Chapter 2. The results of this project were documented in NUREG-1824 (EPRI 1011999),
Verification and Validation of Selected Fire Models for Nuclear Power Plant Applications.
Twenty-six full-scale fire experiments from six different test series were used to evaluate the
models ability to estimate thirteen quantities of interest for fire scenarios that were judged to be
typical of those that might occur in a nuclear power plant (NPP). The results of the study are
summarized in Table 4-1. An explanation of this table is to follow.
Table 4-1. Results of the V&V study, NUREG-1824 (EPRI 1011999).
Output Quantity
HGL Temperature Rise*
FDTs
1.44
0.25
HGL Depth*
N/A
Ceiling Jet Temp. Rise
N/A
FIVE-Rev1
1.56
0.32
N/A
CFAST
MAGIC
FDS
Exp
1.06
0.12
1.01
0.07
1.03
0.07
0.07
1.04
0.14
1.12
0.21
0.99
0.07
0.07
1.84
0.29
1.15
0.24
1.01
0.08
1.04
0.08
0.08
Plume Temperature Rise
0.73
0.24
0.94
0.49
1.25
0.28
1.01
0.07
1.15
0.11
0.07
Flame Height**
I.D.
I.D.
I.D.
I.D.
I.D.
I.D.
I.D.
I.D.
I.D.
I.D.
I.D.
Oxygen Concentration
N/A
N/A
0.91
0.15
0.90
0.18
1.08
0.14
0.05
Smoke Concentration
N/A
N/A
2.65
0.63
2.06
0.53
2.70
0.55
0.17
Room Pressure Rise
N/A
N/A
1.13
0.37
0.94
0.39
0.95
0.51
0.20
Target Temperature Rise
N/A
N/A
1.00
0.27
1.19
0.27
1.02
0.13
0.07
1.32
0.54
1.07
0.36
1.10
0.17
0.10
Radiant Heat Flux
2.02
0.59
1.42
0.55
Total Heat Flux
N/A
N/A
0.81
0.47
1.18
0.35
0.85
0.22
0.10
Wall Temperature Rise
N/A
N/A
1.25
0.48
1.38
0.45
1.13
0.20
0.07
Wall Heat Flux
N/A
N/A
1.05
0.43
1.09
0.34
1.04
0.21
0.10
I.D. indicates insufficient data for the statistical analysis.

N/A indicates that the model does not have an algorithm to compute the given Output Quantity.
Underlined values indicate that the data failed a normality test because of the relatively small sample size.
* The algorithm used to compute the layer temperature and depth for the model FDS is described in NUREG-1824.
** All of the models except FDS use the Heskestad Flame Height Correlation (Heskestad, SFPE Handbook). These models were
shown to be in qualitative agreement with the experimental observations, but there was not enough data to further quantify this
assessment.
Models: Five fire models were selected for the study, based on the fact that they are commonly
used in fire analyses of NPPs in the U.S. Two of the models consist of simplified engineering
4-2

correlations (Fire Dynamics Tools (FDTs) and Fire-Induced Vulnerability Evaluation (FIVERev1)), two are zone models (Consolidated Fire Growth and Smoke Transport (CFAST) model
(CFAST) and MAGIC), and one is a computational fluid dynamics (CFD) model (Fire Dynamics
Simulator (FDS)).
Experiments: Six series of experiments (26 individual fire experiments in all) were selected for
the NRC/EPRI fire model validation study (NUREG-1824/EPRI 1011999). Each series
represented a typical fire scenario (for example, a fire in a switchgear room or turbine hall);
however, the test parameters could not encompass every possible NPP fire scenario. Table 2-5
lists various normalized parameters that can be used to characterize fire scenarios and the
ranges of the validation experiments. These parameters express, for instance, the size of the
fire relative to the size of the room, or the relative distance from the fire to critical equipment.
This information is important because typical fire models are not designed for fires that are very
small or very large in relation to the volume of the compartment or very large in relation to the
ceiling height.
For a given set of experiments and NPP fire scenarios, the user can calculate the relevant
normalized parameters. These parameters will either be inside, outside, or on the margin of the
validation parameter space. Consider each case in turn:
1. If the parameters fall within the ranges that were evaluated in the validation study, then
Table 4-1 can be referenced directly.
2. If only some of the parameters fall within the range of the study, additional justification is
necessary (see Section 2.3.7 for guidance). This is a common occurrence because realistic
fire scenarios involve a variety of fire phenomena, some of which are easier to estimate than
others. A case in point is the burning of electrical cabinets and cables. NUREG-1824 (EPRI
1011999) does not address these fires directly, even though some of the experiments used
in the study were intended as mock-ups of control or switchgear room fires. For scenarios
involving these kinds of fires, the heat release rates (HRR) are often taken from experiments
rather than being predicted by a model. It has been shown, in NUREG-1824 (EPRI
1011999) and other validation studies, that the models can estimate the transport of smoke
and heat with varying degrees of accuracy, but they have not been shown (at least not in
NUREG-1824 (EPRI 1011999)) to estimate the details of the fires ignition and growth.
While this does not eliminate the models from the analysis, it still restricts their applicability
to only some of the phenomena.
3. If the parameters fall outside the range of the study, a validation determination cannot be
made based on the results from the study. The modeler needs to provide independent
justification for using the particular model. For example, none of the experiments
considered in NUREG-1824 (EPRI 1011999) were under-ventilated. However, several of
the models have been independently compared to under-ventilated test data, and the results
have been documented either in the literature or in the model documentation. As another
example, suppose that the selected model uses a plume, ceiling jet, or flame height
correlation outside the parameter space of NUREG-1824 (EPRI 1011999) but still within the
parameter space for which the correlation was originally developed. In such cases,
appropriate references are needed to demonstrate that the correlation is still appropriate,
even if not explicitly validated in NUREG-1824 (EPRI 1011999). It is expected that the V&V
effort, as documented in NUREG-1824 (EPRI 1011999), will be updated over time to include
comparisons with additional test data and new versions of the models. These updates will
expand the validation range shown in Table 2-5 and ensure the availability of V&V
information for the latest versions of the models documented in this report.
4-3

Predicted Quantities: The experimental data for the validation study consisted of
measurements of one or more of the 13 physical quantities listed in the table. The FDTs and
FIVE-Rev1 do not possess algorithms to estimate every quantity; in the cases in which there
was no estimate, the table cell is labeled N/A.
Statistics: For each model and output quantity, a summary plot of the results is presented in
NUREG-1824 (EPRI 1011999). For example, Figure 4-1 compares the measured and predicted
target temperatures for the model FDS. If a particular prediction and measurement are the
same, the resulting point falls on the solid diagonal line. The longer-dashed off-diagonal lines
indicate the experimental uncertainty. Roughly speaking, points within the longer dashed lines
are said to be within experimental uncertainty, and in such cases it is not possible to further
quantify the accuracy of the prediction. Points falling outside the experimental uncertainty
bounds cannot be said to be free of model uncertainty. At the time of the publication of
NUREG-1824 (EPRI 1011999), the writing team decided to use the colors green and yellow to
indicate the degree to which the model predictions are inside or outside of the experimental
uncertainty bounds. However, since the writing of NUREG-1824 (EPRI 1011999), it was
decided to replace the color system with a more quantifiable metric of model accuracy.
700
FDS Target Temperature
Predicted Temperature (C)
600
500
400
300
200
100
0
0
100
200
300
400
500
600
700
Measured Temperature (C)
Figure 4-1. Sample set of results from NUREG-1824 (EPRI 1011999).
Consider again Figure 4-1. To make better use of results such as these, two statistical
parameters8 have been calculated for each model and each predicted quantity. The first
parameter, , is the bias factor. It indicates the extent to which the model, on average, underor over-predicts the measurements of a given quantity. For example, the bias factor for the data
shown in Figure 4-1 is 1.02. This means that the model has been shown to slightly
8
The statistical parameters listed in Table 4-1 are based on the versions of the fire models used in the V&V study, circa 2006. As
the models are improved and new validation data introduced, these values may change.
4-4

overestimate target temperatures by an average 2%, and this is represented by the red dashdot line just above the diagonal. The bias factor for each model and each output quantity is
listed in Table 4-1. Results of the V&V study, NUREG-1824 (EPRI 1011999)..
The second statistic9 in Table 4-1 is the relative standard deviation of the model, , and the
experiments, . These indicate the uncertainty or degree of scatter of the model and the
experiments, respectively. Referring again to Figure 4-1, there are two sets of off-diagonal
lines. The first set, shown as long-dashed black lines, indicate the experimental uncertainty.
The slopes of these lines are 1 2 (it is customary to express uncertainties in the form of 2sigma or 95% confidence intervals). The second set of off-diagonal lines, shown as shortdashed red lines, indicates the model uncertainty. The slopes of these lines are
2 . If the
model is as accurate as the measurements against which it is compared, the two sets of offdiagonal lines would merge. The extent to which the data scatters outside of the experimental
bounds is an indication of the degree of model uncertainty.
The derivation of the statistical parameters in Table 4-1 is provided in the next section, while
their use is described here. Suppose that a model prediction is denoted as . The true value
of the predicted quantity is specified as a normally distributed random variable with a mean,
/ . Using these values, the probability of
/ , and a standard deviation,
exceeding a critical value, , is:
1
erfc
2
(4-1)
Note that the complementary error function is defined as follows:

erfc
(4-2)
It is a standard function in mathematical or spreadsheet programs like Microsoft Excel10.

To summarize, the procedure for determining the probability that a quantity predicted by a
model could exceed a critical value is as follows:
1. Express the model prediction as a rise above its ambient value. Call this number . Note
that the ambient value of most output quantities is zero. Temperature, oxygen
concentration, and smoke layer height are exceptions. For these quantities, express the
predicted value as a temperature rise, oxygen decrease below ambient, and layer depth.
from Table 4-1, compute the mean,
/ , and standard
2. Using the values of and
/ , of the normal distribution for the quantity of interest.
deviation,
3. Use the equation to compute the probability that the predicted quantity could exceed a
critical value, . Remember to also express this critical value as a rise above ambient in
the same way as the predicted value, .
A few examples of this procedure are included in Section 4.3.
For some models/quantities, there was an insufficient amount of data to calculate the relative standard deviation of the distribution,
in which case I.D. is shown in the Table.
10
Excel 2007 does not evaluate erfc
for negative values of , even though the function is defined for all real . In such cases,
use the identity erfc
2 erfc .
4-5
4.2 Derivation of the Model Uncertainty Statistics

This section describes the derivation of the statistics listed in Table 4-1. These values
summarize the results of the NRC/EPRI fire model validation study documented in NUREG1824 (EPRI 1011999). This section is included for information only; there is no need for a
model user to perform this type of calculation. McGrattan and Toman (2011) provide additional
details on the development of these uncertainty calculations.
For each of the fire models and each of the output quantities that were evaluated in the study, a
plot similar to that shown in Figure 4-1. Sample set of results from NUREG-1824 (EPRI
1011999). was produced. For each measurement point, a single experimental measurement
was plotted against a single model prediction. The plot shows all the comparison points. The
calculation of the statistics uses this set of measured and predicted values, along with an
estimate of the experimental uncertainty. The purpose of the calculation is to subtract off, in a
statistical sense, the experimental uncertainty so that the model uncertainty can be estimated.
This calculation is subject to the following constraints:
1. The experimental measurements are unbiased, and their uncertainty is normally distributed
with a constant relative standard deviation,
(that is, the standard deviation as a fraction of
the measured value). Table 4-2 provides estimates of relative experimental uncertainties for
the quantities of interest in terms of 2-sigma (95 percent) confidence intervals.
2. The model error is normally distributed around the predicted value divided by a bias factor,
. The relative standard deviation of the distribution is denoted as .
The computation of the estimated bias and scatter associated with model error proceeds as
follows. Given a set of experimental measurements, , and a corresponding set of model
predictions, , compute the following:
ln
The standard deviation of the model error,
ln
(4-3)
, can be computed from the following equation:
ln
ln
(4-4)
The bias factor is:

exp ln
For a given model prediction, , the true value of the quantity of interest, is a normally
/ .
distributed random variable with a mean of / and a standard deviation of
4-6
(4-5)

Table 4-2. Experimental uncertainty of the experiments performed as part of the validation study
in NUREG-1824 (EPRI 1011999).
Quantity
HGL Temperature Rise
HGL Depth
Ceiling Jet Temperature Rise
Plume Temperature Rise
Gas Concentration
Smoke Concentration
Pressure (no forced ventilation)
Pressure (with forced ventilation)
Heat Flux
Surface or Target Temperature
0.14
0.13
0.16
0.14
0.09
0.33
0.40
0.80
0.20
0.14
There are a few issues to consider when using this procedure:

1. All values need to be positive, and each value needs to be expressed as an increase over
its ambient value. For example, the oxygen concentration should be expressed as a
positive number (i.e., the decrease in concentration below its ambient value). Thus, for an
ambient oxygen concentration of 21 % and a predicted value of 14 %, the predicted
decrease in oxygen concentration is 7 %.
2. If the measurement uncertainty is overestimated, the model error will be underestimated. If
the model error is less than the experimental uncertainty, the latter should be reevaluated.
The model cannot be shown to have less error than the uncertainty of the experiment with
which it is compared.
3. The procedure requires that the quantity ln is normally distributed. This is not
necessarily true, especially in cases where there are an insufficient number of points in the
sample. Figure 4-2 provides two examples in which the normality of the validation data is
tested11. In cases where the data is not normally distributed, only the bias is reported.
4. The validation data used to develop the accuracy statistics may be fairly sparse over a
particular range. For example, in Figure 4-1, for temperatures below 300 C (570 F), there
are a sufficient number of points to derive the statistics. For temperatures above 300 C
(570 F), however, there are less data points and these few points do not necessarily follow
the trend. In such cases, it is left to the discretion of the analyst to determine if it is
appropriate to calculate the model uncertainty using the derived statistics if the temperature
is clearly outside of the range. Depending on the application, it might be sufficient to
indicate the model tends to over-predict these higher temperatures, as is this case, or the
reviewer might request additional evidence that the model can be applied for this range of
temperatures.
11
The Kolmogorov-Smirnov test for normality has been applied using the software package SigmaPlot 10, Systat Software, Inc.
The default P value of 0.05 was used.
4-7
3
Plume Temperature, FDTs
Normality Test: Fail
Wall Temperature
Normality Test: Pass
Number of Data Points
Number of Data Points
40
30
20
10
0
1
10
10
Interval Number
Interval Number
Figure 4-2. Two examples demonstrating how the validation data is tested for normality.
4.3 Calculation of Model Uncertainty

This section contains a few exercises to explain the procedure for calculating model uncertainty.
These examples consider model uncertainty only; that is, the input parameters are not subject
to uncertainty.
4.3.1 Example 1: Target Temperature

Suppose that cables within a compartment fail if their surface temperature reaches 330 C
(625 F). The model FDS predicts that the maximum cable temperature due to a fire in an
electrical cabinet is 300 C (570 F). What is the probability that the cables could fail? For this
exercise, the model input parameters are not subject to uncertainty, only the model itself.
Step 1: Subtract the ambient value of the cable temperature, 20 C (68 F) to determine the
predicted temperature rise. Refer to this value as the model prediction:
300 C
20 C
280
(4-6)
Step 2: Refer to Table 4-1, which indicates that, on average, FDS over-predicts target
temperatures with a bias factor, , of 1.02. Calculate the adjusted model prediction:
280 C
275 C
1.02
Referring again to Table 4-1, calculate the standard deviation of the distribution:
0.13
280
1.02
36 C
(4-7)
(4-8)
Step 3: Calculate the probability that the actual cable temperature would exceed 330 C12:
12
In the result of Equation 4-9, the precise value for and were used rather than the rounded values shown.
4-8

330
1
erfc
2
330
1
erfc
2
20
36
275
0.16
(4-9)
The process is shown graphically in Figure 4-3. The area under the bell curve for
temperatures higher than 330 C (625 F) represents the probability that the actual cable
temperature would exceed that value. Note that this estimate is based only on the model
uncertainty.
Probability Density Function
0.014
0.012
Model Prediction
Adjusted Model Prediction
0.010
0.008
0.006
0.004
P(T>330 C)
0.002
0.000
100
150
200
250
300
350
400
450
500
Temperature (C)
Figure 4-3. Normal distribution of the true value of the cable temperature in a
hypothetical fire.
4.3.2 Example 2: Critical Heat Flux

As part of a screening analysis, the model MAGIC is used to estimate the radiant heat flux from
a fire to a nearby group of thermoplastic (TP) cables. Using, for example purposes, data from
NUREG/CR-6850 (EPRI 1011989), Appendix H, one of the damage criteria for TP cables is a
radiant heat flux to the target cable that exceeds 6 kW/m2. The model, by coincidence, predicts
a heat flux of 6 kW/m2. What is the probability that the actual heat flux from a fire will be
6 kW/m2 or greater? For this exercise, the model input parameters are not subject to
uncertainty, only the model itself.
Step 1: Unlike in the previous example, there is no need to subtract an ambient value of the
heat flux (it is zero). Thus, the model prediction is:
6 kW/m
(4-10)
Step 2: Refer to Table 4-1, which indicates that, on average, MAGIC over-predicts radiant heat
flux with a bias factor, , of 1.07. Calculate the adjusted model prediction:
6
1.07
5.6 kW/m
4-9
(4-11)

Referring again to Table 4-1, calculate the standard deviation of the distribution:
0.36
6
1.07
Step 3: Calculate the probability that the actual heat flux,

heat flux, :
6 kW/m
1
erfc
2
(4-12)
2.0 kW/m
, will exceed the critical value of the
1
6 kW/m
5.6 kW/m
erfc
2
2.0 kW/m 2
0.42
(4-13)
Even though the model predicts a peak radiant heat flux equal to the critical value, there is only
a 42% chance that the actual heat flux would exceed this value. This is mainly due to the fact
that MAGIC has been shown to overestimate the heat flux by about 7%.
It is important to note that this calculation of model uncertainty does not take into account the
input parameters, such as the HRR of the fire. It is only an assessment of how well the model
MAGIC can estimate the radiant heat flux to a target. Another key point is that the damage
threshold, temperature in the first example and heat flux in the second, is known exactly, i.e.,
without uncertainty.
4.4 Parameter Uncertainty

The previous sections describe how to express the uncertainty of a model prediction resulting
from the inherent limitations of the model itself. However, in most cases, the larger source of
uncertainty is the input parameters of the model, not the model itself. This section suggests
ways to assess the impact of this kind of uncertainty on the final prediction.
Parameter uncertainty is addressed in this chapter in two ways: (1) parameter uncertainty
propagation, and (2) sensitivity analysis. It is recognized that one or both approaches may be
necessary for a given application. Depending on the complexity of the model, the number of
input parameters will range from one to several dozen. The choice of which to analyze is based
on (1) the importance of the parameter to the outcome, and (2) its degree of uncertainty. In
almost all cases, the heat release rate of the fire dominates all other parameters in terms of
these two criteria.
The propagation of parameter uncertainty refers to the process of propagating the probability
distributions of one or more input parameters through the model in order to determine the
distributions of the model output quantities. Guidance on uncertainty propagation is provided in
Section 4.4.1.
Sensitivity analysis refers to the process of determining the "rate of change" of the model output
with respect to variations in the magnitude in one or more of the model inputs, usually one
parameter at a time. Guidance on sensitivity analysis is provided in section 4.4.2.
4.4.1 Parameter Uncertainty Propagation

Typically, fire models are run using a discrete set of input parameters that describe a single,
specific fire scenario. However, for some Fire PRA applications, it may be necessary to
consider the range of consequences due to the variability that can result from that specific fire
scenario within a particular compartment. If the key input parameters can be expressed in the
form statistical distributions, then the model output quantities may also be expressed as
4-10

distributions. In this way, it is possible to determine the probability of exceeding a critical
temperature, heat flux, or some other critical value.
Notarianni and Parry (SFPE Handbook, 4th edition) discuss a number of techniques for
propagating parameter uncertainty. The most common are Monte Carlo methods in which the
fire model is run repeatedly with randomly chosen input parameters, based on specified
probability distributions. For simple algebraic models, this technique is relatively simple and
there are various software packages available to help. For zone models, the technique
becomes more complicated because it requires more time to set up and run the model, but it is
still practical if the number of parameters is reduced and the ranges of the parameters are
appropriately discretized into bins. For CFD models, the technique is not practical except in
special cases where the number of model runs can be reduced to a relatively small number.
The increased accuracy afforded by the CFD model is often unwarranted given that the
uncertainty in the input parameters is typically greater than that of the models themselves.
Because the HRR is the most important input parameter in most fire model analyses, and
because NUREG/CR-6850 (EPRI 1011989) provides distributions of the HRR for a variety of
combustibles within an NPP, parameter uncertainty propagation for fire modeling may involve
only the HRR distribution applied within an algebraic model. In fact, Appendix E of NUREG/CR6850 (EPRI 1011989) provides data which can be used to illustrate a simple technique to
propagate the HRR distribution.
Suppose, for example, that as part of a Fire PRA the problem is to determine the probability of
flames extending above an electrical cabinet to a particular height, threatening a cable tray. To
answer this question, the flame height needs to be represented as a probability distribution.
Figure 4-4 displays the distribution13 of peak heat release rates, , for vertical cabinets with
more than one bundle of unqualified cable (NUREG/CR-6850, Appendix E). The probability
density function (pdf) is denoted
; , , where and (2.6 and 67.8) are parameters of this
particular gamma distribution.
13
NUREG/CR-6850 specifies gamma distributions for the various types of combustibles found within an NPP. Microsoft Excel
provides a built-in function (GAMMA.DIST) that calculates the probability density function given the parameters and .
4-11
0.005
HRR Distribution
0.004
0.003
0.002
0.001
0.000
0
200
400
600
Heat Release Rate (kW)
Figure 4-4. Distribution of HRR for an electrical cabinet fire.
For convenience, a spreadsheet can be used to take each value of , from 1 kW to 600 kW,
and compute a corresponding flame height, f (in meters), using Heskestads correlation:
f
0.235
1.02
(4-14)
The diameter of the fire, , is fixed at 0.48 m (1.6 ft), based on the equivalent diameter of the
vent. Whereas Appendix E of NUREG/CR-6850 recommends dividing the range of into 15
bins, it is just as easy for this example to compute the flame height for 600 values of (each
by the
bin has a width of 1 kW). The pdf for the flame height,
f , is related to the pdf for
following expression:
; ,
f
; ,
0.094
(4-15)
This distribution is shown in Figure 4-5. Note that when the derivative in Equation (4-15) is not
easily written in closed form, it is sufficient to calculate the bin width of the model output divided
by that of the model input. In this example, the bin width of the model input parameter, , is
1 kW and the bin width of the model output parameter is the difference in flame heights for two
successive values of .
4-12
1.0
Flame Height Distribution
0.8
0.6
0.4
0.2
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Flame Height (m)
Figure 4-5. Distribution of flame heights for the entire range of cabinet fires.
Once the pdf for the flame height is calculated, it can be used to determine the probability of the
flames reaching a certain height. In this case, the cable tray is 1.5 m (4.9 ft) above the top of
the cabinet. The probability that the flames from a randomly chosen fire will reach the cables is
given by the area beneath the curve in Figure 4-5 for flame heights greater than 1.5 m (4.9 ft).
In this example, it is approximately 0.31.
4.4.2 Sensitivity Analysis

The algebraic calculations described in this report are perhaps the models more amenable to
parameter uncertainty studies in practical applications. The more complex fire models
discussed in this report can require dozens of physical and numerical input parameters for a
given fire scenario. However, only a few of these parameters, when varied over their plausible
range of values, will significantly impact the results. For example, the thermal conductivity of
the compartment walls will not significantly affect a predicted cable surface temperature. Table
4-3 lists the input parameters whose impact on the given output quantity significantly outweighs
all the other parameters. The HRR is almost always one of these.
In Volume 2 of NUREG-1824 (EPRI 1011999), Hamins quantifies the functional dependence of
these key input parameters (see Table 4-3). These relationships are based either on the
governing mathematical equations or on algebraic models. The basic mathematical form of the
relationship is:
Output Quantity
Constant (Input Parameter)
Power
(4-16)
The exact value of the Constant is not important; rather, it is the Power that matters. The larger
its absolute value, the more important the Input Parameter. According to the McCaffrey,
Quintiere, and Harkleroad (MQH) correlation, for example, the hot gas layer (HGL) temperature
rise in a compartment fire is proportional to the HRR raised to the two-thirds power:
/
(4-17)
What is important here is the amount that the HGL temperature, , changes due to a shift in
the HRR, . It is the two-thirds power dependence, as found in Table 4-3, that matters. To
4-13

see why, take the first derivative of
differentials:
with respect to
and write the result in terms of
2
3
(4-18)
This is a simple formula with which one can readily estimate the relative change in the model
output quantity, /
, due to the relative change in the model input parameter, / .
The uncertainty in a measured quantity is often expressed in relative terms14. Suppose that the
uncertainty in the HRR of the fire, / , is 0.15, or 15%. The expression above indicates that a
15% increase in the HRR should lead to a 2/3 x 15 = 10% increase in the prediction of the HGL
temperature. The result is equally valid for a reduction; if the HRR is reduced by 15%, the HGL
temperature is reduced by 10%.
Table 4-3. Sensitivity of model outputs from Volume 2 of NUREG-1824 (EPRI 1011999).
Important Input
Parameters
Power Dependence
HRR
Surface Area
Wall Conductivity
Ventilation Rate
Door Height
2/3
-1/3
-1/3
-1/3
-1/6
Door Height
HRR
Production Rate
1/2
1
HRR
Soot Yield
1
1
Pressure
HRR
Leakage Rate
Ventilation Rate
2
2
2
Heat Flux
HRR
4/3
Surface/Target
Temperature
HRR
2/3
Output Quantity
HGL Temperature
HGL Depth
Gas Concentration
Smoke Concentration
This relationship is based on experimental data, and has nothing to do with any particular
model; however, an effective way to check a fire model is to take a simple compartment fire
simulation, vary the HRR, and ensure that the change in the HGL temperature agrees with the
correlation. Consider the two curves shown in Figure 4-6. For Benchmark Exercise #3 of the
International Collaborative Fire Model Project (ICFMP), Test 3 was simulated with FDS, using
HRR values of 1000 kW and 1150 kW. An examination of the peak values confirms that the
relative change in the HGL temperature (10%) is two-thirds the relative change in the HRR
14
Note that a differential relationship is only approximate. This method of relating input parameters to output quantities is valid for
relative differences that are less than approximately 30% in absolute value.
4-14

(15%), consistent with the empirical result of the MQH correlation. Even though FDS is a much
more complicated model than the simple expression shown above, it still exhibits the same
functional dependence on the HRR.
400
Temperature (C)
Hot Gas Layer Temperature

ICFMP BE #3, Test 3
300
200
100
0
0
10
15
20
25
30
Time (min)
Figure 4-6. FDS predictions of HGL Temperature as a function of time due to a 1,000 kW
fire (solid line) and a 1,150 kW fire (dashed).
This section illustrates the usefulness of sensitivity analysis. As an example, consider that
NFPA 805 uses the term Maximum Expected Fire Scenario (MEFS) to describe a severe fire
that could be reasonably anticipated to occur within a compartment and the term Limiting Fire
Scenario (LFS) to describe a severe fire that exceeds one or more performance criteria. The
analyst is often asked to determine the model inputs for both of these scenarios. For MEFS,
input parameters can be chosen from distributions for a particular percentile value. Gallucci
(2011) discusses the issues important in determining the appropriate choice of HRR distribution.
The development of the LFS is essentially a sensitivity analysis performed to identify which
combinations of input parameters or variables are critical to the analysis. The particular
variables to be evaluated depend entirely on the problem being analyzed. At a minimum, the
following parameters should be varied until failure conditions result: HRR, the fire growth rate or
the flame spread rate, the flame radiative fraction or the radiative power, and the location of the
fuel package relative to the target (if variable).
Suppose, for example, that as part of an analysis the problem is to determine the minimum HRR
needed to cause damage in a particular compartment whose HGL temperature is not to exceed
500 C (930 F). The geometrical complexity of the compartment rules out the use of the
algebraic and zone models, and that FDS has been selected for the simulation.
Step 1: Determine a reasonable, but conservative, estimate of what might be the maximum fire
that could occur in the compartment. Using data from NUREG/CR-6850, for this example,
suppose that a 98th percentile HRR for the electrical cabinet fire, 702 kW, has been chosen for
this representative estimate. Choose a model and calculate the peak HGL temperature.
Step 2: The model chosen is FDS and it predicts 450 C (840 F) for the selected fire scenario.
Adjust the prediction to account for the model bias, (see Table 4-1):
4-15

20
(4-19)
437
1.03
Step 3: Calculate the change in HRR required to increase the HGL temperature to 500 C
(930 F):
3
2
450
20
adj
3
2
adj
500
437
702 kW
437
20
159 kW
(4-20)
This calculation suggests that adding an additional 159 kW to the original 702 kW will produce
an HGL temperature in the vicinity of 500 C (930 F). This result can be double-checked by rerunning the model with the modified input parameters.
Table 4-3 lists several other parameters besides the HRR that can affect the HGL temperature.
Following the example just discussed, similar calculations can be performed in which these
other parameters are varied to determine how else the minimum damage threshold might be
reached. For example, suppose that the surface area, , of the compartment is 400 m2 (4300
ft2). How much would the surface area have to increase or decrease to raise the HGL
temperature to 500 C (930 F)? If the thermal conductivity of the walls, , is 0.1 W/m/K, how
much would it have to change? If the ventilation rate is 1 m3/s (2100 cfm), how much would it
have to change? If the door height, , is 2 m (6.6 ft), how much would it have to change?
Following the example for the HRR, the required changes in these parameters can be
calculated as follows:
3 400 m
adj
adj
(4-21)
500 437
417
0.045 W/m/K
(4-22)
3 1 m s
500 437
417
0.45 m /s
(4-23)
1.8 m
(4-24)
adj
181 m
3 0.1 W/m/K
adj
500 437
417
6 2m
500 437
417
For this example, to increase the HGL temperature by 63 C (145 F), one could increase the
HRR by 159 kW, decrease the surface area of the compartment by 181 m2 (1948 ft2), decrease
the thermal conductivity of the walls by 0.045 W/m/K, decrease the ventilation rate by 0.45 m3/s
(950 cfm), or decrease the door height by 1.8 m (5.9 ft). Of course, some of these options are
not physically possible. Room dimensions and thermal properties are not subject to significant
change, but the HRR and ventilation rates can vary significantly. Also note that if the relative
change in the parameter values exceeds 30 %, it is recommended that further calculations be
performed to confirm the estimated quantity changes.
4-16
4.5 Chapter Summary

