Vous êtes sur la page 1sur 108

!

GUIDANCE NOTES ON
!"#$%&!#'($)*$+',&)!&*)!%%!&,$-$&#+)
./%).'%$)+!.$#0)
JANUARY 2004 (Updated JuIy 2010 - see next page)!
American Bureau of Shipping
Incorporated by Act of Legislature of
the State of New York 1862
Copyright 2004
American Bureau of Shipping
ABS Plaza
16855 Northchase Drive
Houston, TX 77060 USA


!
1234567)
JuIy 2010 consoIidation incIudes:
x! January 2004 version plus Corrigenda/Editorials


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445" iii!
$%&'(%&)*
Background
Fire saIety regulations can have a major impact on many aspects oI the overall design oI ships,
including design layout, aesthetics, Iunction, costs, etc. Rapid developments in modern shipbuilding
technology have oIten resulted in unconventional structures and design solutions. As the physical size
oI ships continuously increases, the complexity oI design and Iunctionality also increases. At the same
time, there have been great strides in understanding oI Iire processes and their interrelationship with
humans and ships. Advancement has been particularly rapid in the areas oI analytical Iire modeling.
Several diIIerent types oI such models, with varying degrees oI sophistication, have been developed
in recent years and are used by engineers in the design process.
The SOLAS regulation II-2/Regulation 17, 'Alternative design and arrangements, along with
supporting MSC/Circ. 1002 entered into Iorce on July 1, 2002, allows a methodology to be used Ior
alternative design and arrangements Ior Iire saIety. It essentially permits the use oI a perIormance-
based Iire engineering approach to achieve an equivalent level oI saIety to the prescriptive
requirements Ior all ship types. This approach Iocuses on the overall perIormance oI speciIic
arrangements and their ability to meet the Iire saIety objectives, enabling enhanced Ilexibility in ship
design and allowing Ior arrangements which traditionally had not been permitted within the
prescriptive Iramework.
As a result, the ship design is no longer restricted to the predeIined conditions within the regulations.
The Iire saIety measures can now be chosen to address the speciIic hazards present in each ship.
Instead oI prescribing exactly which protective measures are required, the perIormance oI the overall
system is presented against a speciIied set oI design objectives (such as stating that satisIactory escape
should be aIIected in the event oI Iire). Fire modeling and evacuation modeling can oIten be used to
assess the eIIectiveness oI the proposed protective measures.
MSC/Circ. 1002, 'Guidelines on Alternative Design and Arrangements Ior Fire SaIety, outline the
methodology Ior the engineering analysis required by SOLAS regulation II-2/17, applying to a
speciIic Iire saIety system, design or arrangements Ior which the approval oI an alternative design
deviating Irom prescriptive requirements oI SOLAS Chapter II-2 is sought.
However, in MSC/Circ. 1002, little inIormation has been provided Ior some crucial parts oI Iire saIety
analysis, Ior example, how to develop the perIormance criteria and how to select the hazard Ior
analysis. In Iact, designers and shipbuilders need the processing guidance on how to carry out the
procedures addressed in MSC/Circ. 1002.
In response to industry need, ABS has developed these Guidance Notes on Alternative Design and
Arrangements for Fire Safetv in order to assist in the understanding oI MSC/Circ. 1002. These
Guidance Notes not only encapsulate the entire contents oI MSC/Circ. 1002, but also provide
supplemental materials to Iurther explain the guidelines in MSC/Circ. 1002, in particular, in the areas
oI identiIying design Iire scenarios, developing trials, selecting design tools Ior trial alternative
designs and developing perIormance-based criteria, etc. ThereIore, these Guidance Notes provide a
practical methodology Ior the situations where an alternative design is being proposed on the premise
that it provides the equivalent level oI saIety to the SOLAS regulations.

iv !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
Contents of the Guidance Notes
These Guidance Notes Iollow the section numbering oI MSC/Circ. 1002 and provide explanatory
appendices Ior the issues addressed in those sections. The contents oI the Guidance Notes take the
Iollowing order:
Section 1: Application. The purpose and the scope oI application oI these Guidance Notes are
addressed in this Section.
Section 2: Definitions. This Section deIines the general terms used with speciIic technical meanings
in these Guidance Notes Ior the purpose oI clariIication. Some terms which are not listed in
MSC/Circ. 1002, but used in these Guidance Notes, are listed in a separate order.
Section 3: Engineering Analysis. This Section emphasizes the necessity oI Iollowing an
established approach Ior the process oI the alternative design and arrangements. This Section also
provides two examples oI established approaches, and deIines the proper phases oI process, i.e.,
preliminary analysis and quantitative analysis.
Section 4: Design Team. This Section emphasizes the qualiIication and the responsibility oI the
design team acceptable to the Administration.
Section 5: Preliminary Analysis in Qualitative Terms. The process oI preliminary analysis is
outlined in detail in this Section. The Subsections contain a great deal oI explanatory material to
supplement the original section oI MSC/Circ. 1002 regarding the deIinition oI scope, the development
oI Iire scenarios, the development oI trial alternative designs and the preliminary analysis report.
Section 6: Quantitative Analysis. The quantitative analysis is the most labor-intensive Irom a Iire
saIety engineering standpoint. It consists oI quantiIying the design Iire scenarios, developing the
perIormance criteria, veriIying the acceptability oI the selected saIety margins and evaluating the
perIormance oI trial alternative designs against the prescriptive perIormance criteria. All oI these
issues are discussed in detail in this Section, and some are Iurther discussed in the corresponding
Appendices oI these Guidance Notes.
Section 7: Documentation. This Section lists the necessary steps oI documentation Ior all design
processes.
Appendix A: Report on the Approval of Alternative Design and Arrangements for Fire Safety.
This Appendix shows the oIIicial submittal Iorm Ior the approval oI alternative design and
arrangements Ior Iire saIety.
Appendix B: Document of Approval of Alternative Design and Arrangements for Fire Safety.
This Appendix shows the oIIicial submittal Iorm Ior the documentation oI alternative design and
arrangements Ior Iire saIety.
Appendix C: Technical Reference and Resources. This Appendix emphasizes the necessary
requirements oI reliable technical reIerences and resources. Some examples oI the technical
reIerences and resources are provided in this appendix.
Appendix D: Identifying Design Fire Scenarios. This Appendix discusses the methodology to
provide minimum design Iire scenarios Ior evaluation oI alternative design and arrangements. Two
types oI the probabilistic and deterministic design techniques are reviewed.
Appendix E: Developing Trial Alternative Designs. This Appendix Iocuses on how to develop
trial alternative designs to represent Iire protection system design alternatives developed to address
design Iire scenarios to achieve the previously established perIormance requirements.

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445" v!
Appendix F: Design Tools for Trial Alternative Designs. This Appendix reviews the
Iundamentals oI Iire modeling in quantitative Iire analysis. It discusses the basics oI Iire modeling,
major assumptions, heat transIer and Iire dynamics, explicabilities and limitations oI the Iire
modeling. The main part oI this Appendix is devoted to zone modeling and Iield (CFD) modeling
techniques. The Iinal portion oI this Appendix lists some available Iire models developed Ior various
Iire protection applications.
Appendix G: Developing Performance-based Criteria. This Appendix states the eIIects oI liIe
saIety and non-liIe saIety criteria on the development oI perIormance criteria. As the consideration oI
tenability oI liIe saIety becomes increasingly important in Iire saIety design, the eIIects oI various
liIe-threatening hazards are discussed in this Appendix.
Appendix H: Example Analysis - Alternative Design and Arrangements for Containership
Cargo Spaces. This Appendix provides an example oI alternative design Ior a carbon dioxide Iire
extinguishing system in a containership cargo hold. This Appendix outlines the procedures to
complete the design process and gives step-by-step illustrations on how the equivalent level oI Iire
protection provided by alternative design can be met against the prescriptive regulations and
requirements.



!

This Page Intentionally LeIt Blank


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! vii!


GUIDANCE NOTES ON
!+,-./!,01-*2-#03/*!/2*!..!/3-4-/,#*
$5.*$0.-*#!$-,6*
CONTENTS
SECTION 1 AppIication.............................................................................. 1

SECTION 2 Definitions............................................................................... 3

SECTION 3 Engineering AnaIysis............................................................. 5
1 Process of Alternative Design and Arrangements.................5
2 Examples of Acceptable Approaches....................................5
3 Phases of Process.................................................................5

SECTION 4 Design Team........................................................................... 7
1 General Requirements...........................................................7
2 Qualifications .........................................................................7
3 Responsibility of Design Team..............................................7
4 Setting the Comparison .........................................................8

SECTION 5 PreIiminary AnaIysis to QuaIitative Terms .......................... 9
1 Definitions of Scope...............................................................9
1.1 Contents of Scope............................................................. 9
1.2 Documenting Regulations ............................................... 11
2 Development of Fire Scenarios ...........................................11
2.1 General ........................................................................... 11
3 Development of Trial Alternative Designs............................13
4 Preliminary Analysis Report.................................................13
4.1 Contents of Report .......................................................... 13
4.2 Submittal of Report ......................................................... 13

TABLE 1 Examples of Fire Safety Goals ..................................10


viii !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
SECTION 6 Quantitative AnaIysis .......................................................... 15
1 General ................................................................................15
1.1 Scope ..............................................................................15
1.2 Risk Assessment .............................................................15
2 Quantification of Design Fire Scenarios ..............................15
2.1 Choosing Models for Quantification.................................15
2.2 Developing Fire Scenarios ..............................................17
2.3 Description of Fire Scenarios ..........................................17
2.4 Consequences of Fire Scenarios ....................................17
3 Development of Performance Criteria .................................17
3.1 General............................................................................17
3.2 Performance Criteria Based Directlyn on SOLAS
Chapter -2 .....................................................................18
3.3 Performance Criteria Developed from a Commonly
Used Acceptable Prescriptive Design .............................18
3.4 Specific Performance Criteria and Safety Margins ..........19
3.5 mpact on Areas not Specifically Part of the Alternative
Design .............................................................................19
3.6 Evaluation........................................................................19
4 Evaluation of Trial Alternative Designs................................19
4.1 Process Flowchart ...........................................................19
4.2 Analysis of Trial Design...................................................19
4.3 Level of Engineering Analysis .........................................20
4.4 Final Alternative Design and Arrangements ....................20

FGURE 1 Alternative Design and Arrangements Process
Flowchart....................................................................20

SECTION 7 Documentation..................................................................... 21
1 Basic Requirements.............................................................21
1.1 Scope of the Analysis or Design......................................21
1.2 Description of Alternative Design(s) or
Arrangement(s) ...............................................................21
1.3 Results of Preliminary Analysis .......................................22
1.4 Results of Quantitative Analysis......................................22
2 Documentation of Approval .................................................23
3 Reporting and Approval Forms............................................23
3.1 Report..............................................................................23
3.2 Documentation ................................................................23
4 Reference in SOLAS Certificates ........................................23

APPENDIX 1 Report on the ApprovaI of AIternative Design and
Arrangements for Fire Safety ............................................. 25

APPENDIX 2 Document of ApprovaI of AIternative Design and
Arrangements for Fire Safety ............................................. 27


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! ix!
APPENDIX 3 TechnicaI References and Resources ............................... 29

APPENDIX 4 Identifying Design Fire Scenarios ...................................... 31
1 General ................................................................................31
2 dentifying Design Fire Scenarios........................................31
3 Probabilistic Design .............................................................32
3.1 Background..................................................................... 32
3.2 Basic Probabilistic Techniques........................................ 33
3.3 Data Required................................................................. 36
3.4 Common Mode Failures.................................................. 37
4 Deterministic Design............................................................38
4.1 Background..................................................................... 38
5 Design Fire Curves ..............................................................40
6 Design Fire Scenarios in NFPA 101....................................42

FGURE 1 Fault Tree and Gate for Case when Lower-level
Events are Dependent ...............................................35
FGURE 2 Fault Tree and Gate for Case when Lower-level
Events are ndependent.............................................36
FGURE 3 Event Tree..................................................................36
FGURE 4 Phases of a Design Fire Curve..................................41

APPENDIX 5 DeveIoping TriaI AIternative Designs................................. 43
1 General ................................................................................43
2 Functional Statements, Performance Requirements and
Trial Designs........................................................................43
3 Subsystems .........................................................................45
3.1 Fire nitiation and Development....................................... 45
3.2 Spread, Control and Management of Smoke.................. 45
3.3 Fire Detection and Alarm ................................................ 46
3.4 Automatic Fire Suppression Systems ............................. 46
3.5 Human Behavior and Egress .......................................... 47
3.6 Passive Fire Protection ................................................... 47
4 Fire Safety Concept Tree (FSCT)........................................48
5 References...........................................................................52

TABLE 1 Examples of Objectives, Functional Statements and
Performance Requirements.......................................44

FGURE 1 Top Gate of FSCT......................................................48
FGURE 2 Prevent Fire gnition Branch of FSCT........................49
FGURE 3 Logic Symbols Used in FSCT....................................49
FGURE 4 Major Branch of Manage Fire mpact ........................49
FGURE 5 Manage Fire Branch of FSCT....................................50
FGURE 6 Manage Exposed Branch of FSCT............................51

x !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
FGURE 7 Fire Prevent in a Computer Facility............................51
FGURE 8 Administration Action Guide.......................................52

APPENDIX 6 Design TooIs for TriaI AIternative Designs ....................... 53
1 ntroduction ..........................................................................53
2 Zone Models ........................................................................54
2.1 Basic Concepts ...............................................................54
2.2 Model Assumptions .........................................................55
2.3 Fires ................................................................................55
2.4 Heat Transfer ..................................................................55
2.5 Vent Flow ........................................................................56
2.6 Plumes and Layers..........................................................56
2.7 Species Concentrations and Depositions........................56
2.8 Predictive Equations........................................................57
2.9 Limitations of Zone models..............................................58
2.10 Current Available Models ................................................58
2.11 Model Selection...............................................................60
3 Field Models.........................................................................61
3.1 Basic Concepts ...............................................................61
3.2 Model Requirements .......................................................61
3.3 Boundary and nitial Conditions.......................................64
3.4 Current Available Models ................................................65
3.5 Limitations of Field Models..............................................66
3.6 Comparisons between Field Models and Zone
Models.............................................................................66
4 Other Special-purpose Programs ........................................67
4.1 Egress Models.................................................................67
4.2 Smoke Control Models ....................................................68
4.3 Fire Endurance Models ...................................................68
4.4 Fire Detection Models .....................................................70
4.5 Fire Suppression Models.................................................70
4.6 Fire Models from NST ....................................................70
5 References...........................................................................71

FGURE 1 Control Volumes Selected in Zone Modeling.............54

APPENDIX 7 DeveIoping Performance-based Criteria ........................... 73
1 Effects of Life Safety Criteria ...............................................73
1.1 Effects of Toxicity ............................................................73
1.2 Effects of Smoke .............................................................74
1.3 Effects of Radiant Heat....................................................74
1.4 Egress Analysis...............................................................74
1.5 Effects of Fire Extinguishing Agents on Occupants.........74
2 Establishing Performance-based Criteria for Life Safety.....75
2.1 Heat.................................................................................75
2.2 Visibility ...........................................................................75

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! xi!
2.3 Egress............................................................................. 75
2.4 Toxicity............................................................................ 76
2.5 Performance-based Criteria............................................ 76
3 Non-life Safety Criteria.........................................................76
3.1 Thermal Effects............................................................... 76
3.2 Fire Spreads ................................................................... 76
3.3 Smoke Damage .............................................................. 76
3.4 Fire Barrier Damage and Structural ntegrity................... 77
3.5 Damage to Exposed Properties ...................................... 77
3.6 Damage to the Environment ........................................... 77
4 References...........................................................................77

APPENDIX 8 AIternative Design and ExampIe AnaIysis -
Arrangements for Containership
Cargo Spaces ....................................................................... 79
ntroduction .......................................................................................79
Goals of Design ................................................................................79
Objectives of Design.........................................................................80
Rule Requirements of Carbon Dioxide Systems ..............................80
Performance Criteria.........................................................................81
Trial Alternative Design and Arrangements......................................81
dentification of Fire Hazards ............................................................82
1. Sources of hazards: ........................................................ 82
2. Consequence.................................................................. 82
3. Mitigation......................................................................... 82
Specification of Design Fire Scenarios.............................................82
Characteristics of fires................................................................... 82
Geometry....................................................................................... 83
Quantity of Carbon Dioxide Extinguishing Media .............................85
a. SOLAS Requirement....................................................... 85
b. MO FP 47/WP.6.2.......................................................... 85
Discharge Simulation........................................................................85
Discharge Rates ...............................................................................85
Computational Tool ...........................................................................86
nput Parameters ..............................................................................86
Quantitative Results..........................................................................86
Average Volumetric Concentrations of Carbon Dioxide ...................87
Evaluation with Fires nside the Cargo Hold.....................................90
Leakage of CO2 through Openings during Discharge......................91
Minimum Amount of Carbon Dioxide in Multiple Discharging
Systems ...............................................................................93
Sensitivity, Uncertainty and Limitations of Numerical Modeling .......94
Conclusions ......................................................................................94
References........................................................................................95


xii !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
FGURE 1 Cargo Hold Fully Loaded with Containers .................84
FGURE 2 Hatch Cover Opening and CO2 Nozzles at Tops......84
FGURE 3 CO
2
Percent Concentrations in a Fully Loaded
Cargo .........................................................................87
FGURE 4 CO
2
Percent Concentrations in Effective Spaces......88
FGURE 5 CO
2
Percent at 120 Seconds in sometric View in
Performance-based Design (test # Fullbb) ................88
FGURE 6 CO
2
Percent at 120 Seconds in Aft View in
Performance-based Design (test # Fullbb) ................89
FGURE 7 Detail Flow at the Corners of Containers...................89
FGURE 8 Temperature at Corners of Container Fire.................90
FGURE 9 CO
2
Concentration in a Design Fire Scenario ...........91
FGURE 10 Loss of CO
2
................................................................92
FGURE 11 Loss Rates of CO
2
.....................................................92
FGURE 12 Minimum CO
2
Percent Concentrations......................93



!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 1!