This chapter describes the three forms of uncertainty related to fire modeling: parameter, model,
and completeness uncertainty. Model and completeness uncertainty are closely related, and it
would be impractical to evaluate them separately. The most practical way to quantify their
combined effect is to compare model predictions with as many experimental measurements as
possible in order to develop a robust statistical description of the models accuracy. The five
models considered in this repor underwent a validation study (NUREG-1824 (EPRI 1011999)) in
which their predictions were compared with measurements from a variety of full-scale
experiments. It is possible to take a given models prediction of a given quantity and determine
a distribution for the true value of this quantity. Rather than reporting the result of a calculation
as a single value, it is preferable to report the probability that the true value of a predicted
quantity will exceed a given critical value.
Treatment of parameter uncertainty is dependent on the application. Some analyses,
probabilistic risk assessments (PRAs) for example, employ relatively simple techniques to
propagate the distribution of the HRR through the model. More complex forms of PRAs can
involve a broad statistical sampling of input parameters from specified distributions. From a fire
modeling perspective, some modeling tools are more amenable to the evaluation of parameter
uncertainty given the complexity and computational times of some of these tools.
Deterministic applications usually consider worst case or bounding analyses, in which
extreme, yet plausible, input parameters are used. In mathematical terms, all of these
applications involve selecting parameters from relatively narrow or broad regions of the
parameter space. It is often prohibitive to consider all possible combinations of input
parameters, which is why a simple form of sensitivity analysis, outlined in this chapter, can be
used to extend the range of outcomes. Algebraic models may initially be used to indicate the
extent to which all of the output quantities of interest are sensitive to changes in the specified
HRR, reducing the need to re-run model simulations for an extensive number of values.
Over time, the V&V effort, documented in NUREG-1824 (EPRI 1011999), will be expanded to
include comparisons with additional test data and new versions of the models. These updates
will expand the validation range shown in Table 2-5 and ensure the availability of V&V
information for the latest versions of the models documented in this report. Model users should
refer to the latest version of the V&V when conducting their fire modeling analyses. In general,
the techniques documented in this chapter for analyzing uncertainty and sensitivity will remain
applicable regardless of the V&V version being used for the analysis.
4-17
4-18
REFERENCES
10 CFR 50, Voluntary Fire Protection Requirements for Light-Water Reactors, 10 CFR Part 50,
Section 50.48(c), RIN 3150-AG48, Federal Register, Volume 69, Number 115, U.S. Nuclear
Regulatory Commission, Washington, D.C., June 16, 2004.
AISC, Steel Construction Manual, 13th edition, American Institute of Steel Construction, New
York, New York, 2006.
ASME/ANS RA-Sa-2009, Standard for Level 1/Large Early Release Frequency Probabilistic
Risk Assessment for Nuclear Power Plant Applications, Addendum A, ASME/ANS RASa-2009, American Society of Mechanical Engineers, 2009.
ASTM E119-10a, Standard Test Methods for Fire Tests of Building Construction Materials,
American Society for Testing and Materials, West Conshohocken, PA, 2010.
ASTM E135505a (2005), ASTM Standard Guide for Evaluating the Predictive Capability of
Deterministic Fire Models, American Society for Testing and Materials, West
Conshohocken, PA, 2005.
Babrauskas, V., Ignition Handbook, Fire Science Publishers/Society of Fire Protection
Engineers, Issaquah, WA, 2003.
Buchanan, A. H., Structural Design for Fire Safety, John Wiley and Sons, LTD, Chichester,
England, 2001.
Drysdale, D., An Introduction to Fire Dynamics, John Wiley and Sons, 3rd Edition, Chichester, pp.
14, 283-284, 2011.
EPRI 1002981, Fire Modeling Guide for Nuclear Power Plant Applications, Electric Power Research
Institute, Palo Alto, CA, August 2002.
Forney, G. P., Users Guide for Smokeview Version 5 - A Tool for Visualizing Fire Dynamics
Simulation Data, Volume II: Technical Reference Guide, NIST Special Publication 10172, National Institute of Standards and Technology, Gaithersburg, MD, October 2010.
Gallucci, R.H.V., Examination of the Efficacy of the NFPA-805 Fire Modeling Approach
(Comparison between Maximum Expected and Limiting Fire Scenarios), ANS PSA
2011 International Topical Meeting on Probabilistic Safety Assessment and Analysis,
Wilmington, NC, March 13-17, 2011, American Nuclear Society, LaGrange Park, IL,
2011.
Gay, L., User Guide of MAGIC Software V4.1.1, EdF HI82/04/022/B, Electricit de France, France,
April 2005.
Gay, L., C. Epiard, and B. Gautier, MAGIC Software Version 4.1.1: Mathematical Model, EdF
HI82/04/024/B, Electricit de France, France, November 2005.
Gay, L., J. Frezabeu, and B. Gautier, Qualification File of Fire Code MAGIC Software Version 4.1.1:
Mathematical Model, EdF HI82/04/024/B, Electricit de France, France, November 2005.
5-1
REFERENCES
Grove, B. S., and J.G. Quintiere, Calculating Entrainment and Flame Height in Fire Plumes of
Axisymmetric and Infinite Line Geometries, Journal of Fire Protection Engineering, Volume
12, August, 2002, pp. 117 - 137.
Gottuk, D. T., and Lattimer, B. Y., Section 2, Chapter 2-5, Effect of Combustion Conditions on
Species Production, SFPE Handbook of Fire Protection Engineering, 4th Edition (P. J.
DiNenno, Editor-in-Chief), National Fire Protection Association and The Society of Fire
Protection Engineers, Quincy, MA, 2008.
Heskestad, G., Section 2, Chapter 2-1, Fire Plumes, Flame Height, and Air Entrainment, SFPE
Handbook of Fire Protection Engineering, 4th Edition (P. J. DiNenno, Editor-in-Chief),
National Fire Protection Association and The Society of Fire Protection Engineers, Quincy,
MA, 2008.
Holman, J. P., Heat Transfer, 7th edition, McGraw-Hill, New York, 1990.
Janssens, M., Section 3, Chapter 3-2, Calorimetry, SFPE Handbook of Fire Protection
Engineering, 4th Edition (P. J. DiNenno, Editor-in-Chief), National Fire Protection Association
and The Society of Fire Protection Engineers, Quincy, MA, 2008.
Jones, W., R. Peacock, G. Forney, and P. Reneke, CFAST: An Engineering Tool for Estimating Fire
Growth and Smoke Transport, Version 5 - Technical Reference Guide, NIST Special
Publication 1030, National Institute of Standards and Technology, Gaithersburg, MD, 2004.
Karlsson, B., and J. Quintiere, Enclosure Fire Dynamics, CRC Press, Boca Raton, Florida, 2000.
Kylu, U. O., and G.M. Faeth, Carbon Monoxide and Soot Emissions from Liquid-Fueled Buoyant
Turbulent Diffusion Flames, Combustion and Flame, 87:61-76, 1991.
McCaffrey, B. J., J. G. Quintiere, and M. F. Harkleroad, Estimating Compartment Temperature and
Likelihood of Flashover Using Fire Test Data Correlation, Fire Technology, Volume 17, No.
2, pp. 98-119, Quincy, MA, 1981.
McGrattan, K. et al., Fire Dynamics Simulator (Version 5) Technical Reference, Volume 3:
Validation, NIST Special Publication 1018-5, National Institute of Standards and
Technology, Gaithersburg, MD, 2010.
McGrattan, K., B. Klein, S. Hostikka, and J. Floyd, Fire Dynamics Simulator (Version 5) Users
Guide, NIST Special Publication 1019-5, National Institute of Standards and Technology,
Gaithersburg, MD, 2009.
McGrattan, K., and B. Toman, Quantifying the predictive uncertainty of complex numerical models,
Metrologia, 48:173-180, 2011.
Mulholland, G. W., and C. Croarkin, Specific Extinction Coefficient of Flame-Generated Smoke,
Fire and Materials, 24:227-230, 2000.
NFPA, Fire Protection Handbook, National Fire Protection Association, 20th Ed., A. E. Cote
(Editor), 2008.
NFPA 70 (NEC 2008), National Electric Code, National Fire Protection Association, Quincy, MA,
2008.
NFPA 805, Performance-Based Standard for Fire Protection for Light Water Reactor Electric
Generating Plants, National Fire Protection Association, Quincy, MA, 2001.
NEI 00-01 (2009), Guidance for Post Fire Safe Shutdown Circuit Analysis, Revision 2, Nuclear
Energy Institute, Washington, D.C., May, 2009.
5-2
REFERENCES
NEI 00-01 (2010), Guidance for Post Fire Safe Shutdown Circuit Analysis, Draft Revision 3,
Nuclear Energy Institute, Washington, D.C., 2010.
NEI 04-02 (2009), Guidance for Implementing a Risk-Informed Performance-Based Fire
Protection Program Under 10 CFR 50.48(c), Rev. 1, Nuclear Energy Institute,
Washington, D.C., September, 2009.
NIST NCSTAR 1-5F, Federal Building and Fire Safety Investigation of the World Trade Center
Disaster: Computer Simulation of the Fires in the World Trade Center Towers, National
Institute of Standards and Technology, Gaithersburg, MD, 2005.
NRC, Regulatory Guide 1.189, Fire Protection for Nuclear Power Plants, U.S. Nuclear
Regulatory Commission, Bethesda, MD, April, 2009.
NUREG-1805, Fire Dynamics Tools (FDTs): Quantitative Fire Hazard Analysis Methods for the U.S.
Nuclear Regulatory Commission Fire Protection Inspection Program, U.S. Nuclear
Regulatory Commission, Washington, D.C., December 2004.
NUREG-1824, Verification and Validation of Selected Fire Models for Nuclear Power Plant
Applications, Volume 1: Main Report, U.S. Nuclear Regulatory Commission, Office of
Nuclear Regulatory Research (RES), Washington, D.C., 2007, and EPRI 1011999, Electric
Power Research Institute (EPRI), Palo Alto, CA.
Applications, Volume 3: Fire Dynamics Tools (FDTs), U.S. Nuclear Regulatory Commission,
Office of Nuclear Regulatory Research (RES), Washington, D.C., 2007, and EPRI 1011999,
Electric Power Research Institute (EPRI), Palo Alto, CA.
Applications, Volume 4: Fire-Induced Vulnerability Evaluation (FIVE-Rev1), U.S. Nuclear
Regulatory Commission, Office of Nuclear Regulatory Research (RES), Washington, D.C.,
2007, and EPRI 1011999, Electric Power Research Institute (EPRI), Palo Alto, CA.
Applications, Volume 5: Consolidated Fire Growth and Smoke Transport Model (CFAST),
U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research (RES),
Washington, D.C., 2007, and EPRI 1011999, Electric Power Research Institute (EPRI), Palo
Alto, CA.
Applications, Volume 6: MAGIC, U.S. Nuclear Regulatory Commission, Office of Nuclear
Regulatory Research (RES), Washington, D.C., 2007, and EPRI 1011999, Electric Power
Research Institute (EPRI), Palo Alto, CA.
Applications, Volume 7: Fire Dynamics Simulator (FDS), U.S. Nuclear Regulatory
Commission, Office of Nuclear Regulatory Research (RES), Washington, D.C., 2007, and
EPRI 1011999, Electric Power Research Institute (EPRI), Palo Alto, CA.
NUREG-1855 Volume 1, Guidance on the Treatment of Uncertainties Associated with PRAs in
Risk-Informed Decision Making, U.S. Nuclear Regulatory Commission, Office of Nuclear
Regulatory Research (RES), Washington, D.C., 2009.
NUREG/CR-4680, Heat and Mass Release Rate for Some Transient Fuel Source Fires: A Test
Report, U.S. Nuclear Regulatory Commission, Washington, D.C., December 2004.
5-3
REFERENCES
NUREG/CR-6738, Risk Methods Insights Gained from Fire Incidents, U.S. Nuclear Regulatory
Commission, Office of Nuclear Regulatory Research (RES), Washington, D.C., 2001.
NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities: Volume 1:
Summary and Overview, U.S. Nuclear Regulatory Commission, Office of Nuclear
Regulatory Research (RES), Washington, D.C., 2005, and EPRI 1011989, Electric Power
Research Institute (EPRI), Palo Alto, CA.
NUREG/CR-6850, EPRI/NRC-RES Fire PRA Methodology for Nuclear Power Facilities: Volume 2:
Detailed Methodology, U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory
Research (RES), Washington, D.C., 2005, and EPRI 1011989, Electric Power Research
Institute (EPRI), Palo Alto, CA.
NUREG/CR-6850 Supplement 1, Fire Probabilistic Risk Assessment Methods Enhancements, U.S.
Nuclear Regulatory Commission, Office of Nuclear Regulatory Research (RES),
Washington, D.C., 2010, and EPRI 1019259, Electric Power Research Institute (EPRI), Palo
Alto, CA.
NUREG/CR-6931, Cable Response to Live Fire (CAROLFIRE), Volume 1: Test Descriptions and
Analysis of Circuit Response Data, U.S. Nuclear Regulatory Commission, Washington,
D.C., 2007.
NUREG/CR-6931, Cable Response to Live Fire (CAROLFIRE), Volume 2: Cable Fire Response
Data for Fire Model Improvement, U.S. Nuclear Regulatory Commission, Washington, D.C.,
2007.
NUREG/CR-6931, Cable Response to Live Fire (CAROLFIRE), Volume 3: Thermally-Induced
Electrical Failure (THIEF) Model, U.S. Nuclear Regulatory Commission, Washington, D.C.,
2007.
NUREG/CR-7010, Cable Heat Release, Ignition, and Spread in Tray Installations during Fire
(CHRISTIFIRE), Phase 1: Horizontal Trays, National Institute of Standards and
Peacock, R., K. McGrattan, B. Klein, W. Jones, and P. Reneke, CFAST Consolidated Model
of Fire Growth and Smoke Transport (Version 6) Software Development and Model
Evaluation Guide, Special Publication 1086, National Institute of Standards and
Peacock, R., W. Jones, P. Reneke, and G. Forney, CFAST: An Engineering Tool for Estimating
Fire Growth and Smoke Transport, Version 6 Users Guide, Special Publication 1041,
National Institute of Standards and Technology, Gaithersburg, MD, 2008.
Quintiere, J. G., Principles of Fire Behavior, Delmar Publishers, 1998.
Quintiere, J. G., Fundamentals of Fire Phenomena, John Wiley, 2006.
Salley, M.H., J. Dreisbach, K. Hill, R. Kassawara, B. Najafi, F. Joglar, A. Hamins, K. McGrattan,
R. Peacock; and B. Gautier, Verification and Validation--How to Determine the
Accuracy of Fire Models, Fire Protection Engineering, vol. 34, pp. 34 - 44, Spring 2007.
Schiffiliti, R., and W. Pucci, Fire Modeling, State of the Art, Fire Detection Institute, 1996.
SFPE, SFPE Engineering Guide to Assessing Flame Radiation to External Targets from Pool
Fires, SFPE Engineering Guide, Society of Fire Protection Engineers, Bethesda, MD,
March 1999.
5-4
REFERENCES
SFPE, SFPE Engineering Guide to Fire Exposures to Structural Elements, SFPE Engineering
Guide, Society of Fire Protection Engineers, Bethesda, MD, November, 2005.
SFPE, SFPE Engineering Guide to Piloted Ignition of Solid Materials Under Radiant Exposure,
SFPE Engineering Guide, Society of Fire Protection Engineers, Bethesda, MD, January,
2002.
SFPE, SFPE Engineering Guide to Predicting Room of Origin Fire Hazards, SFPE Engineering
Guide, Society of Fire Protection Engineers, Bethesda, MD, November, 2007.
SFPE, SFPE Engineering Guide to Substantiating a Fire Model for a Given Application, SFPE
Engineering Guide, Society of Fire Protection Engineers, Bethesda, MD, June, 2010.
SFPE, SFPE Handbook of Fire Protection Engineering, 4th Edition (P. J. DiNenno, Editor-in-Chief),
National Fire Protection Association and The Society of Fire Protection Engineers, Quincy,
MA, 2008.
SFPE, SFPE Engineering Standard on Calculating Fire Exposures to Structures, SFPE S.01.2001,
Society of Fire Protection Engineers, Bethesda, MD, 2011.
Underwriters Laboratories, Single Station Fire Alarm Device, UL 217, Underwriters Laboratories,
Northbrook, Illinois.
Wallis, G. B., Draft Final NUREG-1824, Verification and Validation of Selected Fire Models for
Nuclear Power Plant Applications, Memorandum from Advisory Committee on Reactor
Safeguards to Luis Reyes, ML062980154, U.S. Nuclear Regulatory Commission, October,
2006.
Yuan, L., and G. Cox, An Experimental Study of Some Line Fires, Fire Safety Journal, Volume 27,
1996, pp. 123 139.
5-5
REFERENCES
5-6
INTRODUCTION TO THE APPENDICES

Eight unique fire scenarios that typify different fire modeling applications likely to be
encountered in commercial nuclear power plants (NPP), have been evaluated using the process
and guidance of Chapters 2 4 of this document. The analysis of the eight example fire
scenarios is summarized in Appendices A through H. Table 6-1 provides a summary of the
individual fire scenarios as well as a cross-index between the eight appendix fire scenarios and
the seven generic fire scenarios described in Chapter 3. In the scenarios, one or more aspects
of Chapter 3 generic fire scenarios are incorporated. For example, Appendix B considers both
localized exposure effects near the fire (Chapter 3.2.1) and hot gas layer (HGL) effects in a
simple room (Chapter 3.2.2). Default values for many quantities (e.g., material properties,
numerical parameters, physical parameters, empirical constants, etc.) are referenced
throughout these application examples. Fire model users must understand and justify the use
of default values for their applications. Fire model users are expected to assess the
appropriateness of default values provided in the fire models and make changes or adjust
values as necessary. Fire model user guides, handbooks, and other technical documentation
are useful for this purpose.
Table 6-1. Appendix fire scenario descriptions and cross-index.
Appendix
A
B
C
D
E
F
G
H
Appendix Scenario Description

Determination of the time to main control room (MCR) abandonment given a low
voltage panel fire.
Determination of the potential for a switchgear panel fire to damage cables above the
fire and adjacent to the fire.
Determination of the potential for a lubricant oil fire to damage cables protected with
an electrical raceway fire barrier system (ERFBS).
Determination of the potential for an motor control center (MCC) panel fire to damage
cables via the HGL in an irregularly shaped enclosure.
Determination of the potential for a transient trash fire to damage cables in a stack of
horizontal cable trays located directly above the fire.
Determination of the potential for a lubricant oil fire to cause structural damage.
Determination of the potential for a large transient fuel package fire to damage cables
in an adjacent space.
Determination of the potential for a cable tray fire to damage a nearby raceway with
and without sprinklers.
Chapter 3
Generic Fire
Scenario
3.2.5
3.2.1, 3.2.2
3.2.2
3.2.4
3.2.1
3.2.7
3.2.3
3.2.1, 3.2.6
The structure of each appendix reflects the analysis sequence as well as the recommended
means by which to document each type of fire modeling analysis. In particular, the appendices
use the following format
Fire modeling objective. This section clearly presents the issue that is being evaluated
and the reason the analysis is being performed.
Description of the fire scenario. This section provides a detailed description of the fire
scenario being evaluated, including important parameters such as the room geometry,
6-1
INTRODUCTION TO THE APPENDICES

ventilation, fuel package data, material properties, and scenario specific information
required for a successful analysis.
Selection of the fire models. This section presents the basis for selecting the different
types of fire models used to address the fire modeling objective. This includes
considerations of the model capability as well as the available verification and validation
(V&V) basis. Generally, two or more models are selected in order to highlight the
strengths and weaknesses of the models for the given application. The selected models
are not necessarily the only models that can be used, but the basis provided is broadly
applicable to a given class of fire models.
Estimation of the fire-generated conditions. This section presents model specific inputs
and supporting calculations and identifies applicable model nuances for the scenario
considered.
Evaluation of results. The results of the fire models are presented in terms of the data
necessary to support the fire modeling objective in this section. Model sensitivity and
uncertainty are also described for each output parameter and model considered.
Conclusions. A summary of the fire modeling analysis, including the outcome relative to
the fire modeling goal is presented in this section.
The authors strove to create fire scenarios based on actual commercial NPP configurations in
order to provide realistic fire scenario development and results. Be cautioned that the fire
modeling results are not generic, and different applications will require careful consideration of
the various input parameters, the fire model applicability, and the dominant exposure
mechanism(s). However, the structure presented is generic, and when followed it is expected to
lead to a fire modeling analysis that could be used to support safety related applications.
This report also contains a CD. On the CD you will find the complete NUREG-1934 final report
in pdf format and the input files used for the fire model runs discussed in each appendix. In
addition, the current versions of the FDTs, CFAST and FDS that were used for each appendix
are on the CD. The complete layout and contents of the CD are described in the README file
included on the CD.
6-2
APPENDIX A
CABINET FIRE IN MAIN CONTROL ROOM
A.1 Modeling Objective

The purpose of the calculations described in this appendix is to approximate the length of time
that the main control room (MCR) remains habitable after the start of a fire within a low-voltage
control cabinet. These calculations follow the guidance provided in NUREG/CR-6850
(EPRI 1011989), Volume 2, Chapter 11, Detailed Fire Modeling (Task 11). MCR fire scenarios
are treated differently than fires within other compartments, mainly because it is necessary to
consider and evaluate forced abandonment in addition to equipment damage.
A.2 Description of the Fire Scenario

General Description: A fire ignites within a control cabinet containing XPE/neoprene cables.
The door to the MCR is normally closed, and normal ventilation conditions are in place at the
start of the fire. The typical ambient temperature is 20 C (68 F). Following guidance given in
Chapter 11 of NUREG/CR-6850 (EPRI 1011989), two scenarios are considered, one in which
the ventilation system is turned off and one in which the ventilation system is switched to
smoke-purge mode at the start of the fire.
Geometry: Drawings of the MCR are shown in Figure A-1 and Figure A-2. The compartment
has a variety of control cabinets in addition to typical office equipment, such as computer
monitors on table tops. There is an open-grate suspended ceiling above the floor, a photograph
of which is shown in Figure A-3. This suspended ceiling is specified to be of noncombustible
construction for this scenario, but the combustibility of such ceilings should be verified for other
MCR scenarios. One wall of the compartment is made of 0.9 m (3 ft)-thick concrete with no
additional lining material. The other bounding walls are constructed of 1.6 cm (5/8 in) gypsum
board supported by steel studs. The floor is a slab of concrete covered with low-pile carpet that
is nominally 1.25 cm (0.5 in) thick. The ceiling is concrete with the same thickness as the floor
(0.5 m (1.6 ft)), but with no lining material.
Materials: Thermal properties of various materials in the compartment are listed in Table 3-1.
Carpet is not listed in the table, but, according to Table 6-5 of NUREG-1805, the thermal inertia
(kc) for Carpet (Nylon/Wool Blend) is 0.68 (kW/m2/K)2 s, its ignition temperature is 412 C
(774 F), and its minimum heat flux for ignition is 18 kW/m2. The steel housing of the electrical
cabinets is nominally 1.5 mm (0.06 in) thick.
Ventilation: During normal operation, the mechanical ventilation system provides five air
changes per hour (ACH). As shown in Figure A-1, ventilation is provided through six supply
diffusers and two return vents of nominally the same size. The supply air to the compartment is
equally distributed among the six supply vents, and the return air is drawn equally from the two
returns. A 120 Pa overpressure (relative to the adjacent compartments) is maintained in the
MCR during normal operation. Leakage from the compartment occurs via a 1.3 cm (0.5 in) high
crack under the 0.9 m (3 ft) wide door. All other penetrations are sealed. Smoke-purge mode
provides 25 air changes per hour and is activated manually by an MCR operator in accordance
A-1

with specified procedures, which is known for this control room to be 120 seconds after the start
of a fire.
Figure A-1. Geometry of the main control room.
A-2
Figure A-2. Main control room details.
A-3
Figure A-3. Photograph of a typical open-grate ceiling.
Figure A-4. Photograph of a typical control cabinet.
A-4

Fire: The fire ignites due to an electrical malfunction in bundles of qualified XPE/neoprene
cables inside an isolated control cabinet (Figure A-4), designated as the Fire Origin Cabinet in
Figure A-2. The fire grows according to a t-squared curve to a maximum value of 702 kW in
12 min and remains steady for eight additional minutes, consistent with NUREG/CR-6850 (EPRI
1011989), Appendix G, for a low-voltage cabinet fire involving more than one bundle of qualified
cable. After 20 min, the fires heat release rate (HRR) decays linearly to zero in 19 min. A peak
fire intensity of 702 kW represents the 98th percentile of the probability distribution for the HRR
in cabinets of this general description. The HRR curve is shown in Figure A-5.
Heat Release Rate
800
700
HRR (kW)
600
500
400
300
200
100
0
0
600
1200
1800 2400
Time (s)
3000
3600
Figure A-5. Time history of the HRR used by all models in the MCR scenario.
Based on a physical assessment of the cabinet, it is determined that the exterior panels of the
burning cabinet do not open before or during the fire. The smoke, heat, and flames are
exhausted from an air vent in the side of the cabinet. The top of the air vent is 0.3 m (1 ft) below
the top of the cabinet. The air vent is 0.6 m (2 ft) wide and 0.2 m (8 in) high. The cabinet is
2.4 m (8 ft) tall.
The heat of combustion and product yields for XPE/neoprene cable are taken from Table 3-4.16
of the SFPE Handbook, 4th edition and are listed in Table A-1. When estimating the
composition of the fires combustion products, the jacket and insulation material of the cable are
taken as an equal-parts mixture of polyethylene (C2H4) and neoprene (C4H5Cl), with the
effective chemical formula C3H4.5Cl0.5.
Table A-1. Data for MCR fire based on XPE/neoprene electrical cable.
Parameter
Effective Fuel Formula
Peak HRR
Time to reach peak HRR
Heat of Combustion
CO2 Yield
Soot Yield
CO Yield
Radiative Fraction
Mass Extinction Coefficient
Value
C3H4.5Cl0.5
702 kW
720 s
10,300 kJ/kg
0.63 kg/kg
0.175 kg/kg
0.082 kg/kg
0.53
8700 m2/kg
Source
Combination of polyethylene and neoprene
NUREG/CR-6850 (EPRI 1011989), App. G
SFPE Handbook, 4th ed., Table 3-4.16
Mulholland and Croarkin (2000)
A-5

Habitability: The MCR is manned 24 hours per day during normal plant operations. To assess
habitability of the compartment, the operator position indicated in Figure A-1 is used. According
to NUREG/CR-6850 (EPRI 1011989), Volume 2, Chapter 11, Detailed Fire Modeling, a space
is considered uninhabitable if at least one of the following occurs:
1. The incident heat flux at 1.8 m (6 ft) exceeds 1 kW/m2. A smoke layer temperature of
approximately 95 C (200 F) generates this level of heat flux.
2. The smoke layer descends below 1.8 m (6 ft) from the floor, and the optical density of the
smoke is greater15 than 3 m-1.
A.3 Selection and Evaluation of Fire Models

This section discusses the overall modeling strategy. In particular, it describes the process of
model selection, including a discussion of the validity of these models for the given fire scenario.
A.3.1 Temperature Criterion

The forced ventilation correlation of Foote, Pagni, and Alvares (FPA) that is incorporated in the
Fire-Induced Vulnerability Evaluation (FIVE-Rev1) is used to estimate the compartment
temperatures under smoke purge ventilation conditions. This correlation is not appropriate in
the scenario without ventilation. The Consolidated Fire Growth and Smoke Transport Model
(CFAST) and the Fire Dynamics Simulator (FDS) are used to estimate the hot gas layer (HGL)
temperature for both the smoke purge and unventilated scenarios. One advantage of a
Computational Fluid Dynamics (CFD) model like FDS for this fire scenario is that it can estimate
habitability conditions at the specific location of the operator, whereas a zone model like CFAST
can only estimate average conditions in the hot gas layer.
A.3.2 Heat Flux Criterion

The point source radiation heat flux model is first used to estimate the heat flux to the operator.
FIVE-Rev1 and the Fire Dynamics Tools (FDTs) both contain methods to estimate the heat flux
from a fire to a target located a specified distance away. However, distant targets, including the
operators, may not be exposed directly to the thermal radiation; it is also possible that the
descending hot gas layer will be responsible for some fraction of the heat flux to which the
operator is exposed. Neither FIVE-Rev1 nor the FDTs include calculation methods to account
for this source of thermal radiation. CFAST and FDS will be used to determine whether the
thermal radiation from the HGL reaches the critical value for tenability.
A.3.3 Visibility Criterion

FIVE-Rev1 and the FDTs do not include correlations to estimate visibility in a ventilated
compartment, but correlations from the SFPE Handbook are applied to develop an estimate of
smoke obscuration in the MCR under smoke purge conditions. With a smoke purge ventilation
rate of 25 ACH, a well-mixed smoke environment is expected, and the average smoke
concentration and visibility conditions can be calculated using these methods. CFAST and FDS
are also used to address the visibility criterion. CFAST has an advantage over MAGIC for this
15
The original edition of NUREG/CR-6850 (EPRI 1011989) contains an error in the specification of the optical density (NRC ADAMS
-1
Accession Number ML061630360). The value of 3 m is correct.
A-6

analysis because it computes the smoke obscuration. FDS is particularly useful because it
does not require that the smoke is confined to a uniform descending layer. It is expected that
air currents in the room will mix the smoke to the floor. Zone models typically do not allow the
smoke layer to descend below the height of the base of the fire.
The case of the unventilated enclosure fire scenario is addressed with CFAST and FDS to
estimate the smoke layer interface position and smoke layer optical density as a function of
time. One advantage of a CFD model like FDS for this fire scenario is that it can estimate
habitability conditions at the specific location of the operator.
A.3.4 Validation
The principal source of validation data justifying the use of the fire models discussed above for
this scenario is the U.S. Nuclear Regulatory Commission/Electric Power Research Institute
(NRC/EPRI) verification and validation (V&V) study documented in NUREG-1824 (EPRI
1011999). The National Institute of Standards and Technology (NIST) has expanded the
NRC/EPRI V&V to include the latest versions of CFAST (6.1.1) (Peacock, 2008) and FDS
(5.5.3) (McGrattan, 2010). In particular, a Factory Mutual/Sandia National Laboratories
(FM/SNL) test series was designed specifically as a mock-up of a control room in a nuclear
power plant (NPP). One of these experiments (Test 21) involves a fire within a hollow steel
cabinet.
Table A-2 lists the important non-dimensional parameters that characterize the fire scenario and
the ranges for which the NRC/EPRI validation study is applicable. A few parameters fall outside
the validation parameter space and are addressed individually:
The Fire Froude Number falls outside the range. This parameter is essentially a measure of
the fires heat output relative to its base area. In this example, the fire is specified as
emanating from the side of the cabinet, with the vent opening serving as its base. This
leads to a higher value of
than would be calculated if the fire were burning completely
outside of the cabinet. Thus, the high value of
is the represents more severe fire
conditions than would be expected if the fire were specified as burning partially within the
cabinet.
The relatively low Equivalence Ratio for the compartment is a result of the relatively large
amount of air forced into the room during the smoke purge mode. Twenty-five air changes
per hour is a considerable flow rate, and no validation experiment in NUREG-1824 (EPRI
1011999) involved such a high ventilation rate. However, the results of all the model
simulations indicate that the scenario in which the ventilation is turned off is most likely to
compromise human habitability, and the presence of any level of ventilation reduces room
temperature and heat flux and increases visibility.
For the scenario with no ventilation, the classic definition of the Equivalence Ratio does not
apply because there is no supply of oxygen in the room. However, it can be shown that
there is sufficient oxygen in the room to sustain the specified fire. The total mass of oxygen
in the room is the product of the density of air, , the volume of the room, , and the mass
fraction of oxygen in the air, O :
O ,tot
1.2 kg/m
1945 m
A-7
0.23
537 kg
(A-1)

The mass of oxygen required to sustain the fire is equal to the total energy produced by the
fire divided by the energy released per unit mass oxygen consumed:
702 kW
O ,req
12
8
3
13,100 kJ/kg
60 s/min
19
min
2
69 kg
(A-2)
These calculations show that the quantity of oxygen in the room would be able to sustain the
specified cabinet fire.
The ratio of the Target Distance relative to the Fire Diameter, / , exceeds the range of the
validation study. However, this parameter is only relevant to the point source radiation heat
flux calculation, which is by definition more accurate, as the target moves further from the
source. Thus, although the parameter is outside the validation range, it is not outside of the
methodologys range of validity.
A-8

Table A-2. Normalized parameter calculations for the MCR fire scenario. See Table 2-5 for further
details.
Quantity
Normalized Parameter Calculation
Fire Froude
Number
702 kW
1.2 kg/m
Flame
Height,
,
relative to the
Ceiling
Height,
Ceiling Jet
Radial
Distance, cj ,
relative to the
Ceiling
Height,
Equivalence
Ratio, , of
the room,
based on
Forced
Ventilation of
Purge Mode
Compartment
Aspect Ratio
Target
Distance, ,
relative to the
Fire
Diameter,
1.0 kJ/kg/K 293 K 0.4
3.7
1.02
0.9
0.4 m 3.7
6.2
1.02
702 kW
13,100 kJ/kg 3.7 kg/s
0.23
24.6 m
5.2 m
4.7
1.2 kg/m
13.4 m /s
22
No
0.2 1.0
Yes
1.2 1.7
N/A
0.04 0.6
No
0.6 5.7
Yes
2.2 5.7
No
0.014
3.7 kg/s
16.2 m
5.2 m
8.8 m
0.4 m
0.4 2.4
2.7 m
N/A Ceiling jet targets are not included in simulation.
In
Range?
6.2
9.8 m/s
2.1 m 2.7 m
5.2 m
NUREG1824
Validation
Range
3.1
Notes:
(1)
(2)
The effective diameter of the base of the fire, , is calculated using

4 / ,
where is the area of the cabinet vent.
, is the sum of the height of the fire base above the
The Flame Height,
floor and the fires flame length.
A-9
A.4 Estimation of Fire-Generated Conditions

This section provides details as to how each of the models was set up and run.
A.4.1 Algebraic Models

General: The forced ventilation correlation of FPA is used to estimate the HGL temperature of
the MCR for the smoke purge scenario. The point source method is used to estimate the heat
flux from the fire to the operator for both the smoke purge and non-ventilated scenarios. The
smoke concentration and visibility are calculated for the smoke purge scenario based on quasisteady conditions, as described below.
Temperature in the Smoke Purge Scenario
Figure A-6 is a schematic diagram that illustrates the smoke purge scenario. The FPA
correlation requires room dimensions to be specified in terms of length, width, and height. For
this example, the selected compartment is not a rectangular parallelepiped, so it needs to be
represented with an effective length, width, and height that provide the same volume and
boundary surface area as the actual compartment. The effective compartment height is taken
as the actual height of 5.2 m (17 ft) because it is important to maintain the same compartment
height for smoke filling calculations. Next, the effective length and width are calculated to
maintain the same volume and boundary surface area of the actual compartment. Because the
actual and effective heights are the same, this is equivalent to maintaining the same floor area
( fl ) and wall perimeter ( ) of the enclosure. From Figure A-1, the MCR floor area is 372 m2
(4004 ft2), and the perimeter is 81.6 m (274 ft). Maintaining the total floor area and perimeter
yields effective compartment dimensions of 27.1 m (89 ft) by 13.8 m (45 ft). This is calculated
by solving the following two equations for the effective length and width,
and :
fl
(A-3)
The parameters for the FIVE-Rev1 implementation of the FPA are listed in Table A-3. Note that
the specified time-dependent HRR is used in the calculation, but that the calculation does not
include either the fires elevation above the floor or any other information about the fire. The
walls, ceiling, and floor are all specified as gypsum board rather than concrete because the FPA
correlation only accounts for one type of lining material. Gypsum board was chosen because it
has a lower thermal conductivity than concrete, which results in a slightly higher HGL
temperature.
Table A-3. Summary of input parameters for the FPA calculation of the MCR.
Parameter
Room height (H)
Room effective length (Le)
Room effective width (We)
Room boundary material
Mech. ventilation rate ( )
Ambient temperature (Ta)
Fire parameters
Value
5.2 m
27.1 m
13.8 m
Gypsum board
13.4 m3/s
20 C
A-10
Source
Figure A-1
Equation (A-3)
Equation (A-3)
Table 3-1
Specified (25 ACH)
Specified
Table A-1

Uniform materialpropertiesandthickness
forwalls,floor,andceiling
Constant exhaustinbalance
withsupply
Constant airsupply
Uniform
compartment
temperature
Pointsourcefirewithtime
dependentheatreleaserate
butnospecifiedlocationor
height
Closeddoor
Figure A-6. Schematic diagram of the FPA calculation for the MCR smoke purge scenario.
Heat Flux
The point source model is used to estimate the heat flux from the flames to the operator when
the fire is at its peak HRR. The peak HRR, , is 702 kW, the radiative fraction, , is 0.53, and
the distance from the cabinet vent to the operator is approximately 8.8 m (29 ft). The heat flux
is calculated:
0.53
4
702 kW
8.8 m
0.38 kW/m
(A-4)
While this heat flux prediction is well below the critical value of 1 kW/m2, it does not account for
the thermal radiation from the HGL. Thus, the point source method can be used as a screening
tool, and further analysis can be performed by CFAST and FDS.
Smoke concentration and visibility
Neither the FDTs nor FIVE-Rev1 include methods to calculate smoke concentrations or visibility
in mechanically ventilated enclosure fires, but calculation methods provided in Section 3,
Chapter 9, of the SFPE Handbook, 4th ed. are relatively simple to apply and are based on the
same principles and concepts embodied in zone models. These hand calculations provide an
estimate of the fire-generated smoke concentrations and visibility conditions for this scenario
and will indicate whether more detailed modeling is warranted.
The soot mass generation rate,
, is the product of the soot yield, , and the mass burning
rate of fuel,
. The latter quantity is obtained by dividing the HRR, , by the heat of
combustion, :
0.175
702 kW
10,300 kJ/kg
A-11
0.012 kg/s
(A-5)

The soot mass fraction in the smoke layer,
tot
, is then calculated:
0.012 kg/s
1.2 kg/m
13.4 m s
The extinction coefficient of the smoke,

8700 m kg
0.00075 kg/kg
(A-6)
, is calculated:
1.2 kgm
0.00075 kg/kg
7.8 m
(A-7)
Here
is the mass-specific extinction coefficient listed in Table A-1. By definition, the optical
density of the smoke is related to the extinction coefficient via the expression:
ln 10
7.8 m
2.3
3.4 m
(A-8)
This calculated optical density is then compared with the tenability limit, 3 m-1. This calculation
indicates that the visibility criterion for tenability would be exceeded for the MCR smoke purge
scenario based on the specified parameters and the MCR uniformly filling with smoke.
However, because this analysis is based only on the peak rather than the time-dependent HRR,
further modeling with zone and CFD models is warranted.
A-12
A.4.2 Zone Model

Geometry: CFAST divides the geometry into one or more compartments connected by vents.
For this simulation, the entire compartment is modeled as a single compartment. As with the
algebraic models, zone models simulate fires in compartments with rectangular floor areas. The
strategy for selecting effective room dimensions is the same as for the FPA analysis described
above. Figure A-7 shows the compartment geometry inputs to CFAST.
Figure A-7. Compartment geometry and surface material selection in CFAST for the MCR
Fire.
While there are numerous cabinets and tables in the compartment, they do not significantly
change the overall volume of the room and may be neglected as obstructions. There are no
mechanisms within CFAST to account for the open-grate ceiling. This ceiling provides a
negligible flow resistance to the heat and air that go through it, so it can be ignored. It is
expected that neglecting it will lead to slightly higher HGL temperatures because there is less
resistance for the rising smoke and hot gases.
Fire: In CFAST, a fire is described as a source of heat placed at a specific point within a
compartment that generates combustion products according to user-specified combustion
chemistry. Consistent with typical practice for the use of zone fire models for electrical cabinet
fires, the fire is positioned at the top of the air vent, 0.3 m (1 ft) below the top of the cabinet, at
A-13

the center of the cabinet. The air vent dimensions are 0.6 m (2 ft) wide and 0.2 m (0.7 ft) high.
The effective diameter of the fire is 0.4 m (1.3 ft), as noted in Table A-2.
Combustion chemistry in CFAST is described, at a minimum, by the production rates of CO,
CO2, and soot. The yield of HCl is calculated by the model based on the specified molecular
formula and the conversion of all of the chlorine into hydrochloric acid. Inputs for the fire
specification in CFAST, taken directly from Table A-1, are shown in Figure A-8.
Note: Values for Total Mass, Heat of Gasification, and Volatilization Temperature are set at default values.
Figure A-8. Specification of the fire in CFAST for the MCR Fire.
Materials: CFAST does not include the ability to model individual walls of different materials,
but it can model different materials for walls, floors and ceilings. For this example, the
compartment walls are specified as being entirely made of gypsum wallboard which has a
thermal conductivity lower than the other wall materials. The floor and ceiling are modeled as
0.5 m (1.6 ft) thick concrete.
Ventilation: For the smoke-purge calculation, air is supplied to the MCR via the six supply
vents and exhausted through the two returns. The total ventilation rate is 25 air changes per
hour, or 13.4 m3/s. Mechanical ventilation inputs for CFAST are shown in Figure A-9. A
snapshot of the CFAST simulation is shown in Figure A-10.
A-14
Note: Values for Begin Dropoff At and Zero Flow At are set at default values.
Figure A-9. Mechanical ventilation inputs in CFAST for the MCR fire.
Figure A-10. Smokeview rendering of the CFAST simulation of the MCR fire with
mechanical ventilation.
A-15
A.4.3 CFD Model

Geometry: The entire compartment is included in the computational domain. The exterior
concrete wall coincides with the boundary of the computational domain, meaning that the inside
surface of the concrete wall is flush with the boundary of the computational domain, and the
properties of concrete (including its thickness) are applied to this boundary. The tables (made
out of wood) and the electrical cabinets (made out of steel) are included in the simulation. Note
that the drop ceiling is not modeled because it is open and for this example provides a negligible
resistance to the heat and air that go through it. An FDS/Smokeview rendering of the scenario
is shown in Figure A-12.
The computational mesh consists of a uniform grid of cells that are approximately 0.2 m (0.7 ft)
on a side. A simple grid resolution study demonstrates that because the details of the fire (other
than its specified heat and smoke production rates) within the cabinet are not relevant for this
simulation, there is no need to further refine the grid in the vicinity of the cabinet. An
explanation related to choosing the appropriate grid size can be found in NUREG-1824 (EPRI
1011999), Vol. 7.
Fire: Following the guidance in Chapter 12 of NUREG/CR-6850 (EPRI 1011989),
Supplement 1, the fire is modeled as emanating from the upper vent of the burning cabinet.
The fuel stoichiometry is input to the model as specified above. FDS requires the designation of
a single gaseous fuel molecule via the number of carbon and hydrogen atoms in the surrogate
fuel, plus the number of other atoms in the molecule that play no role in the reaction. The soot
yield and heat of combustion are input directly from Table A-1.
Materials: The cabinets are represented by closed boxes with the specified properties of steel.
The tables are assigned the properties of plywood that is 5 cm (2 in) thick. The table legs are
not modeled because they play little role in the fire or heat transfer calculation to the solids.
Concrete and gypsum properties are applied to the applicable walls and ceiling. The floor is
modeled as a 1 cm (0.4 in) thick carpet over a 0.5 m (1.6 ft) thick concrete slab. The concrete
properties are taken directly as specified. The carpet properties are obtained by choosing
individual values of the thermal conductivity, , density, , and specific heat, , so that their
product equals the given value of
.
0.0017 kW/m/K
200 kg/m
2.0 kJ/kg/K
0.68 kW/m /K
The
used to estimate the individual thermal parameters is derived from an empirical ignition
model, and it may differ somewhat significantly from the true value. However, in this case, the
heat losses to the carpet are not expected to play a significant role in the simulation and the
choice of its thermal properties is not expected to affect the results. If the boundary is expected
to be an important heat loss surface, then a more accurate determination of the thermal
parameters is necessary.
Ventilation: Air is supplied to the MCR via the six supply vents and exhausted through the two
returns. The supply rate is divided equally among the six supply vents, and the return rate is
divided equally among the two returns. The leakage from the compartment is modeled by
specifying a small vent located at the base of the door through which air escapes at a rate
determined by the pressure difference between the MCR and ambient. Note that the door crack
itself is not modeled explicitly, as the numerical grid is not fine enough. Rather, the leak is
spread over a slightly larger area. The volume flow through the leakage area, , is estimated
via the equation:
A-16