S E C T O N 7** !889:;<=:%>*
7* *
These Guidance Notes are based on MSC/Circ. 1002, 'Guidelines on Alternative Design and
Arrangements for Fire Safetv. They incorporate the entire contents oI MSC/Circ. 1002, and are
developed Ior providing additional explanatory materials and a workable example Ior Iire saIety
design and analysis.
?* *
These Guidance Notes are intended to outline the methodology Ior the engineering analysis required
by SOLAS regulation II-2/17, 'Alternative design and arrangements, applying to a speciIic Iire
saIety system, design or arrangements Ior which the approval oI an alternative design deviating Irom
the prescriptive requirements oI SOLAS chapter II-2 is sought.
@* *
These Guidance Notes are not intended to be applied to the type approval oI individual materials and
components.
A* *
These Guidance Notes are not intended to serve as a stand-alone document, but should be used in
conjunction with the Iire saIety engineering design guides and other literature, examples oI which are
reIerenced in Section 3 oI these Guidance Notes.
B* *
For the application oI these Guidance Notes to be successIul, all interested parties, including the
Administration or its designated representative, Owners, operators, designers and classiIication
societies, should be in continuous communication Irom the onset oI a speciIic proposal to utilize these
guidelines. This approach usually requires signiIicantly more time in calculation and documentation
than a typical regulatory prescribed design because oI increased engineering rigor. The potential
beneIits include more options, cost eIIective designs Ior unique applications and an improved
knowledge oI loss potential.


!

This Page Intentionally LeIt Blank


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 3!

S E C T O N ?** 2'C:>:=:%>D*

For the purposes oI these Guidance Notes, the Iollowing deIinitions apply:
i) Alternative Design and Arrangements means Iire saIety measures which deviate Irom the
prescriptive requirement(s) oI SOLAS chapter II-2, but are suitable to satisIy the Iire saIety
objective(s) and the Iunctional requirements oI that chapter. The term includes a wide range
oI measures, including alternative shipboard structures and systems based on novel or unique
designs, as well as traditional shipboard structures and systems that are installed in alternative
arrangements or conIigurations.
ii) Design Fire means an engineering description oI the development and spread oI Iire Ior use in
a design Iire scenario. Design Iire curves may be described in terms oI heat release rate versus
time.
iii) Design Fire Scenario means a set oI conditions that deIines the Iire development and the
spread oI Iire within and through vessel space(s) and describes Iactors such as ventilation
conditions, ignition sources, arrangement and quantity oI combustible materials and Iire load
accounting Ior the eIIects oI Iire detection, Iire protection, Iire control and suppression and
Iire mitigation measures.
iv) Functional Requirements explain in general terms what Iunction the vessel should provide to
meet the Iire saIety objectives oI SOLAS.
v) Performance Criteria are measurable quantities stated in engineering terms to be used to
judge the adequacy oI trial designs.
vi) Prescriptive-based Design or Prescriptive Design means a design oI Iire saIety measures
which comply with the prescriptive regulatory requirements set out in parts B, C, D, E or G oI
SOLAS chapter II-2.
vii) Safetv Margin means adjustments made to compensate Ior uncertainties in the methods and
assumptions used to evaluate the alternative design, e.g., in the determination oI perIormance
criteria or in the engineering models used to assess the consequences oI Iire.
viii) Sensitivitv Analvsis means an analysis to determine the eIIect oI changes in individual input
parameters on the results oI a given model or calculation method.
ix) SOLAS means the International Convention Ior the SaIety oI LiIe at Sea, 1974, as amended.
In additional to the above deIinitions, the Iollowing deIinitions shall also apply:
x) Crew Member means any person onboard a vessel, including the Master, who is not a
passenger.
xi) Deterministic Analvsis means a methodology based on physical relationships derived Irom
scientiIic theories and empirical results that Ior a given set oI initial conditions will always
produce the same results oI prediction. In a deterministic analysis, a single set oI input data
will determine a speciIic set oI output predictions.




Section 2 Definitions

4 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
xii) Fire Model is a physical or mathematical procedure that incorporates engineering and
scientiIic principles in the analysis oI Iire and Iire eIIects to simulate or predict Iire
characteristics and conditions oI the Iire environment.
xiii) Fire Safetv Obfectives mean the descriptions oI the perIormance benchmarks in SOLAS
Chapter II-2/Regulation 2 against which the predicted perIormance oI a design is evaluated.
xiv) Ha:ard means a possible source oI danger that can initiate or cause undesirable consequences
iI uncontrolled.
xv) Model Evaluation means the process oI quantiIying the accuracy oI chosen results Irom a
model when applied Ior a speciIic use.
xvi) Model Jalidation means the process oI determining the correctness oI the assumptions and
governing equations implemented in a model when applied to the entire class oI problems
addressed by the model.
xvii) Model Jerification means the process oI determining the correctness oI the solution oI the
system oI governing equations in a model. With this deIinition, veriIication does not imply
the solution oI the correct set oI governing equations, only that the given set oI equations is
solved correctly.
xviii) Passenger is every person other than the Master and the members oI the crew or other persons
employed or engaged in any capacity onboard a vessel Ior the business oI that vessel.
xix) Probabilitv means the likelihood that a given event will occur. Statistically, this is the number
oI actual occurrences oI a speciIic event divided by the total number oI possible occurrences.
Probabilities are inherently dimensionless and expressed as a number between zero and one,
inclusive.
xx) Probabilistic Analvsis means an evaluation oI the Iire losses and Iire consequences, which
includes consideration oI the likelihood oI diIIerent Iire scenarios and the inputs that deIine
those Iire scenarios.
xxi) Risk means the product oI the potential consequences and the expected Irequency oI
occurrence in the classic engineering sense. Consequences might include occupant death,
monetary loss, business interruption, or environmental damage. The Irequency oI occurrence
could be an estimate oI how oIten the projected loss might occur.
xxii) Stakeholder means the one who has a share or an interest in an enterprise, speciIically, an
individual (or a representative) interested in the successIul completion oI a project. Reasons
Ior having an interest in the successIul completion oI a project might be Iinancial or saIety
related.
xxiii) Trial Design means a Iire protection system design that is intended to achieve the stated Iire
saIety goals and that is expressed in terms that make the assessment oI these achievements
possible.


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 5!

S E C T O N @** ->E:>''&:>E*!><9FD:D*
7* G&%;'DD*%C*!9='&><=:H'*2'D:E>*<>)*!&&<>E'I'>=D*
The process used to show that the alternative design and arrangements provides the equivalent level oI
saIety to the prescriptive requirements oI SOLAS chapter II-2 should Iollow an established approach
to Iire saIety design. This approach should be based on sound Iire science and engineering practice
incorporating widely accepted methods, empirical data, calculations, correlations and computer
models as contained in engineering textbooks and technical literature.
?* -J<I89'D*%C*!;;'8=<K9'*!88&%<;L'D*
Two examples oI acceptable approaches to Iire saIety engineering are listed below:
i) The SFPE Engineering Guide to Performance-Based Fire Protection Analvsis and Design of
Buildings, Society oI Fire Protection Engineers and National Fire Protection Association, 1999.
ii) ISO/TR 13387-1 through 13387-8, 'Fire safetv engineering`, International Standards
Organization, 1999.
Other Iire saIety engineering approaches recognized by the Administration may be used. See
Appendix 3 Ior guidance and a list oI additional technical literature.
@* GL<D'D*%C*G&%;'DD*
The process oI the alternative design and arrangements consists oI two phases: preliminary analysis
and quantitative analysis. The objective oI the preliminary analysis is to review and agree upon the
scope oI the design proposal, identiIy potential Iire hazards, deIine perIormance criteria and speciIy
representative Iire scenarios which are suitable Ior detailed analysis and quantiIication.
The objective oI the quantitative analysis is to demonstrate, using standard tools and methodologies,
that the vessel design meets the perIormance criteria agreed to in the preliminary analysis. The
quantitative analysis should be based on both probabilistic and deterministic methods, including
engineering calculations, computer modeling, Failure Modes and EIIects Analysis, event trees and
scientiIic Iire tests. The Iollowing sections provide more detail regarding completion oI the
equivalency process and the level oI documentation that is expected Ior equivalency determinations.


!

This Page Intentionally LeIt Blank


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 7!

S E C T O N A** 2'D:E>*,'<I*
7* 3'>'&<9*.'MN:&'I'>=D*
A design team acceptable to the Administration should be established by the Owner, builder or
designer and may include, as the alternative design and arrangements demand, a representative oI the
Owner, builder or designer, and expert(s) having the necessary knowledge and experience in Iire
saIety, design, and/or operation as necessary Ior the speciIic evaluation at hand. Other members may
include marine surveyors, vessel operators, saIety engineers, equipment manuIacturers, human Iactors
experts, naval architects and marine engineers.
?* ON<9:C:;<=:%>D*
The level oI expertise that individuals should have to participate in the team may vary depending on
the complexity oI the alternative design and arrangements Ior which approval is sought. Since the
evaluation, regardless oI complexity, will have some eIIect on Iire saIety, at least one expert with
knowledge and experience in Iire saIety should be included as a member oI the team.
@* .'D8%>D:K:9:=F*%C*2'D:E>*,'<I*
The design team should:
i) Appoint a coordinator serving as the primary contact.
ii) Communicate with the Administration Ior advice on the acceptability oI the engineering
analysis oI the alternative design and arrangements throughout the entire process.
iii) Determine the saIety margin at the outset oI the design process and review and adjust it as
necessary during the analysis.
iv) Conduct a preliminary analysis to develop the conceptual design in qualitative terms. This
includes a clear deIinition oI the scope oI the alternative design and arrangements and the
regulations which aIIect the design; a clear understanding oI the objectives and Iunctional
requirements oI the regulations; the development oI Iire scenarios and trial alternative
designs. This portion oI the process is documented in the Iorm oI a report that is reviewed and
agreed upon by all interested parties and submitted to the Administration beIore the
quantitative portion oI the analysis is started.
v) Conduct a quantitative analysis to evaluate possible trial alternative designs using quantitative
engineering analysis. This consists oI the speciIication oI design Iires, development oI
perIormance criteria based upon the perIormance oI an acceptable prescriptive design and
evaluation oI the trial alternative designs against the agreed perIormance criteria. From this
step, the Iinal alternative design and arrangements are selected and the entire quantitative
analysis is documented in a report.




Section 4 Design Team

8 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
vi) Prepare documentation, speciIications and a liIe-cycle maintenance program. The alternative
design and arrangements should be clearly documented, approved by the Administration, and
a comprehensive report describing the alternative design and arrangements and required
maintenance program should be kept onboard the vessel. An operations and maintenance
manual should be developed Ior this purpose. The manual should include an outline oI the
design conditions that should be maintained over the liIe oI the vessel to ensure compliance
with the approved design.
A* #'==:>E*=L'*P%I8<&:D%>*
The Iire saIety objectives in SOLAS regulation II-2/2 and the purpose statements listed at the
beginning oI each individual regulation in chapter II-2 should be used to provide the basis Ior
comparison oI the alternative design and arrangements to the prescriptive regulations.


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 9!

S E C T O N B** G&'9:I:><&F*!><9FD:D*=%*ON<9:=<=:H'*
,'&ID*
7* 2'C:>:=:%>D*%C*#;%8'*
The preliminary analysis may begin with a concept review meeting between the Administration and
the design team. Depending upon the scope and the level oI innovation oI the equivalency, such
meetings may need to be undertaken at a very early stage to agree on the project's scope. Items to be
agreed upon may include a deIinition oI the project scope, level oI analysis necessary Ior this project,
and Iire saIety goals and objectives that the proposed design should meet.
Although much oI the inIormation required Ior the preliminary analysis, as described in Subsection
5/4, may not be known, the design team should be prepared to present a proposed text Ior such a
report at this concept review meeting. The purpose oI such a meeting is to achieve agreement on the
scope oI the proposed equivalency and not Ior the designer to seek out the Administration`s opinion
oI what they need to do.
1.1 Contents of Scope
1.1.1 Elements of Project Scope
The purpose oI this stage is to thoroughly deIine the boundaries oI the problem Ior the
proposed design.
The vessel, vessel system(s), component(s), space(s) and/or equipment subject to the analysis
should be thoroughly deIined. This includes the vessel or system(s) representing both the
alternative design and arrangements and the regulatory prescribed design. Depending on the
extent oI the desired deviation Irom prescriptive requirements, some oI the inIormation that
may be required includes: detailed vessel plans, drawings, equipment inIormation and
drawings, Iire test data and analysis results, vessel operating characteristics and conditions oI
operation, operating and maintenance procedures, material properties, etc. Accordingly, the
project scope deIinition must, at a minimum, contain all oI the relevant inIormation required
by Chapter 4 oI the SFPE Engineering Guide to Performance-Based Fire Protection Analvsis
and Design of Building.
This stage oI the process should be started during the earliest stages oI the design conception,
i.e., at the Iirst point at which the Owner or naval architect acknowledges that an alternative
design or arrangements will be necessary. This stage should also be the opportunity Ior the
naval architect or Owner to arrange the concept review meeting to discuss the scope oI the
alternative design or arrangements with the Administration. II a basic evaluation is required,
then it is likely that all oI the project deIinition can occur during the initial meeting.




Section 5 PreIiminary AnaIysis to QuaIitative Terms

10 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
1.1.2 Defining Goals
Once the project scope has been deIined and agreed upon, the design team should identiIy and
agree upon the Iire saIety goals and objectives. Goals are identiIied through discussions with
the stakeholders and a review oI background materials. The Iollowing list presents Iour
interrelated Iundamental goals Ior Iire saIety:
i) Providing liIe saIety Ior passengers and vessel crews. Minimize Iire-related injuries
and prevent undue loss oI liIe.
ii) Protecting property. Minimize damage to the property Irom Iire.
iii) Providing Ior continuity oI operations. Protect the organization`s ongoing mission,
product or operating capability. Minimize undue loss oI operations and business-
related revenue due to Iire-related damage.
iv) Limiting the environmental impact.
Chapter 5 oI the SFPE Guide contains general guidelines Ior deIining the Iire saIety goals. A
goal is normally deIined in broad terms by the stakeholders. Section 5, Table 1 provides
examples oI diIIerent goals, which the design teams should understand when conducting a
perIormance-based design.
TABLE 1
ExampIes of Fire Safety GoaIs
Fundamental goals
x Minimize Iire-related injures and prevent loss oI liIe
x Minimize Iire-related damage to the vessel, its structures and Iire
integrity
x Minimize loss oI vessel operations and business-related revenue
to Iire-related damage
x Limit the environmental impact oI the Iire and Iire protection
measures
Other possible goals
x Provide suIIicient training and awareness to ensure the saIety oI
the crews and passengers
x Reduce shipbuilding cost while maintaining adequate liIe saIety
measures
x Maximize the Ilexibility oI design and innovation

While the stakeholders might share the same global goals, the engineer must understand that the
priority and relative weight might vary among stakeholders. Further diIIerences might occur when
deIining objectives and perIormance criteria.
1.1.3 Defining Objectives
Once the Iire protection goals have been established and agreed to, the Iire saIety objectives
to meet the goals must be deIined. SOLAS II-2/Regulation 2 includes a statement oI Iire
saIety objectives and the Iunctional requirements to achieve those objectives.
It is recommended that these goals and objectives be the Ioundation Ior developing the
perIormance criteria that are speciIic to the proposal oI alternative design and arrangements.



Section 5 PreIiminary AnaIysis to QuaIitative Terms

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 11!
1.2 Documenting ReguIations
The regulations aIIecting the proposed alternative design and arrangements, along with their
Iunctional requirements, should be clearly understood and documented in the preliminary analysis
report (see Subsection 5/4). This should Iorm the basis Ior the comparative analysis reIerred to in
Subsection 6/4.
?* 2'H'9%8I'>=*%C*$:&'*#;'><&:%D*
2.1 GeneraI
Fire scenarios should provide the basis Ior analysis and trial alternative design evaluation and,
thereIore, are the backbone oI the alternative design process. Proper Iire scenario development is
essential and, depending on the extent oI deviation Irom the prescribed design, may require a
signiIicant amount oI time and resources.
(For each oI the identiIied Iire hazards, a range oI Iire scenarios should be developed. The use oI
event trees is recommended to systematically determine all oI the possible Iire scenarios resulting
Irom a speciIic hazard. Because the alternative design approach is based on a comparison against the
regulatory prescribed design, the quantiIication can oIten be simpliIied. In many cases, it may only be
necessary to analyze one or two scenarios iI this will provide enough inIormation to evaluate the level
oI saIety oI the alternative design and arrangements against the agreed perIormance design. Appendix
4 provides a minimum design Iire scenario that should be considered.)
This process can be broken down into Iour areas:
i) IdentiIication oI Iire hazards
ii) Enumeration oI Iire hazards
iii) Selection oI Iire hazards
iv) SpeciIication oI design Iire scenarios
2.1.1 dentification of Fire Hazards
This step is crucial in the Iire scenario development process as well as in the entire alternative
design methodology. II a Iire hazard or incident is omitted, then it will not be considered in
the analysis and the resulting Iinal design may be inadequate. Fire hazards may be identiIied
using historical and statistical data, expert opinion and experience and hazard evaluation
procedures. There are many hazard evaluation procedures available to help identiIy the Iire
hazards including HAZOP, PHA, FMEA, 'what-iI, etc. Further details on the use oI these
procedures speciIic to identiIying possible Iire hazards is contained in other reIerences
including the SFPE Engineering Guide to Performance Based Fire Protection Analvsis and
Design of Buildings, and the SFPE Handbook of Fire Protection Engineering. As a minimum,
the Iollowing conditions and characteristics should be identiIied and considered:
i) Pre-fire situations. Vessel, platIorm, compartment, Iuel load, environmental
conditions
ii) Ignition sources. Temperature, energy, time and area oI contact with potential Iuels
iii) Initial fuels. State (solid, liquid, gas, vapor, spray), surIace area to mass ratio, rate oI
heat release
iv) Secondarv fuels. Proximity to initial Iuels, amount, distribution
v) Extension potential. Beyond compartment, structure, area (iI in open)



Section 5 PreIiminary AnaIysis to QuaIitative Terms

12 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
vi) Target locations. Note target items or areas associated with the perIormance
parameters
vii) Critical factors. Ventilation, environment, operational, time oI day, etc.
viii) Relevant statistical data. Past Iire history, probability oI Iailure, Irequency and
severity rates, etc.
More details oI characterizing design Iire scenarios can be Iound in Chapter 8 oI the SFPE
Engineering Guide to Performance-Based Fire Protection Analvsis and Design of Building.
2.1.2 Enumeration of Fire Hazards
All oI the Iire hazards identiIied above should be grouped into one oI three incident classes:
localized, major or catastrophic. A localized incident consists oI a Iire with a localized eIIect
zone, limited to a speciIic area. A major incident consists oI a Iire with a medium eIIect zone,
limited to the boundaries oI the vessel. A catastrophic incident consists oI a Iire with a large
eIIect zone, beyond the vessel and aIIecting surrounding vessels or communities. In the
majority oI cases, only localized and/or major Iire incidents need to be considered. Examples
where the catastrophic incident class may be considered would include transport and/or
oIIshore production oI petroleum products or other hazardous materials where the incident
eIIect zone is very likely to be beyond the vessel vicinity. The Iire hazards should be
tabulated Ior Iuture selection oI a certain number oI each oI the incident classes.
2.1.3 Selection of Fire Hazards
The number and type oI Iire hazards that should be selected Ior the quantitative analysis is
dependent on the complexity oI the trial alternative design and arrangements. All oI the Iire
hazards identiIied should be reviewed Ior selection oI a range oI incidents. In determining the
selection, Irequency oI occurrence does not need to be Iully quantiIied, but it can be utilized
in a qualitative sense. The selection process should identiIy a range oI incidents which cover
the largest and most probable range oI enumerated Iire hazards. Because the engineering
evaluation relies on a comparison oI the proposed alternative design and arrangements with
prescriptive designs, demonstration oI equivalent perIormance during the major incidents
should adequately demonstrate the design`s equivalence Ior all lesser incidents and provide
the commensurate level oI saIety. In selecting the Iire hazards, it is possible to lose
perspective and to begin selecting highly unlikely or inconsequential hazards. Care should be
taken to select the most appropriate incidents Ior inclusion in the selected range oI incidents.
2.1.4 Specification of Design Fire Scenarios
Based on the Iire hazards selected, the Iire scenarios to be used in the quantitative analysis
should be clearly documented. The speciIication should include a qualitative description oI
the design Iire (e.g., ignition source, Iuel Iirst ignited, location, etc.), description oI the vessel,
compartment oI origin, Iire protection systems installed, number oI occupants, physical and
mental status oI occupants and available means oI escape. The Iire scenarios should consider
possible Iuture changes to the Iire load and ventilation system in the aIIected areas. The
design Iire(s) will be characterized in more detail during the quantitative analysis Ior each
trial alternative design. Appendix 4 provides the explanatory materials Ior design Iire
scenarios, the methodology oI identiIying design Iire scenarios and some examples oI design
Iire scenarios in NFPA 101.