2
(A-9)
where AL is the actual leakage area (0.9 m (3 ft) by 0.013 m (0.04 ft), or 0.0117 m2 in this
case), is the pressure difference between the inside and outside of the compartment (Pa),
is the ambient air density.
and
Figure A-11. FDS/Smokeview rendering of the MCR, as viewed from above.
A.5 Evaluation of Results

The habitability of the MCR depends on the temperature, heat flux, and smoke concentration to
which the operators would be exposed. According to NUREG/CR-6850 (EPRI 1011989),
Volume 2, Chapter 11, abandonment of the MCR is occurs if the gas temperature 1.8 m (6 ft)
above the floor exceeds 95C (200F) or if the heat flux exceeds 1.0 kW/m2 or if the optical
density exceeds 3 m-1.
Table A-4 summarizes the results of all the models for the three tenability criteria. For each
predicted value, a calculation is performed to determine the Probability of Exceeding the Critical
Value. The procedure for calculating this probability is given in Chapter 4, and it accounts for
the model bias and standard deviation. The purpose of this table is to highlight the criterion that
is most likely to be exceeded, so that further analysis can be focused on this criterion and the
model or models that predict it. Each criterion is discussed in greater detail in the following
subsections.
A-17
Table A-4. Summary of the model predictions of the MCR scenario.
Model
Bias
Factor,
Standard
Deviation,
Ventilation
Critical
Value
Probability
of
Exceeding
70
95
0.000
61
48
82
70
95
95
95
95
0.000
0.000
0.009
0.000
0.4
0.1
0.2
0.6
0.4
1
1
1
1
1
0.000
0.000
0.000
0.228
0.000
7.6
0.5
54
31
3
3
3
3
0.471
0.000
0.912
0.909
Predicted
Value
Temperature (C), Initial Value = 20 C

FIVE-Rev1
(FPA)
CFAST
FDS
CFAST
FDS
1.56
0.32
1.06
1.03
1.06
1.03
0.12
0.07
0.12
0.07
Purge
No Vent.
Heat Flux (kW/m2)
FIVE-Rev1
CFAST
FDS
CFAST
FDS
1.42
0.81
0.85
0.81
0.85
0.55
0.47
0.22
0.47
0.22
Purge
No Vent.
Optical Density (m-1)
CFAST
FDS
CFAST
FDS
2.65
2.7
2.65
2.7
0.63
0.55
0.63
0.55
Purge
No Vent.
A.5.1 Temperature Criterion

The HGL temperature and height predictions are summarized in Table A-4 and are shown in
detail in Figure A-12. None of the analyses imply that the temperature tenability limit would be
exceeded by a fire of this type, regardless of the ventilation system. It is important to note that
neither the FPA correlation nor CFAST estimate the temperature at the operator location
specifically. For the purpose of assessing habitability, the HGL temperature is used to
approximate the flux condition to which the operator would be exposed, regardless of whether
the HGL descends to the operators height. This means that the FPA and CFAST analyses
have an extra level of conservatism built in for this particular case.
The FPA correlation predicts a peak HGL temperature of 70 C (158 F) when the smoke purge
system is on, but, based on the CFAST calculation, it is expected that the layer height would not
descend to the operator level if the purge system were in operation.
CFAST predicts that the HGL temperature reaches just above 80 C (176 F) in 20 min when
the smoke purge system is off. The HGL descends to 2 m (6.6 ft) above the floor in
approximately the same amount of time and thus remains above the head of the operator.
When the smoke purge system is on, CFAST predicts that the peak HGL temperature reaches
A-18

approximately 60 C (140 F), but that the smoke layer does not descend beyond a meter below
the ceiling due to the operation of the smoke exhaust system.
The FDS predictions of HGL temperature are lower than those of the other models because
FDS accounts for the mixing of heat and smoke with ambient air due to the high purging flow,
since it models flow within the compartment in detail.
The HGL temperature is largely a function of the amount of energy from the fire that is carried
aloft in the smoke plume. The model simulations have all used, based on data from the SFPE
Handbook, 4th edition, a convective fraction of the HRR that is relatively low for the kind of
cables under consideration. Typically, approximately 65% of the fires energy is lofted upwards
in the plume, whereas in this case the models have all used a convective fraction of only 47%
(one minus the radiative fraction). As a result, the HGL temperature might be lower than one
would expect from a typical fire because a higher percentage of its energy is radiated away.
Referring to Table 4-3, the HGL temperature rise is proportional to the HRR to the two-thirds
power. For the purpose of this analysis, the HRR can be regarded as the convective HRR. If
the convective HRR were increased by 38%, the HGL temperature rise would increase by
approximately two-thirds of 38%, or 25%. This would have the effect of increasing the CFASTpredicted temperature rise from 60 C (140 F) to 75 C (167 F). Given an ambient
temperature of 20 C (68 F), this means that if CFAST were to use the conventional 35%
radiative fraction (65% convective fraction), its prediction16 of the HGL in the no-ventilation case
would be approximately 95 C (203 F), the critical value for abandonment. A similar argument
can be made for the other HGL predictions.
Temperature near Operator
HGL Height
6
FPA (Purge)
120
CFAST (No Vent.)
100
CFAST (Purge)
FDS (No Vent.)
80
CFAST (No Ventilation)
Height (m)
Temperature (C)
140
FDS (Purge)
60
40
FDS (No Ventilation)
4
3
2
1
20
0
0
0
600
1200
1800
2400
3000
3600
600
1200
1800
2400
3000
3600
Time (s)
Time (s)
Figure A-12. Hot Gas Layer Temperature and Height for the MCR scenario.
A.5.2 Heat Flux Criterion

In the fire scenario that includes the operation of the smoke purge system, none of the models
predict that the heat flux to the operator exceeds the tenability criterion (see Figure A-13). In
fact, CFAST and FDS estimate a peak flux of approximately 0.1 kW/m2, a value that is one-tenth
the critical value. With the smoke purge system turned off, FDS predicts a peak heat flux of
16
The bias factor for the CFAST predictions of HGL temperature is 1.06. Taking this into account, the CFAST prediction of HGL
temperature for the 35% radiative fraction case would be 90 C (194 F) rather than 95 C (203 F). This is close enough to the
critical value to warrant further modeling.
A-19

0.4 kW/m2 and CFAST predicts 0.6 kW/m2. However, as in the case of the HGL temperature
criterion, it is important to consider the ramifications of the decision to use a radiative fraction of
53% rather than a value more typical of most fires, 35%. Table 4-3 suggests that the heat flux
is proportional to the HRR to the four-thirds power. If the models were to use a radiative fraction
of 35% rather than 53%, the convective HRR would be 38% greater, in which case the heat flux
from the HGL layer onto the operator could increase by as much as 4/3 times 38%, or 50%.
Referring to Figure A-13, this would have the effect of increasing the CFAST prediction from
0.6 kW/m2 to 0.9 kW/m2, close to the critical value of 1 kW/m2. In fact, the validation study
documented in NUREG-1824 (EPRI 1011999) indicates that CFAST tends to underpredict the
total heat flux by 19%, on average. Given this fact and the discussion above on the HGL
temperature, it should be noted that in the unventilated case, CFAST predicts that the HGL
would descend to a level comparable in height to the operator (approximately 2 m), and it is
reasonable to conclude that the operator would be exposed to a heat flux comparable to the
habitability threshold.
Heat Flux to Operator
Heat Flux (kW/m2)
1.0
FIVE (Purge)
CFAST (No Vent.)
0.8
CFAST (Purge)
FDS (No Vent.)
0.6
FDS (Purge)
0.4
0.2
0.0
0
600
1200
1800
2400
3000
3600
Time (s)
Figure A-13. Predicted heat flux at the location of the operator.
A.5.3 Visibility Criterion

The optical density results are shown in Figure A-14. As with temperature, the CFAST
prediction is based on its upper layer smoke concentration calculation, whereas that of FDS is
based on the actual operator location. The simple algebraic techniques described in A.4.1 and
CFAST both predict that the optical density will exceed the critical threshold, even when the
purge system is on. FDS, however, predicts a much lower optical density in the purge mode
scenario for two reasons. First, FDS does not limit the transport of smoke to a descending layer
like CFAST; and, second, FDS does not uniformly mix the smoke over the entire compartment
volume like the simple algebraic model. As the operator stands relatively close to two supply
vents, the supplied fresh air keeps this vicinity clearer than other areas.
When the smoke purge system is off, FDS predicts that the visibility tenability criterion will be
exceeded at the operator position in about 12 min. Such conditions would force abandonment
of the MCR. CFAST predicts that the visibility tenability criterion would be exceeded in the
smoke layer in approximately 7 min, but it also predicts that the smoke layer remains above the
operators head throughout the fire simulation, which suggests that the MCR would not need to
be abandoned. The fact that the HGL remains above 2 m (6.6 ft) is partially an artifact of the
A-20

zone model. There is no mechanism in CFAST for the smoke layer to descend below the base
of the fire; a fire with a lower base height could result in a lower HGL elevation.
There is considerable uncertainty in the smoke yield of real fires, especially in cases where the
fire might be under-ventilated inside of a cabinet. A value of 0.175 (kg soot per kg fuel
consumed) was chosen for the smoke yield in the models, even though literature values range
from 0.01 to 0.2 (kg soot per kg fuel consumed), depending on the fuel. In addition to the
uncertainty in the specified input value of the smoke yield, the NRC/EPRI V&V study (NUREG1824 (EPRI 1011999)) indicates that both CFAST and FDS overestimate measured smoke
concentrations, on average, by factors of 2.65 and 2.70, respectively. In light of these
uncertainties in both models and in the input parameters, it is prudent to consider the sensitivity
of the simulation results to the selected value of the smoke yield. Table 4-3 indicates that the
optical density is directly proportional to the smoke yield. This means that if the smoke yield is
doubled, the predicted optical density is doubled as well. The curves in Figure A-14 can easily
be adjusted to show the effect of a variation in the smoke yield, but the predicted abandonment
times in the unventilated scenario do not change significantly with changes in the smoke yield.
Optical Density near Operator
Optical Density (1/m )
60.0
SFPE (Purge)
CFAST (No Vent.)
50.0
CFAST (Purge)
FDS (No Vent.)
40.0
FDS (Purge)
30.0
20.0
10.0
0.0
0
600
1200
1800
2400
3000
3600
Tim e (s)
Figure A-14. Optical density predictions for the MCR scenario.
A.6 Conclusion
A fire modeling analysis has been performed to assess the habitability of the MCR in the event
of a fire within an isolated electrical cabinet. The fire is not expected to spread to other
cabinets. Of the three MCR abandonment criteria, it is most likely that the operators would be
forced to abandon the MCR because the optical density would surpass 3 m-1 approximately 12
minutes after the fire ignites if the smoke purge system is not activated before this time,
according to the FDS analysis. A simple analytical method and the zone model CFAST indicate
that the optical density would exceed the critical value with the smoke purge system on and with
the ventilation system turned off. However, these analyses are based on the use of several
important conservative parameters. For the smoke purge case, the analytical method predicts
that the smoke fills the entire compartment uniformly, even though the FDS analysis shows that
the supply vents maintain visibility in the vicinity of the operator location. CFAST reports the
optical density of the upper layer, but does not predict that the upper layer would descend to the
level of the operator in either the purge or no-ventilation scenario based on the conservative
specifications, at least for a fire having a base height of 2 m (6.6 ft).
A-21
A.7 References
1. Mulholland, G.W., and C. Croarkin, Specific Extinction Coefficient of Flame Generated
Smoke, Fire and Materials, 24:227230, 2000.
2. NIST SP 1018-5, Fire Dynamics Simulator (Version 5), Technical Reference Guide,
Volume 3, Experimental Validation, 2010.
3. NIST SP 1030. CFAST: An Engineering Tool for Estimating Fire Growth and Smoke
Transport, Version 5 - Technical Reference Guide, National Institute of Standards and
Technology, Gaithersburg, Maryland, 2004.
4. NUREG-1805, Fire Dynamics Tools (FDTs) Quantitative Fire Hazard Analysis Methods for
the U.S. Nuclear Regulatory Commission Fire Protection Inspection Program, 2004.
5. NUREG-1824 (EPRI 1011999), Verification and Validation of Selected Fire Models for Nuclear
Power Plant Applications, 2007.
6. NUREG/CR-6850 (EPRI 1011989), EPRI/NRC-RES Fire PRA Methodology for Nuclear
Power Facilities, 2005.
7. SFPE Handbook of Fire Protection Engineering, 4th edition, 2008.
A.8 Attachments (on CD)

1. FDS input files:
a. Main_Control_Room_No_Purge.fds
b. Main_Control_Room_Purge.fds
2. CFAST input files:
a. Cabinet fire in MCR No Ventilation.in
b. Cabinet fire in MCR.in
3. Algebraic calculation input files:
a. FPA_AppA.xlsx
A-22
APPENDIX B
CABINET FIRE IN SWITCHGEAR ROOM
B.1 Modeling Objective

The calculations described in this appendix estimate the effects of fire in a switchgear room
electrical cabinet on overhead cables and other nearby cabinets (targets). These calculations
are part of a larger fire analysis described in Chapter 11 of NUREG/CR-6850 (EPRI 1011989),
Volume 2, Detailed Fire Modeling (Task 11). The switchgear room (SWGR) contains both
Train A and Train B safety-related equipment that is not separated as required by 10 CFR 50,
Appendix R. The lack of separation between the two has been identified as an unanalyzed
condition. The purpose of the calculation is to analyze this condition and determine whether
these targets fail, and, if so, at what time failure occurs.
B.2 Description of the Fire Scenario

General Description: The 4160 V SWGR is located in the auxiliary building and contains three
banks of cabinets (labeled A, B, and C in Figure B-1). Cabinet banks A and C contain safetyrelated equipment. In addition to the cabinets, there are nine cable trays, with three stacks of
three trays each located, directly above each of the cabinet banks. The lower two trays above
the middle bank of cabinets contain control cables for safety-related equipment. The
compartment is not normally occupied. Ambient temperature in the room is 20 C (68 F).
Construction: Plan and section views of the SWGR are shown in Figure B-1. The
compartment floor, ceiling, and walls are concrete, nominally 0.5 m (1.6 ft) thick. The cabinets,
ventilation ducts, and cable trays are made of steel. The cabinet housings are 1.5 mm (0.06 in)
thick. The ducts are 2 mm (0.08 in) thick. The cable tray steel is 3 mm (0.12 in) thick. The
trays are stacked 0.5 m (20 in) apart, bottom to bottom. The tray width is 0.8 m (31 in).
Materials: The thermal properties of the concrete and steel are listed in Table 3-1. The cable
trays are filled with PE-insulated, PVC-jacketed control cables, which have a diameter of
approximately 15 mm (0.6 in), a jacket thickness of approximately 1.5 mm (0.06 in), and seven
conductors. The cable mass per unit length is 0.38 kg/m (0.25 lb/ft), and the mass fraction of
copper is 0.68. The cables are contained in nine stacked cable trays. There are 50 cables in
each tray.
Thermoplastic (TP) cables are considered damaged when their interior17 temperature reaches
205 C (400 F) or the exposure heat flux reaches 6 kW/m2 (NUREG-1805, Appendix A). The
damage criteria for the adjacent cabinet is taken to be equal to that for PE/PVC cable since the
cables inside the cabinet are unqualified.
Ventilation: There are three supply and three return vents located near the side walls, as
shown in Figure B-1. Each vent is 0.5 m (1.6 ft) by 0.6 m (2.0 ft), and each has an airflow rate
of 0.47 m3/s (1000 cfm). The mechanical ventilation is normally on, and normal operations
continue during the fire. The supply air to the compartment is equally distributed among the
supply vents, and the return air is drawn equally from the returns. The compartment has only
17
The cable interior temperature is taken to be that of the insulation material surrounding the conductors.
B-1

one door, which is normally closed. The room temperature is maintained at 20 C (68 F), and
the pressure is comparable to adjacent compartments. Leakage from the compartment occurs
via a 2.5 cm (1 in) high crack under the 0.91 m (3 ft) wide door indicated on the drawing. All
other penetrations are sealed.
Fire: The fire ignites in one cabinet in the middle bank, as specified in Figure B-1. The cabinet
door is closed, but there are vents on the top of the cabinet for air circulation. The cabinet
contains more than one bundle of unqualified cable. The fire grows following a t-squared
curve to a maximum value of 464 kW in 12 min and remains steady for eight additional min,
consistent with NUREG/CR-6850 (EPRI 1011989), page G-5, for a cabinet with more than one
bundle of unqualified cable. After 20 min, the fire heat release rate (HRR) decays linearly to
zero in 12 min. A peak fire intensity of 464 kW represents the 98th percentile of the probability
distribution for HRRs in cabinets with unqualified cable in scenarios where flames propagate
through cable bundles. From a cabinet configuration perspective, this selection is appropriate
for control cables where cable loading is typically higher than in other types of cabinets. From
an applications perspective, the use of the 98th percentile is consistent with the guidance
provided in NUREG/CR-6850 (EPRI 1011989) for evaluating fire conditions with different fire
intensities (including the 98th percentile) within the probability distribution range.
There is an air vent on the top the cabinet. The air vent is 0.6 m (2 ft) wide and 0.3 m (1 ft) long.
The cabinet is 2.4 m (8 ft) tall. Consistent with NUREG/CR-6850 (EPRI 1011989), the fire burns
within the interior of the cabinet, and the smoke, heat, and possibly flames exhaust from the air
vent at the top of the cabinet.
In this scenario, the cables above Cabinet B are not only targets, they are also intervening
combustibles that may add to the overall HRR of the fire. Predicting the ignition and growth of
a cable fire is challenging for all the models, which is why empirical models are used instead. In
this case, the 464 kW cabinet fire described above is supplemented by additional heat from the
cable fire. In this scenario, the cables are ignited by the cabinet fire, after which a relatively
simple model for predicting the growth and spread of a fire within a vertical stack of horizontal
cable trays is applied. The model is referred to as FLASH-CAT, short for Flame Spread over
Horizontal Cable Trays. This scenario follows Appendix R of NUREG/CR-6850 (EPRI
1011989), with some additional information provided by the small- and intermediate-scale
experiments described in NUREG/CR-7010. The FLASH-CAT model makes use of the
following information:
The cable trays are horizontal and stacked vertically.

There are no barriers separating the trays, and the tray tops and bottoms are open.
The cables are not protected with coatings, shielding, or thermal blankets.
There is a fire beneath the lowest tray.
Each tray has at least a single row of cables.
In this scenario, the fire propagates upward through the array of cable trays according to an
empirically determined timing sequence. First, ignition of the cables in the lowest tray occurs
when the internal temperature of a target cable within that tray reaches the failure temperature
of 205 C (400 F). This analysis is based on guidance provided in NUREG/CR-6850 (EPRI
1011989), Appendix R. The calculation of the cables internal temperature is based on the
Thermally-Induced Electrical Failure (THIEF) methodology (NUREG/CR-6931, Volume 3).
Following ignition, the cables in the lowest tray burn at a rate of 250 kW/m2, a value appropriate
for TP cables (NUREG/CR-7010, Volume 1). The width of the burning cable is the same as the
B-2

width of the trays (0.8 m). The lateral extent of burning cable in the lowest tray, before the onset
of lateral spread, is equal to that of the ignition source. The lateral extent of the burning cable in
upper trays, before the onset of lateral spread, is given by the formula:
2
tan 35
(B-1)
where is the length of tray and is the distance (bottom to bottom) between tray
1 and
tray . The 35 upward spread angle is described in NUREG/CR-6850 (EPRI 1011989),
is equal to the length of the vent in the cabinet (0.6 m (2 ft))
Appendix R. In this example,
and is the distance between trays (bottom to bottom).
Following ignition, the fire in the first tray spreads laterally at a rate of 3.2 m/h (NUREG/CR6850 (EPRI 1011989), Appendix R). The fire in the second tray ignites 4 min after the first, and
the lateral extent of the initial fire in the second tray is widened based on Equation (B-1) to
1.3 m. The burning and spread rates of the fire in the second tray are the same as the first.
The fire in the third tray ignites 3 min after the fire in the second, and the initial lateral extent of
the fire is widened yet again following the 35 spread angle to 2.0 m.
Local burnout of the fire occurs when the cable plastic is consumed. The time to burnout is
calculated as follows. First, determine the combustible mass per unit area of tray,
:
1
50
0.32
1 0
0.8 m
0.38 kg/m
7.6 kg/m
(B-2)
is the mass fraction of combustible (i.e., nonwhere is the number of cables per tray,
is the total mass per unit
metallic or plastic) material in the cable, is the residue yield,
length of a single cable, and
is the tray width. Next, calculate the burnout time, :
5 avg /6
7.6 kg/m
20,900 kJ/kg
5/6 250 kW/m
762 s
(B-3)
where is the heat of combustion18, and avg is the average HRR per unit area of tray. The
FLASH-CAT model asserts that the HRR per unit area ramps linearly to its average value over a
time period of /6, remains steady for a time period of 2 /3, and then decreases linearly to
zero over a time period of /6. The linear ramp-up and ramp-down are typical ways of
approximating the time history of an items HRR. Further details of the FLASH-CAT model are
provided in NUREG/CR-7010, Volume 1.
The heat of combustion and product yields for PE/PVC cables are taken from Table 3-4.16 of
the SFPE Handbook, 4th edition. Note that five different types of PE/PVC cables are listed in
the chapter. The values listed in Table B-1 are for a cable with relatively high soot and CO
yields, typical of an under-ventilated fire burning within a closed cabinet. Note also that the nonmetallic components of the cables are a mixture of PE (C2H4) and PVC (C2H3Cl). Because the
mixture consists of approximately the same mass of each, the cable materials with an effective
chemical formula of C2H3.5Cl0.5 have been selected. Table B-1 summarizes the fuel and
combustion parameters for this scenario.
18
By default, the FLASH-CAT model uses a heat of combustion of 16,000 kJ/kg unless there is experimental data that is more
appropriate. In this case, the chosen heat of combustion is based on a measurement of PE/PVC cable.
B-3

Table B-1. Products of combustion for switchgear room cabinet and cable fire.
Parameter
Peak HRR
Heat of Combustion
CO2 Yield
Soot Yield
CO Yield
Radiative Fraction
Value
C2H3.5Cl0.5
464 kW
20,900 kJ/kg
1.29 kg/kg
0.136 kg/kg
0.147 kg/kg
0.49
Source
Combination of polyethylene and PVC
SFPE Handbook, 4th Ed., Table 3-4.16
B-4
Figure B-1. Geometry of the switchgear room.
B-5
B.3 Selection and Evaluation of Fire Models

B.3.1 Temperature Criterion

The algebraic models are used in this scenario to estimate the flame height and plume
temperatures from the cabinet fire, first to determine whether the overhead cable trays would be
damaged by the cabinet fire and then to calculate the average hot gas layer (HGL) temperature
using the Foote, Pagni, and Alvares (FPA) correlation to determine whether cables in the
adjacent cabinets would be damaged by the cabinet fire only. Note that the Fire Dynamics
Tools (FDTs) do not allow the HRR to be input as a function of time. With a constant HRR, the
FDTs are use a conservative approximation of an instantaneous, fully developed fire that
remains at peak HRR for the duration of the fire scenario. Although this approach would be
more likely to estimate a conservative shorter time to failure than tools that utilize a timedependent HRR, it is useful as a screening tool to determine whether failure is likely to occur at
any time. The FPA correlation is used to estimate the average HGL temperature that would
result from the initial cabinet fire alone; this analysis using the FPA correlation does not consider
the potential ignition of cable trays, or their potential contribution to the fires HRR.
The Consolidated Fire Growth and Smoke Transport Model (CFAST) and the Fire Dynamics
Simulator (FDS) estimate damage to and ignition time of the overhead cable trays, and also use
the development of elevated temperature and heat flux from the cabinet fire and any ignited
overhead cables to estimate damage to the adjacent cabinets. CFAST provides timedependent conditions for the cable trays and cabinet, and uses a point source for radiation from
the fire sources. FDS distributes the fire in a more realistic way, which should provide more
accurate estimates of temperature and heat flux to adjacent cabinets.
B.3.2 Heat Flux Criterion

CFAST and FDS are used to estimate heat flux to the overhead cable trays and to the adjacent
cabinets. The heat flux calculation in CFAST includes the contribution of radiation from the fire,
upper and lower gas layers, and bounding surfaces, as well as convection from nearby gases.
Radiation from the fire sources is based on a point source for the radiation calculation.
The relative position of the cabinet fire and cable trays may provide a challenge because the
algorithms used by the zone models to assess target damage are based on a fire radiation point
source. FDS models the fire in much the same way as the zone models, with the fire on top of
the cabinet. However, because it is a CFD model, FDS can estimate local conditions at the
specific location of the target cables and adjacent cabinet. Thus, instead of locating the fire at a
single point, FDS more naturally distributes the fire amongst the trays, and the resulting heat
flux calculation is more realistic.
B.3.3 Validation
NUREG-1824 (EPRI 1011999) contains experimental validation results for CFAST and FDS that
are appropriate for this scenario. The National Institute of Standards and Technology (NIST)
has expanded the U.S. Nuclear Regulatory Commission/Electric Power Research Institute
B-6

(NRC/EPRI) V&V to include the latest versions of CFAST (6.1.1) (Peacock, 2008) and FDS
(5.5.3) (McGrattan, 2010). In particular, the International Collaborative Fire Model Project
(ICFMP) Benchmark Exercise #3 test series was designed specifically as a mock-up of a real
SWGR. These experiments include ventilation effects on, heat fluxes to, and temperatures of
various targets, particularly cables. Fire sizes in these experiments bound those used in this
scenario. Also, the cable failure and cable fire spread algorithms, THIEF and FLASH-CAT, are
developed and validated in NUREG/CR-6931 (Vol. 3) and NUREG/CR-7010 (Vol. 1).
Table B-2 below lists the important non-dimensionalized parameters that characterize this fire
scenario. With the exception of the Fire Froude Number, all of the scenario parameters fall
within the range of the NRC/EPRI fire model validation study (NUREG-1824 (EPRI 1011999)).
The Fire Froude Number is essentially a measure of the fires heat output relative to its base
area. In this example, the fire is located at the cabinets top vent, with the vent opening serving
as the area of the base of the fire. This method leads to a higher value of
than would be
is the
calculated if the fire burns completely outside of the cabinet. Thus, the high value of
result of this method that will lead to more severe fire conditions than would be expected if the
fire were to burn partially within the cabinet. Therefore, the model predictions should be valid
for this scenario. Also, the Fire Froude Number is calculated to be within the validation range
when the heat release rate for the initial cabinet fire and the burning cables are considered
together with an average effective diameter of 1 m (3.2 ft).
B-7

Table B-2. Normalized parameters and their ranges of applicability to NUREG-1824 (EPRI
1011999).
Quantity
NUREG1824
Validation
Range
In
Range?
0.4 2.4
No
0.2 1.0
Yes
1.2 1.7
N/A
0.04 0.6
Yes
0.6 5.7
Yes
2.2 5.7
Yes
464 kW
1.2 kg/m
Fire Froude
Number
1.0 kJ/kg/K 293 K 0.48
2.6
9.8 m/s
1600 kW
Flame Length, Lf ,
relative to the
Ceiling Height, Hc
Ceiling Jet Radial
Distance,rcj ,
relative to the
Ceiling Height, Hc
Equivalence Ratio,
, as an indicator of
the Ventilation Rate
Compartment
Aspect Ratio
Target Distance, r,
relative to the Fire
Diameter, D
3.7
1.4
1.0 kJ/kg/K 293 K 1 . m .

9.8 m/s
2.4 m 2.1 m
0.7
6.1 m
1.2 kg/m
1.02
0.48 m 3.7
2.6
1.02
2.1 m
0.23
kW
kJ/kg
. kg/s
0.23
26.5 m
6.1 m
0.31 (based on peak fire size)
1.2 kg/m
1.4 m /s
18.5 m
6.1 m
4.3
1.5 m
0.48 m
3.1
0.4 kg/s
3.0
Notes:
(1)
(2)
The effective diameter of the fire is determined from the formula,

4 / ,
where is the area of the vent on the cabinet. Fire area varies for the cable fire;
an area of 1 m2 is typical at the peak heat release rate.
, is the sum of the height of the fire from the floor, plus
The Fire Height,
the fires flame length.
B-8
B.4 Estimation of Fire-Generated Conditions

This section provides details specific to each model.
B.4.1 Algebraic Models

General: The general approach to using the algebraic models for this scenario is to first
calculate the flame height of the cabinet fire to determine whether the flame reaches one or
more of the overhead cable trays, then to calculate the plume temperatures from the cabinet fire
to determine which of the overhead cable trays would be damaged by the cabinet fire, and
finally to calculate the average hot gas layer temperature using the FPA correlation to determine
whether the cables in the adjacent cable trays would be damaged by the cabinet fire only. The
general scenario is depicted schematically in Figure B-2.
Cable
Tray C
Cable
Tray B
Cable
Tray A
Bank A
Bank B
Bank C
Figure B-2. Schematic diagram of cabinet fire in switchgear room.
The first step in using algebraic models is to determine whether the cables in the cable trays
located directly above the cabinet fire are likely to be damaged and potentially ignited by the
cabinet fire. As shown in Figure B-1, the top of the cabinet is located at an elevation of 2.4 m
(7.9 ft), and the lowest overhead cable tray is located at an elevation of 3.9 m (12.8 ft), which is
1.5 m (4.9 ft) above the top of the cabinet. As shown in Table B-2, the flame length of the
cabinet fire is calculated to be 2.1 m (6.9 ft) at the cabinet peak heat release rate of 464 kW, so
this empirical correlation for flame length can be used to confirm that the overhead cables would
likely be damaged in this scenario. This calculation supports the scenario that the overhead
cable trays would be ignited by the cabinet fire, as discussed in Section B.2.
The next step is to calculate the fire plume temperatures that develop from the cabinet fire to
determine which of the three cable trays located above the cabinet would be damaged by the
cabinet fire. The Heskestad plume temperature correlation included in the FDTs and the FireInduced Vulnerability Evaluation (FIVE-Rev1) was used to calculate the plume centerline
temperature above the cabinet fire. The results of this calculation are shown in Figure B-3.
These results show that the plume temperature at all three cable trays would exceed the cable
damage threshold temperature of 205 C (400 F). However, the Heskestad plume temperature
B-9

correlation is based on an unobstructed plume. The obstruction caused by the position of the
cable trays within the fire plume would alter the actual fire plume entrainment and temperatures.
Nonetheless, these results demonstrate that the potential for damage and ignition of the cable
trays as a result of the cabinet fire warrants more detailed analysis.
Plumetemperatures Heskestadcorrelation
1000
900
Temperature(C)
800
700
600
Bottom
Middle
Top
500
400
300
200
100
0
0
600
1200
1800
Time(s)
2400
3000
3600
Figure B-3. Plume temperatures at cable trays located above a cabinet fire.
The FPA forced ventilation correlation is used to estimate the average HGL temperature of the
SWGR resulting only from the cabinet fire only, based on the parameters described in the
following subsections and summarized in Table B-3.
Geometry: The FPA correlation requires room dimensions to be specified in terms of length,
width, and height. For this example, the selected compartment is a rectangular parallelepiped,
so its length, width, and height are specified directly from dimensions shown in Figure B-1.
Fire: As applied to this scenario, the FPA correlation is used with the time-dependent HRR
specified for the cabinet fire only. This HRR history is shown in Figure B-4.
Materials: The walls, ceiling, and floor are all specified as concrete, with the thermal properties
specified in Table 3-1.
Ventilation: The ventilation rate of the smoke purge mode is 1.42 m3/s (3,000 cfm). This value
is used as a direct input parameter in the FPA correlation.
B-10
Table B-3. Summary of input parameters for FPA analysis of switchgear room scenario.
Parameter
Room height (H)
Room length (L)
Room effective width (We)
Mech. Ventilation rate ( )
Fire elevation (Hf)
Value
5.2 m
26.5 m
18.5 m
Concrete
1.42 m3/s
2.4 m
Source

Fire parameters
20C
See Table B-1
Figure B-1
Figure B-1
Calculation
Figure B-1. See Table 3-1 for properties.
From scenario description
From scenario description of cabinet height
and vent location.
Specified
Temperature: The FPA HGL temperature correlation for mechanically ventilated spaces is
expressed in non-dimensional terms as:
(B-4)
0.63
is the HGL temperature rise above ambient,

is the ambient air absolute temperature,
is the specific heat of
is the HRR of the fire compartment, is the mass ventilation flow rate,
is the heat transfer coefficient, and
is the total area of the compartment enclosing
air,
surfaces. The results for the cabinet fire are shown in Figure B-4, based on the input
parameters specified in Table B-3 and the HRR history shown in Figure B-4. These results
show that, for the specified parameters, the average HGL temperature reaches a maximum of
approximately 65 C (149 F) at 20 minutes, based on the peak cabinet fire HRR of 464 kW.
These results show that cables in the two adjacent cabinets would not be damaged by the initial
cabinet fire alone. However, further analysis is required to determine the potential impact of
overhead cable ignition on the potential for damage to cables in the adjacent cabinets. CFAST
and FDS are used to perform this more detailed analysis.
B-11

HGLtemperaturecalculation FPAcorrelation
70
500
450
60
350
300
40
250
30
200
HRR(kW)
Temperature(C)
400
50
FPA[C]
HRR[kW]
150
20
100
10
50
0
0
600
1200
1800 2400
Time(s)
3000
0
3600
Figure B-4. Average HGL temperature (red line) from FPA correlation and HRR (blue line)
for SWGR cabinet fire scenario.
B.4.2 Zone Model

Geometry: The CFAST analysis defines the compartment as a single rectangular
parallelepiped with the specified dimensions. While there are a number of cable trays in the
compartment, the compartment is large enough that the cable trays do not occupy a significant
fraction of the total volume, so the compartment dimensions are taken directly from the scenario
description and Figure B-1. Figure B-5 illustrates the scenario as modeled by CFAST. Figure
B-6 shows the CFAST inputs for the geometry.
Figure B-5. Average CFAST/Smokeview rendering of SWGR.
B-12
Figure B-6. CFAST inputs for compartment geometry in SWGR scenario.
Fire: CFAST requires a user-specified, time-dependent HRR and stoichiometry for the
combustion of fuel and oxygen. The HRR is the combined cabinet/cable fire described above.
Figure B-7 shows the CFAST inputs for the fire taken directly from Table B-1.
B-13
Note: Values for Lower Oxygen Limit and Gaseous Ignition Temperature are set at default values.
Figure B-7. CFAST fire specification inputs for the SWGR scenario.
Materials: The walls, floor, and ceiling are specified as concrete in CFAST with properties as
previously described. The target properties are provided directly as input.
Ventilation: Mechanical ventilation and leakage are specified as input to CFAST directly from
the scenario description. CFAST uses three inlet and three outlet vents for the mechanical
ventilation at the heights specified in the scenario description. Horizontal placement of the
mechanical ventilation within the compartment does not affect the zone model calculation and is
not part of the input. Figure B-8 shows the CFAST inputs for the mechanical ventilation. Since
pressure for this scenario peaks approximately at the lower fan cutoff pressure threshold,
default values for these pressures are used as they should not impact the calculations.
B-14
Note: Values for Begin Dropoff At and Zero Flow At are set at default values.
Figure B-8. CFAST mechanical ventilation inputs for the SWGR scenario.
Cable Targets: In CFAST, target temperatures are calculated with a one-dimensional

cylindrical heat transfer calculation based on the material properties and cable diameter, as
specified in the scenario description. CFAST uses the THIEF methodology developed as part of
the Cable Response to Live Fire (CAROLFIRE) program (NUREG/CR-6931, Vol. 3). Each of
the target cables is specified directly in the model. Thermal properties are taken directly from
NUREG/CR-6931. Cable density is calculated from the specified mass per unit length and the
cross-sectional area of the cable as 0.38 kg/m /( x 0.0075 m)2 = 2150 kg/m3. Figure B-9
shows the CFAST inputs for targets in the scenario. Ignition time for the target cables was
calculated with an initial CFAST run with only the cabinet fire. This run estimated this ignition
time (as the time flames reached the secondary cable target) as 480 s.
B-15
Figure B-9. CFAST inputs for cabinet and cable targets for the SWGR scenario.
B.4.3 CFD Model