Section 5 PreIiminary AnaIysis to QuaIitative Terms

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 13!
@* 2'H'9%8I'>=*%C*,&:<9*!9='&><=:H'*2'D:E>D*
At this point in the analysis, one or more trial alternative designs should be developed so that they can
be compared against the developed perIormance criteria. Trial alternative designs are the Iire
protection strategies developed by the design team that are intended to meet the objectives, Iunctional
statements and perIormance requirements. The trial alternative design should also take into
consideration the importance oI human Iactors, operations and management as reIlected in part E oI
SOLAS chapter II-2. It should be recognized that well-deIined operations and management
procedures might play a large part in increasing the overall level oI saIety. Appendix 5 provides the
explanatory details to assist in developing alternative designs by understanding what the objectives
and Iunctional statements are, as well as the perIormance requirements, and looking at various
individual or combinations oI subsystems that will allow one to achieve the perIormance
requirements.
Under certain circumstances, such as designing an alternative or innovative Iire Iighting system, the
trial alternative design should be subject to a test conducted by an accredited independent third party
(i.e., test Iacility or laboratory) or the group undertaking the actual alternative design. The test could
be modeled or Iull-scaled experiment; it could be theoretical or practical. II the test is done by
computer modeling, Iurther testing may be required. Furthermore, iI a practical test is done to the pre-
agreed criteria, then the test house should publish a report and conclusions. The test results oI trial
alternative design should be documented in a report published by the above testing parties. The
purpose oI the report is to provide an open and available set oI conclusive results.
The test report should include all oI the data, wherein the test results oI the trial alternative design as
achieved or obtained are plotted or indicated in a comparative manner in conjunction with the
perIormance criteria, and the results oI which are not mathematically scaled Irom a test oI diIIerent
proportions (i.e., smaller or alternative Iire proportions rather than compartment proportions). For
example, very oIten the area or volume is scaled but the Iire is not, and then the actual test (physical)
being conducted is not indicative oI the same results since the proportions and ratios are in Iact a little
diIIerent.
A* G&'9:I:><&F*!><9FD:D*.'8%&=*
4.1 Contents of Report
A report oI the preliminary analysis should include clear documentation oI all steps taken to this
point, including identiIication oI the design team, their qualiIications, the scope oI the alternative
design analysis, the Iunctional requirements to be met, the description oI the Iire scenarios and trial
alternative designs selected Ior the quantitative analysis.
4.2 SubmittaI of Report
The preliminary analysis report should be submitted to the Administration Ior Iormal review and
agreement prior to beginning the quantitative analysis. The report may also be submitted to the port
State Ior inIormational purposes, iI the intended calling ports are known during the design stage. The
key results oI the preliminary analysis should include:
i) A secured agreement Irom all parties to the design objectives and engineering evaluation
ii) SpeciIied design Iire scenario(s) acceptable to all parties
iii) Trial alternative design(s) acceptable to all parties


!

This Page Intentionally LeIt Blank


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 15!

S E C T O N Q** ON<>=:=<=:H'*!><9FD:D*
7* 3'>'&<9*
The quantitative analysis is the most labor-intensive Irom a Iire saIety engineering standpoint. It
consists oI quantiIying the design Iire scenarios, developing the perIormance criteria, veriIying the
acceptability oI the selected saIety margins and evaluating the perIormance oI trial alternative designs
against the prescriptive perIormance criteria.
1.1 Scope
The quantiIication oI the design Iire scenarios may include calculating the eIIects oI Iire detection,
alarm and suppression methods, generating time lines Irom initiation oI the Iire until control or
evacuation, and estimating consequences in terms oI Iire growth rate, heat Iluxes, heat release rates,
Ilame heights, smoke and toxic gas generation, etc. This inIormation will then be utilized to evaluate
the trial alternative designs selected during the preliminary analysis.
1.2 Risk Assessment
Risk assessment may play an important role in this process. It should be recognized that risk cannot
ever be completely eliminated. Throughout the entire perIormance-based design process, this Iact
should be kept in mind. The purpose oI perIormance design is not to build a Iail-saIe design, but to
speciIy a design with reasonable conIidence that it will perIorm its intended Iunction(s) when
necessary and in a manner equivalent to or better than the prescriptive Iire saIety requirements oI
SOLAS chapter II-2.
?* ON<>=:C:;<=:%>*%C*2'D:E>*$:&'*#;'><&:%D*
2.1 Choosing ModeIs for Quantification
AIter choosing an appropriate range oI Iire incidents, quantiIication oI the Iires should be
accomplished Ior each oI the incidents. QuantiIication will require speciIication oI all Iactors that may
aIIect the type and extent oI the Iire hazard. The Iire scenarios should consider possible Iuture
changes to the Iire load and ventilation system in the aIIected areas. This may include calculation oI
heat release rate curves, Ilame height, length and tilt, radiant, conductive and convective heat Iluxes,
smoke production rate, pool Iire size, duration, timelines, etc. ReIerences on suggested example
correlations and models that may be oI use are listed in Appendix 3.
Models Ior quantiIication oI Iire scenarios shall be chosen based upon an appropriate evaluation and
veriIication process. The methodology Ior evaluating the predictive capability oI Iire models by which
the process can be classiIied into Iour areas oI evaluation as described in ASTM E 1355-97 is
recommended:




Section 6 Quantitative AnaIysis

16 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
2.1.1 Model and Scenario Definition
SuIIicient documentation oI calculation models, including computer soItware, is absolutely
necessary to assess the adequacy oI the scientiIic and technical basis oI the models and the
accuracy oI computational procedures. Also, adequate documentation will help prevent the
unintentional misuse oI Iire models.
Scenario deIinition provides a

complete description oI the scenarios or phenomena oI interest
in the evaluation to Iacilitate appropriate application oI the model, to aid in developing
realistic inputs Ior the model, and criteria Ior judging the results oI the evaluation.
2.1.2 Theoretical Basis for the Model
An independent review oI the underlying physics and

chemistry inherent in a model ensures
appropriate application oI sub-models which have been combined to produce the overall
model. This review should include the assessment oI the completeness oI the documentation,
particularly with regard to the assumptions and approximations, and the assessment oI
whether there is suIIicient scientiIic evidence in the open scientiIic literature to justiIy the
approaches and assumptions being used. Empirical or reIerence data used Ior constants and
deIault values in the code should also be assessed Ior accuracy and applicability in the context
oI the model.
2.1.3 Mathematical and Numerical Robustness
The computer implementation oI the model should be checked to ensure such implementation
matches the stated documentation. The analyses which can be perIormed include analytical
tests, code checking and numerical tests.
Many Iire problems involve the interaction oI diIIerent physical processes, such as the
chemical or thermal processes and the mechanical response. Time scales associated with the
processes may be substantially diIIerent, which can easily cause numerical diIIiculties. Such
problems are called stiII. Some numerical methods have diIIiculty with stiII problems since
they slavishly Iollow the rapid changes even when they are less important than the general
trend in the solution. Special algorithms should be devised Ior those cases with stiII problems.
2.1.4 Model Uncertainty and Accuracy of the Model
2.1.4(a) Model Uncertaintv. Even deterministic models rely on inputs oIten based on
experimental measurements, empirical correlations or estimates, made by engineering
judgment. Uncertainties in the model inputs can lead to corresponding uncertainties in the
model outputs. Sensitivity analysis is used to quantiIy these uncertainties in the model outputs
based upon known or estimated uncertainties in model inputs. The purpose oI conducting a
sensitivity analysis is to assess the extent to which uncertainty in the model inputs is
maniIested to become uncertainty in the result oI interest Irom the model.
2.1.4(b) Experimental Uncertaintv. In general, the result oI measurement is only the result oI
an approximation or estimate oI the speciIic quantity subject to measurement, and thus the
result is complete only when accompanied by a quantitative statement oI uncertainty.
2.1.4(c) Model Evaluation. A model should be assessed Ior a speciIic use in terms oI its
quantitative ability to predict outcomes such as Iire growth and spread, rate oI Ilame spread,
Iire resistance, Iire hazard typiIied by available egress time, tenability, response oI active and
passive Iire protection, and some other property damages, etc. As a result, a model may be
evaluated by comparisons with standard tests, Iull-scaled tests conducted speciIically Ior the
chosen evaluation, previously published Iull-scaled test data, documented Iire experience,
proven benchmark models, etc.



Section 6 Quantitative AnaIysis

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 17!
Obtaining accurate estimates oI Iire behavior using predictive Iire models involves ensuring
correct model inputs appropriate to the scenarios to be modeled, correct selection oI a model
appropriate to the scenarios to be modeled, correct calculations by the model chosen and
correct interpretation oI the results oI the model calculation. Evaluation oI a speciIic scenario
with diIIerent levels oI knowledge oI the expected results oI the calculation addresses these
multiple sources oI potential error.
2.1.5 Choosing Tools for Trial Design Scenarios
It should be noted that when using any oI these or other tools, the limitations and assumptions
oI these models should be well understood and documented. This becomes very important
when deciding on and applying saIety margins. Documentation oI the alternative design
should explicitly identiIy the Iire models used in the analysis and their applicability.
ReIerence to the literature alone should not be considered as adequate documentation. A
summary oI the deterministic Iire models is provided in Appendix 6. The general procedure
Ior speciIying design Iires includes Iire scenario development completed during the
preliminary analysis, timeline analysis and consequence estimation, which are detailed below.
2.2 DeveIoping Fire Scenarios
For each oI the identiIied Iire hazards, a range oI Iire scenarios should be developed. Because the
alternative design approach is based on a comparison against the regulatory prescribed design, the
quantiIication can oIten be simpliIied. In many cases, it may only be necessary to analyze one or two
scenarios iI this provides enough inIormation to evaluate the level oI saIety oI the alternative design
and arrangements against the required prescriptive design.
2.3 Description of Fire Scenarios
A timeline should be developed Ior each oI the Iire scenarios beginning with Iire initiation. Timelines
should include one or more oI the Iollowing: ignition, established burning, Iire detection, Iire alarm,
Iire suppression/control system activation, personnel response, Iire control, escape times (to assembly
stations, evacuation stations and liIeboats, as necessary), manual Iire response, untenable conditions,
etc. The timeline should include Iire size throughout the scenario, as determined by using the various
correlations, models and Iire data Irom the literature or actual Iire tests.
2.4 Consequences of Fire Scenarios
Consequences oI various Iire scenarios should be quantiIied in Iire engineering terms. This can be
accomplished by using existing correlations and calculation procedures Ior determining Iire
characteristics such as heat release rate curves, Ilame height, length and tilt, radiant, conductive and
convective heat Iluxes, etc. In certain cases, live Iire testing and experimentation may be necessary to
properly predict the Iire characteristics. Regardless oI the calculation procedures utilized, a sensitivity
analysis should be conducted to determine the eIIects oI the uncertainties and limitations oI the input
parameters.
@* 2'H'9%8I'>=*%C*G'&C%&I<>;'*P&:='&:<*
3.1 GeneraI
PerIormance criteria are quantitative expressions oI the Iire saIety objectives and Iunctional
requirements oI the SOLAS regulations. The required perIormance oI the trial alternative designs is
speciIied numerically in the Iorm oI perIormance criteria. PerIormance criteria may include tenability
limits such as smoke obscuration, temperature, height oI the smoke and hot gas layer in a
compartment, evacuation time or other criteria necessary to ensure successIul alternative design and
arrangements.



Section 6 Quantitative AnaIysis

18 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
3.2 Performance Criteria Based DirectIy on SOLAS Chapter II-2
Each oI the regulations in SOLAS chapter II-2 state the purpose oI the regulation and the Iunctional
requirements that the regulation meets. Compliance with the prescriptive regulations is one way to
meet the stated Iunctional requirements. In some cases, the perIormance criteria Ior the alternative
design and arrangements should be determined by a direct interpretation oI the regulations, taking into
consideration the Iire saIety objectives, the purpose statements and the Iunctional requirements oI the
regulations. The Iollowing example is an illustration oI this:
'Example oI a perIormance criterion drawn directly Irom the regulations in SOLAS chapter II-2:
Assume that a design team is developing performance criteria for preventing fire spread
through a bulkhead separating a gallev from an accommodation space. Thev are seeking a
numerical form for this criteria.
(e.1) Regulation II-2/2 contains the fire safetv obfective 'to contain, control, and suppress
fire and explosion in their compartment of origin.`
(e.2) One of the functional requirements in which this obfective is manifest is 'separation
of accommodation spaces from the remainder of the ship bv thermal and structural
boundaries.`
(e.3) Regulation II-2/9 contains the prescriptive requirements to achieve this functional
requirement, in particular it requires an "A-60" class boundarv between areas of high fire
risk (like a machinerv space or gallev) and accommodation spaces.
(e.4) Regulation II-2/3 contains the definition of an "A" class division, which includes the
maximum temperature rise criteria of 180C at anv one point, after a 60 minute fire exposure.
(e.5) Therefore, one possible performance criterion for this analvsis is that 'no point on
the other side of the bulkhead shall rise more than 180C above ambient temperature during
a 60 minute fire exposure.`
3.3 Performance Criteria DeveIoped from a CommonIy Used AcceptabIe
Prescriptive Design
II the perIormance criteria Ior the alternative design and arrangements cannot be determined directly
Irom the prescriptive regulations because oI novel or unique Ieatures, they may be developed Irom a
quantitative evaluation oI the intended perIormance oI a commonly used acceptable prescriptive
design, provided that an equivalent level oI Iire saIety is maintained. This is a useIul method Ior
developing perIormance criteria where it is diIIicult to quantiIy the desired perIormance in terms oI
absolute values. By stating the perIormance criteria in terms oI the perIormance oI a regulatory
design, it can be inherently assumed that the Ieatures incorporated in the regulatory design provide an
overall acceptable level oI saIety.
Comparative perIormance criteria should be speciIied in terms oI a comparison to a similar
prescriptive design. Further, since the alternative design will invariably have some diIIerence Irom the
regulatory design, the criteria should also address cases when the regulatory design does not provide a
suIIicient level oI perIormance. As an example: 'The alternative design shall provide the lesser of.
the evacuation time for a vessel built to the requirements of SOLAS II-2/13, or the minimum
evacuation time required bv the assumed design fires for the proposed design`. When the
perIormance criteria are determined through comparative analysis, it will be necessary Ior the design
team to quantiIy the perIormance oI both the alternative design and the regulatory prescribed design
during the quantitative analysis.



Section 6 Quantitative AnaIysis

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 19!
3.4 Specific Performance Criteria and Safety Margins
BeIore evaluating the prescriptive design, the design team should agree on what speciIic perIormance
criteria and saIety margins should be established. Depending on the prescriptive requirements to
which the approval oI alternative design or arrangements is sought, these perIormance criteria could
Iall within one or more oI the Iollowing areas:
3.4.1 Life Safety Criteria
These criteria address the survivability oI passengers and crew and may represent the eIIects
oI heat, smoke, toxicity, reduced visibility and evacuation time.
3.4.2 Criteria for Damage to Vessel Structure and Related Systems
These criteria address the impact that Iire and its eIIluents might have on the vessel structure,
mechanical systems, electrical systems, Iire protection systems, evacuation systems,
propulsion and maneuverability, etc. These criteria may represent thermal eIIects, Iire spread,
smoke damage, Iire barrier damage, degradation oI structural integrity, etc.
3.4.3 Criteria for Damage to the Environment
These criteria address the impact oI heat, smoke and released pollutants on the atmosphere
and marine environment.
Appendix 7 provides a description oI the eIIects oI liIe and non-liIe saIety criteria.
3.5 Impact on Areas not SpecificaIIy Part of the AIternative Design
The design team should consider the impact that one particular perIormance criterion might have on
other areas that might not be speciIically part oI the alternative design. For example, the Iailure oI a
Iire barrier may not only aIIect the liIe saIety oI passengers and crew in the adjacent space, but it may
result in structural Iailure, exposure oI essential equipment to heat and smoke and the involvement oI
additional Iuel in the Iire.
3.6 EvaIuation
Once all oI the perIormance criteria have been established, the design team can then proceed with the
evaluation oI the trial alternative designs (see Subsection 6/4).
A* -H<9N<=:%>*%C*,&:<9*!9='&><=:H'*2'D:E>D*
4.1 Process FIowchart
All oI the data and inIormation generated during the preliminary analysis and speciIication oI design
Iires should serve as input to the evaluation process. The evaluation process may diIIer depending on
the level oI evaluation necessary (based on the scope deIined during the preliminary analysis), but
should generally Iollow the process illustrated in Section 6, Figure 1.
4.2 AnaIysis of TriaI Design
Each selected trial alternative design should be analyzed against the selected design Iire scenarios to
demonstrate that it meets the perIormance criteria with the agreed saIety margin, which in turn
demonstrates equivalence to the prescriptive design.