Geometry: The compartment has a simple rectangular geometry that coincides with the
external boundary of the computational domain. In other words, the exterior walls are not
explicitly declared, but are defined by default to be the external boundaries of the domain with
the surface properties of concrete, given above. The cabinets are modeled simply as boxes
constructed of steel, whose properties are specified above. No attempt is made to model the
interior of the cabinets because the fire has been specified as originating at or near the top of
one of the cabinets. Figure B-10 shows the compartment geometry used in FDS.
The numerical mesh away from the fire consists of uniform grid cells, approximately 0.2 m (8 in)
on a side. In the vicinity of the fire, the mesh cells are approximately 0.1 m (4 in) on a side.
This refinement of the mesh is needed to better locate the actual tray locations. Further
refinement of the mesh is unnecessary because the fires growth and spread are to be specified
based on the empirical model FLASH-CAT.
Materials: The material properties are applied directly as specified to the walls, floor, ceiling,
and cabinet. The cabinets adjacent to the center bank are modeled as hollow steel boxes
whose interiors remain at ambient temperature. For the cables, the thermal properties are
B-16

superseded by the algebraic ignition and flame spread model called FLASH-CAT. In FDS, the
model is applied as follows. First, each cable tray is declared an obstruction, the top and
bottom of which are given the thermal properties of PVC cables from NUREG/CR-6850 (EPRI
1011989), Volume 2, Appendix R.4.1.1 (k=0.192 W/m/K, =1,380 kg/m3, c=1.289 kJ/kg/K).
These properties are not particularly important because the cable failure, ignition, and fire
spread are not dependent on them, but merely provide approximate thermal boundary
conditions for those cables that are not burning. Next, the top surface of each tray is divided
into three sections. The middle section ignites and burns according to the timing sequence of
the FLASH-CAT model. The two adjacent sections ignite at the same time, but only at the edge
abutting the middle section. The fire then spreads laterally at a rate specified by the FLASHCAT model, in this case, 0.9 mm/s.
Fire: The initial fire source is modeled as a 0.6 m (2 ft) x 0.3 m (1 ft) gas burner atop the
central cabinet with the specified HRR. This is meant to represent a fire that burns near the top
of the cabinet and exhausts through the vent. The ignition and growth of the cable fire is based
on the empirical FLASH-CAT model described above. Figure B-11 shows a snapshot of the
burning cable during the simulation.
Ventilation: The door is included in the calculation merely as a surface with different properties
than the default concrete wall. The supply and return vents are specified according to the
drawing and given volume flow rates. Note that because of the relative coarseness of the
underlying numerical grid, the ventilation rate is input directly in terms of the volume flow
rate (m3/s) rather than as a separate vent area (m2) and velocity (m/s). The model automatically
adjusts the dimensions of all objects to conform to the numerical mesh, and it also adjusts the
velocity of the air stream to properly reflect the desired volume flow rate.
B-17
Figure B-10. FDS/Smokeview rendering of the SWGR.
Figure B-11. FDS/Smokeview rendering of the SWGR fire showing localized ignition of
extinction of secondary cable fires resulting from initial cabinet fire.
B-18
B.5 Evaluation of Results

In this example, the objective of the calculations is to estimate the effects of fire in a cabinet in a
SWGR on nearby cable and cabinet targets; that is, to determine whether and when
temperatures and/or heat flux to the cable and adjacent cabinets exceed established critical
values. Cables are considered damaged when the temperature of the cable reaches 205 C
(400 F) or the exposure heat flux reaches 6 kW/m2 (NUREG/CR 6850 (EPRI 1011989),
Appendix H, Table H-1). These criteria are intended to be indicative of electrical failure, but,
based on experimental observation, are routinely also used as ignition criteria.
Table B-4 summarizes the results of all the models for the chosen failure criteria. For each
Predicted Value, a calculation is performed to determine the Probability of Exceeding the Critical
Value. The procedure for calculating this probability is given in Chapter 4, and it accounts for
the model bias and scatter. The purpose of this table is to highlight the criterion that is most
likely to be exceeded, so that further analysis can be focused on this criterion and the model or
models that predict it. Each criterion is discussed in greater detail in the following subsections.
Table B-4. Summary of the model predictions of the cabinet fire scenario.
Model
Bias
Factor,
CFAST
FDS
CFAST
FDS
1.00
1.02
1.00
1.02
CFAST
FDS
0.81
0.85
Standard
Deviation,
Location
Predicted
Value

0.27
335
Cable Tray A
0.13
755
0.27
168
Cabinet A
0.13
136
2
Heat Flux (kW/m )
0.47
5.3
Cabinet A
0.22
4.2
Critical
Value
Probability
of
Exceeding
205
205
205
205
0.937
1.000
0.177
0.000
6
6
0.576
0.159
B.5.1 Cable Ignition and Damage

The algebraic models cannot be used in this case to accurately assess the damage to cables.
FIVE-Rev1 does not have an algorithm that considers the thermal inertia of the cables. FDTs
does, but the model is only applicable when the exposing temperature is constant, which is not
the case for this example. However, the algebraic models can be used for screening purposes,
to show that the plume temperatures resulting from the specified cabinet fire are high enough to
potentially cause damage to and ignition of cables located in cable trays directly above the
cabinet.
CFAST and FDS estimate conditions resulting from the ignition (based on initial model runs of
the initial cabinet fire) and burning of the cables (based on calculations from the FLASH-CAT
model). NUREG/CR-6850 (EPRI 1011989) contains some guidance on modeling cable ignition,
flame spread, and the fires resulting heat release based on a limited set of fire test data. The
differences in HRR between the models (Figure B-12) result from variations in the
implementation of this guidance. Figure B-12(a) shows the HRR from the initial cabinet fire
B-19

source only. Figure B-12(b) shows the overall HRR, including the initial cabinet fire source and
the cables ignited by this initial fire.
Heat Release Rate
1800
500
450
400
350
300
250
200
150
100
50
0
1600
CFAST
CFAST
1400
HRR (kW)
HRR (kW)
Heat Release Rate
FDS
1200
FDS
1000
800
600
400
200
0
600
1200
1800
2400
3000
3600
600
Time (s)
1200
1800
2400
3000
3600
Time (s)
(a) Initial cabinet fire only
(b) Initial cabinet fire and ignited cables
Figure B-12. Heat release rate inputs to CFAST and FDS for a SWGR cabinet fire scenario.
There are several ways to assess the initial fire sources potential to ignite the lowest cable tray
(Cable Tray A). Cables are considered damaged when the temperature reaches 205 C
(400 F) or the exposure heat flux reaches 6 kW/m2 (NUREG-1805, Appendix A). These criteria
are intended to indicate electrical failure, but are routinely also applied as ignition criteria. In
newer studies in NUREG/CR-7010, cable ignition was not observed at fluxes below 25 kW/m2,
and most often only with direct flame impingement. Handbook values for minimum ignition flux
for power and communication cables are reported in the range of 15 kW/m2 to 35 kW/m2 (SFPE
Handbook, Table 3-4.2). For this scenario, CFAST predicts that the flame height reaches the
cable tray in approximately 490 s, quite similar to the temperature-based prediction. Table B-5
shows the lowest cable trays estimated time to ignition for a variety of ignition criteria. For this
simulation, 490 s was chosen.
Table B-5. Estimated time to ignition of lowest cable tray by CFAST for the SWGR cabinet fire.
Ignition Criterion
Gas temperature 205 C
Cable temperature 205 C
Heat flux 6 kW/m2
Heat flux 15 kW/m2
Flame impingement
Time
270 s
860 s
490 s
740 s
490 s
The CFAST and FDS temperature predictions for the Cable Tray A cables are shown in Figure
B-13. FDS predicts cable failure in Tray A at about 495 s, CFAST in about 600 s. Peak
temperatures from both models are well above the failure criteria for the cables, so it can be
expected that the cables will ignite and provide an additional source of fire.
B-20
Cable Tray A Temperature

800
Temperature (C)
700
CFAST
600
500
FDS
400
300
200
100
0
0
600
1200
1800
2400
3000
3600
Time (s)
Figure B-13. Estimated temperatures for Cable Tray A directly above the fire source for a
SWGR cabinet fire scenario.
Qualitatively, the results of the CFAST and FDS predictions are quite different. The radiation
from the fire source in CFAST is calculated based on a point source fire positioned at the base
of the fire. Thus, once the fire grows and the flame height approaches the target cable tray,
CFAST can be expected to underestimate the local cable temperature and heat flux, since the
cable would actually be immersed within the flames. CFAST does include an estimate of the
flame height, which can also be used as an indicator of damage to the cable. For this scenario,
CFAST predicts that the flame height will reach the cable tray in approximately 490 s, quite
similar to the temperature-based prediction. Past this point, CFAST estimates of the local target
temperature are expected to be under-predictions. FDS predictions include the impact of direct
flame impingement and immersion of the target in flames. Thus, the higher temperatures
predicted by FDS are expected.
Upon ignition of the bottom cable tray (Cable Tray A), the higher cable trays are ignited,
consistent with the FLASH-CAT model.
B.5.2 Cabinet Damage

To assess potential damage to adjacent cabinets, both the predicted temperatures and heat
fluxes are evaluated. Because the two adjacent cabinets are equidistant from the fire and have
similar properties, only one is considered here. The critical damage thresholds are the same for
these cabinets as for the cables in trays. Figure B-14 shows estimated temperature and heat
flux on the cabinet surface.
The algebraic models are not capable of estimating the temperature of a target such as an
electrical cabinet, whereas the other models do provide this capability. CFAST and FDS
estimate peak temperatures below 165 C (330 F), which is well below the threshold of 205 C
B-21

(400 F). The somewhat higher cabinet temperature and heat flux predicted by CFAST is
consistent with the point source radiation calculations for the CFAST simulation. CFAST and
FDS both estimate an incident heat flux below about 5 kW/m2, with the difference caused by the
more simple radiation calculation in CFAST.
Cabinet A Heat Flux
Cabinet A Temperature
6
180
CFAST
140
Heat Flux (kW/m2)
Temperature (C)
160
120
FDS
100
80
60
40
20
0
0
600
1200
1800
2400
3000
CFAST
4
FDS
3
2
1
0
3600
Time (s)
600
1200
1800
2400
3000
3600
Time (s)
Figure B-14. Estimated temperature and heat flux to a cabinet adjacent to the fire source
in a SWGR cabinet fire scenario.
B.5.3 Parameter Uncertainty Propagation

The analysis above has shown that a 98th percentile cabinet fire is very likely to ignite cables in
the trays above the cabinet and could potentially damage adjacent cabinets. Within the context
of a PRA, the next step is to calculate the probability of cable damage for any fire within the
cabinet, not just the 98th percentile fire.
Figure B-15 displays the distribution19 of peak heat release rates for cabinets with more than
one bundle of unqualified cable (NUREG/CR-6850, Appendix G). The analysis described above
made use of the 98th percentile fire from this distribution, with a peak of 464 kW.
19
provides a built-in function (GAMMA.DIST) that calculates the probability density function given the parameters and . In this
case, these parameters are 2.6 and 67.8, respectively.
B-22
HRR Distribution
0.005
0.004
0.003
0.002
0.001
0.000
0
100
200
300
400
500
600

Figure B-15. Distribution of HRR for an electrical cabinet fire.
Applying Heskestads flame height correlation to the entire range of HRR, now taken as a
random variable, leads to a distribution of flame height shown in Figure B-16.
Flame Height Distribution
1.0
0.8
0.6
0.4
0.2
0.0
0
0.5
1.5
2.5
Flame Height (m)
Figure B-16. Distribution of flame heights for the entire range of cabinet fires.
The cable tray is 1.5 m (4.9 ft) above the top of the cabinet. The probability that the flames from
a randomly chosen fire will reach the cables is equal to the area beneath the curve in Figure
B-16 for flame heights greater than 1.5 m (4.9 ft), or approximately 0.31. Consistent with the
guidance in NUREG/CR-6850, this resulting probability can be used as the severity factor for
the quantification of corresponding fire ignition frequencies. It is interesting to note that an
analysis that uses the 50th percentile fire would lead to the conclusion that the flame from the
B-23

cabinet fire does not impinge on the cables and that ignition of the cables is not likely to occur.
This changes the outcome of the analysis significantly because an analysis that uses only the
mean HRR without explicit quantification of the uncertainty would result in an overly optimistic
modeling conclusion and an inappropriate evaluation of the scenario within the integrated fire
analysis.
B.6 Conclusion
This analysis has considered the potential that a fire in an electrical cabinet in a 4160 V SWGR
will damage overhead cables and adjacent electrical cabinets. Algebraic equations from the
FDTs and FIVE-Rev1, including the Heskestad flame height correlation and the Heskestad
plume temperature correlation, were used for screening purposes, to evaluate the potential for
damage as well as to determine whether more detailed analysis with CFAST and FDS was
warranted. The algebraic equations demonstrate that the calculated flame height from the
cabinet fire would be high enough to potentially ignite the lowest of the three horizontal cable
trays located directly above the cabinet fire. They also demonstrate that the calculated fire
plume temperatures are high enough at all three horizontal cable trays located directly above
the cabinet fire to potentially damage cables in all three trays. As applied in this scenario, the
algebraic equations demonstrate that a more detailed analysis with CFAST and FDS is
warranted.
The more detailed analyses with CFAST and FDS demonstrate that the cabinet fire is likely to
ignite the electrical cables in the lowest cable tray directly above the cabinet fire in
approximately 10 min. The additional cable trays directly above the lowest tray would then
ignite in turn. In addition, the models indicate that it is possible that the combined cabinet and
cable fire could damage adjacent cabinets. These conclusions are based on an analysis of a
severe (i.e., 98th percentile) fire. A subsequent analysis involving the propagation of the entire
HRR distribution through a simple flame height model indicates that 31 % of all possible fires
within the cabinet would ignite the overhead trays and the resulting fire could damage adjacent
cabinets.
B.7 References
1. Jones, W., R.D. Peacock, G.P. Forney, and P.A. Reneke, CFAST - Consolidated Model of
Fire Growth and Smoke Transport (Version 6), Technical Reference Guide, SP 1026,
National Institute of Standards and Technology, Gaithersburg, MD, 2009.
2. NIST SP 1018-5, Fire Dynamics Simulator (Version 5), Technical Reference Guide,
Volume 3, Experimental Validation, 2010.
4. NUREG-1824 (EPRI 1011999), Verification and Validation of Selected Fire Models for
Nuclear Power Plant Applications, 2007.
B.8 Attachments (on CD)

1.
FDS input file: Switchgear_Room_Cabinet.fds
B-24

2.
CFAST input files:

a. Initial Fire Only.in
b. Cabinet Fire in Switchgear.in
3.
Algebraic calculation input files:

a. FPA_AppB.xlsx
B-25
B-26
APPENDIX C
LUBRICATING OIL FIRE IN PUMP COMPARTMENT
C.1 Modeling Objective

The purpose of the calculations described in this appendix is to determine whether important
safe-shutdown cables within a pump room will fail in the event of a lubricating oil fire, and, if so,
at what time failure occurs. These cables are protected by an electrical raceway fire barrier
system (ERFBS), but there is a concern that the existing barrier system will not provide the
required protection. The impact of opening a door during the fire is also investigated.
C.2 Description of the Fire Scenario

General Description: The compartment is of fire-resistive construction and contains an
emergency core cooling system pump and a single tray containing safe-shutdown cables that
are protected by an ERFBS. The pump is surrounded by a dike designed to contain any
lubricating oil that may leak or spill, with a maximum capacity of 190 L (50 gal). The
compartment contains one smoke detector and one sprinkler. The compartment is
mechanically ventilated. The fire occurs when pump oil leaks into the dike area and ignites.
Large oil fires are likely to cause flashover conditions. Flashover refers to the rapid transition
from the growth period of a fire (pre-flashover) to the fully developed fire (post-flashover). A
flashover condition is typically expected when the hot gas layer (HGL) temperature reaches
500 C (932 F) or greater. Post-flashover conditions are expected in this scenario.
Geometry: The pump room is relatively small and has only one door. As shown in Figure C-1,
the walls are 0.1 m (0.3 ft) thick. The floor and ceiling are 0.9 m (3 ft) thick.
C-1
Figure C-1. Geometry of the pump room.
C-2

Materials: The walls, ceiling, and floor are all constructed of concrete. Nominal values for the
thermal properties of various materials in the compartment are listed in Table 3-1. The single
cable tray in this compartment is filled with PE/PVC cables with copper conductors. The
properties of the cables are listed in Table C-1. The damage criterion is taken to be the point at
which the cable temperature reaches 205 C (400 F) (NUREG-1805, Appendix A). The cable
tray is protected by an ERFBS, which is two layers of ceramic fiber insulation blankets, covered
by 0.0254 mm (1 mil) foil. The properties of the insulation material and cables are listed in
Table C-1. Note that the properties of the ceramic fiber material are typically dependent on
temperature, and that this material undergoes a series of chemical reactions that are designed
to absorb heat and protect the underlying cables. Detailed thermophysical properties for this
particular material are not known. However, the ERFBS has undergone a fire endurance
furnace test in which the average temperature of the electrical raceway was maintained below
121 C (250 F) and the maximum temperature below 163 C (325 F) for an hour when
exposed to the standard ASTM E 119 temperature curve.
The damage criterion is taken to be when the cable surface temperature reaches 205 C
(400 C) (NUREG-1805, Appendix A). Using the temperature on the cable surface rather than
inside the cable jacket is conservative, but is chosen in this case to allow a comparison of
results between MAGIC and FDS at equivalent positions in the two models.
Table C-1. Data for ERFBS and cable insulation.
Material
Ceramic Fiber
Insulation
Cable
Parameter
Thickness (2 layers)
Thermal conductivity
Density
Specific heat
Emissivity
Diameter
Jacket thickness
Insulation/jacket conductivity
Insulation/jacket density
Insulation/jacket specific heat
Mass per unit length
Conductor mass fractions
Value*
5 cm
0.06 W/m/K
128 kg/m3
1.07 kJ/kg/K
0.9
15 mm
2 mm
0.192 W/m/K
1380 kg/m3
1.289 kJ/kg/K
0.4 kg/m
33% PE/PVC, 67% copper
*Source: Product literature (ERFBS) and NUREG/CR-6850 (EPRI 1011989), Volume 2, Appendix R
(PVC cable insulation).
Fire Protection Systems: As shown in Figure C-1, a smoke detector and a sprinkler are
located in the pump room. However, to determine whether the barrier system alone will provide
the required protection, the fire detection and suppression systems are not credited in the fire
scenario under consideration.
Ventilation: There is one supply and one return air vent, each with an area of 0.25 m2 (2.7 ft2),
providing a volume flow rate of 0.25 m3/s (530 cfm). The locations are shown in Figure C-1.
The ventilation system continues to operate during the fire, with no changes brought about by
fire-related pressure effects. This does not imply that the fire does not impact the ventilation
system, but rather that there is typically limited information about the ventilation network that
connects to a given compartment. The pump compartment has one door; it is 1.1 m (3.6 ft)
wide and 2.1 m (6.9 ft) tall. The door is normally closed, but it is opened 10 min after ignition by
C-3

the fire brigade. Before the door opens, leakage from the doorway occurs via a 1.3 cm (0.5 in)
gap under the door.
Fire: The fire starts following an accidental release of 190 L (50 gal) of lubricating oil. The spill
is contained by the dike. Lubricating oil is a mixture of hydrocarbons, mostly alkanes, which
have the chemical formula CnH2n+2 (with n ranging from 12 to 15). For the purpose of modeling,
the fuel is specified to be C14H30. Fuel properties for the lubricating oil are summarized in Table
C-2. The properties obtained from NUREG-1805 correspond to those for transformer oil, based
on the statement in Table 3-4 in NUREG-1805 that lubricating and transformer oils are similar.
Table C-2. Data for lubricating oil fire.
Parameter
Value
Source
CnH2n+2
Specified as C14H30
Mass burning rate
0.039 kg/s.m2 NUREG-1805 Table 3-4
Fuel volume
190 L
Specified
3
Fuel density
760 kg/m
NUREG-1805 Table 3-4
Heat of Combustion
46,000 kJ/kg NUREG-1805 Table 3-4
Heat of Combustion per unit
13,100 kJ/kg Huggett 1980, Average value
mass of oxygen consumed
CO2 Yield
2.64 kg/kg
SFPE Handbook, 4th ed., Table 3-4.16*
Soot Yield
0.059 kg/kg
CO Yield
0.019 kg/kg
Radiative Fraction
0.34
2
8700 m /kg
*Material identified as Hydrocarbon in SFPE Handbook was used to derive the properties.
C-4
C.3 Selection and Evaluation of Models

This section describes the applicability of the models to this scenario.
C.3.1 Fire Sustainability

This fire scenario involves a large fire in a ventilated room that may not have sufficient oxygen
to sustain the fire. An algebraic model is used to determine the oxygen availability within the
room with the door closed.
C.3.2 Temperature Criterion

Algebraic Models: Neither the Fire-Induced Vulnerability Evaluation (FIVE-Rev1) nor the Fire
Dynamics Tools (FDTs) contain correlations to estimate the hot gas layer (HGL) temperatures
within a flashed-over, under-ventilated compartment. Also, the point source radiation heat flux
calculation included within FIVE-Rev1 and the FDTs cannot account for the attenuation of
thermal radiation by the smoke that fills the compartment. Consequently, neither model is used
for this scenario. However, an algebraic calculation is used to examine the availability of
oxygen.
Zone Models: This fire scenario is not a typical application of a zone model because it involves
post-flashover conditions, where the two layers essentially become one. Nevertheless, zone
models can transition to this state when the HGL essentially descends all the way to the floor
and the room becomes a well-stirred reactor. Conservation laws of mass and energy still apply
within the single layer; additionally, the processes governing the heating of a target immersed in
the HGL still apply, even when the HGL fills the entire compartment. The zone model MAGIC
has been selected for this application.
CFD Models: This fire scenario is a challenging application, even for a computational fluid
dynamics (CFD) model. It involves relatively high temperatures, under-ventilated conditions,
and flashover. The primary advantage of a CFD model for this fire scenario is that CFD models
typically include combustion algorithms that estimate near- and post-flashover conditions.
C.3.3 Validation
Table C-3 lists various important non-dimensional parameters and the ranges for which the
validation study NUREG-1824 (EPRI 1011999) is applicable. The only parameters that fall
outside the validation parameter space are the equivalence ratios for the mechanically
ventilated portion of the scenario (first 10 minutes, while the door is closed) and for the natural
ventilation portion of the scenario (after 10 minutes when the door is opened). In both cases,
the high equivalence ratios for the compartment are a result of the relatively large fire and low
airflows.
For MAGIC, a sensitivity case is run, in which double doors, rather than a single door, are
opened after 10 minutes to determine whether an increase in airflow would cause higher
temperatures in the room. With the enlarged opening, the equivalence ratio is 0.5, putting it
within the verification and validation (V&V) range. The results of the sensitivity case, presented
in Section C.5.2, show that the HGL temperatures predicted in the room are not sensitive to the
size of the door opening.
C-5

As part of the work performed at the National Institute of Standards and Technology (NIST) for
the investigation of the World Trade Center disaster, the Fire Dynamics Simulator (FDS) has
been validated against large-scale fire experiments. The experiments involved fairly large fires
in a relatively small compartment, limited ventilation, a liquid fuel spray fire, and the
measurement of the heat flux to and temperatures of insulated steel (similar to the cables
protected by ceramic fiber blankets). The large-scale tests with furniture had equivalence ratios
of approximately 1.4, which provides a validation basis for FDS under conditions similar to the
natural ventilation portion of the scenario (after the door has been opened). The NIST
experiments and the FDS simulations are described in NIST NCSTAR 1-5F.
Table C-3. Normalized parameter calculations for the pump room fire scenario.
Quantity
Fire Froude
Number
4934 kW
1.2 kg/m
Flame Length,
L , relative to the
Ceiling Height,
Hc
1.0 kJ/kg/K 293 K 1.9

4.8 m
4.9 m
/
3.7
1.02
9.8 m/s
0. 93
1.02
Equivalence
Ratio, , based
on natural
ventilation
0.23
0.23 0.5
Target Distance,
r, relative to the
Fire Diameter, D
1.2 kg/m
0.5
1.2 1.7
N/A
0.04 0.6
No
0.04 0.6
No
0.6 5.7
Yes
2.2 5.7
N/A
0.38 kg/s
0.6
N/A
The non-dimensional parameters are explained in Table 2-5.
C-6
Yes
0.99
Notes:
(1)
0.2 1.0
0.07 kg/s
2.31 m 2.1 m
2.8 m
4.9 m
1.9
Yes
5.5
0.25 m /s
4934 kW
13,100 kJ/kg 0.38 kg/s
0.23
9.4 m
4.9 m
Compartment
Aspect Ratios
4934 kW
13,100 kJ/kg 0.07 kg/s
0.23
0.4 2.4
4.8 m
N/A
Equivalence
Ratio, , based
on the
mechanical
ventilation rate
In
Range?
0.9
0.99
1.9 m 3.7
Ceiling Jet
Radial
Distance,rcj ,
relative to the
Ceiling Height,
Hc
NUREG1824
Validation
Range
(2)
(3)
The equivalent fire diameter,

4 / , where is the area of the spilled
lubricating oil.
The compartment aspect ratios are calculated using the equivalent length and
width.
C.4 Estimation of Fire-Generated Conditions

This scenario is modeled using algebraic calculations, the zone model MAGIC, and the CFD
model FDS.
C.4.1 Calculation of Oxygen Availability

At the start of the scenario, the mechanical ventilation is operational, the door is closed, and the
fire output immediately jumps to the peak heat release rate (HRR) with a total spill area of
approximately 2.75 m2 (29.6 ft2), as shown in the hatched area of Figure C-1. The peak HRR,
, is computed from the fuel mass burning rate,
, the heat of combustion, , and the
specified area of the spill, :
0.039 kg/m s
46,000 kJ/kg
2.75 m
4,934 kW
(C-1)
The oxygen needed to sustain the fire is calculated from the following equation:
4934 kW
13,100 kJ/kg
0.377 kg/s
(C-2)
where
is the heat of combustion per unit mass of oxygen consumed. The quantity of
oxygen provided by the ventilation system is calculated by multiplying the oxygen content (0.23)
by the density and the ventilation rate of the air:
0.23
0.23
1.2 kg/m
0.25 m /s
0.069 kg/s
(C-3)
The oxygen provided by the ventilation system is much lower than the amount needed to
sustain the fire. The oxygen initially in the room can provide the additional oxygen needed for
combustion for a short time. The available oxygen in the room, calculated from the room
dimensions (Table C-4), is:
0.23
0.23
1.2 kg/m
2.81
9.39
4.9 m
35.7 kg
(C-4)
The oxygen initially in the room can sustain the fire for an amount of time equal to the oxygen
quantity in the room divided by the consumption rate minus the ventilation supply rate, as shown
below:
35.7 kg
0.377 kg/s 0.069 kg/s
C-7
116 s
(C-5)

Equation C-4 requires that all the oxygen within the room be consumed by the fire. This
establishes an upper limit to the burning duration before the fire becomes ventilation-limited.
After 116 s, the size of the fire is maintained only by the ventilation system and is limited to:
0.069 kg/s
13,100 kJ/kg
904 kW
(C-6)
These results show that the oxygen supply available to the room will only allow a fire of reduced
size to burn until the door is opened (under-ventilated condition).
C.4.2 Zone Model

The following paragraphs outline the data utilized to model the scenario using MAGIC. Figure
C-2 provides an illustration of the scenario, as rendered by MAGIC.
Geometry: To model this scenario with MAGIC, the pump compartment is modeled as a single
compartment having the same total volume and surface area as the actual enclosure. This
allows the volume in which the HGL develops and the surface area through which energy is
transferred from the compartment to be maintained. Maintaining the total volume and surface
area while leaving the ceiling height unchanged at 4.9 m (16 ft) yields an effective compartment
size of 9.39 m (30.8 ft) by 2.81 m (9.2 ft). The modification to the geometry can be seen by
comparing Figure C-1 and Figure C-2. All other aspects of the geometry are relatively
unchanged.
Table C-4. Calculated input for lubricating oil fire.
Parameter
Effective Length
Effective Width
Fire Diameter
Peak Heat Release Rate
Fire Duration
Mass of Fuel (kg)
Stoichiometric Mass-Oxygen
Specific Area
C-8
Value
2.81 m
9.39 m
1.87 m
4,934 kW
1345 s
144.4
3.5 g/g
513 m2/g
Figure C-2. MAGIC view of the pump room.
Fire: For this scenario, the lubricating oil is preheated prior to the spill, such that the HRR
reaches the peak immediately upon fire initiation, as shown in the HRR curve plotted in Figure
C-3. The lower oxygen level is specified to be 10%. Using the specified spill area and volume,
the spill depth is calculated as 0.069 m (0.23 ft).
The fire is modeled as a single circular area of equivalent diameter. The actual entrainment for
the pool fire is proportional to the perimeter of the fire, which is significantly greater than the
perimeter of the circular area. However, the enclosure is small and the smoke filling rates are
expected to be short regardless of the fire shape.
The fire duration, , is determined from the pool depth, , density, , and burning rate,
0.069 m 760 kg/m

0.039 kg/m /s
C-9
1345 s
22.4 min
:
(C-7)
Heat Release Rate

6000
5000
HRR (kW)
Input HRR
4000
3000
2000
1000
0
0
600
1200
1800
2400
3000
3600
Time (s)
Figure C-3. Heat release rate curve for lubricating oil fire.
The location of the fire is placed at the edge of the dike, closest to the target cable. The total
mass of fuel is 144.4 kg, calculated from the volume multiplied by the density from Table C-2.
The stoichiometric mass-oxygen-to-fuel ratio, , is calculated using Equation 22 from Chapter
3-4 of SFPE Handbook, 4th Edition and the values from Table C-1, as follows:
46,000 kJ/kg
13,100 kJ/kg
3.5 kg/kg
(C-8)
where is heat of combustion per unit mass of fuel consumed and O is the heat of
combustion per unit mass of oxygen consumed. One of the inputs required by MAGIC is the
specific area, s, a measure of the smoke generation, which is calculated as described in
NUREG-1824 (EPRI 1011999), Volume 6, Section 3.2.7:
0.059
8,700 m /kg
513 m /kg
(C-9)
where
is the mass extinction coefficient and is the soot yield, as listed in Table C-2. The
pyrolysis rate (g/s) is calculated for input to MAGIC by dividing the HRR values (4934 kW) at
each time step by the heat of combustion (46,000 kJ/kg). Other inputs needed for MAGIC are
listed in Tables 3-1, C-1, C-2, and C-4. Figure C-4 is a screenshot of the source fire in the
MAGIC input file.
C-10
Figure C-4. MAGIC fire input screen for pump fire.
C-11

In MAGIC, multi-conductor cables, composed of jacket, insulation, and conductor (copper, in
this case), are modeled as single-conductor cables, as shown in Figure C-5.
Jacket
Insulation
Conductor
Figure C-5. Modeling multi-conductor cables in MAGIC. Source: NUREG-1824

(EPRI 1011999), Volume 6, Figure 3-3.
The radius of the conductor in an equivalent single-conductor cable is needed for input to
MAGIC. The mass of the conductor per unit length is calculated from the mass fraction of the
conductor multiplied by the mass per unit length of the multi-conductor cable: 0.67 x 0.4 kg/m =
0.27 kg/m (values from Table C-1). To determine the conductor radius, , the mass per unit
length (0.27 kg/m) is set equal to the cross-sectional area times the density of copper, 8954
. Rearranging the equation to solve for conductor radius
kg/m3 (Table 3-1) or 0.27
results in the following:
0.27
8954
3.1 mm
(C-10)
The insulation thickness is calculated by cable radius jacket thickness copper thickness =
7.5 mm 2 mm 3.1 mm = 2.4 mm. Since the jacket and the insulation are both composed of
PE/PVC, the thicknesses are added together for a total thickness of 4.4 mm. As a result, the
ERFBS protected cable raceway is modeled with three layers (Figure C-6): ceramic fiber
blanket (5 cm), PE/PVC (4.4 mm), and copper (3.1 mm). The input screen for the layers of the
ERFBS and the cables is shown in Figure C-7.
C-12
Ceramic Fiber Blanket (ERFBS)

Combined Cable Jacket
Cooper Conductor
Figure C-6. Representation of the ERFBS protected cable raceway for MAGIC.
Figure C-7. MAGIC input screen for ERFBS layers.
Ventilation: Mechanical ventilation is maintained constantly during the simulation, using the
values provided above. MAGIC uses circular ducts, so the rectangular ducts seen in Figure C-1
are modeled as circular areas with equivalent diameter of 0.56 m (1.8 ft). As noted above, the
door is normally closed, but it is opened 10 minutes after ignition by the arriving fire brigade.
Before the door opens, leakage due to the doorway occurs via a 1.3 cm (0.5 in) gap under the
door. The MAGIC input screen for the doorway is shown in Figure C-8.
C-13
Figure C-8. MAGIC input screen for ventilation through door.
C.4.3 CFD Model

The following paragraphs outline the data utilized to model the scenario using FDS. Figure C-9
provides an illustration of the scenario, as rendered by Smokeview.
Geometry: The compartment is modeled as shown in Figure C-1, except that the pump itself is
modeled as two rectangular boxes. A single uniform, rectangular mesh spans the entire
compartment, plus the hallway outside the door. The numerical mesh consists of approximately
0.2 m (0.7 ft) grid cells. A finer calculation with 0.1 m (0.3 ft) cells was performed with similar
results. The latter calculation requires roughly a week of computing time on a single processor
computer (2008 vintage), whereas the more coarsely gridded calculation requires about 10
hours.
Materials: All material properties are as specified above. The protected cable tray is modeled
as a rectangular box with the same dimensions as the tray wrapped in a blanket. The box is
made solely of 5 cm (2 in) of ceramic fiber insulation. The tray is neglected. A cable target is
positioned within the box pointing downwards, as this is the hottest surface of the box. The
exact dimensions of the box are not an issue; what matters is that the cable within the box is
exposed to the heat that penetrates the thermal blanket. The cable temperature is computed
using the Thermally-Induced Electrical Failure (THIEF) methodology (NUREG/CR-6931).
C-14
Figure C-9. FDS/Smokeview rendering of the pump room scenario at the

early stage of the fire, before the compartment becomes under-ventilated.
Fire: Due to the limited amount of validation data available for scenarios of this type and the
considerable uncertainties involved, the approach taken is to specify, rather than attempt to
predict, the burning rate of the fuel, even though the FDS model does provide the physical
mechanisms to estimate burning rates. The fire is specified in the diked area surrounding the
pump. Although FDS has a liquid fuel burning model, it is not used here because there is not
enough information about the fuel, and, more importantly, it lacks the exact geometry of the
pump and diked area. Typically, FDS would expect that the oil has formed a relatively deep
pool with relatively little influence by the surrounding solids. This is not the case here. Instead,
the specified burning rate, 0.039 kg/m2/s, is applied directly to the model over an area of
2.75 m2 (29.6 ft2), yielding a burning rate of 0.107 kg/s. The density of the oil is 0.76 kg/L, which
means that the oil burns at a rate of 0.141 L/s. At this rate, 190 L (50 gal) will require 1,348 s to
burn out. Note that this is slightly different from the burning duration of 1345 s computed for the
MAGIC input. The fire duration computation for FDS converts the mass data to volumetric data,
C-15

thus introducing an additional step and some rounding. The slight difference in the burning
durations is due to this additional rounding and is not significant. The vaporized fuel is a mixture
of various hydrocarbons, but FDS uses only one fuel molecule in the combustion sub-model.
For this calculation the fuel molecule is modeled as C14H30.
Ventilation: The volume flow rates are applied as specified.
C.5 Evaluation of Results

The primary purpose of these calculations is to assess whether the Kaowool ERFBS applied to
the critical cables within the pump room would be damaged in the event of a lubricating oil fire.
The results of the zone model MAGIC and the CFD model FDS are consistent in their HRR and
compartment temperatures. This is expected because the models use the same specified
burning rate, the same fuel stoichiometry, and the same basic rules of gas phase flame
extinction based on oxygen and temperature levels in the vicinity of the fire.
Table C-5 summarizes the predicted cable temperatures from MAGIC and FDS, including an
assessment of the model uncertainty. Note that the results are based only on a direct
calculation of the cable temperature and do not include an assessment of the sensitivity studies
that are discussed in the next section.
Table C-5. Summary of the model predictions of the pump room scenario.
Model
Bias
Factor,
Standard
Deviation,
Predicted
Value
Critical
Value
Probability of
Exceeding
Cable Temperature (C)

MAGIC
1.19
0.27
135
205
0.000
FDS
1.02
0.13
145
205
0.000
C.5.1 The Fire

The HRR curves predicted by the hand calculations and the MAGIC and FDS models are
shown in Figure C-10 for the entire simulation (first plot) and the first five minutes (second plot).
The figures show the pronounced drop in the HRR soon after the start of the fire, which
demonstrates that there is insufficient air (i.e., oxygen) within the compartment to sustain the
postulated fire. Based on the calculations, the drop in HRR does not occur until after about
120 s, which is later than the predictions by MAGIC and FDS because the algebraic calculations
allow all of the oxygen in the room to be consumed. MAGIC uses a lower oxygen limit of 10%,
and FDS uses a lower oxygen limit that depends on temperature. At high temperatures, FDS
expects all of the oxygen is consumed. After 120 s, but before the door is opened at 600 s, the
value of the HRR calculated by the algebraic calculations (approximately 900 kW) is higher than
the values predicted by MAGIC and FDS (approximately 350 kW) due to the differences in the
lower oxygen level. The sudden jump in the HRR, predicted by FDS at 600 seconds, is caused
by the unburned fuel igniting as the door is opened. Note that none of the models has an
algorithm capable of determining whether or not the fire would be sustained at this reduced
burning rate until the time when the door is opened.
C-16
Heat Release Rate

Entire Simulation
HRR (kW)
16000
6000
MAGIC
5000
MAGIC
FDS
4000
FDS
3000
Hand Calculation
HRR (kW)
20000
Heat Release Rate

First 5 Minutes
12000
8000
2000
1000
4000
0
0
600
1200
1800
2400
3000
3600
30 60 90 120 150 180 210 240 270 300

Time (s)
Time (s)
Figure C-10. HRR predicted by algebraic methods, MAGIC, and FDS for the pump room
fire scenario.
C.5.2 Temperature Criterion