Section 6 Quantitative AnaIysis

20 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
4.3 LeveI of Engineering AnaIysis
The level oI engineering rigor required in any particular analysis will depend on the level oI analysis
required to demonstrate equivalency oI the proposed alternative design and arrangements to the
prescriptive requirements. Obviously, the more components, systems, operations and parts oI the
vessel that are aIIected by a particular alternative design, the larger the scope oI the analysis.
4.4 FinaI AIternative Design and Arrangements
The Iinal alternative design and arrangements should be selected Irom the trial alternative designs that
meet the selected perIormance criteria and saIety margins.
FIGURE 1
AIternative Design and Arrangements Process FIowchart
Preliminary
Analysis
Fire scenario
inIormation
QuantiIy prescriptive
system perIormance
QuantiIy proposed
system perIormance
Evaluate perIormance oI
prescriptive vs. alternative
PerIormance oI
proposed design
acceptable?
All scenarios
evalauted?
Select Iinal
design
Yes
Yes
No
No



!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 21!

S E C T O N R** 2%;NI'>=<=:%>*
7* "<D:;*.'MN:&'I'>=D*
Because the alternative design process may involve substantial deviation Irom the regulatory
prescribed requirements, the process should be thoroughly documented. This provides a record that
will be required iI Iuture design changes to the vessel are proposed or the vessel transIers to the Ilag
oI another State, and will also provide details and inIormation that may be adapted Ior use in Iuture
designs. The Iollowing inIormation should be provided Ior approval oI the alternative design or
arrangements:
1.1 Scope of the AnaIysis or Design
The scope oI the analysis or design might include the Iollowing:
i) Design intents (e.g., new construction, renovation or upgrade oI an existing Iacility, or repair
oI a damaged structure, etc.)
ii) Project constraints (e.g., eIIects on other vessel operations, limitations, etc.)
iii) Stakeholders (vessel owners, Ilag state, classiIication society, insurers, design and
construction team organization, etc.)
iv) Project schedules (e.g., length oI project)
v) Applicable regulations (SOLAS, ABS, etc.)
1.2 Description of AIternative Design(s) or Arrangement(s)
Description oI the alternative design(s) or arrangements(s), including drawings and speciIications:
i) Qualitative goals oI design(s) and arrangement(s), which might include:
x Protection oI liIe saIety (e.g., minimize Iire-related injuries, and prevent undue loss oI
liIe, etc.)
x Protection oI property (e.g., minimize damage to vessel structure Irom Iire and exposure
to and Irom adjacent spaces)
x Providing Ior continuity oI operations oI vessels due to Iires.
ii) Objectives oI design(s) and arrangement(s)
x Fire saIety goals and objectives agreed between the engineers and other stakeholders
should be included.
x The method by which the design objectives are developed, including any uncertainty and
saIety Iactors, should be included.
iii) All drawings oI the alternative design(s) or arrangement(s)
iv) Detailed vessel plans




Section 7 Documentation

22 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
v) Material properties
vi) Characteristics oI occupants
x Response characteristics
II a Iire-extinguishing media is used in the alternative design and arrangements, the
hazard assessment and toxic potency oI the media on occupants shall be provided Ior
approval.
x Location
x Number oI occupants
x StaII assistance
x Emergency response personnel
x OII-site condition
1.3 ResuIts of PreIiminary AnaIysis
Results oI the preliminary analysis are to include:
i) Members oI the design team (including qualiIications), a resume and other inIormation
supporting the qualiIications oI the engineer(s) perIorming the analysis should be provided
ii) Description oI the trial alternative design and arrangements being evaluated
iii) Discussion oI aIIected SOLAS chapter II-2 regulations and their Iunctional requirements
iv) Fire hazard identiIication
Description oI the procedures oI Iire hazard identiIications should be provided.
v) Enumeration oI Iire hazards
vi) Selection oI Iire hazards
vii) Description oI design Iire scenarios
1.4 ResuIts of Quantitative AnaIysis
Results oI quantitative analysis:
i) Design Iire scenarios:
x Critical assumptions
x Amount and composition oI Iire load
x Engineering judgments
x Calculation procedures
x Test data
x Sensitivity analysis
x Timelines
ii) PerIormance criteria
iii) Evaluation oI trial alternative designs against perIormance criteria
iv) Description oI Iinal alternative design and arrangements
v) Test, inspection and maintenance requirements



Section 7 Documentation

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 23!
vi) ReIerences.
x The stakeholders should be provided with suIIicient documentation to support the
validity, accuracy, relevance and precision oI the supposed methods.
x The engineering standards, calculation methods and other Iorms oI scientiIic inIormation
shall be appropriate Ior the particular application and methodologies used.
?* 2%;NI'>=<=:%>*%C*!88&%H<9*
Documentation oI approval by the Administration and the Iollowing inIormation should be
maintained onboard the vessel at all times:
i) Scope oI the analysis or design, including the critical design assumptions and critical design
Ieatures
ii) Description oI the alternative design and arrangements, including drawings and speciIications
iii) Listing oI aIIected SOLAS chapter II-2 regulations
iv) Summary oI the results oI the engineering analysis and basis Ior approval
v) Test, inspection and maintenance requirements
@* .'8%&=:>E*<>)*!88&%H<9*$%&ID*
3.1 Report
When the Administration approves alternative design and arrangements Ior Iire saIety, pertinent
technical inIormation about the approval should be summarized on the reporting Iorm given in
Appendix 1 and should be submitted to the International Maritime Organization Ior circulation to the
Member Governments.
3.2 Documentation
When the Administration approves alternative design and arrangements on Iire saIety, documentation
should be provided as indicated in Appendix 2.
A* .'C'&'>;'*:>*#5+!#*P'&=:C:;<='D*
A reIerence to the approved alternative design and arrangements should be included in the appropriate
SOLAS certiIicate.


!

This Page Intentionally LeIt Blank


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 25!

A P P E N D X 7** .'8%&=*%>*=L'*!88&%H<9*%C*
!9='&><=:H'*2'D:E>*<>)*
!&&<>E'I'>=D*C%&*$:&'*#<C'=F*

REPORT ON THE APPROVAL OF ALTERNATIVE DESIGN AND
ARRANGEMENTS FOR FIRE SAFETY
The Government oI .......... has approved on ....... an alternative design and
arrangement in accordance with provisions oI regulation II-2/17.5 oI the International Convention Ior SaIety
oI LiIe at Sea (SOLAS), 1974, as amended, as described below:

Name oI Ship .................................................................................................
Port oI registry .................................................................................................
Ship type .................................................................................................
IMO Number .................................................................................................

1. Scope oI the analysis or design, including the critical design assumptions and critical design Ieatures:
2. Description oI the alternative design and arrangements:
3. Conditions oI approval, iI any:
4. Listing oI aIIected SOLAS chapter II-2 regulations:
5. Summary oI the result oI the engineering analysis and basis Ior approval, including perIormance
criteria and design Iire scenarios:
6. Test, inspection and maintenance requirements:


!

This Page Intentionally LeIt Blank


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 27!

A P P E N D X ?** 2%;NI'>=*%C*!88&%H<9*%C*
!9='&><=:H'*2'D:E>*<>)*
!&&<>E'I'>=D*C%&*$:&'*#<C'=F*

DOCUMENT OF APPROVAL OF ALTERNATIVE DESIGN AND
ARRANGEMENTS FOR FIRE SAFETY
Issued in accordance with provisions oI regulation II-2/17.4 oI the International Convention Ior SaIety oI LiIe
at Sea (SOLAS), 1974, as amended, under the authority oI the
Government oI ............................................. by................................................................................
(name oI state) (person or organization authorized)

Name oI Ship .................................................................................................
Port oI registry .................................................................................................
Ship type .................................................................................................
IMO Number .................................................................................................
THIS IS TO CERTIFY that the Iollowing alternative design and arrangement applied to the above ship has
been approved under the provisions oI SOLAS regulation II-2/17.
1. Scope oI the analysis or design, including the critical design assumptions and critical design Ieatures:
2. Description oI the alternative design and arrangements:
3. Conditions oI approval, iI any:
4. Listing oI aIIected SOLAS chapter II-2 regulations:
5. Summary oI the result oI the engineering analysis and basis Ior approval, including perIormance
criteria and design Iire scenarios:
6. Test, inspection and maintenance requirements:
7. Drawings and speciIications oI the alternative design and arrangement:

Issued at .......................................... on................................................................................
..........................................................................................
(Signature oI authorized oIIicial issuing the certiIicate)
(Seal oI stamp oI issuing authority, as appropriate)

!

This Page Intentionally LeIt Blank


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 29!

A P P E N D X @** ,';L>:;<9*.'C'&'>;'D*<>)*
.'D%N&;'D*
7* *
Section 3 oI these Guidance Notes states that the Iire saIety engineering approach should be 'based on
sound Iire science and engineering practice incorporating widely accepted methods, empirical data,
calculations, correlations and computer models as contained in engineering textbooks and technical
literature. There are literally thousands oI technical resources that may be oI use in a particular Iire
saIety design. ThereIore, it is very important that Iire saIety engineers and other members oI the
design team determine the acceptability oI the sources and methodologies used Ior the particular
applications in which they are used.
?* *
When determining the validity oI the resources used, it is helpIul to know the process through which
the document was developed, reviewed and validated. For example, many codes and standards are
developed under an open consensus process conducted by recognized proIessional societies, code-
making organizations or governmental bodies. Other technical reIerences are subject to a peer review
process, such as many oI the available technical and engineering journals. Also, engineering
handbooks and textbooks provide widely recognized and technically solid inIormation and calculation
methods.
@* *
Additional guidance on selection oI technical reIerences and resources, along with lists oI subject-
speciIic literature, can be Iound in:
1. The SFPE Engineering Guide to Performance-Based Fire Protection Analvsis and Design of
Buildings, Society oI Fire Protection Engineers and National Fire Protection Association,
1999.
2. ISO/TR 13387-1 through 13387-8, 'Fire safetv engineering`, International Standards
Organization, 1999.




Appendix 3 TechnicaI References and Resources

30 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
A* *
Other important reIerences include:
1. SFPE Handbook of Fire Protection Engineering, 3
rd
Edition, P. J. DiNenno, ed., The Society
oI Fire Protection Engineers, Boston, MA, 2002.
2. Fire Protection Handbook, 19
th
Edition, A. E. Cote, ed., National Fire Protection
Association, Quincy, MA, 2003.
3. Custer, R.L.P., and Meacham, B.J., Introduction to Performance-Based Fire Safetv, Society oI
Fire Protection Engineers, USA, 1997.
4. NFPA 550, Guide to the Use of the Fire Safetv Concepts Tree, National Fire Protection
Association, 1995.
5. ASTM E 1355 97, Standard Guide for Evaluating the Predictive Capabilitv of
Deterministic Fire Models, American Society Ior Testing Materials, 1997.


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 31!

A P P E N D X A** 0)'>=:CF:>E*2'D:E>*$:&'*#;'><&:%D*
7* 3'>'&<9*
The purpose oI this Appendix is to provide minimum design Iire scenarios Ior evaluation oI proposed
designs. This is necessary to ensure a consistently applied 'design load (i.e., the minimum design Iire
that the vessel design should be able to withstand while meeting the perIormance objectives and
criteria).
?* 0)'>=:CF:>E*2'D:E>*$:&'*#;'><&:%D*
Design Iire scenarios are at the core oI the Iire saIety engineering methodology. The methodology is
based on analyzing particular design Iire scenarios and then drawing inIerences Irom the results with
regard to the adequacy oI the proposed Iire saIety system to meet the perIormance criteria that have
been set. IdentiIication oI the appropriate scenarios requiring analysis is crucial to the attainment oI a
vessel that IulIills the Iire saIety perIormance objectives.
Given the large number oI possible Iire scenarios Ior a given perIormance-based design project, it is
usually necessary to reduce the possible Iire scenario populations to a manageable number oI design
Iire scenarios Ior evaluating trial designs. Generally, possible Iire scenarios can be Iiltered into design
Iire scenarios using the engineer`s judgment on what Iires will bound the potential hazards. In
addition, the development oI Iire scenarios may include both probabilistic and deterministic
approaches, iI calculations are necessary.
Each design Iire scenario (which is highly speciIic to support the hazard analysis calculation) is part
oI a scenario group (which is more general to support the Irequency calculation) and is meant to be
representative oI that group. The scenario group must collectively include all potential scenarios to
take into consideration reasonableness, Irequency and severity and should cover low Irequency/high
consequence Iires, high Irequency/low consequence Iires and special challenge Iires, where
applicable.
Once design Iire scenarios have been identiIied Irom the list oI possible design Iire scenarios, then the
signiIicant aspects oI the crews, vessel and Iire characteristics Ior the selected design Iire scenarios that
will aIIect the outcome should be Iurther quantiIied. These parameters will Iunction as inputs during
the analysis stage.
OIten there are neither the resources nor data available to quantiIy every aspect oI a design Iire
scenario. The detailed analysis and quantiIication should be limited to the more signiIicant aspects.
SigniIicant aspects might include a range oI diIIerent Iire types (including smoldering Iires), Iire
growth rates, compartment ventilation rates, etc. In addition, depending on what the particular
design or analysis is intended to determine, various aspects may or may not be required to be deIined.
For instance, assessment oI a smoke management system using the clear layer method may not
need details regarding soot yields or visibility criteria to be deIined, or the details oI the decay
phase to be deIined. Various analysis methods including sensitivity and uncertainty
analyses may need to be perIormed to show which aspects are pertinent and need to be appropriately
addressed.




Appendix 4 Identifying Design Fire Scenarios

32 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
@* G&%K<K:9:D=:;*2'D:E>*
3.1 Background
3.1.1 General
Probabilistic procedures exist to quantiIy ignition, Iire growth, Ilame spread, the movement oI
combustion products, the movement oI people, the reaction to Iire and eIIect on Iire oI vessel
systems and Ieatures, and the consequences oI Iire Ior the vessel and its occupants.
These procedures are based on Iire incident and Iield survey data, as well as a variety oI
techniques Ior producing best subjective estimates. More oIten, 'probabilistic procedures use
a combination oI probabilistic methods Ior phenomena such as ignition and system reliability
with deterministic methods Ior phenomena such as Iire growth and development and eIIects
on people and property. A probabilistic design analysis involves the use oI these procedures
to calculate the perIormance oI a design in a Iorm that can be compared to probabilistic
criteria.
There are some advantages and disadvantages to probabilistic procedures vs. deterministic
procedures. At a Iundamental level, probabilistic procedures provide a basis Ior addressing
and considering all types oI Iire scenario. Deterministic procedures may mislead iI a design is
unusually vulnerable to a scenario that is:
i) Slightly less probable but much more severe than any considered in the analysis;
ii) Slightly less severe but much more probable than any considered in the analysis; or
iii) More probable and/or more severe but more unusual (e.g., in location) than any
considered in the analysis.
By the extensive use oI Iire incident and Iield survey data, probabilistic procedures are better
able to reIlect all oI the aspects oI real Iires, including the oIten complex interactions among
Iactors. Probabilistic procedures are also better adapted to quantiIy uncertainties.
Disadvantages oI probabilistic procedures include gaps in needed data that require either
expensive data collection procedures or extensive use oI subjective estimates, with associated
large uncertainties. Also, probabilistic procedures oIten lack the technical detail and the Iull
use oI Iire science Iundamentals Iound in deterministic procedures. This can make them
diIIicult to use Ior design.
3.1.2 Probabilistic Techniques
Basic probabilistic techniques oI Iault trees and event trees are brieIly described later in this
Subsection. More detailed descriptions may be Iound in a number oI reIerences, including the
SFPE Handbook for Fire Protection Engineering.
3.1.3 Fire Scenarios
As already pointed out, the interaction oI Iire, vessels and people can give rise to a very
complex system, which means a nearly inIinite number oI possible Iire scenarios. Full
analysis oI all scenarios would be impossible, so it is necessary to identiIy a manageable
group oI scenarios Ior analysis. In probabilistic techniques, these selected scenarios must be
chosen so that they collectively represent all oI the possible Iire scenarios. Each detailed
scenario is speciIic enough to permit calculation oI its consequences, or anticipated loss, but
each detailed scenario is also associated with the other scenarios that resemble it, and
probability is estimated Ior the larger set oI scenarios.



Appendix 4 Identifying Design Fire Scenarios

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 33!
The scenarios should be grouped by similar type oI hazard. A group should be deIined so that
a design Ieature that aIIects one scenario in the group will aIIect all oI the scenarios in that
group in similar Iashion. For example, Iires originating in the same or similar locations will
tend to respond to detection, suppression and compartment Ieatures in the same way, across a
wide range oI initial sizes and speeds oI growth oI the Iires. Choose the most representative or
typical Iire scenario in each group, and those will be the Iire scenarios selected Ior analysis,
with the probabilities calculated Ior the associated groups. Each scenario will be suIIiciently
diIIerent Irom the other selected scenarios as to justiIy separate assessment. Each scenario
will be speciIic enough that it can be deIined in suIIicient detail Ior quantitative evaluation.
This detailed speciIication is called a 'design Iire.
In conducting the risk assessment, it will be possible to ignore many Iactors and
characteristics oI Iires that can be shown to have negligible eIIect on probability and severity.
Some Iactors that cannot be ignored will be diIIicult to quantiIy, and Ior these it is important
to use assumptions that are neither conservative nor typical oI all vessels and passengers, but
rather that are typical oI vessels and passengers involved in Iires. Only in this way will the
resulting risk assessment properly reIlect patterns oI Iire development.
3.1.4 Limits of Application and Sensitivity Analysis
Probabilistic techniques are subject to the same limits due to experimental-scale eIIects,
uncertainties oI data extrapolation and uncertainties oI model validity and applicability, as
described Ior deterministic techniques in Appendix 4, Subsection 4.
The probabilistic models themselves can be adapted to quantiIy uncertainty by the use oI
probability distributions Ior the probabilities in the models. This approach, oIten called
Bayesian analysis, is described in greater detail in any reIerence on probabilistic modeling.
3.2 Basic ProbabiIistic Techniques
3.2.1 General
Probabilistic risk analysis begins with a deIinition oI the risk as a Iunction oI the probabilities
and consequences oI scenarios:
Risk f (probability, consequence oI a given scenario), Ior all scenarios.
There are two commonly used Iunctions deIining risk. One is the 'expected value or
average-consequence deIinition oI risk:
Risk f (probability u consequence oI a given scenario), Ior all scenarios.
The other is the probability that consequences will exceed a speciIied saIety threshold:
Risk f (probability oI a given scenario), Ior all scenarios where the
consequences exceed the
speciIied saIety threshold.
The complementary deIinition oI saIety is the inverse oI risk, i.e.,
SaIety Risk
-1

A probabilistic risk assessment, using a particular deIinition oI risk, will include the Iollowing
steps:
i) Determining what Iire scenarios can occur
ii) Dividing these Iire scenarios that can occur into groups and selecting speciIic Iire
scenarios Ior analysis Irom each group
iii) Estimating or calculating the probability oI each scenario group



Appendix 4 Identifying Design Fire Scenarios

34 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
iv) Estimating or calculating the eIIects and consequences oI each Iire scenario selected
Ior analysis
v) Calculating the total risk associated with Iire
vi) II step v) identiIies unacceptable risks, identiIying the extra measures required to
reduce that risk
Items i), iv) and vi) should be considered in detail during the design, just as in a deterministic
calculation. Items ii), iii) and v) are unique to probabilistic risk assessment, but have
analogous steps in deterministic approaches.
3.2.2 Fault Trees
Fault trees are logic diagrams showing the logical dependence oI events on one another. Fault
trees are most suitable when risk is deIined as the probability that the consequences will
exceed a certain threshold, including cases like the example where risk is deIined as the
probability oI an unacceptable event (e.g., structural collapse). The unacceptable event or,
more generally, the event oI the consequences exceeding the threshold is shown as a 'top
event deIined as Iailure, hence the name 'Iault tree and the Iault tree is constructed to
show what combinations oI events would lead to Iailure. More detailed descriptions oI Iault
trees may be Iound in a number oI reIerences listed in these Guidance Notes.
II two or more lower-level events must all occur in order Ior a higher-level event to occur, the
Iault tree uses an AND gate (see Appendix 4, Figure 1). II the lower-level events are
'independent (i.e., the probability that one will occur is unaIIected by knowledge oI whether
the other lower-level event(s) has(ve) occurred), then the probability oI the higher-level event
is equal to the product oI the probabilities oI the lower-level events.
II any one oI two or more lower-level events will lead to a higher-level event, the Iault tree
uses an OR gate (see Appendix 4, Figure 2). II the lower-level events are independent, then
the probability oI the higher-level event is equal to the sum oI the probabilities oI the lower-
level events.
The methodology may be illustrated by a compartment Iire example, in which risk is deIined
as the probability oI an unacceptable consequence, and the unacceptable consequence is
deIined as structural Iailure.
Suppose Iurther that the only Iactors capable oI preventing structural Iailure are prevention oI
ignition, restriction oI Iuel load, Iire resistance oI the structure and Iire sprinklers. Suppose
that the Iirst two are not treated as design elements, but as uncontrollable random Iactors:
(a) Did a Iire start which was capable oI reaching room burn-out?
x II no, then structural Iailure is avoided.
x II yes, continue.
(b) Were sprinklers present?
x II no, go to (e).
x II yes, continue.
(c) Were the sprinklers operational (a reliability question)?
x II no, go to (e).
x II yes, continue.



Appendix 4 Identifying Design Fire Scenarios

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 35!
(d) Was the Iire scenario one that would render sprinklers ineIIective (e.g., a large initial
explosion)?
x II no, then structural Iailure is avoided.
x II yes, continue.
(e) Was the structural Iire resistance intact (both a reliability and a scenario question)?
x II no or yes, continue.
(f) Based on the answers to the above questions, what was the critical Iuel load such that
a room burn-out would result in a Iire oI suIIicient intensity and duration as to cause
structural Iailure, and was that critical Iuel load present?
x II no, then structural Iailure is avoided.
x II yes, then structural Iailure occurs.
QuantiIication oI the analysis can be illustrated by going through the branching. The
probability oI a Iire capable oI reaching room burn-out can be estimated Irom Iire incident
data (e.g., as the probability oI a Iire in an unsprinklered enclosure having Ilame spread
beyond the room oI origin). The question oI whether sprinklers are present or not is a design
question, and the analysis should be run both ways, with yes and no answers to the question.
Reliability data will answer question (c), but it is important to include the human errors that
can render sprinklers non-operational (e.g., the Iact that the sprinkler valve had been turned
oII), as they are more common than mechanical Iailures. Question (d) can also be answered
using an estimate Irom Iire incident data. Some oI the scenarios that disable sprinklers can
also damage the structure or its Iire resistance, but question (e) will mostly be a reliability
question, depending upon workmanship and maintenance. Like question (c), it can be
answered by Iield surveys. Question (I) requires a deterministic calculation or use oI Iire tests
to determine the critical Iuel loads in each situation (e.g., the critical Iuel load with damaged
Iire resistance would be less than with intact Iire resistance). Then a Iield survey is needed to
determine the probability oI that critical Iuel load being present. The answer to each question
is a probability, and the risk Ior that scenario group is the product oI the probabilities Ior the
respective questions.

FIGURE 1
FauIt Tree and Gate for Case when
Lower-IeveI Events are Dependent
P
2
AND P
v
P
1
P
3
P
v
P
1
u P
2
u P
3




Appendix 4 Identifying Design Fire Scenarios

36 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
FIGURE 2
FauIt Tree and Gate for Case when
Lower-IeveI Events are Independent
P
2
OR P
v
P
1
P
3
P
v
P
1
P
2
P
3

3.2.3 Event Trees
Event trees (see Appendix 4, Figure 3) are diagrams showing events in time in Iire
development, movement oI people, response oI systems, etc. Event trees are most suitable
when risk is deIined as an expected value.
FIGURE 3
Event Tree
Fire
Starts
Occupant
Detection
Smoke
Detection
Heat
Detection
P
1
P
2
P
3
P
4
P
5
P
6
P
7
OUTCOME
DETECTION
DETECTION
DETECTION
NO DETECTION


3.3 Data Required
3.3.1 General
The acquisition oI reliable data can be one oI the most important tasks in perIorming any risk
assessment.
The type oI inIormation required can be broadly classiIied into Iour main groups:
i) Deterministic data
ii) Fire statistics
iii) Vessel data
iv) System reliability data



Appendix 4 Identifying Design Fire Scenarios

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 37!
3.3.2 Deterministic Data
Deterministic inIormation regarding the development and possible consequences oI Iire may
be evaluated on the basis oI the deterministic procedures (see Appendix 4, Subsection 4) and
the subsystems.
3.3.3 Fire Statistics
Fire statistics include statistics that identiIy the most likely areas oI ignition, items Iirst
ignited and the likelihood oI spread beyond the space oI Iire origin. Other statistics data,
including past Iire history, Iire Irequency and Iire ignition Irequency, can also be included.
The past history data include historical data Irom Iires in a particular existing vessel or group
oI vessels or in similar types oI equipment, contents and other items. The Iire Irequency is the
number oI times a Iire occurs within a speciIic time interval.
3.3.4 Vessel Data
Survey data such as Iractal Iire loads and occupancy levels are available. The continued
development oI a Iire and the potential consequences will depend upon a number oI Iactors
such as:
i) The availability oI combustibles and the Iractal Iire load
ii) The imposed structural loads
iii) The number oI occupants present and their condition at any given time
Where data are lacking, it is possible to make assumptions regarding occupancy, Iire load, etc.
However, the use oI reliable statistical data will assist in the perIormance oI a realistic risk
assessment.
3.3.5 System Reliability Data
All Iire protection systems may on occasion Iail Ior reasons such as lack oI maintenance,
random mechanical Iailures or inability to cope with an unusually high Iire severity.
ManuIacturers may be able to provide data on Irequencies oI mechanical or electrical Iailure
and on severity oI Iire conditions required to overpower the system. Fire incident data or other
published Iield survey statistics may be able to provide data on the Irequency oI Iire
conditions with the severity speciIied by the manuIacturers and the Irequency oI Iailure due to
human error (e.g., the Iact that the sprinkler valve had been closed).
Examples oI aspects oI automatic Iire detection and control systems Ior which reliability data
may be required are:
i) Detection system response
ii) Smoke control system operation
iii) Extinguishing system operation
iv) Breaches oI compartmentation (e.g., insuIIicient Iire stopping, doors being propped
open at time oI Iire, etc.)
3.4 Common Mode FaiIures
In some instances, the Iailure oI one part oI the system can have an adverse eIIect on the eIIiciency oI
another Iire protection measure, e.g., an open Iire door will not only be an ineIIective barrier to Iire
spread, but may also lead to Iailure oI a gaseous extinguishing system due to loss oI agent. Particular
care must be taken by the design team and those responsible to ensure that any such common mode
Iailures are identiIied and accounted Ior in the analysis.



Appendix 4 Identifying Design Fire Scenarios

38 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
A* 2'='&I:>:D=:;*2'D:E>*
4.1 Background
4.1.1 General
Deterministic procedures exist to quantiIy ignition, Iire growth, Ilame spread, the movement
oI combustion products, the movement oI people, the reaction to Iire and eIIect on Iire oI
vessel systems and Ieatures, and the consequences oI Iire Ior the vessel and its occupants.
These procedures are based on physical, chemical, thermodynamic, hydraulic, electrical or
behavioral relationships derived Irom scientiIic theories and empirical methods, or Irom
experimental research. A deterministic design analysis involves the use oI these procedures to
calculate the perIormance oI a design in a Iorm that can be compared to deterministic criteria.
4.1.2 Deterministic Techniques
In deterministic models, a complete set oI diIIerential equations based on laws oI physics and
chemistry can compute the conditions produced by Iire at a given time in a speciIied volume
oI air in a well-deIined physical scenario. Deterministic Iire models can range Irom simple
one-line correlation oI data to highly complex models. More detailed descriptions may be
Iound in a number oI reIerences, including the SFPE Handbook for Fire Protection
Engineering.
4.1.3 Fire Scenarios
The interaction oI Iire, vessel enclosures and people can give rise to a very complex system,
which means a nearly inIinite number oI possible Iire scenarios. Full analysis oI all scenarios
would be impossible, so it is necessary to identiIy a manageable group oI scenarios Ior
analysis. These selected scenarios should be chosen so that a vessel design shown to deliver
acceptable saIety Ior these scenarios can be depended upon to deliver acceptable saIety Ior all
oI the unanalyzed scenarios as well.
A deterministic design will be evaluated using a hazard assessment, which will assess
perIormance against deterministic criteria. ThereIore, in selecting scenarios, the Iirst
consideration is the type and severity oI hazard oI each scenario. For many scenarios (e.g., a
discarded cigarette on a concrete Iloor), it may be apparent without analysis that the scenario
will not produce a level oI hazard that would be unacceptable under the criteria. These
scenarios can be ignored.
Some scenarios with an unacceptably large hazard may be excluded, either because oI very
low probability or because neither their probability nor their severity can be signiIicantly
aIIected by design decisions (e.g., a thermonuclear blast). Such exclusions should be made
cautiously. To be excluded due to low probability, these scenarios must have very low
probability not only individually but also collectively. And Ior many severe scenarios (e.g., a
bomb in a parking garage in a high-rise oIIice building), loss can be signiIicantly mitigated
through design, even iI it cannot be entirely prevented.
The scenarios that remain all having suIIicient probability and severity to justiIy attention
should be grouped by similar type oI hazard. A group should be deIined so that a design
Ieature that aIIects one scenario in the group will aIIect all oI the scenarios in that group in
similar Iashion. For example, Iires originating in the same or similar locations will tend to
respond to detection, suppression and compartment Ieatures in the same way, across a wide
range oI initial sizes and speeds oI growth oI the Iires. The most severe Iire in each group
should be chosen, and those will be the 'worst credible Iire scenarios. Each scenario will be
suIIiciently diIIerent Irom the other selected scenarios as to justiIy separate assessment, in
order to make sure the design is acceptably saIe overall. Each scenario will be suIIiciently
speciIic so that it can be deIined in enough detail Ior quantitative evaluation. This detailed
speciIication is called a 'design Iire.



Appendix 4 Identifying Design Fire Scenarios

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 39!
In conducting the hazard assessment, it will be possible to ignore many Iactors and
characteristics oI Iires that can be shown to have negligible eIIect on probability and severity.
Some Iactors that cannot be ignored will be diIIicult to quantiIy, and Ior these, it is important
to use simpliIying assumptions that are conservative. However, iI too many conservative
assumptions are used, the overall assessment will be too conservative and may, in Iact, be
incompatible with any practical design. An iterative process should be used in deIining
scenarios, so that the degree oI conservatism is diminished Ior each assumption as the number
oI conservative assumptions increases. Typical current designs that are acceptable to
authorities under existing regulations should also be Iound acceptable under the hazard
assessment. ThereIore, the hazard assessment can be applied to such designs as a way oI
calibrating the necessary level oI conservatism in the assessment.
4.1.4 Limits of Application
OIten, the experimental work used to develop empirical relationships is carried out in scaled-
down Iacilities in research establishments. It is important to appreciate that the application oI
the models resulting Irom such work may be limited by the degree oI extrapolation that can be
made, e.g., in terms oI the size oI the room or the range oI Iactors that have been examined.
This must be careIully considered iI extrapolation oI test data is unavoidable.
Deterministic techniques provide a useIul indication oI the development and eIIects oI a Iire,
but the nature oI Iire is such that the results are unlikely to be precise. Normally, well-
Iormulated models would be expected to provide conservative predictions within their range
oI application.
However, in some cases there may be no Iactor oI saIety inherent within the model, and the
technique should be used with care. In all situations where there is any doubt as to the validity
oI a model, the user should establish Irom the literature how the experimental work was
carried out and decide whether the design situation is markedly diIIerent. II so, Iactors oI
saIety should be applied.
4.1.5 Sensitivity Analysis
Deterministic design may involve uncertainties. Usually, these can be dealt with by taking a
conservative approach, e.g., selecting a Iire growth rate that is Iaster than would normally be
expected. However, iI this approach is not suitable, then the primary sources oI uncertainty
should be addressed. These are associated with:
i) The input parameters, i.e., uncertainties associated with the initial qualitative
interpretation oI the problem;
ii) The simpliIication needed to develop the deterministic techniques and hence make
the analysis more tractable.
An indication oI sensitivity may be gained by investigating the response oI the output
parameters to changes in the individual input parameters. This will act as a guide to the level
oI accuracy required oI the input data.
The objective oI a sensitivity study should not be simply to check the accuracy oI the results,
but also to investigate the criticality oI individual parameters, For example, it may be
important to establish how critical a sprinkler system is to the Iinal consequences. II a single
system or assumption is shown to be critical to the overall level oI saIety achieved, such as
heat release rates as input in some Iire scenarios, consideration should be given to providing a
degree oI redundancy in the design oI carrying out a probabilistic study.



Appendix 4 Identifying Design Fire Scenarios

40 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
The simpliIications and assumptions made in the input data to aid the Iull analysis should be
tested Ior their criticality to the Iire saIety design. For example, it may have been assumed Ior
a comparative study with existing codes that a compartment remains a compartment, and that
the possibility oI an open door may be ignored. However, an alternative scenario would
include the open-door assumption. Thus, a sensitivity test on the qualitative components oI
Iire saIety design is possible.
4.1.6 Common Mode Failures
In some instances, the Iailure oI one part oI the system can have an adverse eIIect on the
eIIiciency oI another Iire protection measure: e.g., an open Iire door will not only be an
ineIIective barrier to Iire spread but may also lead to Iailure oI a gaseous extinguishing
system due to loss oI agent. Particular care must be taken by the design team to ensure that
any such common mode Iailures are identiIied and accounted Ior in the analysis.
4.1.7 Property Protection
Property protection objectives may be stated in terms oI monetary losses or spatial extent oI
damage Irom Iire and its eIIects. Monetary-loss measures are easier to use in combination
with inIormation on the costs oI design alternatives, but calculation methods and Iire tests can
only produce estimates oI spatial damage. Data on the monetary value oI property damage per
area or space damaged, by type oI damage (e.g., char, smoke deposition), are not generally
available, but will need to be developed iI calculations oI spatial damage are to be translated
into predictions oI monetary loss.
The extent oI acceptable damage is deIined by the design team Ior speciIic objects or zones,
and the calculated deterministic values Ior heat and smoke spread should not exceed these.
Predicting damage caused by IireIighting water Irom either Iire suppression systems (e.g.,
sprinklers) or the contaminations Irom Iire-Iighting activities, in either spatial or monetary
terms, is much more diIIicult than predicting or calculating damage Irom Iire and its eIIects. It
is recommended that the analysis not attempt to include such damage, as the associated
uncertainty is likely to be so large as to render the analysis results unusable.
4.1.8 Environmental Protection
The amount oI damage done to the atmosphere local to the vessel on Iire may be calculated
using a large Iire plume model capable oI predicting the trajectory and dispersion oI the Iire
gases. Contamination oI the land and ground water, however, is not easy to calculate.
The extent oI acceptable contamination oI the air, land and water will have been set Ior the
project during the design process. Calculated contamination values should not exceed the
environmental limits.
B* 2'D:E>*$:&'*PN&H'D*
Part oI the characterization oI the design Iire scenario is characterizing design Iire curves. A design
Iire curve describes the heat release rate (HRR) history oI a Iuel package. A Iuel package may
be one or more combustible items. Appendix 4, Figure 4 depicts the phases oI a design Iire curve that
may need to be typically deIined.