MAGIC and FDS estimate the temperature of the hot gas layer as a function of time, as shown
in Figure C-11. As expected, the HGL temperature changes in accordance with the altered
(oxygen-starved) HRR. Once the door opens at 600 seconds, the increased HRR causes the
HGL temperature to rapidly increase until the fire consumes the available fuel. After the fire
burns out, the HGL temperature slowly drops as heat leaves the HGL through the bounding
surfaces and open door.
Temperature (C)
HGL Temperature
1000
900
800
700
600
MAGIC
FDS
ASTM E-119
500
400
300
200
100
0
0
600
1200
1800
2400
3000
3600
Time (s)
Figure C-11. HGL Temperature Predicted by MAGIC and FDS for the pump room fire
scenario.
C-17

There are two strategies for assessing the viability of the protected cables. The first is to
compare the predicted HGL temperature to the standard fire endurance curve under which the
ERFBS was tested to determine whether the predicted thermal exposure is comparable to the
qualification test. The second is to calculate the temperature of the cable directly, using the
nominal thermal properties of the fiber insulating blanket. Each strategy will be assessed in turn
below.
Comparison to the Standard Fire Endurance Temperature Curve
Figure C-11 includes the standard ASTM E119 temperature curve to which the ERFBS was
subjected during its qualification test. The predicted HGL temperatures of both MAGIC and
FDS fall below this curve during most of the hour-long simulation, but there is a period near the
beginning of the fire where the models predicted temperatures exceed the standard curve. In
order to compare the relative exposure of the ERFBS, it is necessary to consider the integrated
incident heat flux corresponding to the model HGL predictions and the ASTM E 119
temperature curve. The integrated heat flux is given by the following formula:
(C-11)
is the integrated heat flux received by the ERFBS, is the HGL temperature,
is the
Here,
ambient temperature (20 C), is the convective heat transfer coefficient (about 0.025 kW/m2/K
in fully developed fires), and is the Stefan-Boltzmann constant (5.67 10-11 kW/m2/K4). Note
that Equation C-11 is the total energy transferred from a thermally thick (emissivity of 1) hot gas
to a cold target. It is intended only to compare the different temperature curves. In reality, the
net heat transferred to a target in the compartment decreases as the target heats up.
Applying Equation (C-11) to each of the HGL temperature curves in Figure C-11 yields values of
346 MJ/m2 for the ASTM E119 curve and approximately 40 MJ/m2 for both FDS and MAGIC.
This 40 MJ/m2 exposure corresponds to an approximately 14 min exposure within the standard
test furnace. Table C-6 lists the thermal exposure as a function of time in the standard test
furnace. It is also significant to note that the maximum predicted exposure temperature remains
lower than the maximum exposure temperature that the ERFBS protected raceway was
exposed to during the ASTM E119 fire test.
C-18

Table C-6. Integrated thermal exposure of an object subjected to the ASTM E119 temperature
curve.
Thermal
Time
Exposure
(min)
(MJ/m2)
5
6
10
23
15
47
20
75
25
104
30
135
35
167
40
200
45
235
50
270
55
307
60
346
Direct Calculation of Cable Temperature
MAGIC and FDS have heat conduction algorithms to account for the multiple layers of insulation
and cable materials. The surface temperature predictions of the cables protected by the
ERFBS (ceramic fiber insulation in this case) are shown in Figure C-12. MAGIC predicts a
maximum cable surface temperature of approximately 135 C (275 F). FDS predicts a
maximum cable surface temperature of approximately 145 C (293 F). Note that although the
HGL temperature drops and then increases dramatically when the door opens, as shown in
Figure C-11, the cable temperature slowly rises. This is due to the thermal lag caused by the
ERFBS.
Cable Surface Temperature
Inside ERFBS
Temperature (C)
160
140
MAGIC
120
FDS
100
80
60
40
20
0
0
600
1200
1800
2400
3000
3600
Time (s)
Figure C-12. Cable surface temperature predicted by MAGIC and FDS for the pump room
fire scenario.
C-19

Sensitivity of the ERFBS Construction
Comparing Figure C-11 and Figure C-12 shows that the ERFBS has a large impact on the
temperature of the target cable. To determine the sensitivity of the target cable temperature to
the insulation installation technique, two additional MAGIC cases are run. In the first case (file:
Pump_Room_thinner_wrapping.cas.), the thickness of the ceramic insulation blanket is reduced
by 25% to 0.0375 m. In the second case (file: Pump_Room_tighter_wrapping.cas.), the
thickness of the ceramic insulation blanket is reduced by 25% while the density is increased to
171 kg/m3, such that the mass per area remains constant, which simulates a tighter installation
of the insulation. The results, plotted in Figure C-13, show that both cases led to a higher cable
temperature.
Inside ERFBS
160
Temperature (C)
140
120
100
80
MAGIC Base Case
60
40
25% Tighter
Wrapping
20
25% Thinner
Wrapping
0
0
600
1200
1800
2400
3000
3600
Time (s)
Figure C-13. Cable surface temperature predicted by MAGIC for changes to insulation
wrapping.
Sensitivity of the Door Size

As mentioned in Section C.3.3, the equivalence ratio for the pump room scenario falls outside of
the validation range. As a sensitivity test, MAGIC was run with the door area doubled, such that
the equivalence ratio falls within the applicable validation range (0.04 0.6) for the portion of the
simulation when the doors are open (file: Pump_Room_2Doors.cas), as calculated below:
O
0.23 0.5
0.23
0.5
4.62 m 2.1 m
4934 kW
13,100 kJ/kg 0.77 kg/s
0.77 kg/s
0.5
(C-12)
(C-13)
Figure C-14 shows the temperature comparison for the HGL and the cable surface temperature
(measured inside the ERFBS) for the base case and for the case with double doors. The plots
show that the results for both cases are very similar, indicating that the door size does not
significantly affect the results. Nevertheless, it is consistent with experimental data that the
scenario with the equivalence ratio closest to unity produces the highest enclosure temperature.
C-20

Inside ERFBS
HGL Temperature
700
160
500
MAGIC Base Case
140
MAGIC Base Case
Double Doors
120
Double Doors
Temperature (C)
Temperature (C)
600
400
300
200
100
100
80
60
40
20
0
0
600
1200
1800
2400
3000
3600
Time (s)
600
1200
1800
2400
3000
3600
Time (s)
Figure C-14. Temperature predicted by MAGIC for increased door size.
Sensitivity of HRR Profile

As discussed in Section C.5.1, both MAGIC and FDS predict that the HRR decreases rapidly
soon after the fire starts, due to an insufficient oxygen supply within the compartment. The
models indicate that the HRR decreases from about 4900 kW to about 350 kW in approximately
2 min (Figure C-10). When the oxygen is insufficient to maintain the fire, MAGIC continues to
vaporize the unburned liquid fuel, but these fuel gases do not combust (Gay, 2005). FDS also
continues to vaporize the unburned fuel, and it continues to transport the fuel gas until the door
is opened after 10 min, at which time this excess fuel mixes with incoming air and burns. This
rapid burning of built-up excess fuel gas in an under-ventilated compartment is known as a
backdraft, and it is apparent from the HRR plot in Figure C-10. However, much of the heat
from this rapid burning of fuel is immediately exhausted from the compartment and does not
significantly affect the temperature of the ERFBS.
During the period of underventilation, both models specify that the liquid fuel contained in the
dike continues to evaporate at the specified rate. In reality, the reduced temperature and
oxygen within the compartment will probably reduce the evaporation rate of the fuel, the extent
of which is difficult to predict. To account for this possibility, the fuel evaporates at a constant
rate can be changed so that the evaporation rate during the time period between 2 min and
10 min is reduced to support a 350 kW fire only. The excess fuel that does not evaporate in this
time period is added to the end of the specified burning period so that the total fuel mass is
conserved. As shown in the plot to the right in Figure C-15, a revised HRR curve (labeled as
Extended HRR) was specified as input for MAGIC. The HGL temperature for this case reaches
640 C (1184 F), compared to 580 C (1076 F) for the base case (Figure C-14), which is still
significantly lower than the ASTM E119 temperature curve (Figure C-11). However, the
predicted cable surface temperature is 200 C (392 F), falling just below the failure criterion of
205 C (400 F). This five-degree margin suggests that further validation may be needed to
ensure that the thermal properties of the ERFBS are accurate. As shown in the previous
sensitivity cases, changes in the thermal properties of the ERFBS led to an increase of more
than 10 C in ERFBS exposure temperature. Therefore, further validation of the thermal
C-21

properties of the ERFBS is needed in this case to reduce the impact of parameter uncertainty
on the surface temperature calculation.
The sensitivity case shows that (1) based on comparison to the ASTM E119 temperature curve,
the ERFBS system is not expected to fail under the predicted exposure temperatures, and (2)
based on the predicted cable surface temperature, further validation of the thermal properties of
the ERFBS is warranted as the surface temperature of the cable is close to the damage criteria.
Base Case Heat Release Rate
Sensitivity Case Heat Release Rate
6000
6000
Input HRR
5000
Extended
HRR
5000
4000
3000
2000
MAGIC
4000
HRR (kW)
HRR (kW)
MAGIC
3000
2000
1000
1000
0
0
600
1200
1800
2400
3000
3600
600
1200
Time (s)
1800
2400
3000
3600
Time (s)
Figure C-15. HRR for base case and HRR sensitivity case.

Inside ERFBS
HGL Temperature
250
600
MAGIC Base
Case
500
Extended
HRR
Temperature (C)
Temperature (C)
700
400
300
200
100
0
0
600
1200
1800
2400
3000
MAGIC Base Case
200
Extended HRR
150
100
50
0
3600
Time (s)
600
1200
1800
2400
Time (s)
Figure C-16. Temperature for base case and HRR sensitivity case.
C-22
3000
3600
C.6 Conclusion
This analysis has considered the potential for a relatively large lubricating oil spill fire in a
relatively small enclosure to damage a cable tray protected by an ERFBS. Algebraic
calculations, the zone model MAGIC, and the CFD model FDS were all used to evaluate the fire
conditions within the enclosure. MAGIC and FDS were used to calculate the thermal response
of the cables to these calculated fire conditions.
Based on the specified lubricating oil spill area and burning characteristics, a fire of
approximately 5 MW is expected. However, after the rapid consumption of the limited quantity
of air in the room, the mechanical ventilation to the enclosure could only support a HRR of less
than 1 MW before the door to the enclosure opens after 10 min. This analysis suggests that to
avoid rapid fire escalation, doors to such rooms should not be opened until firefighters are
prepared to suppress the fire, and, even then, the potential for rapid fire escalation should be
considered.
Two different strategies were applied to assess the integrity of the ERFBS. Because the
thermal and chemical properties of the insulating material are only partially known, it is
necessary to implement an alternative technical approach of comparing the predicted HGL
temperatures from the models with the standard temperature curve under which the ERFBS
received an hour rating. Since the predicted HGL temperatures do not lie completely within the
standard curve, a simple integrated heat flux calculation was performed to demonstrate that the
ERFBS received approximately 10 times the thermal exposure in the standard fire endurance
test than is predicted by the two models.
A second strategy for assessing the integrity of the ERFBS was to directly calculate the heat
penetration through the insulating blankets using the thermal material properties of the cables
and the ERFBS. Both models predicted cable temperatures below the reported critical values.
Based on the two approaches to determine its performance, the ERFBS is expected to prevent
the cables from reaching temperatures that would limit their functionality in the event of a fire
involving burning spilled lubricating oil. This conclusion is based on certain expected burning
behavior of the lubricating oil during the under-ventilated stages. A sensitivity study on the
burning behavior of the lubricating oil concluded that the results could change if the burning rate
decreases during the under-ventilated stage. The results are also shown to be sensitive to the
thermal properties of the ERFBS material. Further research or testing of the ERFBS thermal
properties may be necessary to confirm the initial conclusion.
C.7 References
1. Gay, L., C. Epiard, and B. Gautier, MAGIC Software Version 4.1.1: Mathematical Model,
EdF HI82/04/024/B, Electricit de France, France, November 2005.
2. Gay, L., and J. Frezabeu, Qualification File of Fire Code MAGIC version 4.1.1, EdF HI82/04/022/A, Electricit de France, France, December 2004.
3. Hugget, C., Estimation of Rate of Heat Release by Means of Oxygen Consumption
Measurements, Fire and Materials, 4:61-65, 1980.
4. Mulholland, G. W., and Croarkin, C., Specific Extinction Coefficient of Flame Generated
Smoke, Fire and Materials, 24:227230, 2000.
C-23

5. NIST NCSTAR 1-5F, Federal Building and Fire Safety Investigation of the World Trade
Center Disaster: Computer Simulation of the Fires in the World Trade Center Towers, 2005.
6. NIST SP 1018-5, Fire Dynamics Simulator (Version 5), Technical Reference Guide, Volume
3, Experimental Validation, 2010.
10. NUREG/CR-6931, Cable Response to Live Fire (CAROLFIRE), Volume 3, ThermallyInduced Electrical Failure (THIEF) Model, 2007.
C.8 Attachments (on CD)

1. FDS input file: Pump_Room.fds
2. MAGIC input file:
a.
b.
c.
d.
e.
Pump_Room.cas
Pump_Room_tighter_wrapping.cas
Pump_Room_thinner_wrapping.cas
Pump_Room_2Doors.cas
Pump_Room_extendedHRR.cas
C-24
APPENDIX D
MOTOR CONTROL CENTER FIRE IN A SWITCHGEAR

ROOM
D.1 Modeling Objective

The calculations described in this appendix estimate the likelihood that a motor control cabinet
(MCC) fire will damage various cables and an adjacent cabinet within a switchgear room
(SWGR).
D.2 Description of the Fire Scenario

General Description: The SWGR is located in the reactor building for a Boiling Water Reactor
(BWR). The compartment contains multiple MCCs and some other switchgear cabinets.
Geometry: The layout of the compartment is shown in Figure D-1. Figure D-2 shows the
equipment typically contained in the compartment, and Figure D-3 shows the large elevation
change between the high and low ceiling areas.
Materials: Property values for the relevant materials are listed in Table 3-1. The SWGR
boundaries are made of concrete that is at least 0.6 m (2 ft) thick. The cabinet housing is
1.5 mm (0.06 in) thick steel.
Cables: The cable trays are filled with cross-linked polyethylene (XPE or XLPE) insulated
cables with a neoprene jacket. These are considered thermoset (TS) materials. These cables
have a diameter of approximately 1.5 cm (0.6 in), a jacket thickness of approximately 2 mm
(0.79 in), 3 conductors, and a mass per unit length of 0.4 kg/m. Tray locations are shown in the
compartment drawing. These particular cables have been shown to fail when the temperature
just underneath the jacket reaches approximately 400 C (750 F) (NUREG/CR-6931, Vol. 2,
Table 5.1020). A second criterion for damage is exposure to a heat flux that exceeds 11 kW/m2
(NUREG-1805, Appendix A, Section A.5.4). Damage criteria for the adjacent cabinet are the
same as for the cable trays because the cables within the cabinet are subjected to similar
thermal exposure conditions as the steel cabinet housing.
Fire: A fire originates within a MCC cabinet. The cabinet is closed and contains more than one
bundle of qualified cable. The fire grows following a t-squared curve to a maximum value of
702 kW in 12 min and remains steady for 8 more minutes, consistent with NUREG/CR-6850
(EPRI 1011989), Appendix G. After 20 min, the heat release rate (HRR) decays linearly to zero
in 19 min. A peak fire intensity of 702 kW represents the 98th percentile of the probability
distribution for HRR in cabinets of this general description.
The top of the cabinet contains a louvered air vent, 0.6 m (2 ft) long and 0.3 m (1 ft) wide. The
cabinet is 2.4 m (8 ft) tall. The fire burns within the interior of the cabinet, and the smoke and
20
The cable failure temperature is based on experiments conducted with Cable #14, an XLPE/CSPE, 3 conductor control cable.
D-1
MOTOR CONTROL CENTER FIRE IN A SWITCHGEAR ROOM

flames exhaust from this vent, which has an area of 0.18 m2 (2 ft2) and an effective diameter21 of
0.48 m (1.6 ft).
The heat of combustion and product yields for XLPE/neoprene cable are obtained from Table 34.16 of the SFPE Handbook, 4th edition, and are listed in Table D-1. When estimating the
composition of the fires exhaust products, the jacket and insulation material of the cable are
taken as an equal-parts mixture of polyethylene (C2H4) and neoprene (C4H5Cl), with an effective
chemical formula of C3H4.5Cl0.5.
Table D-1. Products of combustion for the MCC fire.
Parameter
Peak HRR
Heat of Combustion
CO2 Yield
Soot Yield
CO Yield
Radiative Fraction
Value
C3H4.5Cl0.5
702 kW
720 s
10,300 kJ/kg
0.63 kg/kg
0.175 kg/kg
0.082 kg/kg
0.53
Source
Combination of polyethylene and neoprene
Ventilation: The compartment is normally supplied with three air changes per hour (ACH).
The supply and return vents are indicated on the drawing. The two doors are normally closed.
Normal heating, ventilation, and air conditioning (HVAC) operations continue during the fire, and
the doors remain closed. The volume of the compartment is 882 m3 (31,150 ft3); thus, three air
changes per hour is equivalent to a volume flow rate of 0.735 m3/s.
21
The effective diameter is calculated as
4 / , where
is the area of the vent opening.
D-2
Figure D-1. Geometry of the MCC/SWGR in a BWR.
D-3
Figure D-2. Typical electrical cabinet in the lower part of the SWGR.
Figure D-3. View of the high ceiling space.
D-4
D.3 Selection and Evaluation of Fire Models

This section describes the overall modeling strategy, the selection of models, and a discussion
of the validation exercises justifying the use of these models for this scenario. The discussion
also separately addresses the prediction of the temperature and heat flux to the cabinet and
cable targets.
D.3.1 Temperature Criterion

The primary temperature criterion of interest for this scenario is the cable temperature in Tray A,
which is located 0.2 m (8 in) directly above the specified cabinet fire. The Heskestad flame
height correlation included in the Fire Dynamics Tools (FDTs) and the Fire-Induced Vulnerability
Evaluation (FIVE-Rev1) is used to show if the cables in Tray A would be engulfed in the cabinet
fire flames, and would therefore be expected to fail.
The hot gas layer (HGL) temperature correlations included in the FDTs and FIVE-Rev1 are
based on compartment fire experiments with a relatively uniform ceiling height. They are also
based on fires located near floor level without flame impingement on the ceiling. For these
reasons, the McCaffrey, Quintiere, and Harkleroad (MQH) and Foote, Pagni, and Alvares (FPA)
temperature correlations are not appropriate for this scenario. Because the zone models
Consolidated Fire Growth and Smoke Transport Model (CFAST) and MAGIC are designed
primarily for compartments with relatively flat ceilings, the Fire Dynamics Simulator (FDS) will be
used as a check on these simpler models. The differing ceiling heights are not an issue for
FDS; the compartment geometry is input as is, with no need for further simplification. In a case
like this, it may be convenient to use two rectangular meshes instead of one. The two meshes
conform well to the actual geometry and enable the calculation to be run in parallel on two
processors instead of one if desired.
D.3.2 Heat Flux Criterion

For this application, the point source radiation model included in the FDTs and FIVE-Rev1 can
be used to estimate the radiation heat flux to the adjacent cabinet. However, this heat flux
calculation does not consider the influence of flame extension beneath the ceiling, so it is useful
primarily for screening purposes. The zone and/or the computational fluid dynamics (CFD)
model are needed to calculate the heat flux from the HGL in the lower ceiling space if the HGL
temperature calculation of any model indicates that the temperature of the layer is sufficiently
high to cause damage.
D.3.3 Validation
The principal source of validation data justifying the use of the above-listed fire models for this
scenario is the U.S. Nuclear Regulatory Commission/Electric Power Research Institute
(NRC/EPRI) verification and validation (V&V) study documented in NUREG-1824 (EPRI
1011999). The National Institute of Standards and Technology (NIST) has expanded the
NRC/EPRI V&V to include the latest versions of CFAST (6.1.1) (Peacock, 2008) and FDS
(5.5.3) (McGrattan, 2010). Also, CFAST and FDS utilize the cable failure algorithm, THIEF,
which was developed and validated in NUREG/CR-6931 (Vol. 3).
D-5

Table D-2 lists various important model parameters and the ranges for which the validation
study is applicable. Two of the parameters, the Fire Froude Number and the Fire Height to
Ceiling Height ratio, fall outside the listed validation ranges. The Fire Froude Number is
essentially a measure of the fires heat output relative to its base area. In this example, the fire
is specified as originating at the top of the cabinet, with the vent opening serving as its base.
for this scenario will be higher than would be calculated if the fire was burning completely
outside of the cabinet. Thus, the high value of
is the result of specifying conditions that will
produce a more severe fire conditions than would be expected if the fire were to burn partially
within the cabinet.
With the fire located on top of a 2.4 m (7.9 ft) high cabinet under a 3.0 m (9.8 ft) high ceiling, the
flame impinges on the ceiling and consequently falls outside of the NUREG-1824 (EPRI
1011999) validation range. The extension of the flame beneath the ceiling would have an
influence on the heat flux to adjacent cabinets, which needs to be addressed. The plume
correlations used within the spreadsheets and the zone models are not valid when the fire
impinges on the ceiling. The MQH and FPA compartment temperature correlations do not
explicitly address fire elevation, but are based on experiments in which the fire source was
located near floor level; they have not been validated for scenarios like this one, which have
significant flame impingement on the ceiling. Zone models use plume air entrainment
correlations in estimating the temperature and depth of the HGL, so they have limited validity for
scenarios with significant flame impingement. For this reason, FDS is used to check the results
of the simpler models.
The FDS Validation Guide (NIST SP 1018) lists two sets of experiments that involve flame
impingement on the ceiling. The first set, performed in support of the NIST investigation of the
collapse of the World Trade Center, involved 3 MW heptane spray fires under a 3.8 m (12.5 ft)
ceiling. The value of
was 2.7, and the ratio of flame to ceiling height was 1.2. A second set
of experiments was performed by the Swedish fire test laboratory SP. The HRR from the
propane burner was 450 kW, and the burner was elevated 0.65 m off the floor. The burner area
was only 0.3 m by 0.3 m (1 ft by 1 ft), and
was approximately 6. The flame to ceiling height
ratio was 1.2, but, because the fire was in the corner, the degree of flame impingement was
enhanced significantly. Predicted HGL temperatures for these experiments were within 10 % of
the reported measurements. Predicted target temperatures were within 20 %.
The second important issue in regard to model validation is the two-tiered ceiling. Although
none of the experiments used in the NRC/EPRI validation study have a similar ceiling
configuration, Benchmark Exercise #2 of this study provides validation data to evaluate the
models ability to estimate the plume and HGL temperature/depth of smoke and hot gases filling
a fairly large, open hall with an angled roof. Predicted temperatures were within approximately
10 % of the reported measurements.
D-6

Table D-2. Normalized parameter calculations for the MCC fire scenario.
Quantity
Fire Froude
Number
702 kW
1.2 kg/m
Flame
Length,
,
relative to the
Ceiling
Height,
Ceiling Jet
Radial
Distance,rcj ,
relative to the
Ceiling
Height, Hc
Equivalence
Ratio, , as
an indicator
of the
Ventilation
Rate
Compartment
Aspect Ratio
(Lower |
Upper)
Target
Distance, r,
relative to the
Fire
Diameter, D
1.0 kJ/kg/K 293 K 0.5
2.4 m 2.5 m
3.0 m
/
3.7
1.02
0.48 m 3.7
9.8 m/s
8.5 m
3.0 m
0.23
2.8 ;
0.23
8.5 m
3.0 m
3.6
1.02
0.735 m /s
8.6 m
9.1 m
2.8
1.1 m
0.5 m
2.2
0.4 2.4
No
0.2 1.0
No
1.2 1.7
N/A
0.04 0.6
Yes
0.6 5.7
Yes
2.2 5.7
Yes
1.6
702 kW
13,100 kJ/kg 0.2 kg/s
1.2 kg/m
In
Range?
3.6
2.5 m
N/A There are no targets like sprinklers or smoke detectors under

consideration in this example.
NUREG1824
Validation
Range
0.9 ;
0.3
0.2 kg/s
8.5 m
9.1 m
0.9
Notes:
(1)
(2)
(3)
The effective diameter of the fire is determined from the formula,

4 / ,
where is the area of the vent on the cabinet.
, is the sum of the height of the fire from the floor and
The Flame Length,
the fires calculated flame length.
The ceiling height, , is the lower of the two ceiling heights.
D-7
D.4 Estimation of Fire-Generated Conditions

This section describes how each of the models is used in the analysis, including specific
approximations unique to the particular model.
D.4.1 Algebraic Models

Fire: The FDTs use a steady-state HRR in both the flame height and radiation heat flux
calculation. A constant HRR of 702 kW is used for both. A fire diameter of 0.48 m (1.6 ft) is
calculated from the effective vent area atop the cabinet of 0.18 m2 (2 ft2). Table D-2 indicates
that the Heskestad flame height correlation yields a calculated flame height of 2.5 m (8.2 ft).
Consequently, the cables located directly above the cabinet would be engulfed in flame and
would therefore be expected to fail. This flame height calculation also shows that there would
be significant flame extension beneath the ceiling, which is located just 0.6 m (2 ft) above the
base of the fire.
The point source radiation model predicts the peak heat flux to the side of the adjacent cabinet,
which is approximately 1.1 m (3.6 ft) from the center of the vent on top of the burning cabinet:
0.53
4
702 kW
1.1
24.5 kW/m
(D-1)
This estimate does not include any contributions to the heat flux by the HGL or by the flame
extension beneath the ceiling. However, it does indicate that the heat flux to the adjacent
cabinet could exceed the critical heat flux by a relatively large margin. Consequently, this
scenario would warrant more detailed analysis with either a zone model or a CFD model.
D.4.2 Zone Models

Geometry: Zone fire models subdivide the space of interest into one or more compartments
connected by vents. With CFAST, the single, large compartment is modeled as two adjacent
compartments, connected by a vertical vent. Figure D-4 and Figure D-5 show the geometry and
input parameters of the CFAST calculation.
Fire: Following guidance in NUREG/CR-6850 (EPRI 1011989), the fire is placed near the top of
the cabinet. It is positioned directly below Cable Tray A to maximize exposure. The current
version of CFAST only uses the plume centerline temperature. The specified fire area, HRR,
and species yields are input directly into the model. Figure D-6 shows the fire inputs for the
CFAST calculation.
Cables: CFAST uses the Thermally-Induced Electrical Failure (THIEF) methodology developed
as part of the Cable Response to Live Fire (CAROLFIRE) program (NUREG/CR-6931, Vol. 3).
The thermal conductivity and specific heat are fixed constants. Cable density is calculated from
the specified mass per unit length, , and the cross-sectional area of the cable, :
0.4 kg/m
0.015/2 m
D-8
2264 kg/m
(D-2)

Electrical functionality is lost when the temperature just inside of the 2 mm (0.08 in) jacket
reaches 400 C (752 F). Figure D-7 shows the target inputs for the CFAST calculation.
Figure D-4. Smokeview rendering of the geometry of two-height ceiling SWGR, as

modeled in CFAST.
D-9
Figure D-5. CFAST inputs for compartment geometry for SWGR.
Figure D-6. CFAST fire inputs for two-height ceiling SWGR scenario.
D-10
Figure D-7. CFAST target inputs for two-height ceiling SWGR scenario.
Ventilation: The two individual compartments used to model the entire room are connected by
a single large vent. Although the size of this vent relative to the compartment size is not typical
of a zone model application, the simple two-compartment geometry of the space and the more
dominant mechanical ventilation flow from one side of the SWGR to the other should minimize
any uncertainty in the calculation resulting from the large connecting vent. Mechanical
ventilation is included at the specified height and with the specified volume flow applied to the
single supply (in the low-ceiling space) and return (in the high-ceiling space). Additionally, since
zone fire models use compartments that are completely sealed unless otherwise specified, a
typical leakage vent, 13 mm (0.5 in) in height, is included at the bottom of each closed doorway
to reflect the fact that the doorways are not totally airtight. Figure D-8 shows the CFAST inputs
for these natural vents.
D-11
Figure D-8. CFAST inputs for vents connections in two-height ceiling SWGR scenario.
D.4.3 CFD Model

Geometry: The entire compartment is included in the computational domain. Multiple meshes
are used, one for the low-ceiling space and three for the high-ceiling space. The FDS Users
Guide (NIST SP 1019) contains detailed instructions for running the simulation on multiple
computers. The concrete walls are essentially the boundaries of these two meshes. The
electrical cabinets and cables are included in the simulation as simple rectangular solids, and
their dimensions have been approximated to the nearest 10 cm (4 in). There is no attempt to
model the details of either the cable trays or cabinets because the grid resolution is not fine
enough. This is appropriate because the cables and cabinets are merely targets for which it is
sufficient to know their bulk thermal properties. An FDS/Smokeview rendering of the scenario is
shown in Figure D-9.
The numerical mesh consists of uniform grid cells, 10 cm (4 in) on a side. Even with this
relatively fine grid, there is considerable uncertainty in the exact nature of the fire relative to the
cabinet and the cables just above. This uncertainty mainly has to do with specifying that the fire
originates at the top of the cabinet rather than from within.
Fire: The fire burns over an area of 0.6 m (2 ft) by 0.3 m (1 ft) on top of the cabinet, with a
maximum HRR per unit area of 3,900 kW/m2, yielding a peak HRR of 702 kW.
Cables: The cables are modeled as 1.5 cm (0.6 in) cylinders with uniform thermal properties.
Following the THIEF methodology in NUREG/CR-6931, Vol. 3, electrical functionality is lost
D-12

when the temperature just inside of the 2 mm (0.08 in) jacket reaches 400 C (752 F). Note
that no attempt is made in the simulation to predict ignition and spread of the fire over the
cables, which is why the in-depth heat penetration calculation is focused on a single cable. For
this scenario, at least one cable per tray is expected to be relatively free of its neighbors and
would heat up more rapidly than those buried deeper within the pile. This is conservative and
simplifies the heat transfer calculation.
Ventilation: Three ACH are achieved with a volume flow of 0.735 m3/s applied to the single
supply and return vents. No other penetrations are included in the model.
Figure D-9. FDS/Smokeview representation of the MCC/SWGR scenario.
D-13
D.5 Evaluation of Results

The purpose of the calculations described above is to predict if and when various components
within the compartment become damaged due to a fire in the MCC. XLPE cables are expected
to be damaged when their internal temperature exceeds 400 C (750 F) or the exposing heat
flux exceeds 11 kW/m2. Damage criteria for the adjacent cabinet are equivalent to those for the
cables because the cables within the cabinet come in contact with the heated metal housing,
and are therefore exposed to similar thermal conditions. The targets of interest are three cable
trays, labeled A, B, and C, and a single electrical cabinet adjacent to the burning MCC (Figure
D-1).
Table D-1 summarizes the model predictions of the target temperature and heat flux. Chapter 4
describes how to calculate the probability of exceeding the critical value. The purpose of the
model summary is to readily identify the most likely targets to be damaged, and the model
inputs underlying that assessment.
Table D-3. Summary of the model predictions of the MCC fire scenario.
Model
Bias
Factor,
CFAST
FDS
CFAST
FDS
CFAST
FDS
CFAST
FDS
1
1.02
1
1.02
1
1.02
1
1.02
CFAST
FDS
CFAST
FDS
CFAST
FDS
CFAST
FDS
0.81
0.85
0.81
0.85
0.81
0.85
0.81
0.85
Standard
Deviation,
Target
Predicted
Value
Critical
Value
Surface Temperature (C), Initial Value = 20 C

0.27
390
400
Cabinet
0.13
170
400
0.27
705
400
Cable A
0.13
620
400
0.27
305
400
Cable B
0.13
280
400
0.27
40
400
Cable C
0.13
65
400
Heat Flux (kW/m2)
0.47
24.3
11
Cabinet
0.22
6.0
11
0.47
104
11
Cable A
0.22
75.0
11
0.47
15.8
11
Cable B
0.22
23.0
11
0.47
0.2
11
Cable C
0.22
2.5
11
Probability of
Exceeding
0.460
0.000*
0.950
0.997
0.112
0.000
0.000
0.000
0.911
0.006*
0.974
1.000
0.823
0.997
0.000
0.000
* These results require closer scrutiny. See discussion below.
D.5.1 Damage to Cabinet

The predicted heat flux to and temperatures of the cabinet adjacent to the MCC are shown in
Figure D-10. The cabinet is located approximately 1.1 m (3.6 ft) from the center of the flaming
vent. The point source radiation calculation included in CFAST and the FDTs predicts a peak
heat flux of 24.5 kW/m2 to the nearest point on the cabinet. FDS predicts the peak heat flux
D-14

(and resulting surface temperature) to be significantly lower because the fire is partially
obscured by the overhead cable tray and the burning MCC. FDS also accounts for the
orientation of the adjacent cabinet top and side relative to the fires location. However, the heat
flux to the top of the MCC near the adjacent cabinet is substantially greater than the critical
value, and a small change in the position of the fire could result in a much higher heat flux to the
target. Given the sensitivity of the predicted heat flux and surface temperature to a minor
change in the fire dynamics, the FDS prediction for the cabinet ought to be discounted in this
case.
Cabinet Temperature
500
25
400
20
Temperature (C)
Heat Flux (kW/m2)
Cabinet Heat Flux

30
CFAST
FDS
15
FDTs
10
5
CFAST
FDS
300
200
100
0
0
0
600
1200
1800
2400
3000
3600
600
1200
1800
2400
3000
3600
Time (s)
Time (s)
Figure D-10. Heat flux and temperature predictions for the adjacent cabinet.
D.5.2 Cable Damage Based on Temperature Alone

The predicted cable temperatures for the three trays are shown in Figure D-11. CFAST and
FDS estimate cable temperatures using the THIEF methodology (NUREG/CR-6931, Vol. 3).
Both models predict that the cables in Tray A are likely to fail.
Neither model predicts that the cables in Tray B will reach the failure temperature of 400 C
(750 F), but the CFAST prediction of 300 C (572 F) suggests that there is a 9% probability
that the cable temperature could be as high as the critical value. Note that these predictions are
sensitive to the exact location of the target cable within the tray, its view of the fire, and the HGL
temperature. In this case, the cables in Tray B are heated primarily by convection and radiation
from the HGL. Given that the HRR is the most important parameter controlling the temperature
of the HGL, how much would the HRR have to increase to increase the CFAST prediction from
300 C (572 F) to 400 C (752 F)? Table 4-3 indicates that the rise in the HGL temperature is
proportional to the HRR to the 2/3 power. Following the methodology in Section 4.4.1, in order
to increase the predicted HGL temperature by 100 C (212 F), the peak HRR, , must increase
by approximately:
3
2
3
702 kW
2
100 C
300 C 20 C
376 kW
(D-3)
Both FDS and CFAST predicted cable temperatures for Tray C indicate that the cables are
unlikely to fail.
D-15
D.5.3 Cable Damage Based on Incident Heat Flux

The predictions of heat flux to the cables in the three trays are shown in Figure D-11. The
critical value is 11 kW/m2. Flame height correlations predict that the fire will impinge on Tray A,
and both CFAST and FDS indicate that the heat flux to these cables would be well in excess of
the critical value.
Cable A Heat Flux
Cable A Tem perature

120
800
CFAST
100
FDS
FDS
Heat Flux (kW/m 2)
400
200
80
60
40
20
0
0
600
1200
1800
2400
3000
3600
600
1200
Tim e (s)
2400
3000
3600
Cable B Heat Flux
Cable B Tem perature

25
300
Heat Flux (kW/m 2)
CFAST
Tem perature (C)
1800
Tim e (s)
400
FDS
200
100
CFAST
20
FDS
15
10
5
0
0
0
600
1200
1800
2400
3000
3600
600
1200
Tim e (s)
1800
2400
3000
3600
Tim e (s)
Cable C Temperature
Cable C Heat Flux
100
2.5
CFAST
80
CFAST
Heat Flux (kW/m2)
Temperature (C)
Tem perature (C)
CFAST
600
FDS
60
40
20
0
0
600
1200
1800
2400
3000
3600
2.0
FDS
1.5
1.0
0.5
0.0
0
Time (s)
600
1200
1800
2400
3000
3600
Time (s)
Figure D-11. Summary of the cable temperature and heat flux predictions for the
MCC/SWGR.
D-16

For the cables in Tray B, both FDS and CFAST indicate a relatively high probability of failure,
based partly on the fact that both models have been shown to under-predict heat flux in the
NRC/EPRI fire model validation study (NUREG-1824).
For the cables in Tray C, none of the models predicts a heat flux that approaches the critical
value. Note that the significant difference between FDS and CFAST is based on the fact that
FDS outputs the maximum heat flux and cable temperature over the entire tray of cables,
whereas CFAST predicts these values at a single target location.
D.6 Conclusion
The purpose of the calculations in this example is to predict if and when various components
within a compartment will become damaged due to a fire in the MCC. The fire model analyses
performed for this scenario indicate that the fire would damage the cables in Tray A because all
the models (FDTs, CFAST, FDS) predict that the flames would directly impinge on the cables
themselves.
CFAST and FDS predict that the cables in Tray B are likely to be damaged based on the
heat flux criterion. However, neither model predicts that the interior cable temperatures are
likely to be high enough to cause failure.
Neither FDS nor CFAST predicts that the cables in Tray C would be damaged.
A point source heat flux analysis indicates that the adjacent cabinet housing would be
exposed to a heat flux that would cause damage. Even though FDS does not predict
damage, its predictions of heat flux to surfaces very near the adjacent cabinet are
sufficiently high to cast doubt on the conclusion that the cabinet would not be damaged.
Small changes in the positions of various obstructions could easily change the predicted
heat flux by an order of magnitude. Even though the point source method tends to overpredict the heat flux to targets close to the fire, there is too much uncertainty in the
geometric configuration to accept the validity of the more detailed calculation.
D.7 References
2. NIST SP 1019, Fire Dynamics Simulator (Version 5) Users Guide, 2007.
3. NIST SP 1086, CFAST Consolidated Model of Fire Growth and Smoke Transport (Version
6), Software Development and Model Evaluation Guide, 2008.
7. NUREG/CR-6931, Cable Response to Live Fire (CAROLFIRE), Volume 2: Cable Fire
Response Data for Fire Model Improvement, 2007.
8. NUREG/CR-6931, Cable Response to Live Fire (CAROLFIRE), Volume 3: ThermallyInduced Electrical Failure (THIEF) Model, 2007.
D-17