Appendix 4 Identifying Design Fire Scenarios

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 41!
FIGURE 4
Phases of a Design Fire Curve
Fully Developed
H
e
a
t

R
e
l
e
a
s
e

R
a
t
e
Growth
Ignition Established
Burning
Time


Although HRR is the basic input to most Iire eIIect prediction methods and Iire models and would
Iorm the basis Ior the design Iire curve, other characteristics such as mass loss rate can be used.
However, the inIormation related to HRR or mass loss rates is typically limited to an individual commodity
and/or simply arranged Iuel packages, and care should be taken in understanding where this data
came Irom (i.e., Iull-scale tests, small scale tests, theoretical derivation, etc) and how it applies to
the speciIic analysis and to what degree inIormation can be extrapolated.
Fuel loads oIten involve composite Iuel packages with various types oI combustible materials and may
be contained in a complex geometry and require careIul application oI the available data, as these
will eIIect how they burn. Frequently, however, the data will not be complete or directly
applicable to the Iuel package selected. Input data must thereIore be applied in a manner that is
consistent with the way in which it was generated. II the inIormation comes Irom actual tests, the user
should consider the applicability oI those tests to the expected scenario under the proposed design. II
the inIormation comes Irom theoretical analysis, then again the user should consider the
applicability, usually by looking at the underlying assumptions and/or test data Ior the theoretical
analysis, and determine the proper manner oI use Ior that analytical method.
Various aspects oI the Iire curve can be calculated to obtain approximations Ior predicted behavior.
Other aspects can be roughly estimated or may require subjective testing. At the present time, there is
no overall Iramework that provides exact solutions oI the entire design Iire. ThereIore, the Iire
protection engineer should determine which portions oI the design Iire curve are important, as it
may not always be necessary to quantiIy each aspect oI the design Iire curve.
In developing the design Iire curve, the engineer needs to Iocus on the intent oI the analysis, the
damage mechanisms (smoke, toxicity, thermal, corrosion), the perIormance requirements that will be
evaluated in the given design Iire scenario and the Iire characteristics oI the burning Iuel package(s) to
determine which aspects are critical. For example, iI in a perIormance-based design, the response oI an
alternative automatic Iire suppression system to standard sprinklers is being examined Ior
equivalence, the design Iire scenario might stop at the point oI activation oI the suppression system,
or at complete extinguishment. Or, the growth phase may be oI interest in detection actuation analysis,
while completion oI the Iully developed phase may be required to determine whether or not
structural Iailure will occur.



Appendix 4 Identifying Design Fire Scenarios

42 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
A thorough review oI potential and typical Iuel packages and ignition sources Ior the vessel should
thereIore be perIormed and presented to the authorities. It is oIten diIIicult to obtain speciIic
inIormation about vessel contents (i.e., outIitting, stored materials, etc.) during the design stages oI a
project. However, an attempt should be made to understand what combustibles will be in the vessel, as
well as what may be there in the Iuture. II assumptions are made regarding speciIic burning
characteristics oI materials, they should be documented and incorporated into Iinal project
speciIications.
Q* 2'D:E>*$:&'*#;'><&:%D*:>*/$G!*7S7*
NFPA 101, 'The LiIe SaIety Code (2000 Edition) deIines which types oI scenarios are to be used
and which major assumptions must be made. In all, eight scenarios are discussed where scenarios
selected as design Iire scenarios could include, but should not be limited to, those speciIied below.
Scenario 1. An occupancy-speciIic scenario representative oI a typical Iire Ior the
occupancy. The scenario shall explicitly account Ior occupant activities, number and location;
room size; Iurnishings and contents; Iuel properties and ignition sources; and ventilation
conditions. The Iirst item ignited and its location shall be explicitly deIined.
Scenario 2. An ultraIast developing Iire (i.e., Ilammable liquid Iire) in the primary means oI
egress with interior doors open at the start oI the Iire. This scenario shall address the concern
oI reducing the number oI available means oI egress.
Scenario 3. A Iire starting in a normally unoccupied room that can potentially endanger a
large number oI occupants in a large room or other area. This scenario shall address the
concern oI a Iire starting in a normally unoccupied room and migrating into a space that can,
potentially, hold the greatest number oI occupants in the vessel.
Scenario 4. A Iire originating in a concealed wall- or ceiling-space adjacent to a large
occupied room. This scenario shall address the concern oI a Iire originating in a concealed
space that does not have either a detection system or suppression system and the Iire
spreading into the room within the vessel that can, potentially, hold the greatest number oI
occupants.
Scenario 5. A slow developing Iire shielded Irom Iire protection systems, in close proximity
to a high occupancy area. This scenario shall address the concern oI a relatively small ignition
source causing a signiIicant Iire.
Scenario 6. An ultraIast developing Iire resulting Irom the largest possible Iuel load
characteristic oI the normal operation oI the vessel. This scenario shall address the concern oI
a rapidly developing Iire with occupants present.
Scenario 7. Outside exposure Iire. This scenario shall address the concern oI a Iire starting
remotely Irom the area oI concern and either spreading into the area, blocking escape Irom the
area or developing untenable conditions within the area.
Scenario 8. A Iire originating in ordinary combustibles in a room or area with each passive
or active Iire protection system independently rendered ineIIective. This scenario shall
address the concern oI a Iire protection system or Ieature being either unreliable or
unavailable.
The probabilistic elements should be integrated by requiring certain types oI Iires (e.g., ultraIast) and
assumptions about operability oI systems (e.g., detection and suppression system Iailure). This
approach could be readily modiIied Ior shipboard use, and strengthened with respect to deIined
Iactors oI saIety and additional scenarios which require Iailure assumptions about other Iire saIety
Ieatures (e.g., passive systems Iailure).


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 43!

A P P E N D X B** 2'H'9%8:>E*,&:<9*!9='&><=:H'*
2'D:E>D*
7* 3'>'&<9*
Once objectives and Iunctional statements, perIormance requirements and design Iire scenarios have
been identiIied, then trial alternative designs should be developed. Trial alternative designs represent
Iire protection system design alternatives developed to address design Iire scenarios to achieve the
previously established perIormance requirements.
Trial alternative designs can be developed on a subsystem or system level, depending on the depth oI
the analysis required. Trial designs may involve comparison with a prescriptive system design
requirement or on a system perIormance basis. Results can be assessed on a comparative basis (i.e.,
perIormance oI prescriptive requirement relative to the proposed trial design) or compared to the
perIormance requirements.
Trial designs that are developed Ior assessment using the perIormance requirement basis should be
developed using design Ieatures that address the perIormance requirements. Trial design subsystems
can include Iire detection and alarm, Iire suppression, occupant behavior and egress, passive Iire
protection, Iire initiation and development and smoke management. Some or all oI these may
comprise the various trial designs. These systems interact with each other to provide an overall level
oI saIety Ior the vessel. It is possible to assess the perIormance oI individual subsystems, however, the
interaction between various subsystems should also be assessed to help reduce the chance that other
subsystems may negatively impact the perIormance oI other subsystems.
?* $N>;=:%><9*#=<='I'>=DT*G'&C%&I<>;'*.'MN:&'I'>=D*<>)*
,&:<9*2'D:E>D*
When selecting perIormance requirements and trial designs Ior a given Iunctional statement, there
may be more than one set oI requirements to achieve each Iunctional statement. For instance, Ior an
objective oI no loss oI liIe outside the room oI origin, perIormance requirements and trial designs
could be developed around: preventing Ilashover in the room oI origin, containing Iire and smoke
within the room oI origin or maintaining tenable conditions outside the room oI origin.
In developing trial designs, it is thereIore necessary to Iirst understand what Iunctional statements and
perIormance requirements must be achieved, then to develop trial designs comprised oI various
subsystems to meet these. Appendix 5, Table 1 provides examples oI objectives, Iunctional statements
and perIormance requirements.
Functional statements and perIormance requirements, once agreed upon by all stakeholders, become
the design Iocus and benchmark Ior measuring the proposed solutions. It is thereIore diIIicult to
eIIectively evaluate trial designs Ior conIormance without a clearly deIined set oI Iunctional
statements and measurable perIormance requirements.




Appendix 5 DeveIoping TriaI AIternative Designs

44 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
TABLE 1
ExampIes of Objectives, FunctionaI Statements
and Performance Requirements
Fire Protection
Obfectives
Functional
Statements
Performance
Requirements
Minimize Iire-related
injuries and prevent undue
loss oI liIe.
No loss oI liIe outside oI the
room or compartment oI
Iire origin.
COHb level not to exceed
12 percent.
Visibility greater than 7
meters.
Minimize Iire-related
damage to the building, its
contents, and its historical
Ieatures and attributes.
No signiIicant thermal
damage outside oI the room
or compartment oI Iire
origin.
Upper layer temperature not
greater than 200C.
Minimize undue loss oI
operations and business-
related revenue due to Iire-
related damage.
No process downtime
exceeding eight hours.
HCl not greater than 5 ppm.
Particulate not greater than
0.5 g/m
3
.
Limit environmental
impacts oI Iire and Iire
protection measures.
No groundwater
contamination by Iire
suppression water runoII.
Impoundment capacity at
least 1.20 times the design
discharge.

To assist in developing trial designs and achieving the desired Iunctional statements, one could use
NFPA 550, 'The Fire SaIety Concepts Tree (FSCT). While incorporating the logic and structure oI a
Iault tree described in Appendix 4, FSCT describes paths leading to success rather than Iailure. FSCT
assists in showing various elements that should be considered in developing trial designs and their
interrelationship with each other.
For example, one oI the more common uses oI perIormance-based design is to extend travel distances.
Assuming it may be diIIicult to 'Prevent Fire Ignition Ior this space, the 'Manage Fire Impact
branch is used. Under this branch, one can 'Manage Fire or 'Manage Exposed. Hence, one may
develop a trial design using the 'Control Combustion Process sub-branch and control the Iuel by
limiting Iuel quantity. In addition, one would also want to manage the exposed, or the occupants, to
evacuate them saIely. ThereIore, the 'SaIeguard Exposed and 'Move Exposed sub-branches could
be used, which would recommend use oI detection and alarm systems to notiIy occupants, and
providing adequate egress Iacilities to allow the occupants to evacuate to a saIe location. As another
alternative, should the stakeholders want to maintain Ilexibility in the space and allow some
combustibles, then under the 'Manage Fire branch, the 'Control Fire by Construction sub-branch
could be used to control the movement oI smoke by either conIining/containing the smoke, iI
appropriate, to the space, or venting the smoke to maintain tenable conditions.
A trial design should express expected Iire growth and spread in the context oI the Iire hazard,
available ventilation and compartment geometry. One oI a number oI trial designs might include
strategies to conIine a Iire to a room or compartment oI origin wherein occupants might reasonably be
expected to Iind a saIe egress prior to untenable conditions being reached (see 'ConIine/Contain Fire
sub-branch under the 'Control Fire by Construction branch oI the Fire SaIety Concepts Tree). This
may or may not include the integration with other subsystems, including additional active Iire
protection to achieve the perIormance requirements.
As seen, the Fire SaIety Concept Tree can be used to develop various alternatives. In addition, some
oI these alternatives incorporate multiple subsystems which are Iurther described below.



Appendix 5 DeveIoping TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 45!
@* #NKDFD='ID*
When developing trial designs, there are various subsystems that can be used alone or in combination
with other systems. Trial designs are developed by understanding what the objectives and Iunctional
statements are, as well as the perIormance requirements, and looking at the various individual or
combinations oI subsystems that will allow one to achieve the perIormance requirements. More than
one trial design can be developed to meet a given set oI perIormance requirements.
A subsystem is a grouping oI similar Iire protection strategies (i.e., detection, alarm, suppression,
compartmentation, etc.). A proposed perIormance-based design could include none, one or many
subsystems as Iire protection strategies to deal with the prevention, control or impact oI a Iire as part
oI a solution. These subsystems (redundant) can act independently oI one another or in concert to
achieve the desired eIIects. Grouping Iire protection strategies into subsystems is intended to Iacilitate
the analysis oI trial designs. The Iollowing Paragraphs provide an overview oI some oI these
subsystems.
The Iunctional statements range Irom controlling the size or eIIects oI a Iire to managing the impact
oI a Iire on a Iacility and its occupants. Typical Iunctional statements might include providing early
warning oI a developing Iire to all Iacility occupants prior to a prescribed level oI smoke in an area or
controlling growth oI a Iire through automatic suppression to prevent Ilashover.
3.1 Fire Initiation and DeveIopment
The Iire initiation and development subsystem can be used to either assist in Iire prevention or to
control the development oI the Iire once it has started. Fire prevention is intended to reduce the
likelihood that ignition will occur. Various concepts that can be employed to achieve this may include
controlling ignition sources, controlling materials, selecting materials that are inherently resistant to
ignition or implementing Iire saIety management procedures to assist in controlling ignition sources
or accumulations oI combustible materials. These concepts are also covered in the Prevent Ignition`
branch oI the Fire SaIety Concepts Tree.
Controlling Iire development can be used to assist in reducing the development rate oI a Iire and its
associated smoke and heat production. Concepts oIten employed to assist include selection and
placement oI contents, selection oI interior Iinishes and construction materials, limiting the quantity oI
materials and controlling the size and geometry oI a compartment and its ventilation.
3.2 Spread, ControI and Management of Smoke
This subsystem assists in addressing the hazards resulting Irom smoke by limiting its production,
controlling its movement and/or reducing the amount oI it. This subsystem concept can be used to
either control materials to exclude those that produce large quantities oI smoke and toxic gases, or to
manage the smoke through various methods including containment, extraction or pressurization, as
well as inclusion oI suppression systems to reduce the amount oI smoke that is being produced.
Various guides are available providing additional inIormation on smoke management, including SFPE
Handbook of Fire Protection Engineering,

NFPA 92A and NFPA 92B.
Use oI this option oIten entails management procedures to control the quantities and types oI
combustible materials allowed in various spaces. InterIaces with other subsystems that oIten need to
be evaluated in parallel include suppression systems with regards to their potential eIIect on the
design Iire size and duration, as well as detection systems to help determine the activation time oI the
smoke management system.
InIormation should be provided to the code oIIicial when using the smoke management subsystem
that includes detector activation times, Ian start up times, controls, interIaces with building
management systems, supply air, extract rates, ducting and Ian design criteria.



Appendix 5 DeveIoping TriaI AIternative Designs

46 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
3.3 Fire Detection and AIarm
Fire detection can assist in providing detection oI a Iire to notiIy occupants or emergency responders.
Detection can also be used as a means oI activating ancillary Iire protection systems (i.e., smoke
management systems, special suppression systems, etc.). Detection can be provided manually by
people or by automatic initiating devices.
Detectors usually sense Iires through various means including heat, smoke or radiant emissions. In
speciIying the type oI detectors to be used, inIormation should be provided as to the type oI Iire
signature being produced by a Iire during the phase where detection is intended to occur. For instance,
diIIerent Iire signatures are produced/available during the smoldering phase (e.g., smoke) versus the
Iully developed phase (heat, radiant emissions, etc.) and will impact the ability oI the detector to
perIorm as intended.
Detector location should also be addressed since the conIiguration and geometry oI the space
(volume, ceiling height), as well as conIiguration oI the ceiling (sloped, beams, etc.) can have an
impact on the time and ability oI the Iire signature to reach the detector. Further guidance regarding
perIormance-based designs oI detection and alarm systems can be Iound in NFPA 72, Appendix B,
and the SFPE Handbook for Fire Protection Engineering.
NotiIication systems may be initiated either manually or by automatic means. They may be provided
by audible and/or visual means.
NotiIication may also include provision oI inIormation to the emergency responders once on-site to
assist them in determining the location and possible extent oI the Iire.
Overall, when assessing the Iire detection and alarm subsystem, inIormation should be provided to the
code oIIicial indicating the Iire signatures that the detection system can detect, as well as the location
oI the initiating devices in relation to the location oI the Iires. In addition, delays oI detection systems
in sensing Iire signatures, alarm veriIication and system processing times, and delays in sending
signals to emergency responders, including via intermediate monitoring Iacilities, should be
understood and included in the descriptions oI the trial designs.
3.4 Automatic Fire Suppression Systems
Fire suppression systems are provided to either extinguish or at least control the development oI a Iire.
Suppression can be either by manual or automatic means.
Automatic suppression systems require no human interaction and typically entail sprinkler, Ioam or
gaseous suppression systems. DiIIerent types oI Iires may require diIIerent types oI suppression
agents. For instance, some Ilammable liquid Iires are better addressed by Ioam than water. In addition,
the size oI the Iire at the desired point oI suppression/extinguishments is important in selecting a
suppression system. For instance, in computer/telecommunication rooms where early detection and
suppression are desired, an early detection system activating a special suppression system would
typically provide earlier suppression (i.e., smaller Iire) than an automatic sprinkler system.
Some oI these systems are dependent on activation oI a Iire detection system, and hence, assessment
oI detection time and time to discharge oI the suppression agent is important. The trial design should
thereIore provide details on the interIace and pertinent Ieatures oI these other integrated systems that
impact its eIIectiveness to activate in a suIIicient amount oI time and discharge an appropriate
quantity and type oI suppressant, so they can be appropriately assessed.
The characteristics oI the room/space should be included in the assessment to determine the
eIIectiveness oI the suppression system in activating and perIorming as desired. This should include
the size and geometry oI the space. Sprinklers provided on a high ceiling, Ior instance in an atrium,
may not only be delayed in activating, but also have diIIiculty in providing suIIicient quantities oI
water on the Iire below once the Iire has grown to a size suIIicient to cause activation.



Appendix 5 DeveIoping TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 47!
II it is desired to use manual suppression, whether by crews or by internal Iire brigades, various
aspects should be part oI the assessment, including notiIication, response time to site, access to
site/Iacility/Iire area, number oI emergency responders, equipment and Iire Iighting Ieatures provided
at the Iacility, including water supply.
It is important to understand that some types oI suppression systems are used to control Iires (i.e.,
sprinklers, water mist systems) while others are intended to extinguish Iires (gaseous systems, early
suppression Iast response sprinklers). In choosing one Ior a trial design, it should be clear what the
suppression system is intended to do. II it is only controlling the Iire, then the resultant on-going Iire
induced conditions should continue to be assessed to ensure Iunctional statements and perIormance
requirements are still achieved.
3.5 Human Behavior and Egress
When developing perIormance requirements and trial designs to meet the objectives and Iunctional
statements, it is critical to deIine the characteristics oI the passengers and crew members and their
anticipated behavior during a Iire, as well as the egress Ieatures and vessel characteristics.
The design team needs to consider several general principles regarding the passengers and crew
members and egress Ieatures as they relate to their surroundings in developing a trial design:
x What is the minimum and maximum number oI people expected/permitted to be in the structure,
Iacility or speciIic portions thereoI?
x What is the maximum length oI time the structure is occupied?
x How mobile are the passengers? Do people normally sleep or might they be expected to sleep in
their cabins or the Iacility?
x Can passengers reasonably be expected to be Iamiliar with the vessel layout and means oI egress?
x What percentage oI crew members and passengers can be considered members oI a vulnerable
population (e.g., children, elderly, disabled, incapacitated persons, etc.)?
x Are the egress Iacilities adequate?
x What is the nature oI the hazard in the vessel and what are the expected responses oI the crew
members and passengers?
Once the relevant characteristics oI passengers and crew members, egress Ieatures and resulting risk
Iactors are assessed, appropriate trial designs can be developed based upon managing Iire impact
strategies, i.e., whether to provide suitable protected egress routes, deIend in place, provide early
notiIication and assisted egress, etc. While a certain level oI knowledge regarding egress and human
behavior currently exists to evaluate egress in certain trial designs, the design team should undertake
various 'what iI assessments to help provide appropriate alternatives (i.e., What iI the passenger
loads are higher? What iI an exit is blocked by Iire, etc.).
3.6 Passive Fire Protection
Passive Iire protection is intended to address two components:
i) Structural stability
ii) Issues related to limiting Iire and smoke spread in a Iacility
The structural stability subsystem addresses preventing premature collapse oI part or all oI a Iacility.
Various approaches and methods are available to assess the necessary protection Ior structural
members to limit the chance oI structural Iailure due to the anticipated thermal loading imposed by
the design Iire scenarios. In perIorming these assessments, the inherent stability oI an unprotected
element may be suIIicient, whereas in others, protection in addition to that which is required by code
is needed.



Appendix 5 DeveIoping TriaI AIternative Designs

48 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
In undertaking these assessments, issues including Iire perIormance oI structural systems Ior well-
deIined design Iire loads, Iire perIormance oI Iire protective materials Ior agreed Iire loads,
connection ductility, protection oI connections, eIIect oI load transIers, composite actions oI Iloor
slabs and Irames and susceptibility to progressive collapse may need to be considered.
The subsystem oI limiting Iire and smoke spread through passive means includes concepts that can be
used independently oI each other or integrated together to limit the spread oI Iire and smoke in a
space. These Ieatures include compartmentation, Iire barriers, protection oI openings, prevention oI
external Iire spread and controlling the Iire by means such as automatic or manual suppression.
Non-Iire-rated glazing, glass partitions and unrated construction may all provide some limited Iire
endurance, but the trial design should thoroughly evaluate the consequences when these are exposed
to credible design Iire scenarios. It is important to note that there are multiple strategies available that
may be considered in developing trial designs that include combinations oI active and passive Iire
protection, one or the other or none at all.
While International Code Ior Application oI Fire Test Procedures (FTP Code) has historically
measured Iire endurance in the context oI components tested to a standard time-temperature exposure
in a standardized test Iacility with Iixed laboratory conditions, the perIormance oI these components
or systems in the Iield will vary. This is due to changes in Iires and temperature-induced conditions to
that which is used in the test Iurnaces and can either be more or less severe depending on the credible
Iires Ior a speciIic Iacility. In addition, changes in compartment characteristics will aIIect the transIer
oI heat Irom the Iire to the structural or compartment components and thus also aIIect its ability to
perIorm Ior its anticipated time. In addition, when structural elements/components are interconnected
to other vessel structures and systems, their perIormance will be aIIected by heat transIer to these
other components, as well as the ability Ior various components to redistribute their loads. ThereIore,
these should be included when undertaking a perIormance-based design that involves passive Iire
protection.
A* $:&'*#<C'=F*P%>;'8=*,&''*U$#P,V*
The concept oI NFPA 550, Fire Safetv Concept Tree (FSCT), is a useIul systematic approach to
providing an overall structure with which to analyze the potential impact oI various codes and
standards on a particular Iire saIety problem.
FIGURE 1
Top Gate of FSCT
+
Prevent
fire
ignition
Manage
fire
impact
Fire safety
objective(s)


|Reprinted with permission Irom NFPA 550, Guide to the Fire Safetv Concepts Tree. Copyright 2002 National Fire
Protection Association, Quincy, MA 02269. This reprinted material is not the complete and oIIicial position oI the National
Fire Protection Association on the reIerenced subject, which is represented only by the standard in its entirety.|




Appendix 5 DeveIoping TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 49!
FIGURE 2
Prevent Fire Ignition Branch of FSCT
+
Prevent
fire
ignition
Control
heat-energy
source(s)
Control
rate of
heat-energy
release
Control
fuel
Control
source-fuel
interactions
+
Eliminate
heat-energy
soyrce(s)
Control
heat-energy
transfer
processes
Control
heat-energy
source
transport
Control
fuel
transport
Eliminate
fuel(s)
Control
fuel
ignitibility
+
Control
convection
Control
radiation
Control
conduction
Provide
separation
Provide
barrier
Control the
environment
Control fuel
properties
Provide
barrier
Provide
separation
+ + +


FIGURE 3
Logic SymboIs Used in FSCT
Key
+ = "OR" gate
= "AND" gate


FIGURE 4
Major Branch of Manage Fire Impact
+
Manage
fire
Manage
exposed
Manage
fire
impact


|Reprinted with permission Irom NFPA 550, Guide to the Fire Safetv Concepts Tree. Copyright 2002 National Fire
Protection Association, Quincy, MA 02269. This reprinted material is not the complete and oIIicial position oI the National
Fire Protection Association on the reIerenced subject, which is represented only by the standard in its entirety.|




Appendix 5 DeveIoping TriaI AIternative Designs

50 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
FIGURE 5
Manage Fire Branch of FSCT
+
Manage
fire
Control
conbustion
process
Control the
environment
Control
fire by
construction
Control
source-fuel
interactions
+
Control
fuel
Automatically
suppress
fire
Confine/
contain
fire
Vent
fire
+
+
+ +
Limit
fuel
quantity
Control
fuel
distrubution
Control
fuel
properties
Control
physical
properties of
environment
Control
chemical
composition of
environment
Control
movement
of fire
Provide
structural
stability
Manually
suppress
fire
Detect
fire
Communicate
signal
Decide
action
Respond
to site
Apply
sufficient
suppressant
Apply
sufficient
suppressant
Detect
fire


|Reprinted with permission Irom NFPA 550, Guide to the Fire Safetv Concepts Tree. Copyright 2002 National Fire
Protection Association, Quincy, MA 02269. This reprinted material is not the complete and oIIicial position oI the National
Fire Protection Association on the reIerenced subject, which is represented only by the standard in its entirety.|




Appendix 5 DeveIoping TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 51!
FIGURE 6
Manage Exposed Branch of FSCT
+
Safeguard
exposed
Move
exposed
Cause
movement
of exposed
Provide
movement
means
Provide
instruction
Provide
capacity
Provide
route
completeness
Provide
proetcted
path
Provide
route
access
Signal
need
Detect
need
Defend
against fire
product(s)
Provide
structural
stability
Defend
the place
Restrict
movement
of exposed
Maintain
movement
environment
Provide
safe
destination
Go to
A
Defend
exposed
in place
A
Limit
amount
exposed
Manage
exposed
+
Go to
A


FIGURE 7
Fire Prevent in a Computer FaciIity
+
Prevent
fire
ignition
Control
heat-energy
source(s)
Control
rate of
heat-energy
release
Control
fuel
Control
source-fuel
interactions
+
Eliminate
heat-energy
soyrce(s)
Control
heat-energy
transfer
processes
Control
heat-energy
source
transport
Control
fuel
transport
Eliminate
fuel(s)
Control
fuel
ignitibility
+
Control
convection
Control
radiation
Control
conduction
Provide
separation
Provide
barrier
Control the
environment
Control fuel
properties
Provide
barrier
Provide
separation
+ + +
Key A = Above standard
B = Standard
C = Below standard
N = Nonexistent
S
S
S S
S
S
S
N N S
N
N
N N N N N N N N


|Reprinted with permission Irom NFPA 550, Guide to the Fire Safetv Concepts Tree. Copyright 2002 National Fire
Protection Association, Quincy, MA 02269. This reprinted material is not the complete and oIIicial position oI the National
Fire Protection Association on the reIerenced subject, which is represented only by the standard in its entirety.|



Appendix 5 DeveIoping TriaI AIternative Designs

52 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
FIGURE 8
Administration Action Guide
+
+
Accomplish by
features of
design
+
Adopt
legislation
Educate
user
nspect
properly
Enforce
law
Adopt
legislation
Educate
user
nspect
properly
Enforce
law
Motivate
user
Educate
user
nspect
properly
Motivate
user
Educate
user
nspect
properly
Accomplish by
control of
human action
Accomplish by
features of
design
Accomplish by
voluntary
human action
Accomplish by
mandatory
action
Accomplish by
voluntary
action
Accomplish by
administration
action


|Reprinted with permission Irom NFPA 550, Guide to the Fire Safetv Concepts Tree. Copyright 2002 National Fire
Protection Association, Quincy, MA 02269. This reprinted material is not the complete and oIIicial position oI the National
Fire Protection Association on the reIerenced subject, which is represented only by the standard in its entirety.|

B* .'C'&'>;'D*
1. Milke, J., 'Smoke Management in Covered Malls and Atria, the SFPE Handbook of Fire
Protection Engineers, 3
rd
Edition, Section 4-13, National Fire Protection Association, Quincy,
MA: 2002.
2. NFPA 550: Guide to the Fire Safetv Concept Tree, 1995 Edition.
3. NFPA 92A: Smoke-Control Svstems, 2000 Edition.
4. NFPA 92B: Smoke Management Svstems in Malls, Atria, and Large Areas, 2000 Edition.
5. NFPA 72: National Fire Alarm Code, 1999 Edition.
6. SchiIiliti, R., Meacham, B., and Custer, R. 'Design oI Detection Systems, The SFPE
Handbook of Fire Protection Engineering, 3
rd
Edition, Section 4-1, National Fire
Protection Association, Quincy, MA: 2002.
7. International Code Ior Application oI Fire Test Procedures (FTP Code), IMO, 1998.
8. Guidelines Ior Chemical Process Quantitative Risk Analysis, Center Ior Chemical Process
SaIety oI the American Institute oI Chemical Engineers (AIChE), AIChE, New York: 1989.


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 53!

A P P E N D X Q** 2'D:E>*,%%9D*C%&*,&:<9*!9='&><=:H'*
2'D:E>D*
7* 0>=&%)N;=:%>*
Fire is a dynamic process oI interacting physics and chemistry, and Iire phenomena include a larger
range oI time and space scales. Time ranges Irom the picoseconds involved in molecular
rearrangement and vibration transitions to hours needed Ior the collapse oI steel-reinIorced barriers.
Space scales range Irom microns in polymer connections to meters in constructions. At present, it is
impossible to include the entire range oI phenomena in a comprehensive model oI such a process.
ThereIore, the models with a range oI approximations are used to describe Iire dynamics and the
consequence aIter the Iire.
The Iundamental conservation equations Ior Iire dynamics include the governing equations oI Iluid
dynamics, heat transIer and combustion, and enormous progress has been made toward the numerical
solutions Ior Iluid and thermal applications. The simplest methods to predict Iire phenomena are the
algebraic equations. OIten developed wholly or in part Irom correlation to experimental data, they
represent, at best, estimates with signiIicant uncertainty. Yet, under the right circumstances, they have
been demonstrated to provide useIul results.
However, it is inappropriate to rely solely on such estimation techniques Ior Iire development or
smoke Iilling calculations. Only Iire models should be used. However, due to the inherent complexity
oI the Iire dynamics problem, the practical mathematical models oI Iire are relatively recent. The
diIIiculties revolve about at least three issues: First, there are enormous numbers oI possible Iire
scenarios to consider due to their accidental nature. Second, the physical insight and computing power
necessary to perIorm all necessary calculations Ior most Iire scenarios are limited. Any
Iundamentally-based study oI Iires must consider at least some aspects oI bluII body gas dynamics,
multi-phase Ilows, turbulent mixing and combustion and radiative transport, all oI which are active
research areas in their own right. Finally, the 'Iuel in most Iire was never intended as such. Thus, the
mathematical models and data needed to characterize the degradation oI condensed phase materials
that supply the Iuel may not be available. The mathematical modeling oI the physical and chemical
transIormations oI the real materials as they burn is still in the preliminary stage.
Currently there are two Iundamentally diIIerent approaches to Iire modeling: (1) probabilistic and (2)
deterministic. The probabilistic or stochastic approach involves the assessment oI probable Iire risk in
an enclosure by associating Iinite probabilities with all Iire-inIluencing parameters, such as
distributions oI Iuel, numbers oI vents and openings, and human behavior, etc. Little or no physics is
included in probabilistic-based models. This approach, while useIul in suggesting likelihood oI a Iire
in a given enclosure, provides little inIormation about the distribution oI Iire production, temperature
proIile and smoke propagation.
In deterministic models, a complete set oI diIIerential equations based on laws oI physics and
chemistry can compute the conditions produced by Iire at a given time in a speciIied volume oI air in
a well-deIined physical scenario. Deterministic Iire models can range Irom simple one-line correlation
oI data to highly complex models. The more complex models are typically divided into two classes:
(1) zone models and (2) Iield models, based on the strategy used to solve the equations representing
the physical processes associated with the Iire.



Appendix 6 Design TooIs for TriaI AIternative Designs

54 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
?* W%>'*4%)'9D*
2.1 Basic Concepts
The zone modeling concept divides the hypothetical burning enclosure into two spatially
homogeneous volumes, i.e., a hot upper layer and a cool lower layer. This two-layer approach has
evolved Irom the observation oI such layering in real-scale Iire experiments. Hot gases collect at the
ceiling and Iill the compartment Irom the top. While these experiments show some variation in
conditions within the layer, these are small compared to the diIIerences between the layers. Thus,
zone models can provide a Iairly realistic simulation under most conditions.
Mass and energy balances are enIorced Ior each layer, with additional models describing other
physical processes appended as diIIerential or algebraic equations, as appropriate. Examples oI such
phenomena include Iire plume, Ilows through windows, ceilings and vents, radiative and convective
heat transIer and solid Iuel pyrolysis rate, etc.
In the zone model, the conservation equations Ior the upper and lower gas zones are developed either
by using Iundamental equations oI energy, mass and momentum transport in control volume Iorm
applied to the zones, or by using diIIerential equations that represent the conservation laws and
integrating them over the zones. However, the momentum equations will not be explicitly applied
since inIormation needed to compute velocities and pressure is based on assumptions and speciIic
applications oI momentum principles at vent boundaries oI the compartments. For example, the mass
transIer between control volumes results Irom the air entrainment oI the Iire plume and vent Ilows
through openings.
Appendix 6, Figure 1 illustrates a typical zone model Ior a compartment Iire process. The velocity oI
the control volume along the interIace, w, is equal to the Iluid velocity, v . The properties oI the upper
and lower zones are assumed to be spatially uniIorm, but can vary with time.
FIGURE 1
ControI VoIumes SeIected in Zone ModeIing
T, p
CV
1
CV
2
m
s
(Iuel)
w 0
m
e
w v
m

|Figure taken Irom ReI. 6.|




Appendix 6 Design TooIs for TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 55!
2.2 ModeI Assumptions
The most dominant characteristic assumption oI the zone models is that it divides the room(s) into a
hot upper layer and a cool lower layer. The model calculations provide estimates oI key conditions Ior
each oI the layers as a Iunction oI time. Other assumptions in the application oI the conservation oI
laws in the zone model may also include:
x No diIIusion or conduction occurs at the Iree boundary surIace oI the control volume.
x Exchange oI mass at Iree boundaries is due to bulk transport oI Iluid and due to pressure or shear
mixes.
x DiIIusion can occur at solid boundaries, but is generally ignored.
x The plume instantly arrives at the ceiling.
x The mass or heat capacity oI the room contents is ignored compared to the enclosure structural
wall, ceiling and Iloor elements, i.e., the wall or ceiling can be treated as heat sinks or heat
sources as Ilame spreads.
x The horizontal cross-section oI the enclosure is a constant.
x The pressure in the enclosure is considered uniIorm in the energy equation, but hydrostatic
variations account Ior pressure diIIerences at Iree boundaries oI the enclosure, i.e., P ~~ UgH.
x Mass Ilow into the Iire plume is due to the turbulent entrainment.
x Fluid Irictional eIIects at solid boundaries are ignored in most models.
2.3 Fires
A Iire is a source oI Iuel which is released at a speciIied rate. This Iuel is converted into enthalpy and
mass as it burns. A Iire is constrained iI the enthalpy conversion depends on the oxygen
concentration; otherwise, it is unconstrained. Burning can take place in the portion oI the plume in the
lower layer (iI any), in the upper layer or in a door jet. For an unconstrained Iire, the burning will all
take place within the Iire plume. For a constrained Iire, burning will take place where there is
suIIicient oxygen available. When insuIIicient oxygen is entrained into the plume, unburned Iuel will
successively Ilow into and burn in places such as the upper layer oI the Iire compartment, the plume
in the doorway to the next compartment, the upper layer oI the next compartment, the plume in the
doorway to the third compartment and so Iorth until it is consumed or gets to the outside.
Most zone models include the ability to independently track multiple Iires in one or more enclosures.
These Iires are treated as totally separate entities, i.e., with no interaction oI the plume or radiative
exchange between Iires in the enclosure. These Iires are generally reIerred to as 'objects and can be
ignited at a speciIied time, temperature or heat Ilux.
2.4 Heat Transfer
Gas layers exchange energy with surroundings via convective and radiative heat transIer. While
diIIerent material properties can be used Ior the ceiling, Iloor and walls Ior each compartment,
material thermophysical properties are mostly assumed to be constant, although we know that they
vary somewhat with temperature. Radiative heat transIer occurs among the Iire(s), gas layers and
compartment surIaces (ceiling, walls and Iloor). This transIer is a Iunction oI the temperature
diIIerences and the emissivity oI the gas layers, as well as the compartment surIaces. Some models
ignore the heat conduction through the compartment wall surIaces, whereas some apply a one-
dimensional heat-conduction equation to estimate the heat transIer through the wall surIaces.