D.8 Attachments (on CD)

1. FDS input file: Switchgear_Room_MCC.fds
a. MCC in Switchgear.in
D-18
APPENDIX E
TRANSIENT FIRE IN CABLE SPREADING ROOM
E.1 Modeling Objective

The calculations in this appendix estimate the impact that a fire in a trash receptacle inside a
Cable Spreading Room (CSR) would have on safe-shutdown cables. These calculations are
part of a larger fire analysis described in Chapter 11 of NUREG/CR-6850 (EPRI 1011989),
Volume 2, Detailed Fire Modeling (Task 11). The CSR contains a large quantity of redundant
instrumentation and control cables needed for plant operation. Transient combustibles have
been identified as a possible source of fire that may impact the cables. The purpose of the
calculation is to analyze this condition and determine whether the cable targets will fail, and, if
so, at what time failure occurs. The time to smoke detector activation is also estimated. The
calculation will provide information for a decision on the hazard and risk for this scenario.
E.2 Description of the Fire Scenario

General Description: The CSR contains a large quantity of redundant instrumentation and
control cables needed for plant operation. The cables are installed in either ladder-back trays or
conduits.
Geometry: Figure E-1 illustrates the geometry of the CSR; Figure E-2 shows a photograph of
the CSR. In addition to cables, the CSR contains a fully enclosed computer compartment,
ductwork, and large structural beams. There is no high- or medium-voltage equipment
(switchgears or transformers) in the compartment. As shown in Figure E-3, the top 2.2 m
(7.2 ft) of the compartment is filled with cable trays containing cables, or ductwork, or large
structural beams.
Construction: The walls, floor, and ceiling of the CSR are constructed of normal-weight
concrete. The ductwork is made of 2 mm (0.08 in) thick steel.
Materials: Thermal properties of various materials in the compartment have been taken from
Table 2-3 of NUREG-1805 and are listed in Table 3-1. The important cables for this calculation
are located in the third and sixth trays above the fire source, which are filled with PE-insulated,
PVC-jacketed control cables important to safe shutdown. These cables have a diameter of
approximately 1.5 cm (0.6 in), a jacket thickness of approximately 1.5 mm (0.06 in), and 7
conductors. The cable mass per unit length is 0.4 kg/m. There are approximately two rows of
cables per tray. These cables are damaged when the internal cable temperature reaches
205 C (400 F) or the exposure heat flux reaches 6 kW/m2 (NUREG-1805, Appendix A). Cable
insulation thermal properties are as follows: the density is 1380 kg/m3, the thermal conductivity
is 0.192 W/m/K, and the specific heat is 1.289 kJ/kg-K (NUREG/CR-6850, Volume 2,
Appendix R). The copper in the cables has the following properties: the density is 8954 kg/m3,
the thermal conductivity is 386 W/m/K, and the specific heat is 0.3831 kJ/kg-K (SFPE
Handbook, 4th Edition). The copper mass fraction of the cable is 0.67.
E-1

The bottom cable tray is protected on the lower surface by a solid metal barrier. The bottom
and side surfaces of all cable trays in the CSR are of solid metal construction; top surfaces are
open. The tray steel thickness is 3 mm (0.12 in).
Detection System: Smoke detectors are located on the ceiling at locations shown in Figure
E-1. The detectors are UL-listed with a nominal sensitivity of 4.9 %/m.
Suppression System: A total flooding CO2 fire suppression system is initiated automatically by
cross-zoned smoke detection in the compartment, or can be operated manually. In order to
maintain a proper concentration of suppression agent, CO2 discharge causes fire dampers to
close and mechanical ventilation fans to stop. Activation of the CO2 system is not modeled in
this example.
Ventilation: The CSR has two doors on the east wall that are normally closed. Each door is
2 m (6.6 ft) wide by 2 m (6.6 ft) tall, with a 1 cm (0.4 in) gap along the floor. Standard procedure
calls for an operator to investigate the fire within 600 s (10 min) of an alarm condition. For this
reason, the door that is farthest from the fire is opened for this investigation.
There are two supply and two return vents for mechanical ventilation, each with an area of
0.25 m2 (2.7 ft2). The total air supply rate is 1.4 m3/s (3,000 cfm). All vents are located 2.4 m
(8 ft) above the floor. Upon smoke detector activation, the mechanical ventilation fans stop and
the dampers close.
Fire: A trash fire ignites within a cylindrical steel waste bin, 0.8 m (2.6 ft) high and 0.6 m (2.0 ft)
in diameter, containing 5 kg (11 lb) of trash. The heat release rate (HRR) of the transient fire is
estimated using NUREG/CR-6850 (EPRI 1011989). The fire grows following a t-squared
curve to a maximum value of 317 kW (the 98th percentile value from Table G-1 in NUREG/CR6850 (EPRI 1011989)) in 480 s (consistent with NUREG/CR-6850 (EPRI 1011989) Supplement
1 for a transient fire growth rate contained within a trash can). The fire burns at its maximum
HRR value until the trash is consumed. To determine the duration of the fire, the total energy of
the fuel, , is calculated as the product of the fuel mass and the heat of combustion:
(5 kg)(30,400 kJ/kg) = 152,000 kJ. The heat of combustion is described below. The time to
consume the fuel is found by integrating the heat release rate over the time of the burn:
152,000 kJ
317 kW
480
480 s
3
480 s
(E-1)
Here,
is the peak HRR of 317 kW, which is reached following a t-squared growth curve in
480 s. Solving for yields a total burning time of 799 s, which is rounded up to 800 s.
While the exact composition of the trash is unknown, typical waste at the plant includes wood
scraps and polyethylene protective suits. For this scenario, the trash is specified as being
comprised of equal parts of each. The radiative fraction22 and product yields, like the heat of
combustion, are taken to be averages of values for red oak and polyethylene (SFPE Handbook,
4th ed., Table 3-4.16) and shown in Table E-1. For the purpose of modeling, the fuel molecule is
specified as C4H7O2.5.
22
The fraction of the fires total energy emitted as thermal radiation.
E-2

Table E-1. Products of combustion for CSR fire.
Parameter
Peak HRR
Heat of Combustion
CO2 Yield
Soot Yield
CO Yield
Radiative Fraction
Value
C4H7O2.5
317 kW
480 s
30,400 kJ/kg
2.0 kg/kg
0.038 kg/kg
0.014 kg/kg
0.40
Source
Specified
E-3
Figure E-1. Geometry of the CSR.
E-4
Figure E-2. Photograph of the CSR used for analysis. Note that the cables are located in
the trays in the overhead.
Figure E-3. Geometric detail of the CSR.
E-5
E.3 Selection and Evaluation of Fire Models

This scenario is a typical application of both zone models and computational fluid dynamics
(CFD) models. With a single fire contained within a rectangular compartment, the application is
straightforward. The general strategy is to use a plume correlation to estimate temperature at
the height of the important cables. If this result indicates that the cables are likely to fail, a more
detailed estimate of cable temperature will be made with zone and CFD models.
Cable failure is typically judged on both elevated temperature and heat flux criteria. These
cables are damaged when the internal cable temperature reaches 205 C (400 F) or the
exposure heat flux reaches 6 kW/m2 (NUREG-1805, Appendix A).
E.3.1 Temperature Criterion

The algebraic models are used in this scenario to estimate the flame height and plume
temperatures from the transient trash can fire to determine whether the overhead cable trays
would be damaged by the trash fire. The Foote, Pagni, and Alvares (FPA) correlation is used to
estimate the average hot gas layer (HGL) temperature that would result from the trash fire
alone; this analysis does not consider either the potential ignition of cable trays or their potential
contribution to the fire HRR.
The Consolidated Fire Growth and Smoke Transport Model (CFAST) and the Fire Dynamics
Simulator (FDS) are used to estimate damage to the overhead cable trays and development of
elevated temperature and heat flux from the fire and any ignited overhead cables to estimate
damage to adjacent cables. CFAST imposes time-dependent conditions on the cable trays, but
uses a point source for radiation from the fire sources. FDS will spread the fire in a more
realistic way, which should provide more accurate estimates of temperature and heat flux to
adjacent cables.
E.3.2 Heat Flux Criterion

CFAST and FDS are used to estimate heat flux to the overhead cable. The heat flux calculation
in CFAST includes the radiation from the fire, upper and lower gas layers, and bounding
surfaces, as well as the convection from nearby gases. Radiation from the fire sources uses a
point source for the radiation calculation.
FDS models the fire in much the same way as the zone models, with the fire at the top of the
trash receptacle. However, because it is a CFD model, FDS can estimate local conditions at
the specific location of the target cables.
E.3.3 Validation
NUREG-1824 (EPRI 1011999) contains experimental validation results for CFAST and FDS that
are appropriate for this scenario. These experiments include ventilation effects on, heat fluxes
to, and temperatures of various targets, particularly cables. Fire sizes in these experiments
E-6

bound those used in this scenario. For CFAST, the Software Development and Model
Evaluation Guide, NIST SP 1086, includes updated validation results for the newest version of
the model used for this calculation. This includes all of the validation comparisons from
NUREG-1824 (EPRI 1011999), plus additional comparisons for experiments not included in the
U.S. Nuclear Regulatory Commission (NRC) guide. Plume temperature calculations have been
validated for a broad range of fire sizes and distances above the fire source in NIST SP 1086.
Also, CFAST and FDS use the cable failure algorithm, THIEF, developed and validated in
NUREG/CR-6931 (Vol. 3).
Table 2-5 of Volume 1 of NUREG-1824 (EPRI 1011999) lists various important model
parameters and the ranges for which the validation study is applicable. Table E-2 below lists
the values of these parameters for this fire scenario, along with their ranges of applicability.
Of these parameters, only the compartment aspect ratio is outside the range of tests included in
NUREG-1824 (EPRI 1011999). In this scenario, the compartment width to height ratio is well
within limits, but the length to height ratio is higher than those included in NUREG-1824 (EPRI
1011999). For a zone model, this longer than typical compartment is mainly a concern early in
the fire development, before a reasonably uniform layer has formed. Thus, predictions of
events that occur early in the fire (such as smoke detection) may be expected to have a higher
uncertainty if they are located farther from the fire source than those that occur later in the fire
(such as ignition of cables above the initiating fire source) once the fire is more fully developed.
For this scenario, smoke detectors are included throughout the compartment, but the primary
ones of concern are those which would naturally respond faster (i.e., those nearest the fire).
E-7

Table E-2. Normalized parameters and their ranges of applicability to NUREG-1824.
Quantity
Fire Froude
Number
317 kW
1.2 kg/m
0.8 m 1.6 m
4.0 m
Flame Length,
Lf , relative to the
Ceiling Height,
Hc
Ceiling Jet
Radial
Distance,rcj ,
relative to the
Ceiling Height,
Hc
1.0 kJ/kg/K 293 K 0.6
3.7
1.02
0.6 m 3.7
0.23
9.8 m/s
0.23
40 m
4.0 m
Compartment
Aspect Ratio
Target Distance,
r, relative to the
Fire Diameter, D
1.2 kg/m
1.0
1.02
2.3 m
0.6 m
3.8
Yes
0.2 1.0
Yes
1.2 1.7
N/A
0.04 0.6
Yes
0.6 5.7
No
2.2 5.7
Yes
1.6 m
0.06
1.4 m /s
18.5 m
4.0 m
10
0.4 2.4
0.6
317 kW
13,100 kJ/kg 0.4 kg/s
In
Range?
1.0
Equivalence
Ratio, , as an
indicator of the
Ventilation Rate
NUREG1824
Validation
Range
4.6
0.4 kg/s
Notes:
(1)
The Fire Height,

the fires flame length.
, is the sum of the height of the fire from the floor and
E-8
E.4 Estimation of Fire-Generated Conditions

This section provides details specific to each model.
E.4.1 Algebraic Models

General: The general approach to using the algebraic models for this scenario is to first
calculate the flame height of the transient trash fire to determine whether the flame reaches one
or more of the overhead cable trays, then to calculate the plume temperatures from the trash
fire to determine which of the overhead cable trays would be damaged by the trash fire alone.
The general scenario is depicted schematically in Figure E-4.
Air Supply
(2 Vents)
Smoke
Detector
Tray B
Air Return
(2 Vents)
Tray A
Bottom
Figure E-4. Schematic diagram of transient trash fire in CSR (not to scale).
The first step to estimate the fire-generated conditions using algebraic models is to determine
whether the cables in the cable trays located directly above the trash fire are likely to be
damaged and potentially ignited by the trash fire. As shown in Figure E-3, the top of the trash
receptacle is located at an elevation of 0.8 m (2.6 ft), and the lowest overhead cable tray is
located at an elevation of 1.8 m (5.9 ft), which is 1.0 m (3.3 ft) above the top of the trash
receptacle. As shown in Table E-2, the flame length of the cabinet fire is calculated to be 1.6 m
(5.2 ft) at the peak heat release rate of 317 kW, so this empirical correlation for flame length can
be used to confirm that at least some of the overhead cables would likely be damaged in this
scenario. However, due to the metal barrier on the bottom of the lowest cable tray, algebraic
models cannot be used to predict whether the cables in the lowest cable tray would ignite.
The next step is to calculate the fire plume temperatures that develop from the trash fire to
determine whether cable trays A or B located above the trash fire would be damaged by the
trash fire alone. The Heskestad plume temperature correlation included in the Fire Dynamics
Tools (FDTs) and the Fire-Induced Vulnerability Evaluation (FIVE-Rev1) was used to calculate
the plume centerline temperature above the trash fire. The results, shown in Figure E-5, show
that the plume temperature of all the cable trays would exceed the cable damage threshold
temperature of 205 C (400 F). However, the Heskestad plume temperature correlation is
based on an unobstructed plume. The obstruction caused by the position of the cable trays
within the fire plume would alter the actual fire plume entrainment and temperatures.
Nonetheless, these results demonstrate that the potential for damage and ignition of the cable
trays located above the transient trash fire warrants more detailed analysis.
E-9
Plumetemperatures Heskestadcorrelation
1400
Temperature(C)
1200
1000
800
Bottom
TrayA
TrayB
600
400
200
0
0
600
1200
1800
Time(s)
2400
3000
3600
Figure E-5. Plume temperatures at cable trays located above a transient trash fire.
The FPA forced ventilation correlation is used to estimate the average HGL temperature of the
CSR resulting only from the transient trash fire, based on the parameters described in the
following subsections.
Geometry: The FPA correlation requires room dimensions to be specified in terms of length,
width, and height. For this example, the selected compartment is a rectangular parallelepiped,
as shown in Figure E-1, so its length, width, and height are specified directly.
Fire: As applied to this scenario, the FPA correlation is used with the time-dependent HRR
specified for the transient trash fire. This HRR history is shown in Figure E-6.
Materials: The walls, ceiling, and floor are all specified as concrete, with the thermal properties
specified in Table 3-1.
Ventilation: The ventilation rate in the smoke purge mode is 1.4 m3/s (3,000 cfm). This value
is used as a direct input parameter in the FPA correlation.
Table E-3. Summary of input parameters for FPA analysis of CSR scenario.
Parameter
Room height (H)
Room length (L)
Room width (W)
Mech. Ventilation rate ( )
Fire elevation (Hf)
Fire parameters
Value
4.0 m
40.0 m
18.5 m
Concrete
1.4 m3/s
0.8 m
20C
See Table E-1
Source
Figure E-1
Figure E-1
Figure E-1
Figure E-1 See Table 3-1 for properties.
Specified
E-10

Temperature: The FPA HGL temperature correlation for mechanically ventilated spaces is
included in both the FDTs and FIVE-Rev1. The FPA results for the trash fire alone are shown in
Figure E-6, based on the input parameters specified in Table E-3 and the HRR history shown in
Figure E-6. These results show that, for the specified parameters, the average HGL
temperature reaches a maximum of approximately 49 C (120 F) at 800 seconds, based on the
peak trash fire HRR of 317 kW. These results show that cables in adjacent cable trays would
not be damaged by the trash can fire alone. However, further analysis is required to determine
the potential impact of overhead cable ignition on the potential for damage to cables in the
adjacent cable trays. CFAST and FDS are used to perform this more detailed analysis.
60
350
50
300
250
40
200
30
150
20
HRR(kW)
Temperature(C)
HGLtemperaturecalculation FPAcorrelation
FPA[C]
HRR[kW]
100
10
50
0
0
600
1200
1800 2400
Time(s)
3000
0
3600
Figure E-6. Average HGL temperature (red line) from FPA correlation and heat release rate (blue
line) for the CSR trash fire scenario.
E.4.2 Zone Model

General: In CFAST, the CSR is modeled as a single compartment with obstructions accounted
for by modifying the cross-sectional area of the compartment as a function of height, as
described in the geometry section below. Figure E-7 shows the scenario, as modeled by
CFAST.
E-11
Figure E-7. CFAST/Smokeview rendering of the CSR scenario.
Geometry: Since zone models are concerned with volumes and not physical length and width,
the volume of the computer compartment, as well as the numerous cable trays, ductwork, and
beams, was modeled in CFAST with a cross-sectional area that varies with height. Table E-4
shows the cross-sectional area as a function of the height calculated from the compartment
geometry shown in Figure E-1. Figure E-8 shows the CFAST inputs for compartment geometry.
Table E-4. Cross-sectional area as a function of height used for CFAST calculation.
Floor Level
Bottom of Cable Trays
Bottom of Obstructions
HVAC Ductwork
Top of Obstructions
Top of Cable Trays
Ceiling Level Obstructions
Ceiling Level
E-12
Height
(m)
0
1.8
2.2
2.4
2.8
3.2
3.6
4
Area
(m2)
700.04
635.74
483.74
514.89
634.74
699.04
291.46
291.46
Figure E-8. CFAST inputs for compartment geometry for the CSR scenario.
Fire: The specified fire is input directly. The combustion chemistry in CFAST is described, at a
minimum, by the production rates of CO and soot. Figure E-9 shows the CFAST inputs for the
fire location in the scenario taken directly from Table E-1.
Cables: In CFAST, target temperatures are calculated using a one-dimensional cylindrical heat
transfer calculation based on the material properties and cable diameter, as specified in the
scenario description. Following the Thermally-Induced Electrical Failure (THIEF) methodology
in NUREG/CR-6931, Vol. 3, electrical functionality is lost when the temperature just inside the
jacket of a thermoplastic cable reaches 205 C (400 F). Specific heat and thermal conductivity
for the cables are taken to be 1.5 kJ/kg/K and 0.2 W/m/K, as specified in the THIEF model.
Cable density is determined from the mass per unit length and cross-sectional area of the cable
from the scenario description, 0.4 0.015 2 2264 kg/m3. To account for the shielding
of the cables on the lower surface of the cable tray, the CFAST input for the normal vector from
the cable surface is directed upwards. This effectively shields the cables from the fire below
while exposing them to the surrounding gas temperature for convection and to the hot upper
gas layer for radiation. Figure E-10 shows the CFAST inputs for the three cable targets above
the fire.
2
E-13
Figure E-9. CFAST inputs for the fire in the CSR scenario.
E-14
Figure E-10. CFAST inputs for cable targets above the fire source in the CSR scenario.
Ventilation: The supply and return airflow rates are input directly into CFAST. Upon smoke
detector activation, mechanical ventilation fans stop and dampers are closed. Thus, before a
stop time for the fans can be specified, the time to smoke detector activation is needed. This
requires that CFAST be run with the fans on for the entire time to find the first smoke detector
activation. The model is then re-run using the smoke detector activation time as the fan stop
time. Figure E-11 shows the CFAST inputs for the natural ventilation.
E-15
Figure E-11. CFAST inputs for natural ventilation for the CSR scenario. Note the opening of the
doorway at 170 s, the time estimated for smoke detectors to activate.
Fire Detection: Although there are multiple smoke detectors in the space, the closest detector
is the only one that needs to be modeled to determine time to detection, based on the fact that
the nearest detector will be exposed to the greatest concentration of smoke products and the
highest gas temperatures, thus leading to the earliest response. There are no geometric or
ventilation features that would prevent this from being the case in the example considered. In
CFAST, there is no direct way of calculating smoke density for smoke detector activation.
Instead, the smoke detector is modeled as a sprinkler with a very low activation temperature
and response time index (RTI). An activation temperature of 30 oC (86 oF) and an RTI of
5 (m/s)1/2 were used for this scenario. Figure E-12 shows the CFAST inputs for the detectors in
the scenario.
E-16
Figure E-12. CFAST inputs for smoke detectors in the CSR scenario.
E.4.3 CFD Model

General: This scenario is notable because it includes a considerable amount of clutter, that
is, the space has a relatively large number of obstructions. Figure E-13 illustrates the FDS
simulation with all the blockages. Because the cable trays are regularly spaced in both the
horizontal and vertical directions, it is easy in FDS to simply replicate a single tray as many
times as necessary. Another interesting feature of the scenario is the automatic shutdown of
the ventilation system at the time of any smoke detector activation. FDS models this by
associating the creation or removal of obstructions or the activation/deactivation of a vent with
actions taken by any number of fire protection devices.
Figure E-13. FDS/Smokeview rendering of the CSR scenario.
Geometry: The interior of the compartment is modeled, and all obstructions have been
included. To get increased resolution in the area of interest, multiple meshes are used. The
finest mesh has a 10 cm (4 in) resolution and spans a volume surrounding the trash can, which
E-17

is 6 m (20 ft) long, 3 m (10 ft) wide, and 4 m (13 ft) high. Coarse meshes cover the remainder of
the compartment and adjacent hallway with cells of 20 cm (8 in). Because the objective of the
calculation is to estimate time to failure for cables within stacked trays, it is important to have at
least a 10 cm (4 in) resolution, the typical dimension of the rails of conventional cable trays.
Fire: The trash can is modeled with a square, rather than round, cross section with an
equivalent area to the round cross section and a height equal to the height of the trash can.
The local flow features around the fire are not capable of affecting the outcome of this example
calculation; therefore, the transformed square geometry is appropriate for this application. The
specified HRR is applied to the top of the trash can. The duration of the fire is 635 s, as was
computed for the zone model input. There is no need to model the interior of the can.
Materials: The thermal properties of the walls are applied directly, as specified.
Cables: The primary objective of the calculation is to estimate the potential damage to the
cables within the trays. FDS is limited to only 1-D heat transfer into either a rectangular or
cylindrical obstruction. In this simulation, the cables are modeled as 1.5 cm (0.6 in) cylinders
with uniform thermal properties, given above. Following the THIEF methodology in
NUREG/CR-6931, Vol. 3, electrical functionality is lost when the temperature just inside the
jacket of a thermoplastic (TP) cable reaches 205 C (400 F). Note that no attempt is made in
the simulation to estimate ignition and spread of the fire over the cables. The THIEF
methodology does not account for the effects of bundled cables, which may reduce the overall
heat-up of a single cable.
For the bottommost cable tray, the bottom surface of the tray is protected by a solid metal
surface. To model this in FDS, which models the trays as rectangular obstructions, individual
properties for the top, sides, and bottom are specified. The top has the same properties as the
cables; the sides and bottom are specified as having the same metal properties.
Detection: FDS has a smoke detection algorithm that predicts the smoke obscuration within
the detection chamber based on the smoke concentration and air velocity in the grid cell in
which the detector is located. The detector itself is not modeled, as it is merely a point within
the computational domain. The two parameters needed for the model are the obscuration at
alarm, which is given by the manufacturer, and an empirically determined length scale from
which a smoke entry time lag is estimated from the outside air velocity. The SFPE Handbook,
4th Edition, provides a nominal value of 1.8 m (5.9 ft) for this length scale. The obscuration at
alarm is 4.9 %/m, a typical sensitivity for smoke detectors.
Ventilation: The supply and return airflow rates are input directly into FDS. The ducts are
represented by rectangular obstructions with thin plates just below (one grid cell) the vent itself
to represent the diffusing effect of the grill. The resolution of the grid is not fine enough to
capture this effect directly. FDS is capable of stopping the ventilation system upon the
activation of any smoke detector.
E-18
E.5 Evaluation of Results

The purpose of the calculations described above is to estimate smoke detector activation times
and potential cable damage from a trash can fire in the CSR.
The compartment itself is relatively large, and the relatively small fire (317 kW) does not
substantially heat it up. Figure E-14 shows the HRR and estimated HGL temperature for the
CSR scenario for CFAST and FDS. Differences between the two models likely result from
FDSs ability to locally account for all the blockages in the room. HGL temperature in CFAST is
a spatially average value intended to represent the bulk conditions throughout the compartment,
while the FDS values are calculated based on a single vertical profile of temperature at a fixed
location within the room (in this scenario placed several meters away from the fire location to
eliminate local effects of the fire plume on the temperature profile).
HGL Temperature
400
50
300
40
Temperature (C)
HRR (kW)
Heat Release Rate
CFAST
200
FDS
100
0
CFAST
30
FDS
20
10
600
1200
1800
2400
3000
3600
Time (s)
600
1200
1800
2400
3000
3600
Time (s)
Figure E-14. HRR and estimated HGL temperature for CSR scenario.
Because of the uncertainty in all models smoke detector activation predictions and the
uncertainty associated with the possible ignition of cables in the trays just above the fire, it is
difficult to predict whether or not the CO2 suppression system would be activated in time to
prevent possible cable ignition. No validation results are available that address time to detector
activation. Thus, the analysis does not consider activation of the suppression system.
Table E-5 summarizes the results of all the models for the chosen damage criteria. For each
predicted value, a calculation is performed to determine the probability of exceeding the critical
value. The procedure for calculating this probability is given in Chapter 4, and it accounts for
the model bias and scatter. The purpose of this table is to highlight the criterion that is most
likely to be exceeded so that further analysis can be focused on this criterion and the model or
models that predict it. Each criterion is discussed in greater detail in the following subsections.
E-19

Table E-5. Summary of the model predictions for the CSR scenario.
Model
CFAST
FDS
CFAST
FDS
CFAST
FDS
CFAST
FDS
CFAST
FDS
CFAST
FDS
Bias
Factor,
1
1.02
1
1.02
1
1.02
Standard
Deviation,
M
Location
Predicted
Value
Critical
Value
Probability
of
Exceeding
205
205
205
205
205
205
0.893
0.000
0.472
0.000
0.003
0.000
6
6
6
6
6
6
0.367
0.001
0.091
0.000
0.000
0.001

0.27
298
Bottom
Cable
0.13
54
0.27
202
Cable A
0.13
36
0.27
126
Cable B
0.13
61
0.81
0.85
0.81
0.85
0.81
0.85
0.47
0.22
0.47
0.22
0.47
0.22
Heat Flux (kW/m2)

4.2
Bottom
Cable
1.0
3.0
Cable A
0.3
2.0
Cable B
0.8
E.5.1 Smoke Detection

Table E-6 shows the models predictions for smoke detector activation. The models provide
similar estimates of the detector activation time. CFAST models smoke detector actuation as a
heat detector with a relatively low thermal inertia and activation temperature. However, there is
no consensus in the fire literature for the appropriate RTI value and activation temperature.
Given the presence of beam pockets and obstructions, even a CFD model like FDS, which uses
actual smoke concentration rather than temperature in its detector algorithm, is subject to
significant uncertainty. For this reason, no credit is given for smoke detection in the final design,
and there is no need to consider either its uncertainty or any further impact on the design.
Table E-6. Smoke detector activation times, CSR.
Model
CFAST
FDS
Time (s)
170 s
160 s
E.5.2 Temperature Criterion

Figure E-15 shows the estimated impact of the fire on the cable trays above the fire. The
bottom cable is located at least 1 m (3.3 ft) above the base of the waste bin fire. With an
estimated flame height of 1.7 m (5.7 ft), ignition may occur from flame impingement. The
algebraic tools and CFAST both predict a temperature well above the cable failure temperature.
The plume temperature estimated by the algebraic tool is naturally higher than the CFAST
E-20

calculation, since CFAST is able to account for the heat transfer into the cable and thus provide
an estimate of the actual cable temperature. However, both of these calculations are based on
the cable being directly exposed to the high gas temperature of the fire plume directly above the
fire and do not account for the protection afforded by the solid metal lower surface of the cable
tray. FDS calculations take this into account and show a much lower temperature of the
bottommost cable.
For the upper cables, CFASTs predicted temperature and heat flux are higher than FDSs
because CFAST does not account for the fact that the cable trays of interest are shielded by
trays below or that the burning spreads outward from the ignition point.
E.5.3 Heat Flux Criterion

Heat flux estimates for both CFAST and FDS in Figure E-15 show values well below the critical
value of 6 kW/m2. Still, the CFAST-estimated heat flux to the bottom cable tray of 4.2 kW/m2
(Table E-5), combined with the estimated flame height (Figure E-15), does show the importance
of the solid lower surface of the cable tray.
It should also be noted that these damage criteria are intended to indicate electrical failure, but
are routinely used as ignition criteria. In newer studies in NUREG/CR-7010, cable ignition was
not observed at fluxes below 25 kW/m2, and most often only with direct flame impingement.
Handbook values for minimum ignition flux for power and communication cables are reported in
the range of 15 kW/m2 to 35 kW/m2 (SFPE Handbook, 4th ed., Table 3-4.2).
E-21
Flame Height
Bottom CableTemperature
2.0
350
CFAST
FDS
200
CFAST
1.5
250
Flame Height (m)
Temperature (C)
300
150
100
50
1.0
0.5
0.0
0
0
600
1200
1800
2400
3000
3600
600
1800
3000
3600
Cable A Heat Flux
Cable A Temperature
250
CFAST
150
FDS
Heat Flux (kW/m)
200
100
50
0
0
600
1200
1800
2400
3000
CFAST
FDS
2
1
0
3600
600
1200
Time (s)
1800
2400
3000
3600
Time (s)
Cable B Heat Flux
Cable B Temperature
140
2.0
120
CFAST
CFAST
1.5
100
Heat Flux (kW/m)
Temperature (C)
2400
Time (s)
Time (s)
Temperature (C)
1200
FDS
80
60
40
20
0
FDS
1.0
0.5
0.0
600
1200
1800
2400
3000
3600
Time (s)
600
1200
1800
2400
Time (s)
Figure E-15. Estimated cable conditions for the CSR.
E-22
3000
3600
E.5.4 Parameter Uncertainty Propagation

The analysis above has shown that a 98th percentile trash fire is unlikely to damage cables in
certain trays overhead because of the presence of a protective barrier beneath the lowest tray.
However, for some PRA applications, it may be necessary to calculate the probability of cable
damage for any fire, not just the 98th percentile fire, with or without a protective barrier. To
illustrate this concept, lets consider a simple plume temperature correlation.
Figure E-16 displays the distribution23 of peak heat release rates for transient combustibles
(NUREG/CR-6850, Appendix E). The analysis described above made use of the 98th percentile
fire from this distribution, whose peak is 317 kW.
HRR Distribution
0.010
0.008
0.006
0.004
0.002
0.000
0
100
200
300
400
500
Figure E-16. Distribution of HRR for a trash fire.
Following the methodology described in Chapter 4 and Appendix E of NUREG/CR-6850,

Heskestads plume temperature correlation24 is applied to the entire range of HRR, resulting in
distributions of plume temperatures at the locations of Tray A and Tray B, as shown in Figure
E-17. Tray A is 1.5 m above the top of the trash bin. The probability that the plume
temperature at this height from a randomly chosen fire will exceed the cable failure temperature
(205 C; 400 F) is equal to the area beneath the left hand curve in Figure E-17 for
temperatures greater than 205 C, or approximately 0.42. Similarly, the probability that the
plume temperature at Cable Tray B (2.3 m above the bin) will exceed 205 C is approximately
0.08. Consistent with the guidance in NUREG/CR-6850, this resulting probability can be used
as the "severity factor" for the quantification of corresponding fire ignition frequencies.
Note that this analysis neglects the effect of the protective barrier or cable mass. It just
estimates the likelihood that gas temperatures in the vicinity of the critical components could
reach levels that might cause damage.
23
provides a built-in function (GAMMA.DIST) that calculates the probability density function given the parameters and . In this
case, these parameters are 1.8 and 57.4, respectively.
24
A value of 0.35 was used for the radiative fraction in determining the convective heat release rate.
E-23
Plume Temperature Distribution at Tray A
Plume Temperature Distribution at Tray B

0.010
0.010
0.008
0.006
0.004
0.002
0.000
0.008
0.006
0.004
0.002
0.000
100
200
300
400
500
600
Plume Temperature (C)
100
200
300
400
500
Figure E-17. Distribution of plume temperatures at Trays 3 and 6, respectively.
E.6 Conclusion
This analysis considers the potential impact of a transient trash fire on a stack of horizontal
cable trays located directly above the transient fire source; it also considers the potential for the
cables in these trays to ignite.
Algebraic equations from the FDTs and FIVE-Rev1, including the Heskestad flame height
correlation and the Heskestad fire plume temperature correlation, were used to evaluate the
potential for the transient trash fire to ignite or damage cables in the cable trays located directly
above this fire source. These calculations showed that the unobstructed flame height would
reach multiple cable trays at the bottom of the stack; they also showed that the unobstructed
plume temperatures would exceed the cable damage temperature in all trays located above the
fire source. However, these calculations do not account for the impact of the cable trays on the
actual flame height and plume temperatures, so more detailed analysis with CFAST and FDS is
needed.
The FPA correlation included in the FDTs and FIVE-Rev1 was used to evaluate whether the
transient trash fire alone would cause damage to cable trays not located directly above this fire
source. This calculation showed that the HGL temperature from the transient trash fire alone is
well below the damage temperatures for the cables.
The more detailed analysis with CFAST and FDS shows that a 317 kW waste bin fire beneath a
vertical array of horizontal cable trays is unlikely to damage cables in the trays located three
(Tray A) and six (Tray B) levels above the fire. Both CFAST and FDS estimate peak
temperatures and heat fluxes below the failure criteria for cables in the third tray from the
bottom. Based on FDS calculations, temperatures and heat fluxes for the protected lowest
cable tray are well below critical values. Estimates from CFAST for unprotected cables
demonstrate the importance of the protection afforded by the solid metal lower surface of the
cable trays.
E-24
600
E.7 References
1. NIST SP 1086, Consolidated Model of Fire Growth and Smoke Transport, CFAST (Version
6), Software Development and Model Evaluation Guide, 2008.
5. NUREG/CR-6931, Cable Response to Live Fire (CAROLFIRE) Volume 3: ThermallyInduced Electrical Failure (THIEF) Model, 2008.
E.8 Attachments (on CD)

1. FDS input files: Cable_Spreading_Room.fds
a. Trash Fire In Cable Spreading Room.in
3. Algebraic model input files:
a. FPA_AppE.xlsx
E-25
E-26
APPENDIX F
LUBRICATING OIL FIRE IN A TURBINE BUILDING
F.1 Modeling Objective

The purpose of the calculations described in this appendix is to estimate the temperatures of
unprotected structural steel members due to a large lubricating oil fire in a turbine building. For
this scenario, a catastrophic failure of the turbine results in a large lubricating oil spill fire. The
purpose of this example is to evaluate the turbine building structural steel response to internal
lubricating oil fire in one of two potential curbed locations in the turbine building. The
calculations provide information for a decision on the hazard posed to the structural steel for
each potential fire location, thus serving as a basis for a plant modification.
F.2 Description of the Fire Scenario

General Description: The turbine building is a large structure that is approximately 100.3 m
(329 ft) long by 99.5 m (326 ft) wide by 21 m (69 ft) tall, as shown in Figure F-1. The ambient
temperature is 36 C (97 F), inside and outside the building. Although the building has multiple
levels, for this example, only two levels the main turbine deck and the level just below it are
evaluated. The lowest portion of the lower level floor is about 1.2 m (4 ft) below grade in the
area near the fire, but most of the lower level floor elevation is at the 0 m elevation. The ceiling
height of the lower level is 4.6 m (15 ft) above grade. The floor of the turbine deck is at the
5.6 m (18 ft) elevation, and the ceiling for this upper level is at the 19.8 m (65 ft) elevation. The
turbine deck has a smaller floor area than the lower level; the turbine deck level is
approximately 90 m (295 ft) long by 70 m (230 ft) wide. The building contains the turbine
generators (see Figure F-1) and a heating, ventilating, and air conditioning (HVAC) room.
Each turbine generator contains about 3,000 L (792.5 gal) of lubricating oil in a single tank, as
illustrated in Figure F-2. The proposed curbed is 6.1 m (20 ft) long by 4.6 m (15 ft) wide. Two
potential locations have been identified on the lower level where the curbing could be installed.
These locations are shown in Figure F-1. Either of the curbed areas would be designed to
contain the entire volume of 3,000 L (792.5 gal) from one of the turbine generators main
lubricating oil tanks.
There are 40 unprotected steel support columns in a rectangular configuration (four rows of ten
each) around the lubricating oil tank. Figure F-3 shows a typical unprotected steel column. The
columns are all W14 145 standard wide flange members, as shown in Detail A of Figure F-1
(American Institute of Steel Construction, 2006). The six columns identified for analysis are
denoted in Figure F-1 as A, B, C, D, E, and F. A structural analysis indicates that the loss of
any of these six columns could lead to partial collapse of the turbine building.
F-1
Figure F-1. Geometry of the turbine building.
F-2
Figure F-2. Main turbine lubricating oil tanks in the turbine building.
Figure F-3. Typical steel column in the turbine building.
F-3

Geometry: Plan and elevation views of the turbine building are shown in Figure F-1. The area
of interest involves the two levels shown in Figure F-1, which are separated by the concrete
turbine deck. There are several stairwell, hatch, and exhaust vent penetrations throughout the
turbine deck. These penetrations are identified in Figure F-1 by the H symbol for a hatch and
the S symbol for a stairwell.
Construction: The turbine deck is made of 1 m (3.3 ft) thick normal-weight concrete. The floor
and walls of the lower level are constructed of 0.3 m (1 ft) thick normal-weight concrete.
Numerous areas and landings in the turbine building are made of metal grating. The floor in the
area of the lubricating oil tank is 1 m (3.3 ft) thick normal-weight concrete. The walls and ceiling
of the upper level of the turbine building are made of 3 mm (0.12 in) thick corrugated steel. The
structural columns, steel grating, and corrugated steel are fabricated from steel containing
0.5 percent carbon.
Materials: Thermal properties of the normal-weight concrete and steel are listed in Table 3-1.
The damage criteria for the structural steel are listed in Table F-1; these criteria are based on
the acceptance criteria for a standard fire resistance test (ASTM E119-10a, 2010). The
maximum cross-sectional temperature threshold normally applies to unprotected structural steel
because of its high thermal conductivity.
Table F-1. Structural steel failure criteria (ASTM E119-10a).
Member
Maximum Cross-Section
Average Temperature
C (F)
Beam
593 (1,099)
Column
538 (1,000)
Detection System: There is no safety-related equipment in the turbine building, thus, no

detection or suppression systems are credited for this analysis. If under-deck sprinklers were
installed, they would not be credited when considering the potential for worst-case structural
failure and the possible need for passive structural fire protection.
Ventilation: The turbine building is an open area configuration with all forced ventilation
intentionally shut down at the start of the fire for reasons unrelated to the fire. There are
eighteen roof-mounted horizontal exhaust vents around the perimeter of the turbine deck level,
as shown in Figure F-1. The exhaust vents each measure 5.0 m (16.4 ft) long by 2.5 m (8.2 ft)
wide. There are no other internal or external openings beyond those already noted in the
drawings.
Fire: A large, confined spill fire involving 3,000 L (792.5 gal) of lubricating oil from the main
turbine lubricating oil tank is postulated on the lower level. The lubricating oil has been
preheated prior to the spill, such that the growth rate of the fire would be relatively short
compared to the total time required to burn the spill volume. The total spill area is
approximately 28.1 m2 (300 ft2), as shown in Figure F-1; the spill depth is calculated to be
0.11 m (0.35 ft), based on the specified spill area and fuel volume.
The lubricating oil is dominated by alkanes, that have the chemical formulae of CnH2n+2, where n
ranges between 12 and 15 (centered around 14) and a soot yield of 0.059. Fuel properties for
F-4

the lubricating oil, obtained from Tables 3-2 and 3-4 of NUREG-1805 and from Table 3-4.16 of
the SFPE Handbook, 4th edition, are summarized in Table F-2. The Handbook identifies the fuel
as a Hydrocarbon. NUREG-1805 lists properties and states that lubricating and transformer
oils are similar.
Table F-2. Data for lubricating oil fire.
Parameter
Value
CnH2n+2
Mass burning rate

Fuel volume
Density
Heat of Combustion
CO2 Yield
Soot Yield
CO Yield
Radiative Fraction
Source
Developed from fuel chemistry (n in range of
12-15)
Specified
0.039 kg/s.m2
3,000 L
760 kg/m3
46,000 kJ/kg
2.64 kg/kg
0.059 kg/kg
0.019 kg/kg
0.34
8,700 m2/kg
The peak heat release rate (HRR), , is computed from the fuel mass burning rate,
of combustion, , and the specified area of the spill, :
0.039 kg/m s
46,000 kJ/kg
28.1 m
, the heat
50,400 kW
The fire duration, , is determined from the pool depth, , density, , and burning rate,
0.11 m 760 kg/m

0.039 kg/m /s
F-5
2,144 s
35.7 min
(F-1)
:
(F-2)
F.3 Selection and Evaluation of Fire Models

This section describes the process of model selection, including a discussion of their validity for
this scenario. To the temperature of the steel it is necessary to calculate the heat flux from the
fire to and the resulting temperature of the steel. It is also necessary to determine whether
flame extension beneath the ceiling or a hot gas layer (HGL) would develop and contribute
significantly to the column heating.
F.3.1 Heat Flux and Target Temperature Algorithms

The turbine building fire under consideration is not a typical compartment fire scenario because
it exhibits both pre- and post- flashover conditions simultaneously. The 50 MW fire is expected
to fully engulf the lower deck from floor to ceiling in the vicinity of the curbed areas and
resemble a compartment fire after the onset of flashover. Away from the curbed areas, a
distinct smoke layer is expected to form and the space would exhibit pre-flashover fire
conditions. The empirical correlations and zone models are only capable of modeling one or the
other condition, not both. However, the heat flux and target temperature algorithms used by
these simpler models may be applicable. In particular, the point source and solid flame radiant
heat flux models can be used to estimate the thermal exposure condition to the near-field and
far-field columns, respectively.
The CFD model, FDS, does have the necessary physics to model the scenario, but it is
necessary first to assess if it has been validated for such conditions.
F.3.2 Validation
Table F-3 provides a summary of the normalized parameter calculations for the turbine building
fire scenario. A number of parameters fall outside of the parameter space of the NRC/EPRI
V&V study (NUREG-1824 (EPRI 1011999)), including the flame length/ceiling height ratio, the
compartment aspect ratios and some of the target distance to fire diameter ratios. In addition,
the equivalence ratio cannot be calculated with the simple equations that have been used for
various other examples.
The calculation of the equivalence ratio is challenging because natural ventilation is provided
through the 18 roof vents located around the perimeter of the turbine deck level. To evaluate
the potential impact of ventilation on the fire for this scenario, the quantity of oxygen available in
the turbine building is compared to the amount of oxygen that would be consumed by the
specified lubricating oil fire. Given a total volume of approximately 209,600 m3, the mass of
oxygen within the turbine building is estimated to be:
O ,tot
1.1 kg/m
209,600 m
0.23
53,030 kg
(F-3)
The mass of oxygen required to burn all the fuel is estimated to be:
O ,req
50,400 kW
2,144 s
13,100 kJ/kg
F-6
8,249 kg
(F-4)

Thus, the specified fire would consume less than 16% of the oxygen available within the turbine
building. Consequently, the fire would not be expected to be ventilation-limited on a global
basis, even without ventilation from the outside environment through the roof vents.
F-7
F-8
8.5 m
6.0 m
28.0 m
6.0 m
4.7
1.4
1.02
1.1 kg/m
N/A
6.0 m 3.7
0. 52
1.0 kJ/kg/K 309 K 6.0

11.0 m
2.4
4.6 m
1.02
4.5
1.2
18.8 m
6.0 m
8.8 m
6.0 m
100.3 m
4.6 m
1.5
3.1
21.8 ;
18.3 m
6.0 m
3.9 m
6.0 m
0.7
3.1
99.5 m
4.6 m
7.2
6.1
11 m
2.2 5.7
0.6 5.7
0.04 0.6
1.2 1.7
0.2 1.0
0.4 2.4
NUREG1824
Validation
Range
Yes/No
No
Yes
N/A
No
Yes
In
Range?
is the curbed area of the dike.
13.5
13.1
0.524
78. m
6.0 m
80. m
6.0 m
9.8 m/s
4 / , where
36.5 m
6.0 m
43.3 m
6.0 m
21.6
See Section F.3.2 for discussion of this parameter.
3.7
7.2 m
6.0 m
26.9 m
6.0 m
50,400 kW
Notes: The effective diameter of the fire is determined from the formula,
Target
distance to
fire diameter
(Columns
A,B,C,D,E,F)
Compartment
aspect ratios
Equivalence
ratio based
on opening
area
Ceiling jet
radius
relative to the
ceiling height,
Hc
Flame length,
Lf, relative to
ceiling height,
Hc
Fire Froude
Number
Quantity
Table F-3. Normalized Parameter Calculations for the Turbine Building Fire Scenario.

The flame length to ceiling height ratio of 2.4 calculated for this scenario indicates that the
flames would be impinging on the ceiling and spreading out over a relatively large distance
beneath the ceiling. This configuration is beyond the capability of the FDTs, FIVE-Rev1, and
CFAST, and could affect the plume entrainment, the HGL temperature and depth, and the
radiant heat flux to a target. In fact, the fire in this scenario is so large that for columns within a
few (say 3) fire diameters, the fire can be taken as fully-engulfing the columns. For columns at
greater distances, the fire can be viewed as a point source of radiation.
As part of the work performed at NIST for the investigation of the World Trade Center (WTC)
disaster, FDS was validated against large-scale fire experiments involving liquid fuel fires
impinging on the ceiling of a large compartment with a flat ceiling. Measurements included the
heat flux to and the temperatures of bare and insulated steel structural members. The NIST
experiments and the FDS simulations are described in NIST NCSTAR 1-5F. The WTC
validation work did not involve a compartment with aspect ratios as large as those in the
scenario under consideration here, but other validation exercises have shown that the model
can handle hallway flows with comparable length to height ratios. The FDS Validation Guide
(NIST Special Publication 1018-5) includes the details of these validation studies, including the
Bureau of Alcohol, Tobacco and Firearms (ATF) Corridor Experiments and the National Bureau
of Standards (NBS) Multi-room Experiments.
F.4 Estimation of Fire-Generated Conditions

This section provides details of how the different models are applied to the scenario.
F.4.1 Algebraic Models

General: The fire is depicted schematically in Figure F-4. As discussed in Section F.3, the
flame length is expected to be approximately 2.4 times the ceiling height, indicating that
significant flame extension will occur beneath the turbine deck.
The concept of ceiling flame impingement is illustrated in Figure 3-8 of NUREG-1805 and is
discussed in Karlsson and Quintiere (2000). This reference indicates that the radial length of
flame beneath a ceiling is approximately equal to the length of flame truncated by the ceiling, as
illustrated in Figure F-4. This estimate is based on data from Heskestad and Hamada (1993).
With a calculated flame length of 11 m (36 ft) and a ceiling height of 4.6 m (15 ft), this suggests
that the flame beneath the ceiling would have a radial extension of approximately 6.4 m (21 ft)
from the centerline of the fire.
F-9
Unobstructed
flame height
Effective flame
length beneath
ceiling
Figure F-4. Schematic diagram of the fire impinging on the ceiling.
Figure F-5 shows the approximate radial flame extension of 6.4 m (21 ft) superimposed (yellow
circle) on a detail of Figure F-1 near the two proposed curb locations. As shown in Figure F-5,
Column B is at the edge of the estimated flame extension for Curb Location 1, and Column D is
well within the estimated flame extension for Curb Location 2, which means that the upper part
of Column D would be engulfed in flames. Column A is near the edge of the estimated flame
extension for Curb Location 1, and Column C is near the edge of the estimated flame extension
for Curb Location 2.
F-10
Figure F-5. Detail from Figure F-1 with estimated flame extension beneath ceiling
superimposed.
The heat flux from the fire to a nearby column can be estimated using one of the FDTs
(NUREG-1805, 2005), specifically:
05.1_Heat_Flux_Calculations_Wind_Free.xls (Point Source)
05.1_Heat_Flux_Calculations_Wind_Free.xls (Solid Flame 1)
For the point source method, the estimated peak HRR is 50,400 kW, the radiative fraction is
0.33, and the horizontal distance from the center of the lubricating oil pool to the nearest column
(Column D) is approximately 4.2 m (13.8 ft):
0.34
4
50,400 kW
4.2 m
75.0 kW/m
(F-5)
The solid flame calculation in the FDTs is based on the flame being approximated as a
cylindrical, blackbody, homogeneous radiator with an average emissive power (Beyler, SFPE
Handbook, 4th edition). However, Beyler points out that the solid flame calculation represents
the average emissive power over the whole flame and is significantly less than the emissive
power that can be attained locally. Thus, the solid flame calculation is inappropriate for
estimating the heat flux to columns that are potentially engulfed in flames. In fact, neither
calculation is appropriate for estimating the heat flux to columns that are potentially in flames.
Neither of these simple radiation calculations account for the flame extension beneath the
ceiling, so it is likely that they underestimate the actual heat flux at the column, particularly for
the case of the solid flame heat flux calculation.
F-11

Even though the simple heat flux calculation methods are questionable for close targets, it is still
worthwhile to estimate the time for the column to heat up to temperatures that could potentially
cause it to fail. If so, then it makes sense to continue analysis with a more detailed model like
FDS.
In order to estimate an approximate time for a column to reach the specified failure temperature
of 538 C (1,000 F) when subjected to a given radiant heat flux, a simple energy balance is
used to calculate the steels rate of temperature:
(F-6)
The subscript refers to steel. The temperature of the steel is modeled as changing uniformly
through the steel cross-section due to its relatively high thermal conductivity. This is sometimes
called a lumped capacitance or low Biot number approximation. For a constant net heat flux,
this differential equation can be readily integrated to yield the steel temperature as a function of
time:
(F-7)
/
To calculate the time, crit , when the steel failure temperature is reached, this equation is
rearranged, with the critical steel temperature, crit , inserted for the steel temperature.
/
crit
crit
crit
(F-8)
The term / is sometimes called the section factor, and is the effective thickness of the steel
member; it is calculated as the cross-sectional area of a steel member divided by the heated
perimeter of the member. In the U.S., it is more common to use a parameter referred to as the
W/D ratio, which is simply the section factor multiplied by the steel density. For a W14x145
steel column, the W/D ratio has a value of approximately 96.2 kg/m2 (1.64 lb/ft/in). Using this
value for the W/D ratio, the time to reach the critical steel temperature for the column can be
estimated using the radiant heat flux estimated in Equation F-5:
crit
0.465 kJ/kg/C 96.2 kg/m

75.0 kW/m
538 C
36 C
300 s
(F-9)
Because this time is much shorter than the estimated burning duration of the lubricating oil fire
(~2,100 s), this analysis suggests that it may be necessary to protect the columns nearest the
proposed curb locations with fire resistant coatings to prevent the columns from reaching the
specified steel failure temperature.
Because a number of conservative approximations were used for this analysis using algebraic
models, FDS is also used to analyze this scenario.
F-12
F.4.2 CFD Model

Geometry: The entire turbine hall is included in the simulation (see Figure F-6). One mesh
covers the lower deck, and one covers the upper turbine deck. The numerical mesh consists of
uniform grid cells with a resolution of 1 m (3.3 ft). While this mesh appears to be fairly coarse,
the fire is so large that the ratio of
(the characteristic fire diameter) to the cell size is about
five.
The columns cannot be resolved on the relatively coarse grid, and are approximated as steel
plates with the given thickness of the actual columns. The column obstructions are one cell
thick, which allows the boundary on the surface opposite the fire to be exposed to the room
conditions. Even though the column obstruction is one cell thick in the domain mesh, the
thickness of the steel surface through which heat is transferred is equal to the thickness of the
column web. Note that FDS only performs a one-dimensional heat transfer calculation within
solid obstructions, which is why there is little to be gained by resolving the column further. The
neglect of lateral heat conduction within the solid tends to produce a slight over-prediction of the
average column cross section temperature, but, because the heat flux from the fire is expected
to be fairly uniform over the width of the column, a more detailed thermal conduction calculation
is not warranted.
Hatchesandstairs
betweenlevels
Upperlevel
HVACroom
Lowerlevel
ColumnE
ColumnB
ColumnA
ColumnD
ColumnC
Plant
north
Ventstoexterior
Curb
Location1
Curb
Location2
Stairopening
Figure F-6. FDS geometry for the turbine building fire scenario.
Materials: The material properties are applied directly as specified to the walls, floor, and
ceiling.
F-13

Fire: The fire is specified in the curbed areas, as described in Section F.2. The HRR, soot
yield, and molecular weight are as described in Section F.2, and are provided directly as inputs
to FDS. A ten-second growth rate is used to allow the flows to develop over a finite time
interval.
Ventilation: The openings to the exterior and between the upper and lower level are modeled
at the locations, as shown in Figure F-6. It should be noted that the point of including the lower
and upper levels of the turbine building in the simulation is to check whether there would be
sufficient make-up air drawn through the various vents to sustain a steady-state 50 MW fire.
F.5 Evaluation of Results

The purpose of the calculations described above is to estimate the steel temperature of six large
columns in the turbine building to determine whether any would lose the ability to carry their
design load in the event of a large fire in the curbed area around a tank of lubricating oil. A
structural steel column is considered to fail if the average cross section temperature of the steel
exceeds 538 C (1,000 F), as described in Section F.2. The results of the FDS simulations are
summarized in Table F-4.
Table F-4. Summary of results for the turbine building fire scenarios.
Model
Bias
Factor,
FDS
FDS
FDS
FDS
FDS
FDS
1.02
1.02
1.02
1.02
1.02
1.02
FDS
FDS
FDS
FDS
FDS
FDS
1.02
1.02
1.02
1.02
1.02
1.02
Standard
Deviation,
Target
Predicted
Value
Critical
Value
Surface Temperature (C), Initial Value = 36 C

Curb Location 1
0.13
Column A
270
538
0.13
Column B
260
538
0.13
Column C
170
538
0.13
Column D
150
538
0.13
Column E
90
538
0.13
Column F
50
538
Curb Location 2
0.13
Column A
130
538
0.13
Column B
120
538
0.13
Column C
400
538
0.13
Column D
620
538
0.13
Column E
75
538
0.13
Column F
50
538
Probability of
Exceeding
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.001
0.828
0.000
0.000
F.5.1 Column Heat Flux and Column Temperature

The predicted column temperatures for Curb Location 1 and Curb Location 2 are shown in
Figure F-7 and Figure F-8. According to the FDS analyses, only Column D is threatened
significantly by a fire at Curb Location 2.
F-14
Temperature of Column B
Temperature of Column A
400
400
350
FDS
300
Temperature (C)
Temperature (C)
350
250
200
150
100
50
FDS
300
250
200
150
100
50
0
0
0
600
1200
1800
2400
3000
3600
600
1200
Time (s)
3000
3600
Temperature of Column D
Temperature of Column C
400
350
350
FDS
300
Temperature (C)
Temperature (C)
2400
Time (s)
400
250
200
150
100
50
FDS
300
250
200
150
100
50
0
0
600
1200
1800
2400
3000
3600
600
1200
Time (s)
1800
2400
3000
3600
Time (s)
Temperature of Column E
Temperature of Column F
400
400
350
350
FDS
300
Temperature (C)
Temperature (C)
1800
250
200
150
100
50
FDS
300
250
200
150
100
50
0
0
600
1200
1800
2400
3000
3600
0
0
600
1200
1800
Time (s)
Time (s)
Figure F-7. Temperatures of columns for Curb Location 1.
F-15
2400
3000
3600
Temperature of Column A
Temperature of Column B
800
800
700
FDS
600
Temperature (C)
Temperature (C)
700
500
400
300
200
100
FDS
600
500
400
300
200
100
0
0
0
600
1200
1800
2400
3000
3600
600
1200
Time (s)
3000
3600
Temperature of Column D
Temperature of Column C
800
700
700
FDS
600
Temperature (C)
Temperature (C)
2400
Time (s)
800
500
400
300
200
100
FDS
600
500
400
300
200
100
0
0
600
1200
1800
2400
3000
3600
600
1200
Time (s)
1800
2400
3000
3600
Time (s)
Temperature of Column E
Temperature of Column F
800
800
700
700
FDS
600
Temperature (C)
Temperature (C)
1800
500
400
300
200
100
FDS
600
500
400
300
200
100
0
0
600
1200
1800
2400
3000
3600
0
0
600
1200
1800
Time (s)
Figure F-8. Temperatures of columns for Curb Location 2.
F-16
2400
3000
3600
F.6 Conclusion
This analysis has addressed the potential for a relatively large lubricating oil fire to damage
exposed structural steel in a turbine building. The analysis is complicated by the significant
flame impingement on the ceiling caused by an oil fire spread over a relatively large area. This
type of fire behavior is beyond the validation range addressed in NUREG-1824 (EPRI 1011999).
Algebraic calculations were performed to estimate the extent of flame extension beneath the
ceiling. These algebraic calculations indicate that at least one of the columns (Column D) would
be engulfed in the flames extending from the fire at Curb Location 2 These calculations also
indicate that other columns would be located near the outer extent of flames from Curb
Locations 1 and 2. Algebraic calculations were also performed to estimate the time to reach the
critical steel temperature of the nearest column. These calculations indicate that damage could
occur within a time frame of approximately five minutes. These calculations indicate that a more
detailed analysis is warranted. The CFD model, FDS, was used to perform this more detailed
analysis because zone models do not have the necessary physical models to simulate the
postulated fire.
Based on the FDS simulation of this scenario, a 50 MW lubricating oil fire in Curb Location 1 is
not predicted to cause the steel columns to exceed a temperature of 538 C (1,000 F). This is
not the case for the proposed Curb Location 2, which is located closer to Column D.
Consequently, the recommendation for the design package is to install the curbed area at Curb
Location 1.
F.7 References
1. American Institute of Steel Construction, Steel Construction Manual, 13th Edition, New York,
2006.
2. ASTM E119-10a. Standard Test Methods for Fire Tests of Building Construction Materials,
American Society for Testing and Materials, West Conshohocken, PA, 2010.
3. Heskestad, G., and Hamada, T., Ceiling jets of strong fire plumes, Fire Safety Journal, Vol.
21, No. 1, pp. 69-82, 1993.
4. Karlsson, B., and J.G. Quintiere, Enclosure Fire Dynamics, CRC Press LLC, Boca Raton,
FL, 2000.
5. Mulholland, G.W., and C. Croarkin, Specific Extinction Coefficient of Flame Generated
Smoke, Fire and Materials, 24:227-230, 2000.
6. NIST NCSTAR 1-5F, Federal Building and Fire Safety Investigation of the World Trade
Center Disaster: Computer Simulation of the Fires in the World Trade Center Towers, 2005.
7. NIST SP 1018-5, Fire Dynamics Simulator (Version 5), Technical Reference Guide, Vol. 3,
Experimental Validation, 2010.
F-17
F.8 Attachments (on CD)

1. FDS input files:
a. Lube_oil_fire_in_TB_Location_1.fds
b. Lube_oil_fire_in_TB_Location_2.fds
2. Algebraic model files:
a. 05.1_Heat_Flux_Calculations_Wind_Free_AppF.xls (Point Source)
b. 05.1_Heat_Flux_Calculations_Wind_Free_AppF.xls (Solid Flame 1)
F-18
APPENDIX G
TRANSIENT FIRE IN A MULTI-COMPARTMENT

CORRIDOR
G.1 Modeling Objective
The calculations described in this appendix predict room temperatures in a multi-compartment
corridor after the accidental start of a transient fire in a stack of burning pallets with two trash
bags on top that were left after a plant modification. The purpose of the calculation is to
determine whether cables in adjacent compartments will fail, and, if so, at what time failure
occurs. The time to smoke detector activation is also estimated.
G.2 Description of the Fire Scenario

General Description: After a plant modification, a fire ignites in a stack of pallets with two trash
bags on top, in a corner of a corridor containing multiple compartments, with different door
heights and soffits. Various important cables are routed through the corridor compartments.
Geometry: This area consists of interconnected compartments and corridors on the same
level. All boundary surfaces are 0.5 m (1.6 ft) thick, as shown in Figure G-1. Figure G-1 and
Figure G-2 illustrate the geometry.
Materials: The walls, ceiling, and floor are made of concrete. The thermal properties of the
materials in the compartment are listed in Table 3-1. The cable trays contain cross-linked
polyethylene (XPE or XLPE)-insulated cables with a Neoprene jacket. These thermoset (TS)
type cables are considered damaged when the internal temperature just underneath the jacket
reaches 330 C (625 F) (NUREG-6850 (EPRI 1011989), Table H-1). The tray locations are
shown in Figure G-2.
Fire Protection Systems: There are nine smoke detectors, located as shown in Figure G-1.
The detectors are Underwriters Laboratory (UL) listed, with a sensitivity of 4.9%/m (UL 217,
2006). There is no automatic fire suppression.
Ventilation: The ventilation system supplies the combined space at a rate of 1.67 m3/s
(3,54 ft3/min). The supply and return vents are shown in the drawing. There are three doors
leading into the space, all of which are closed during normal operation but each has a 1.3 cm
(0.5 in) gap between the floor and its base.
G-1
TRANSIENT FIRE IN A MULTI-COMPARTMENT CORRIDOR
Figure G-1. Geometry of the multi-compartment corridor.
G-2
Figure G-2. Geometry details of the multi-compartment corridor.
G-3

Fire: The fire, a stack of four wood pallets with two trash bags, is located in the corner, as
shown in Figure G-1 and Figure G-2. The pallet stack is 0.44 m (1.4 ft) tall. The fire HRR
follows a t-squared curve to a maximum value of 2,500 kW in seven minutes and remains
steady for eight additional minutes. The heat release rate (HRR) is estimated by combining
separate estimates for the stack of wood pallets and the trash bags and using data from the
SFPE Handbook, 4th edition. After that, the fires HRR decays linearly to zero in eight minutes.
The heat of combustion and product yields are based on values for red oak, and are listed in
Table G-1. The HRR curve is shown in Figure G-3.
Table G-1. Products of combustion for a wood pallet fire.
Parameter
Peak HRR
Heat of Combustion
Heat of Combustion per unit
mass of oxygen consumed
CO2 Yield
Soot Yield
CO Yield
Radiative Fraction
Value
C6H10O5
2,500 kW
420 s
17,100 kJ/kg
13,100 kJ/kg
Source
Cellulose
SFPE Handbook, 4th Ed., Figs. 3-1.65, 3-1.100
SFPE Handbook, 4th Ed., Figs. 3-1.64
Hugget 1980, Average value
1.27 kg/kg
0.015 kg/kg
0.004 kg/kg
0.37

Heat Release Rate

3000
HRR (kW)
2500
2000
1500
1000
500
0
0
600
1200
1800
2400
3000
Time (s)
Figure G-3. HRR for the corridor fire scenario.
G-4
3600
G.3 Selection and Evaluation of Models

G.3.1 Temperature Criterion

The McCaffrey, Quintiere, and Harkleroad (MQH) correlation incorporated in the Fire-Induced
Vulnerability Evaluation (FIVE-Rev1) is used to estimate the hot gas layer (HGL) temperature in
the fire room. The correlation applies only to a single room; however, if the temperature in the
fire room is less than the damage temperature, the conclusion can be made that the
temperature will be even lower in the adjacent compartment, where the closest cable tray is
located.
The time to smoke detection can be calculated using FIVE-Rev1 with the ceiling jet temperature
unconfined flow correlation of Alpert and the confined flow correlation of Delichatsios.
Zone models can calculate the time-dependent HGL properties in multi-compartment scenarios,
as well as the activation times of smoke detectors. Although the geometry in this scenario is
somewhat complex, it can be handled by zone modeling, since it is largely a group of
interconnected compartments. The zone model MAGIC is used in this scenario.
G.3.2 Validation
A source of validation data justifying the use of the fire models discussed above for this scenario
is the U.S. Nuclear Regulatory Commission/Electric Power Research Institute (NRC/EPRI)
verification and validation (V&V) study documented in NUREG-1824 (EPRI 1011999). The
National Bureau of Standards (NBS, now the National Institute of Standards and Technology
(NIST)) Multi-Room Test Series consists of 45 fire experiments in two relatively small rooms
connected via a relatively long corridor. The fire source, a gas burner, was located against the
rear wall of one of the small compartments, and fire tests of 100 kW, 300 kW, and 500 kW were
conducted. The present scenario has a larger fire in a larger, longer compartment. The V&V
study assessed the MQH HGL temperature correlation, the Alpert ceiling jet correlation, and the
zone model, MAGIC. The unconfined ceiling jet correlation of Delichatsios is included in this
analysis to assess the sensitivity of various input parameters.
Table G-2 lists various important model parameters and the ranges for which the NRC/EPRI
validation study is applicable. The calculations in Table G-2 are for the fire room
(compartment 8, as shown in Figure G-4). The ceiling jet radial distance relative to the ceiling
height is not within the range of validation. As a sensitivity case, an additional detector was
added at a distance that would fall within the validation range (6.8 m (22.3 ft)). In addition, the
room-width geometry is not within the range of validation. A sensitivity case with the height
reduced to a value within the validation range (5 m (16.4 ft)) was performed. The remaining
model parameters are within the applicable validation ranges.
G-5
Table G-2. Normalized parameter calculations for the multi-compartment corridor fire scenario.
Quantity
Fire Froude
Number
2500 kW
1.2 kg/m
Flame Length,
L , relative to
the Ceiling
Height, Hc
1.0 kJ/kg/K 293 K 1.29
3.7
1.02
Ceiling Jet
Horizontal
Radial
Distance,rcj ,
relative to the
Ceiling Height,
Hc
Equivalence
Ratio, , as an
indicator of the
Ventilation Rate
Compartment
Aspect Ratios
Target
Distance, r,
relative to the
Fire Diameter,
D
0.23
9.8 m/s
0.7
1.29 m 3.7
1.2
1.02
1.2 kg/m
1.67 m /s
3.0 m
6.1 m
2.49 ;
N/A
In
Range?
0.4 2.4
Yes
0.2 1.0
Yes
1.2 1.7
No
0.04 0.6
Yes
0.6 5.7
No
2.2 5.7
N/A
1.2
3.8 m
0.8
2500 kW
13,100 kJ/kg 0.46 kg/s
0.23
15.2 m
6.1 m
0.44 m 3.8 m
6.1 m
4.46 m
6.1 m 0.44 m
NUREG1824
Validation
Range
0.4
0.46 kg/s
0.49
Notes:
(1)
(2)
(3)
The non-dimensional parameters are explained in Table 2-5.

The effective diameter of the base of the fire, , is calculated using
4 / ,
where is the area of the pallets.
, is the sum of the height of the fire from the floor and the
The Fire Height,
fires flame length.
G-6
G.4 Estimation of Fire-Generated Conditions

G.4.1 Algebraic Models
This scenario concerns the prediction of cable damage at a location outside the compartment of
fire origin. The temperature of the HGL in the compartment of fire origin can be modeled as a
potential screening tool. If the HGL temperature within the compartment of origin is not likely to
cause damage to cables in that compartment, damage to cables outside the fire compartment is
even more unlikely. As a part of this approach, the cable surface temperature is modeled as
exactly matching the HGL temperature (i.e., heat-up of the cable is immediate).
FIVE-Rev1 was used for the MQH room temperature analysis. The inputs to the model are
found in Table 3-1, Table G-1, and Table G-3. The calculation is applied to the fire room only,
with the opening to the next compartment treated as an opening with an area (height width)
equal to 6.1 3 = 18.3 m2 (197 ft2). To correct the MQH temperature correlation for a fire in the
corner, the results from FIVE-Rev1 are multiplied by a factor of 1.7, as suggested by Karlsson
and Quintiere (Equation 6.23).
For the time to detection, the Alpert ceiling jet temperature calculation is used. The approach is
to calculate the time at which the ceiling jet temperature at the heat detector is 30 C (86 F).
The additional inputs for this correlation are the horizontal radial distance from the centerline of
the fire plume to the detector, which is 4.5 m (14.8 ft), and the fire location factor of 4, because
of the fire in the corner. Because the fire room is corridor-shaped, the flow is likely to be
confined; therefore, the confined flow correlation by Delichatsios is also used. The additional
input is the corridor half-width of 1.5 m (4.9 ft).
G.4.2 Zone Model

This is a classic application of a zone fire model with a fire in one compartment connected to a
number of additional compartments with doorway-like vents. Outputs of primary interest in the
simulation include temperatures in the compartments, activation of smoke detectors in the
compartments, and the temperature of cable targets in the compartments. This scenario was
modeled using the zone model MAGIC (Gay et. al., 2005).
Geometry: To simplify the process of modeling the multi-compartment geometry, the layout
was divided into eight areas, as illustrated in Figure G-4. Note that the small indentation in
compartment 1 was ignored for the MAGIC calculations. Compartments were connected by a
door (compartments 5 to 6), a soffit (compartments 2 to 3), or were left open by using a full-wall
opening (compartments 1 to 2, 3 to 4, 4 to 5, 3 to 7, and 7 to 8).
Table G-3 summarizes the compartment dimensions used for zone modeling. A graphical
depiction of the scenario, as modeled in MAGIC, is shown in Figure G-5.
G-7
5
4
7
1
Figure G-4. Effective corridor layout for implementation in zone models (not to scale).
Table G-3. Compartment dimensions for corridor scenario.

Comp.
Length
m (ft)
Width
m (ft)
Area
m2 (ft2)
8.1 (26.6)
4.1 (13.5)
33.2 (357)
2.0 (6.6)
23.4 (76.8)
46.8 (504)
45.1 (148)
4.1 (13.5)
184.9 (1990)
8.1 (26.6)
6.0 (19.7)
48.6 (523)
10.3 (33.8)
6.6 (21.7)
68.0 (732)
10.3 (33.8)
6.6 (21.7)
68.0 (732)
12.2 (40)
8.2 (26.9)
100.0 (1076)
3 (9.8)
15.2 (49.9)
45.6 (491)
G-8
Figure G-5. MAGIC rendering of the corridor scenario.
Fire: The fire is modeled as a 1.3 m2 (14 ft2) source (equivalent diameter of 1.29 m (4.2 ft)) at
an elevation of 0.44 m (1.4 ft) (see Figure G-2). The stoichiometric oxygen to fuel ratio, , is
calculated using Equation 22 from Chapter 4 in Section 3 of the SFPE Handbook, 4th edition
and the values from Table G-1, as follows:
17.1 kJ/g
13.1 kJ/g
1.3 g/g
(G-1)
where is heat of combustion per unit mass of fuel consumed and O is the heat of
combustion per unit mass of oxygen consumed. One of the inputs required by MAGIC is the
specific area, (NUREG-1824, Volume 6, Section 3.2.7):
0.015
7,600 m
g
114 m /g
(G-2)
is a constant and is the soot yield, as listed in Table G-1. Products of combustion
where
for a wood pallet fire.. The pyrolysis rate (g/s) is calculated for input to MAGIC by dividing the
HRR values (kW) at each time step (as shown in Figure G-3) by the heat of combustion
G-9

(17.1 kJ/g). Other inputs needed for MAGIC are listed in Table G-1. Products of combustion for
a wood pallet fire.. Figure G-6 is a screenshot of the source fire in the MAGIC input file.
Figure G-6. MAGIC screen capture of the fire in corridor scenario.
G-10

Ventilation: The ventilation rate is given in Section G.2. The vents in Figure G-1 are square,
but, because MAGIC uses round vents, an equivalent diameter of 1.13 m (3.7 ft) was used as
input. In room-to-room connections with the same ceiling height, a shallow (0.1 m) soffit was
added to allow smoother model execution. Finally, the only leakage from the space occurs via a
1.3 cm (0.5 in) crack under each of the three doors. The inputs for the supply air vent are
shown in Figure G-7.
Figure G-7. MAGIC screen capture for supply vent specification.
G-11

Fire Protection Systems: In MAGIC, there is no direct way of calculating smoke density for
smoke detector activation. Consistent with NUREG-1805, the recommended approach given by
the developers is to model the smoke detector as a sprinkler with a very low activation
temperature and response time index (RTI). An activation temperature of 30 C (86 F) and an
RTI of 5 (m/s)1/2 was selected. The location of the smoke detector closest to the fire was
entered into the input file, as shown in Figure G-8.
Figure G-8. MAGIC screen capture of the smoke detector specification.
G-12
G.5 Evaluation of Results

The purpose of the calculations is to determine whether a stack of burning pallets plus two trash
bags in a corridor could generate gas temperatures in adjacent compartments that are capable
of damaging cables and electrical equipment. Smoke detector activation is also estimated. The
algebraic model FIVE-Rev1 and the zone model MAGIC were used for this scenario. In
general, the results demonstrate that the fire is not capable of generating damaging conditions,
even in the compartment of fire origin; as a result, there is no need for detailed modeling of the
targets in remote locations. The following sections describe the results in greater detail.
Table G-4 summarizes the results of the models for the damage criteria. For each predicted
value, a calculation is performed to determine the probability of exceeding the critical value.
The procedure for calculating this probability is given in Chapter 4, and it accounts for the model
bias and scatter. The purpose of this table is to highlight the criterion that is most likely to be
exceeded so that further analysis can be focused on this criterion and on the model or models
that predict it. The criterion is discussed in greater detail in the following subsections.
Table G-4. Summary of the model predictions of the corridor scenario.
Model
Bias
Factor,
Standard
Deviation,
Ventilation
Predicted
Value
Critical
Value
Probability
of
Exceeding
HGL Temperature (C), Initial Value = 20 C

FIVE-Rev1
(MQH)
MAGIC
1.56
0.32
Natural
256
330
0.001
1.01
0.07
Mechanical
240
330
0.000
G.5.1 Temperature Criterion

The MQH correlation in FIVE-Rev1 predicts a peak temperature of 256 C (493 F) in the
corridor where the fire is located. MAGIC predicts 240 C (464 F) (see Figure G-9). Both
predictions are below the cable damage temperature threshold of 330 C (625 F). MAGIC
results show that the HGL temperatures for the other corridors are substantially lower; for
example, the next closest corridor (compartment 7) reaches a peak temperature of 160 C
(320 F), as shown in Figure G-9. A comparison of Figure G-9 and Figure G-3 shows that the
timing and the shape of the HGL temperature curve closely follows the HRR curve. The HRR
produced by MAGIC is unmodified from the one based on the input (i.e., there is no oxygen
starvation).
G-13
HGL Temperature Fire

Room
HGL Temperature Compartment 7
300
200
MAGIC
160
FIVE (MQH)
Temperature (C)
Temperature (C)
250
200
150
100
50
120
80
40
0
0
600
1200
1800
2400
Time (s)
600
1200
1800
2400
Time (s)
Figure G-9. HGL temperature predictions by MAGIC for the corridor scenario.
Chapter 4, Model Uncertainty, provides guidance on how to express the uncertainty of the
MAGIC predictions. MAGIC predicted a temperature rise of M = 240 20 = 220 C (428 F).
As shown in Table 4-1, the NRC/EPRI V&V study (NUREG-1824 (EPRI 1011999)) found that
MAGIC predictions of HGL temperature rise are, on average, 1% greater than corresponding
measurements, with a relative standard deviation of 7%. The adjusted model prediction is =
220/1.01 = 218, and the standard deviation is = 0.07 (220/1.01) = 15.2 C. Therefore, the
probability that the cable temperature would exceed 330 C (625 F) is:
330
1
330 20 218
erfc
2
15.22
(G-3)
The same uncertainty calculation for the MQH correlation in FIVE-Rev1 results in a probability
of exceeding the damage temperature of 0.001, or 0.1%. In other words, there is a near-zero
probability of exceeding the damage temperature threshold for cables within the compartment of
fire origin based on a surrounding HGL temperature, according to the FIVE-Rev1 and MAGIC
predictions. This demonstrates that detailed analyses of the cables outside the compartment of
origin are not warranted.
Sensitivity of the Room Width to Height Ratio
As shown in Table G-2, the room width to height ratio is not within the validation bounds. A
sensitivity case is run in which the room height is reduced until the ratio is within the bounds. All
inputs are the same, except that the height is reduced from 6.1 m (20 ft) to 5 m (16.4 ft) in the
fire room. The results, plotted in Figure G-10, show that although the temperatures are slightly
higher in the fire room, there is no change in the adjacent room, which contains the target. This
shows that the room width to height ratio is not a significant influence on the results in this
scenario and that the MAGIC results are applicable, even though the ratio falls outside of the
validation bounds.
G-14
HGL Temperature Fire

Room
HGL Temperature Compartment 7

200
250
Base Case
(H=6.1 m)
200
Ceiling
Height=5 m
Temperature (C)
Temperature (C)
300
150
100
160
Base Case
(H=6.1)
120
Height=5 m
Fire Room
50
0
0
600
1200
1800
80
40
0
2400
600
1200
Time (s)
1800
2400
Time (s)
Figure G-10. HGL temperature for reduced ceiling height by MAGIC.
G.5.2 Smoke Detection

The smoke detector activation time in the corridor containing the fire is based on the time for the
ceiling jet temperature to reach 30 C (86 F) at the detector location. The results, plotted in
Figure G-11, show that the two correlations from FIVE-Rev1 produce identical results of 50 s,
and MAGIC predicts 40 s.
Temperature (C)
80
FIVE (Alpert)
70
FIVE (Delichatsios)
60
MAGIC
50
40
30
20
10
0
0
50
Time (s)
100
150
Figure G-11. Detector temperature prediction by MAGIC for fire corridor.
G-15

Sensitivity of the Ceiling Jet Radial Distance Relative to the Ceiling Height
As shown in Table G-2, the ceiling jet radial distance relative to the ceiling height is not within
the validation bounds. A sensitivity case is performed in which a detector is placed at a further
distance so that the ratio is within the validation bounds. The results, plotted in Figure G-12,
show that the difference in time to detection varies from 40 s for the base case to 50 s for the
detector located slightly further away. This small difference in time indicates that for this
scenario, the results are valid, even though the ratio of radius to ceiling height falls outside of
the validation bounds.

80
Base Case (r=4.5)
70
Radial distance=6.8 m
Temperature (C)
60
50
40
30
20
10
0
0
50
Time (s)
100
150
Figure G-12. Detector temperature for two radial distances predicted by MAGIC.
G.6 Conclusion
This analysis addresses a transient fire in a relatively long, narrow room, identified as a corridor,
which is connected with other rooms in a multi-room arrangement. The targets include cables in
rooms located beyond the corridor where the fire is specified to occur. The MQH correlation
included in FIVE-Rev1 was used to estimate temperature conditions within the corridor, and it
was found that the HGL temperature predicted by the MQH correlation was lower than the cable
damage temperature. Because the cables are located in rooms beyond the corridor where the
fire occurs and it can be reasoned that the temperatures in these rooms will be lower than in the
corridor, cable damage would not be expected.
The zone model MAGIC was used to predict the HGL temperatures in all of the interconnected
compartments, resulting from the stack of burning pallets and trash bags. These MAGIC
calculations also demonstrate that the HGL temperature is not high enough to cause cable
damage in any compartment or corridor, including the corridor of fire origin. These predictions
account for model uncertainty in the temperature predictions and the sensitivity to variations in
the heat release rate. Based on a simplified method for smoke detector activation, smoke
detector operation is expected to occur between 40 s and 50 s after the ignition of the fire.
G-16
G.7 References
1. Gay, L., C. Epiard, and B. Gautier, MAGIC Software Version 4.1.1: Mathematical Model,
EdF HI82/04/024/B, Electricit de France, France, November 2005.
2. Huggett, C., Estimation of Rate of Heat Release by Means of Oxygen Consumption
Measurements, Fire and Materials, 4:61-65, 1980.
3. Karlsson, B., and Quintiere, J. G., Enclosure Fire Dynamics, CRC Press, Boca Raton, 2000.
8. UL 217, Single and Multiple Station Smoke Alarms, 6th edition, Underwriters Laboratories,
Inc., 2006.
G.8 Attachments (on CD)

1. Algebraic model file:
a. Appendix G FIVE.xlsm
b. MQH_AppG.xlsx
2. MAGIC input files:
a. Corridor.cas
b. Corridor_reduced_H.cas
G-17
G-18
APPENDIX H
CABLE TRAY FIRE IN ANNULUS
H.1 Modeling Objective

The calculations described in this appendix assess the potential for damage to redundant safeshutdown cables due to a fire in an adjacent cable tray in the annulus region of a pressurized
water reactor (PWR) containment building. In addition, the calculations provide information on
the effectiveness of a sprinkler system installed in a non-standard configuration.
H.2 Description of the Fire Scenario

General Description: The annulus is the region between the primary containment steel
structure and the secondary concrete containment (shield) building. The primary and secondary
containments are cylindrical with domes on top. The annulus space contains a variety of
penetrations from the reactor to the external support systems. One of these penetrations
contains two trays with redundant cables that control systems in both trains of safety equipment.
A fire starts in one tray and spreads vertically and horizontally. The purpose of the calculation is
to determine if a fire in one train of a cable tray (e.g., Train A) can damage the redundant train
of equipment (e.g., Train B).
Geometry: The layout of the annulus is shown in Figure H-1. The exterior wall is made of
concrete, while the interior wall and cable trays are made of steel. The cable tray locations are
shown in Figure H-2. The trays are 0.6 m (2 ft) wide.
Materials: Property values for the relevant materials are listed in Table 3-1. The annulus wall
thicknesses are indicated in the drawing. The cable tray steel is approximately 2 mm (0.079 in)
thick.
Cables: The cable trays are filled with PE-insulated, PVC-jacketed control cables. These
cables have a diameter of approximately 1.5 cm (0.6 in), a jacket thickness of approximately
1.5 mm (0.06 in), and 7 conductors. There are approximately 120 cables in each tray. The
mass of each cable is 0.4 kg/m. The mass fraction of copper is 0.67. These cables fail when
the internal temperature just underneath the jacket reaches approximately 205 C (400 F) or
when the exposure heat flux exceeds 6 kW/m2 (NUREG-1805, Appendix A).
Fire: The heat of combustion and product yields for PE/PVC cables are taken from Table 34.16 of the SFPE Handbook, 4th edition and listed in Table H-1. Note that the non-metallic
components of the cables are a mixture of PE (C2H4) and PVC (C2H3Cl). Because the mixture
consists of approximately the same mass of each, the cable materials are modeled using an
effective chemical formula of C2H3.5Cl0.5.
The fire ignites at the base of the lower cable train in the vicinity of the bend at the inner wall.
From the results of the CHRISTIFIRE project (NUREG/CR-7010, Vol. 1), the heat release rate
(HRR) per unit area of this thermoplastic (TP) cable is estimated to be 250 kW/m2. The fire
spreads vertically at a rate of 258 mm/s (10 in/s) and horizontally at a rate of 0.9 mm/s
H-1

(0.035 in/s) (NUREG/CR-6850 (EPRI 1011989), Appendix R). The peak HRR is 945 kW once
all of the cables in the first tray are burning (see Figure H-3).
Table H-1. Products of combustion for a PE/PVC cable fire.
Parameter
Heat of Combustion
CO2 Yield
Soot Yield
CO Yield
Radiative Fraction
Value
20,900 kJ/kg
1.29 kg/kg
0.136 kg/kg
0.147 kg/kg
0.49
Source
Local burnout of the fire occurs when the cable plastic is consumed. The time to burnout is
:
calculated as follows. First, determine the combustible mass per unit area of tray,
1
120
0.33
1 0
0.6 m
0.4 kg/m
26.4 kg/m
(H-1)
where is the number of cables per tray,

is the mass fraction of combustible (i.e., nonmetallic or plastic) material in the cable, is the residue yield (which is conservatively set to
is the total mass per unit length of a single cable, and
is the tray
zero in this calculation),
width. Next, calculate the burnout time, :
5 avg /6
26.4 kg/m
20,900 kJ/kg
5/6 250 kW/m
2648 s
(H-2)
where is the heat of combustion and avg is the average HRR per unit area of tray. The HRR
per unit area ramps linearly to its average value over a time period of /6, remains steady for a
time period of 2 /3, and then decreases linearly to zero over a time period of /6. The linear
ramp-up and ramp-down are typical ways of approximating the time history of an items HRR.
H-2
Figure H-1. Geometry of the annulus.
H-3
Figure H-2. Geometry details of redundant cable trays located in the annulus.
H-4
HRR (kW)
Heat Release Rate

1000
900
800
700
600
500
400
300
200
100
0
0
600
1200
1800
2400
3000
3600
Time (s)
Figure H-3. HRR for a cable fire in the annulus.
Suppression System: Standard response sprinklers are located on the inner wall, as shown in
Figure H-2. The sprinklers have a response time index (RTI) of 130 (ms)1/2 and activate at a
temperature of 100 C (212 F) (NUREG-1805, Chap. 10). Each sprinkler is topped by heat
collectors designed to trap heat from a fire. The ambient temperature within the annulus region
is typically 35 C (95 F).
H.3 Selection and Evaluation of Fire Models

This section describes the overall modeling strategy, the selection of models, and a discussion
of the validation exercises justifying the use of these models for this scenario. The discussion
separately addresses the prediction of the heat flux to and temperature of the cable targets, and
the sprinkler activation.
H.3.1 Damage to Cables

The point source radiation heat flux model included in the Fire Dynamics Tools (FDTs) and FireInduced Vulnerability Evaluation (FIVE-Rev1) provides useful screening estimates for cable
damage. A simple extension of the point source heat flux model, in which the sources of heat
are distributed at discrete points within the burning tray, provides a refinement of the single point
source method. Given the close proximity of the two cable trays, the distributed point source
model is expected to provide a more accurate estimate of the heat flux to the adjacent cable
tray.
This fire scenario can be categorized as an open rather than a compartment fire. For this
reason, zone models are not particularly useful. Computational Fluid Dynamics (CFD) can
refine the estimates of the simple heat flux calculations to account for tray geometry and
orientation.
H-5
H.3.2 Sprinkler Activation

Although the geometry of this scenario is unlike the mostly rectangular compartments found in a
nuclear power plant (NPP), it is not particularly difficult to model in the Fire Dynamics Simulator
(FDS). In fact, the containment building is so large that the curvature of the walls has little effect
on the results of the calculation. For this reason, FDS can provide an estimate of the sprinkler
link temperature to determine its potential for activation. Empirical correlations are not reliable
because the fire burns along a horizontal and a vertical tray and thus has no well-defined
circular base, and the plume abuts the cylindrical wall.
H.3.3 Validation
Only one of the non-dimensional parameters that have been used to characterize the fire
scenarios is applicable here (Table H-2), mainly because the other parameters address
phenomena unique to compartment fires. The only relevant parameter, / , indicates the
relative distance separating the fire from the target cables. However, the effective fire diameter,
, is not well-defined when the fire is expected to spread vertically and horizontally within the
cable trays. The lateral distance between the burning tray and the target tray is 1.7 m (5.6 ft).
The trays are each 0.6 m (2 ft) wide. For International Collaborative Fire Model Project (ICFMP)
Benchmark Exercise #3, which is described in NUREG-1824, the relative distance between the
fire and the heat flux gauges in some of the tests was similar to this scenario. However, the
accuracy of the solid flame and point source algorithms included in the FDTs varied
considerably in the validation study. The point source heat flux model loses accuracy when the
target is relatively close to the fire. The solid flame model is designed to improve the accuracy
in the near-field. However, the implementation within the FDTs is based on the fire being
modeled as a pool fire with a nearly circular base and cylindrical shape. This is not the case in
the scenario under consideration. In short, the simple heat flux models included in the FDTs
have been assessed in NUREG-1824 for experiments of comparable scope to the given
scenario, but the models have not been demonstrated to be particularly accurate. For this
reason, the models will be used to provide screening estimates, but the predictions will be
supplemented by CFD calculations.
The FDS sprinkler activation algorithm was validated using a variety of experimental test series
(NIST SP 1018-5). The plume algorithm was also assessed in NUREG-1824. The cable failure
algorithm, THIEF, was developed and validated in NUREG/CR-6931 (Vol. 3).
H-6

Table H-2. Normalized parameter calculations for the annulus fire scenario.
Quantity
NUREG-1824
Validation
Range
In
Range?
Fire Froude
Number
N/A The fire does not conform to classic fire plume theory.
0.4 2.4
N/A
N/A The fire does not conform to classic fire plume theory.
0.2 1.0
N/A
Ceiling Jet Radial

Distance, rcj ,
relative to the
Ceiling Height, Hc
N/A The ceiling height is essentially infinite.
1.2 1.7
N/A
Equivalence Ratio,
, as an indicator
of the Ventilation
Rate
N/A The scenario is outside of a clearly defined

compartment.
0.04 0.6
N/A
Compartment
Aspect Ratio
N/A The scenario is outside of a clearly defined

compartment.
0.6 5.7
N/A
Target Distance, r,
relative to the Fire
Diameter, D
See discussion in Section H.3.3.
2.2 5.7
Yes
Fire Height,
, relative
to the Ceiling
Height,
H.4 Estimation of Fire-Generated Conditions

This section provides specific details on how each model is set up and run.
H.4.1 Algebraic Models

The FDTs contain several correlations for estimating the heat flux at a fixed distance from a fire:
05.1_Heat_Flux_Calculations_Wind_Free.xls (Point Source)
05.1_Heat_Flux_Calculations_Wind_Free.xls (Solid Flame 1)
For the point source method, the estimated peak HRR, , is 945 kW, the radiative fraction, ,
is 0.49, and the horizontal distance from the center of the burning tray to the edge of the target
tray, , is 2 m (6.6 ft). The calculated heat flux is:
ps
0.49
4
945 kW
2.0 m
9.2 kW/m
(H-3)
The solid flame calculation is based on the fire having a roughly circular base with which it
employs a flame height correlation to estimate the vertical extent of the luminous flame region.
H-7

This is not the case for the vertical and horizontal trays filled with burning cables. However, it is
possible to use a variation of the point source method to get a refined heat flux estimate that is
similar in scope to a solid flame model. Essentially, the fire can be divided into several point
sources along the length of the burning cable tray so that the radiative energy is not emanating
from just one point. Suppose that the 4.6 m (15 ft) vertical tray segment is divided into four
segments, each one 1.15 m (3.8 ft) long and 0.6 m (2 ft) wide. The horizontal tray is the fifth
segment, 1.7 m (5.6 ft) long and 0.6 m (2 ft) wide. Each segment generates energy at a rate of
250 kW/m2. This leads to 172.5 kW for each of the four vertical segments and 255 kW for the
horizontal segment. The sum is 945 kW. The distance between the center points of the
segments and a point at the intersection of the horizontal and vertical trays nearest to the
burning cable trays can be calculated, and the combined heat flux from this set of distributed
sources can be calculated:
dps
0.49 255
4
2.9
172.5
2.4
172.5
2.0
172.5
2.2
172.5 kW
2.9
m
6.2 kW/m
(H-4)
Much like the solid flame method, the distributed point source method provides a refined
estimate of the heat flux, based on the fact that the fire is distributed over the two parts of the
tray and is not concentrated at a single point. The end result is a lower estimate of the heat flux
to a given target point. The fire could be distributed over more than five points, but the answer
would not significantly change. Neither the point source nor the distributed point source method
accounts for changes in the orientation of the target. The target cables are modeled as having
a direct view of the fire with no obstructions, such as the sides of the trays, that can reduce the
heat flux to the target cables. The CFD model can provide a more refined estimate.
H.4.2 CFD Model

Geometry: Only the section of the annulus encompassing the cables and relevant targets is
included in the computational domain. This volume is 9.6 m (31.5 ft) wide, 2.5 m (8.2 ft) deep,
and 12.8 m (42 ft) high. Extra depth is needed to accommodate the slight curvature of the
bounding walls. The top, bottom, and sides of the computational domain are specified as open,
that is, open to an infinitely large volume. Since the volume of the annulus is very large, neither
smoke build-up nor pressure effects would influence the region near the cables. Both the
internal and external walls of the annulus are included in the model. Since FDS only allows
rectilinear obstructions, a series of obstructions approximately 20 cm (7.9 in) thick approximate
the curved walls. The numerical grid conforms to this stair-stepped geometry.
Fire: The fire ignites near the base of the vertical portion of the cable tray attached to the inner
wall. The spread rates of 258 mm/s (10 in/s) in the vertical direction and 0.9 mm/s (0.035 in/s)
in the horizontal are input by using an FDS feature in which a surface is designated as having a
fire spread over it at a designated rate. In this case, a surface is specified along the side of the
vertical tray and along the top of the horizontal tray with the respective spread rates. The HRR
per fire unit area is specified directly and not predicted by the model. A Smokeview rendering of
the FDS simulation is shown in Figure H-4.
Cables: One of the objectives of the calculation is to estimate the potential damage to the
cables within the redundant train. FDS is limited to only 1-D heat transfer into either a
rectangular or cylindrical obstruction. In this simulation, the cables are modeled as 1.5 cm
H-8

(5.9 in) cylinders. Following the Thermally-Induced Electrical Failure (THIEF) methodology in
NUREG/CR-6931, Vol. 3, electrical functionality is lost when the temperature just inside the
1.5 mm (0.06 in) jacket reaches 205 C (400 F). Since the objective of this calculation is to
estimate time to failure of the redundant cables, ignition and spread of the fire over the second
set of cables is not considered. The in-depth heat penetration calculation is focused on a single
cable that is relatively free of its neighbors and that would heat up more rapidly than those
buried deeper within the pile.
Sprinkler Activation: FDS uses the conventional RTI concept to predict sprinkler activation.
In this scenario, a steel plate has also been added just above the location of the sprinkler to
simulate the effect of the actual deflector. Note that the sprinkler itself is just a point in the
model, and its activation is determined by the time history of the temperature and the velocity of
the hot gases within the numerical grid cell in which the sprinkler is located.
H-9
Figure H-4. FDS/Smokeview rendering of the annulus fire scenario.
H-10
H.5 Evaluation of Results

The calculations described above assess the potential for damage to redundant safe-shutdown
cables due to a fire in an adjacent tray in the annulus region of the containment building. In
addition, a CFD calculation is used to determine whether the fire would activate a sprinkler. The
results of the calculations are summarized in Table H-3.
Table H-3. Summary of model predictions for the annulus fire scenario.
Model
Bias
Factor,
Standard
Deviation,
Predicted
Value
Critical
Value
Probability
of
Exceeding
1.42
Heat Flux (kW/m2)

0.55
9.2
6.0
0.553
1.42
0.55
6.2
6.0
0.248
6.0
0.000
205.0
0.000
100.0
0.001
Point Source
Distributed Point
Source
FDS
1.10
FDS
1.02
FDS
1.15
Target
Cables
0.17
2.5
Target Temperature (C)
0.13
Cables
120.0
0.11
Sprinkler
90.0
H.5.1 Heat Flux and Temperature

The simple point source heat flux calculations are used as part of a screening analysis to
determine whether more detailed calculations are warranted. Using the peak HRR as a
constant input, the point source model estimates the heat flux to be 9.2 kW/m2. A slightly more
refined estimate using distributed point sources decreases this estimate to 6.2 kW/m2. Both of
these estimates suggest a potential for damage to the redundant train of cables.
A more detailed calculation is performed with FDS, which yields a predicted peak heat flux of
approximately 2.5 kW/m2 (see Figure H-5). The reason that the FDS prediction of heat flux is
significantly lower than the point source method predictions is that (1) FDS accounts for the side
of the tray that blocks much of the thermal radiation, and (2) the radiation transport equation
solved by FDS accounts for the orientation of the target tray relative to the fire. In other words,
FDS does not model the target cables as directly facing the fire, whereas the point source
models implicitly do.
As with the heat flux, the predicted interior cable temperatures predicted by FDS indicate a very
low probability for cable damage. However, it should be noted that these low probabilities are
based on the particular configuration of the cables within the trays. If some of the target cables
were to have a direct view of the fire, the heat flux would more likely be comparable to that
predicted by the distributed point source model.
H-11
Redundant Cable Heat Flux
Redundant Cable Temperature

10
150
120
Heat Flux (kW)
Temperature (C)
90
FDS
60
7
6
Point Source
Distributed Point Source
FDS
5
4
3
2
30
1
0
0
0
600
1200
1800
2400
3000
3600
600
1200
1800
2400
3000
3600
Time (s)
Time (s)
Figure H-5. Summary of simulation results for the annulus.
H.5.2 Fire Protection Systems

The sprinkler link temperature predicted by FDS is shown in Figure H-6. The sprinkler is not
predicted to activate in this scenario because the link temperature is predicted to increase to
approximately 90 C (194 F), less than the activation temperature of 100 C (212 F). It should
be noted, however, that the sprinkler is located just outside the fire plume. It is expected that for
a real fire of this type, the natural air movements within such a large space as the containment
annulus would almost certainly bend the plume from the vertical in a way that would be difficult
to replicate with a model that is not accounting for the air movements throughout the entire
facility. A slight lean in the plume towards the sprinkler could result in activation.
Sprinkler Link Temperature
120
FDS
Temperature (C)
100
80
60
40
20
0
0
600
1200
1800
2400
3000
3600
Time (s)
Figure H-6. Predicted sprinkler link temperature for the annulus fire scenario.
It is possible to determine how large a fire is needed to increase the plume temperature by
10 C. Table 4-3 indicates that the HGL temperature rise is proportional to the HRR to the 2/3
power. The plume temperature behaves similarly. Following the methodology in Section 4.4, in
H-12

order to increase the predicted plume temperature by 10 C, the peak HRR, , must increase
by approximately:
3
2
3
945 kW
2
10 C
90 C 35 C
258 kW
(H-5)
In other words, the peak HRR of the fire would have to be approximately 945 + 258 = 1203 kW
to bring the plume temperature into a range to cause sprinkler activation.
H.6 Conclusion
Simple point source heat flux calculations indicate that a fire in one of the cable trays within the
annulus region of the containment building might damage the cables in an adjacent tray.
However, an additional analysis using FDS indicates that cable damage is unlikely due to the
orientation of the target cables and the blockage of thermal radiation by the cable tray itself.
This suggests that the details of the cable tray location, orientation, and configuration can
significantly impact potential for damage.
FDS predicts that sprinkler activation above the fire is unlikely. However, its prediction is
sensitive to the exact location of the sprinkler relative to a fire plume that may be subject to
unpredictable air movements throughout the entire facility. Alternative protection strategies,
such as shielding between trays or other thermal barriers, should be considered to ensure the
protection of the redundant cables.
H.7 References
5. NUREG/CR-6931, Cable Response to Live Fire (CAROLFIRE), Volume 3: ThermallyInduced Electrical Failure (THIEF) Model, 2008.
6. NUREG/CR-7010, Cable Heat Release, Ignition, and Spread In Tray Installations during
Fire (CHRISTIFIRE), Volume 1: Horizontal Trays, 2012.
H.8 Attachments (on CD)

1. FDS input file: Annulus.fds
H-13
H-14
INDEX
Administrative controls ............................................................................................................... 2-7
Algebraic models .................................... 1-7, 1-15, 1-16, 2-10, 2-11, 2-13, 3-18, 3-21, 3-23, 4-17
ASTM E1355...................................................................................................................... 1-4, 5-1
Ceiling jet ................................................................................ 2-24, 2-30, A-9, B-8, E-8, F-8, F-17
CFAST .xvii, xviii, xxv, 1-3, 1-7, 1-15, 1-16, 2-11, 2-13, 2-18, 2-19, 2-23, 2-32, 2-33, 3-4, 3-21, 42, 4-3, 5-2, 5-3, 5-4, 6-2, A-6, A-7, A-11, A-13, A-14, A-15, A-18, A-19, A-20, A-21, A-22, B-6,
B-11, B-12, B-13, B-14, B-15, B-16, B-19, B-20, B-21, B-24, B-25, D-5, D-8, D-9, D-10, D-11,
D-12, D-14, D-15, D-16, D-17, D-18, E-6, E-11, E-12, E-13, E-14, E-15, E-16, E-17, E-19, E20, E-21, E-24, E-25, F-9
CFD model ... 1-9, 1-10, 2-11, 2-19, 2-20, 2-22, 2-30, 2-33, 3-2, 3-4, 3-10, 3-12, 3-13, 3-15, 3-16,
3-17, 3-19, 3-21, 3-23, 4-11, A-7, A-12, B-6, C-5, C-7, C-16, C-23, D-8, E-6, E-20, F-6, F-17,
H-8
Compartment aspect ratio................................................................................................ 2-30, F-8
Compartment geometry ................................................................................................... 3-2, A-13
Detection ..................................................................... 3-20, 5-4, E-2, E-16, E-18, E-20, F-4, G-15
Documentation ...................................................................................................... 2-14, 2-18, 2-34
Equivalence ratio ............................................................................................................. 2-30, F-8
Fire Dynamics Simulator (FDS) . xvii, 1-3, 1-11, 2-11, 2-19, 4-3, 5-3, A-6, B-6, C-6, D-5, E-6, H-6
Fire Froude number ................................................................................................................. 2-25
Fire location ............................................................................................................................... 3-2
Fire PRA ... xvii, xix, xxiii, 1-1, 1-3, 1-14, 1-15, 2-5, 2-10, 4-10, 4-11, 5-4, A-22, B-24, C-24, D-17,
E-25, G-17, H-13
Heat release rate ................................................................................................. 2-17, B-20, C-10
Hot Gas Layer (HGL) ....................................................................................................... 2-6, 2-23
Intervening combustibles ........................................................................................................... 2-9
Limiting Fire Scenario (LFS) .................................................................................................... 4-15
MAGIC .. xvii, xxiii, 1-3, 1-7, 1-16, 2-11, 2-13, 2-19, 2-23, 2-33, 3-21, 4-2, 4-3, 4-9, 4-10, 5-1, 5-3,
A-6, C-3, C-5, C-7, C-8, C-9, C-10, C-11, C-12, C-13, C-14, C-15, C-16, C-17, C-18, C-19, C20, C-21, C-23, C-24, D-5, G-5, G-7, G-9, G-10, G-11, G-12, G-13, G-14, G-15, G-16, G-17
Main control room .................................................................................................. xxvi, 1-15, 3-18
Maximum Expected Fire Scenario (MEFS) .............................................................................. 4-15
Model uncertainty....................................................................................................................... 4-1
NFPA 805 ................................................ iii, 1-1, 1-11, 1-12, 1-13, 1-14, 2-4, 2-22, 4-2, 4-15, 5-2
Normalized parameter .......................................................2-22, A-9, B-8, C-6, D-7, E-8, G-6, H-7
Parameter uncertainty ...................................................................................................... 4-1, 4-10
Plume .......................................................... 2-15, 2-17, 2-23, 3-8, 4-2, 4-7, B-10, E-7, E-10, H-11
Secondary combustibles ............................................................................................................ 2-9
Sensitivity analysis ................................................................................................................... 4-10
Smoke production ...................................................................................................................... 1-8
Smokeview.1-10, 1-11, 2-19, 5-1, A-15, A-16, B-12, B-18, C-14, D-9, D-12, D-13, E-12, E-17, H8
Structural steel .......................................................................................................................... F-4
Suppression ...................................................................................................................... E-2, H-5
Index-1
INDEX
Target. 1-6, 1-7, 2-8, 2-15, 2-23, 3-12, 3-13, 4-2, 4-7, 4-8, 4-14, A-8, A-9, B-8, C-6, D-7, D-14, E8, F-6, F-8, F-14, G-6, H-7, H-11
Uncertainty ................................................................... 2-33, 4-1, 4-6, 4-8, 4-10, B-22, E-23, G-14
Ventilationxxv, 1-14, 2-8, 2-14, 3-1, 3-3, 3-4, 4-14, A-1, A-9, A-14, A-16, A-18, A-22, B-1, B-8, B10, B-11, B-14, B-17, C-3, C-13, C-16, D-2, D-7, D-11, D-13, E-2, E-8, E-10, E-15, E-18, F-4,
F-14, G-1, G-6, G-11, G-13, H-7
Zone model ............................... 1-8, 2-11, 3-10, 3-13, 3-15, 3-17, 3-19, 3-21, 3-23, A-7, D-6, G-5
Index-2
NRC FORM 335
U.S. NUCLEAR REGULATORY COMMISSION
(12-2010)
NRCMD 3.7
1. REPORT NUMBER
(Assigned by NRC, Add Vol., Supp., Rev.,

and Addendum Numbers, if any.)
BIBLIOGRAPHIC DATA SHEET
NUREG-1934
(See instructions on the reverse)
2. TITLE AND SUBTITLE
3. DATE REPORT PUBLISHED

MONTH
Nuclear Power Plant Fire Modeling Analysis Guidelines (NPP FIRE MAG)
YEAR
November
2012
4. FIN OR GRANT NUMBER
5. AUTHOR(S)
6. TYPE OF REPORT
D. Stroup (US NRC), R. Wachowiak (EPRI), K. McGrattan (NIST), R. Peacock (NIST),

F. Joglar (Kleinsorg), S. LeStrange (Kleinsorg), F. Mowrer (California Polytechnic State
University), S. Hunt (Hughes), C. Worrell (Westinghouse), J. Milke (University of
Maryland), D. Birk (SAIC), B. Najafi (Kleinsorg), K. Zee (ERIN)
Technical
7. PERIOD COVERED (Inclusive Dates)
8. PERFORMING ORGANIZATION - NAME AND ADDRESS (If NRC, provide Division, Office or Region, U. S. Nuclear Regulatory Commission, and mailing address; if
contractor, provide name and mailing address.)
U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research, Washington, DC 20555-0001
Electric Power Research Institute, 3420 Hillview Avenue, Palo Alto, CA 94303
National Institute of Standards and Technology, 100 Bureau Drive, Gaithersburg, MD 20899
Kleinsorg Group Risk Services, a Division of Hughes Associates, 3610 Commerce Drive, Baltimore, MD 21227
9. SPONSORING ORGANIZATION - NAME AND ADDRESS (If NRC, type "Same as above", if contractor, provide NRC Division, Office or Region, U. S. Nuclear Regulatory
Commission, and mailing address.)
Division of Risk Analysis

Office of Nuclear Regulatory Research
Washington, DC 20555-0001

Palo Alto, CA 94303
10. SUPPLEMENTARY NOTES
M.H. Salley, NRC Project Manager
11. ABSTRACT (200 words or less)
There is a movement to introduce risk-informed and performance-based (RI/PB) analyses into fire protection engineering
practice, both domestically and worldwide. This movement exists in both the general fire protection and the nuclear
power plant (NPP) fire protection communities. The Nuclear Regulatory Commission (NRC) has used risk-informed
insights as a part of its regulatory decision making since the 1990s. In 2001, the National Fire Protection Association
(NFPA) issued NFPA 805, Performance-Based Standard for Fire Protection for Light-Water Reactor Electric Generating
Plants, 2001 Edition. In July 2004, the NRC amended its fire protection requirements in Title 10, Section 50.48 of the
Code of Federal Regulations to permit existing reactor licensees to voluntarily adopt requirements contained in
NFPA 805 as an alternative to the existing deterministic fire protection requirements. One key element in RI/PB fire
protection is the availability of verified and validated (V&V) fire models that can reliably estimate the effects of fires.
The U.S. NRC, together with the Electric Power Research Institute (EPRI) and the National Institute of Standards and
Technology (NIST), conducted a research project to verify and validate five fire models that have been used for NPP
applications. The results of this effort are documented in the seven-volumes of NUREG-1824 (EPRI 1011999),
Verification & Validation of Selected Fire Models for Nuclear Power Plant Applications. NUREG-1934 (EPRI 1023259)
describes the implications of the V&V results for fire model users. The features, limitations, and uses of the five fire
models, documented in NUREG-1824, are discussed relative to NPP fire hazard analysis applications.
12. KEY WORDS/DESCRIPTORS (List words or phrases that will assist researchers in locating the report.)
Fire, Performance-Based, Verification and Validation (V&V), Fire Hazard Analysis (FHA),
Risk-Informed Regulation, Fire Safety, Fire Protection, Nuclear Power Plant (NPP),
Probabilistic Risk Assessment (PRA), Fire Modeling
13. AVAILABILITY STATEMENT
unlimited
14. SECURITY CLASSIFICATION
(This Page)
unclassified
(This Report)
unclassified
15. NUMBER OF PAGES
16. PRICE
NRC FORM 335 (12-2010)
UNITED STATES
NUCLEAR REGULATORY COMMISSION

WASHINGTON, DC 20555-0001
-------------------OFFICIAL BUSINESS
NUREG-1934
Nuclear Power Plant Fire Modeling Analysis Guidelines (NPP FIRE MAG)
November 2012

Fire Modeling

Uploaded by

Copyright:

Available Formats

Fire Modeling

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fire Modeling

Uploaded by

Copyright:

Available Formats

NUREG-1934

Nuclear Power Plant

U.S. Nuclear Regulatory Commission

Electric Power Research Institute

AVAILABILITY OF REFERENCE MATERIALS

Non-NRC Reference Material

As of November 1999, you may electronically access

Documents available from public and special technical

NRC publications in the NUREG series, NRC

Copies of industry codes and standards used in a

Nuclear Power Plant

U.S. Nuclear Regulatory Commission

Electric Power Research Institute (EPRI)

U.S. NRC-RES Project Manager

EPRI Project Manager

DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES

3.2.7.2 Modeling Strategy........................................................................................ 3-22

B.6 Conclusion ................................................................................................................... B-24

E.4.2 Zone Model .......................................................................................................... E-11

H.4.1 Algebraic Models ................................................................................................... H-7

Characteristics of compartment fires.................................................................... 1-5

Summary of common fire model tools................................................................ 2-13

Summary of results for the turbine building fire scenarios. ................................F-14

Electric Power Research Institute (EPRI)

National Institute of Standards and

Kleinsorg Group Risk Services

California Polytechnic State University

Individual Plant Examination of External Events

smoke layer interface position above base of fire (m)

fire growth coefficient (kW/s2)

Estimates of the ranges of normalized parameters to be expected in nuclear plant

General knowledge of the behavior of compartment fires

Ability to understanding the basis of mathematical models involving algebraic and

1.3.2 Training Resources

1.4 Fire Modeling Theory

Thermophysical properties include thermal conductivity, specific heat, and density.

Figure 1-1. Characteristics of compartment fires.

Rate of smoke production

1.5 Fire Modeling Tools

1.5.2 Zone Models

1.5.3 CFD Models

Figure 1-3. A Smokeview visualization of a CFD model of a compartment fire experiment.

1.5.4 Fire Model Verification and Validation (V&V)

1.6 Fire Modeling Applications

1.6.1 License Amendments and Exemptions

1.6.2 Fire Induced Circuit Failures

1.6.2.1 Deterministic Application

1.6.2.2 Risk-Informed, Performance-Based Application

1.6.3 NFPA 805 Performance-Based Applications

1.6.3.1 Fire Modeling

The application of fire models within their range and limitations

Fire models shall be verified and validated

Existing in-situ combustibles in the fire area

Types and amounts of transient combustibles that industry experience and

Active and passive fire protection features operating as designed

1.6.3.2 Fire Risk Evaluations

1.6.4 Fire Modeling in Support of Fire PRA

1.7 Organization of the Guide

THE FIRE MODELING PROCESS

Figure 2-1. Fire modeling process.