Appendix 6 Design TooIs for TriaI AIternative Designs

56 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
2.5 Vent FIow
Flow through vents is a dominant component oI any Iire model because it is sensitive to small
changes in pressure and transIers the greatest amount oI enthalpy on an instantaneous basis Ior all oI
the source terms (except Ior the Iire and plume). Its sensitivity to environmental changes arises
through its dependence on the pressure diIIerence between compartments, which can change rapidly.
In the realm oI the zone model, there are two distinct types oI Ilow: horizontal Ilow through vertical
vents (doors, windows, etc.) and vertical Ilows through horizontal vents (ceiling holes, hatches, etc.).
Vertical Ilow is particularly important in two disparate situations: a ship, and the role oI Iire Iighters
doing rooI venting.
Atmospheric pressure is about 100,000 Pa, Iires produce pressure changes Irom 1 Pa to 1000 Pa, and
mechanical ventilation systems typically involve pressure diIIerentials oI about 1 Pa to 100 Pa. The
pressure variables should be solved to a higher accuracy than other solution variables because oI the
subtraction (with resulting loss oI precision) needed to calculate vent Ilows Irom pressure diIIerences.
2.6 PIumes and Layers
A plume is Iormed above any burning object. It acts as a pump to transIer mass and enthalpy Irom the
lower layer into the upper layer. A correlation is used to predict the amount oI mass and enthalpy that
is transIerred. Two sources exist Ior moving enthalpy and mass between the layers within and
between compartments. Within the compartment, a Iire plume provides one source. The other source
oI mixing between the layers occurs at vents, such as doors and windows. The degree oI mixing is
based on an empirically driven mixing relationship.
As enthalpy and mass are pumped into the upper layer by the Iire plume, the upper layer expands in
volume, causing the lower layer to decrease in volume and the interIace to move downward. II the
door to the next compartment has a soIIit, there can be no Ilow through the vent Irom the upper layer
until the interIace reaches the bottom oI the soIIit. Thus, in the early stages, the expanding upper layer
will push down on the lower layer air and Iorce it into the next compartment through the vent by
expansion.
Once the interIace reaches the soIIit level, a door plume Iorms and Ilow Irom the Iire compartment to
the next compartment is initiated. As smoke Ilow Irom the Iire compartment Iills the second
compartment, the lower layer oI air in the second compartment is pushed down. As a result, some oI
this air Ilows into the Iire compartment through the lower part oI the connecting doorway or vent.
Thus, a vent between the Iire compartment and connecting compartments can have simultaneous,
opposing Ilows oI air. All Ilows are driven by pressure and density diIIerences that result Irom
temperature diIIerences and layer depths. The key to getting the correct Ilows is to correctly describe
the Iire and plume`s mass and enthalpy between the layers.
2.7 Species Concentrations and Depositions
When layers are initiated at the start oI the simulation, they are set to the ambient conditions. These
are the initial temperatures speciIied by the user, and 23 by mass (20.8 by volume) oxygen, and
77 by mass (79 by volume) nitrogen, a mass concentration oI water speciIied by the user as a
relative humidity, and a zero concentration oI all other species.
As Iuel is pyrolyzed, the various species are produced in direct relation to the mass oI Iuel burned.
Since oxygen is consumed during burning, the 'yield oI oxygen is negative, and is set internally to
correspond to the amount oI oxygen needed to burn the Iuel. Hydrogen cyanide and hydrogen
chloride are assumed to be products oI pyrolysis, whereas carbon monoxide, carbon dioxide, water
vapor and soot are products oI combustion. Each unit mass oI species produced is carried in the Ilow
to the various compartments and accumulates in the layers. The Iire model keeps track oI the mass oI
each species in each layer.



Appendix 6 Design TooIs for TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 57!
No zone model oI Iire growth and smoke transport incorporates a complete combustion scheme. It is
simply not practical at this time. Rather than try to capture the development oI all species, it has been
deemed more practical to use empirical methods, measure the rate oI production oI species and use
these in the predictive model. For Iires, a combustion chemistry scheme based on a carbon-hydrogen-
oxygen balance is commonly used. The scheme needs to be applied to at least three places. The Iirst is
burning in the portion oI the plume which is in the lower layer oI the Iire origin. The second is in the
portion oI the upper layer, also in the compartment oI Iire origin. The third is in the vent Ilow which
entrains air Irom a lower layer into an upper layer in an adjacent compartment. This is equivalent to
solving the conservation equations Ior each species independently.
There are two signiIicant limitations oI zone models inherent in this prescription. One is that it is
diIIicult to capture the eIIect oI transitioning through the layer interIace, which is one oI the sources
oI carbon monoxide. The other is the transient nature oI the plume, especially in the initial phase oI a
Iire when the plume is developing Irom a small cloud to a complete plume envisioned by Mortem,
Taylor and Turner in their classiIications on plumes.
2.8 Predictive Equations
Zone Iire models solve a set oI equations in the Iorm oI an initial value problem Ior a mixed system oI
diIIerential and algebraic equations. These equations are derived Irom a conservation oI mass and
energy. Subsidiary equations are the ideal gas law and deIinitions oI density and internal energy.
These conservation laws are invoked Ior each zone or control volume.
The basic element oI the model is a zone. The basic assumption oI a zone model is that properties
such as temperature can be approximated throughout the zone by some uniIorm Iunction. The usual
approximation is that temperature, density and so on are uniIorm within a zone. The assumption oI
uniIorm properties is reasonable and yields good agreement with experiments. In general, these zones
are grouped within compartments.
There are two reasonable conjectures which dramatically improve the ease oI solving these equations.
The Iirst is that momentum is ignored within a compartment; the other is that the pressure is
approximately uniIorm within a compartment. However, the hydrostatic variation in pressure is taken
into account in calculation oI the pressure diIIerence between compartments.
Many Iormulations based upon these assumptions can be derived. One Iormulation can be converted
into another, using the deIinitions oI density, internal energy and the ideal gas law. Though equivalent
analytically, these Iormulations diIIer in their numerical properties.
Each Iormulation can be expressed in terms oI mass and enthalpy Ilow. These rates represent the
exchange oI mass and enthalpy between zones due to physical phenomena such as plumes, natural
and Iorced convective and radiative heat transIer, and so on. For example, a vent exchanges mass and
enthalpy between zones in connected rooms, a Iire plume typically adds heat to the upper layer and
transIers entrained mass and enthalpy Irom the lower to the upper layer, and convection transIers
enthalpy Irom the gas layers to the surroundings.
The numerical characteristics oI the various Iormulations are easier to identiIy iI the underlying
physical phenomena are decoupled. For example, CFAST |5| uses the Iormulation in terms oI the
rates oI mass and enthalpy, and assumes that these rates may be computed in terms oI zone properties
such as temperature and density.
Many approximations are necessary when developing physical sub-models Ior mass and enthalpy
terms. For example, most Iire models assume that (1) the speciIic heat terms c
p
and c
v
are constant
even though they are temperature dependent, (2) hydrostatic terms can be ignored in the equation oI
state (the ideal gas law) relating density oI a layer with its temperature. The detailed derivations oI
equations Ior various zone models can be Iound in their user manuals or technical reIerences.



Appendix 6 Design TooIs for TriaI AIternative Designs

58 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
2.9 Limitations of Zone modeIs
The basic assumption oI all zone Iire models is that each room can be divided into a small number oI
control volumes, each oI which is internally uniIorm in temperature and compositions. Beyond the
basic assumptions, a model typically involves a mixture oI established theory (e.g., conservation
equations), empirical correlation where there are data but no theory (e.g., air entrainment coeIIicients,
pre-Ilashover coeIIicients) and approximations (e.g., post-Ilashover combustion theory), etc.
The major limitations oI zone models are directly related to the modeling assumptions. The generality
and uncertainty oI the results can be signiIicant in some particular scenarios. For example, it is oIten
important during the analysis oI a design to determine the exact results oI small-scale phenomena at a
particular location in a compartment. The Iield model will determine the vertical temperature gradient
through the upper layer, whereas a zone model generally gives only average upper layer temperature.
The zone models give the same results regardless oI the location oI Iire origin as long as it is with the
same layer. Thus, the uncertainty oI the modeling results can be signiIicant iI the temperature proIiles
are sensitive to the height oI Iire origin below the ceiling in a real Iire scenario.
Another shortcoming oI the zone model is the assumption oI instantaneous plume spread upon
impingement oI the plume with the ceiling. II the compartment is suIIiciently large (e.g., a warehouse)
or long (e.g., a corridor), the assumption oI instantaneous volume Iilling may be violated. It is well
documented that a lag time exists between plume impingement upon a ceiling and arrival oI the
ceiling jet Iront at the end oI the corridor. While many assumptions and limitations exist in a zone
model, it is ultimately up to the engineer and designer perIorming the analysis to understand and
document to ensure that none oI the assumptions have been violated, or that the assumption violation
has not subsequently invalidated the resulting conclusions.
The development oI zone models will be dependent not only on the advancement oI computer coding
techniques, but also the experimental research needed Ior the improvement in the model.
2.10 Current AvaiIabIe ModeIs
The Iollowing models either have a signiIicant number oI users or are currently used:
Model Countrv Descriptions
ARGOS Denmark Multi-compartment zone model
ASET US One room zone model with no ventilation
ASET-B US ASET in BASIC instead oI Fortran
BRANZFIRE New Zealand Multi-room zone model, including Ilame spread, multi Iires, and
mechanical ventilation
BRI-2 Japan/US Two-layer zone model Ior multistory, multi-compartment smoke
transport
CALTECH US PreIlashover zone model
CCFM.VENTS US Multi-room zone model with ventilation
CFAST/FAST US Zone model with a suite oI correlation programs-CFAST is the solver,
FAST is the Iront-end
CFIRE-X Germany Zone model Ior compartment Iires, particularly liquid hydrocarbon
pool Iires
CiFi France Multi-room zone model
COMPBRN-III US Compartment zone model
COMF-2 US Single-room postIlashover compartment model



Appendix 6 Design TooIs for TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 59!
Model Countrv Descriptions
DACFIR-3 US Zone model Ior an aircraIt cabin
DSLAYV Sweden Single compartment zone model
FASTlite US Feature limited version oI CFAST
FFM US PreIlashover zone model
FIGARO-II Germany Zone model Ior determining untenability
FIRAC US Uses FIRIN, includes complex vent systems
FireMD US One room, two zone model
FIREWIND Australia Multi-room zone model with several smaller submodels (update oI
FIRECALC)
FIRIN US Multiroom zone model with ducts, Ians, and Iilters
FIRM US Two zone, single compartment model
FIRST US One room zone model, includes ventilation
FMD US Zone Iire model Ior atria
HarvardMarkVI US Earlier version oI FIRST
HEMFAST US Furniture Iire in a room
HYSLAV Sweden PreIlashover zone model
IMFE Poland Single room zone model with vents
MAGIC France Two-zone model Ior nuclear power stations
MRFC Germany Multi-room zone model Ior calculation oI smoke movement and
temperature load on structures
NAT France Single compartment zone model with attention to responses oI
structures
NBS US PreIlashover zone model
NRCC1 Canada Single room compartment zone model
NRCC2 Canada Large oIIice space with vents
OSU US Single room compartment zone model
OZONE Belgium Zone model with attention to responses oI structures
POGAR Russia Single compartment zone model
RADISM UK Zone model incorporating an immersed ceiling jet within the buoyant
layer, sprinklers and vents
RFIRES US PreIlashover zone model
R-VENT Norway Single room smoke ventilation zone model
SFIRE-4 Sweden PostIlashover zone model
SICOM France Single compartment zone model
SMKFLW Japan One-layer zone model Ior smoke transport in buildings



Appendix 6 Design TooIs for TriaI AIternative Designs

60 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
Model Countrv Descriptions
SmokePro Australia Zone model Ior single compartment smoke layer interIace position
SP UK PreIlashover zone model
WPI-2 US Single compartment zone model
WPIFIRE US Multi-room zone model
ZMFE Poland Single compartment zone model

Some other models, such as PRETool (US) are based on correlation. SpeciIic applications models,
such as ASCOS and Contam, are Ior designing smoke control systems. Some models, such as
LAVENT and JET programs, are Ior designing sprinkler responding systems.
2.11 ModeI SeIection
To select an appropriate Iire model Ior speciIic applications, proper validation and veriIication
procedures such as speciIied in ASTM E 1355-97 should be Iollowed. The theoretical basis oI the
model should be reviewed by knowledgeable but independent and recognized experts. A key issue in
selecting a model is model validation. Comparison oI model results with experimental data is valuable
Ior determining the applicability oI a model to a particular situation. The model user should careIully
examine the model validation comparisons beIore selecting a model. They will include the
comparisons with standard tests, the comparisons with Iull-scale tests conducted speciIically Ior the
chosen evaluation, the comparisons with previously published Iull-scale data, the comparisons with
documented Iire experience, and the comparisons with a proven benchmark model, etc.
Only models which are rigorously evaluated and documented should be allowed in any applications
involving design, legal consideration such as code enIorcement, or litigation. It is simply not
appropriate to rely on the model developer`s words that the physics is proper. This means that the
model should be supplied with a technical reIerence guide which includes a detailed description oI the
included physics and chemistry with proper literature reIerences, and estimates oI the accuracy oI the
resulting predictions based on comparisons with experiments. Public exposure and review oI the exact
basis Ior a model`s calculations, empirical or reIerence data used Ior constants and deIault values in
the code, and assumptions are necessary Ior it to have credibility in a regulatory application.
No zone Iire model is 'best Ior all applications. The selection oI a zone Iire Ior a particular
application depends on a number oI Iactors. While most oI the zone models are based on the same
Iundamental principles, there is signiIicant variation among diIIerent models. The decision to use a
model should be based on the understanding oI the assumptions and limitations Ior the particular
model.
When using a computer model, it is always a good idea to test the sensitivity oI the model. Such a test
will help the user understand how changes in model input parameters aIIect the results generated by
the model, determine the dominant variables in the model, deIine the acceptable range oI values oI
each input variables, and quantiIy the sensitivity to provide inIormation and cautions on selection oI
input variables.



Appendix 6 Design TooIs for TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 61!
@* $:'9)*4%)'9D*
3.1 Basic Concepts
Field (or CFD) models represent the other alternatives oI deterministic analysis. This approach is
based on basic local conservative laws Ior physical quantities such as mass, momentum, energy and
species concentrations. These equations are solved with spatial and temporary resolutions to yield the
distributions oI the variables oI interest. The set oI equations, reIerred to as the Navier-Stokes
equations, consist oI three-dimensional, time-dependent, non-linear partial diIIerential equations.
Theoretically, this numerical approach should provide the whole history oI Iire evolution including
local characteristics at any given point.
Due the turbulent characteristics oI thermally driven Ilows, the biggest challenge that arose in using
CFD methodology is how to properly handle turbulence. Field models are classiIied based on methods
by which they treat turbulence phenomena. The two major groups may be identiIied as Reynolds-
Averaged Navier-Stokes (RANS) models and Large Eddy Simulations (LES) models.
The RANS model computes time-averaging solutions Ior diIIerent variables oI the modeled equations
over a long time scale. Only the mean Ilow Iield will be described by the time-averaged RANS model,
whereas local Iluctuations and turbulent structures are integrated in the mean quantities and these
structures are no longer to be described in the simulation. The time scale used by RANS is long when
compared to the turbulent motion within the system; the variable data obtained Irom this technique
provide mean quantities diIIerent Irom the instantaneous ones. Strong unsteady mixing eIIects,
resulting Irom the rolling up oI shear layers, are observed in turbulent Ilames, but the knowledge oI
steady statistical mean quantities is not always suIIicient to describe the turbulent combustion. It is
evident that one oI the major limitations oI RANS models is the inadequate treatment oI turbulent
Ilow. An alternative is to use LES technique.
The object oI LES is to explicitly compute the largest structures oI the Ilow (typically, the structures
larger than the computational mesh size). The large scale eddies are generally more energetic in
turbulent combustion. The assumption behind the space-averaged LES is that only the large scales
that carry most oI the energy need to be directly resolved in order to present the Ilow accurately, and
that energy dissipation into smaller scales can be modeled. The LES technique is inherently time-
dependent, since N-S equations are not time-averaged. Transient problems can be solved quickly
using this approach.
3.2 ModeI Requirements
Although a very wide range oI engineering problems can be addressed by CFD models, there is no
single CFD code that can incorporate all oI the physical and chemical processes oI importance. There
exist, thereIore, only a handIul oI CFD codes that can be used Ior problems involving Iires. These, in
turn, use a number oI diIIerent approaches to the subprocesses that need to be modeled. Some oI the
most important oI these subprocesses include turbulence modeling, radiation and soot modeling and
combustion modeling, etc. Since the modeling oI these subprocesses are still very active in their own
research perspective, the use oI CFD codes in Iire saIety engineering design is limited to the expert
knowledge required Ior the processes listed above.
3.2.1 Turbulence Modeling
The Ilow occurring in room Iires is turbulent, generating eddies or vortices oI varying sizes.
The energy contained in large eddies cascades down to smaller and smaller eddies until it
diIIuses into heat. Such eddies exist down to the sizes where the viscous Iorces dominate over
inertial Iorces and energy is dissipated into heat. For typical Iires, the length scale oI eddies is
down to a millimeter or so. Thus, the control volume size to discrete N-S equations should be
consistent with this scale. Additionally, the Iluctuations can occur very Iast and can have a
Irequency in the order oI 10 kHz.



Appendix 6 Design TooIs for TriaI AIternative Designs

62 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
A direct solution oI the time-dependent N-S equations oI Iully turbulent Ilows at high
Reynolds numbers, thereIore, requires extremely Iine geometric grids and extremely small
time steps. Thus, the computing requirements Ior direct solution are truly phenomenal and
impractical Ior Iire modeling applications. Certain assumptions must thereIore be made to
avoid the need to predict the eIIects oI each and every eddy in the Ilow.
Several such turbulent modeling approaches have been used and depend mostly on the type oI
engineering problem to be solved. These models can be separated into two broad classes: (1)
eddy viscosity models and (2) second-order closure models. Eddy viscosity models speciIy
the Reynolds stresses and Iluxes algebraically in terms oI known mean quantities. Second-
order closure models solve diIIerential transport models Ior the turbulent Iluxes.
The k H model, an example oI eddy viscosity model based on the time-averaged Reynolds
equations, is widely used to develop the time-averaged approximation to the conservation
equations oI Iluid dynamics. This model results in two additional partial diIIerential equations
per control volume. The Iirst equation governs the distribution oI turbulent kinetic energy, k,
while the second describes rate oI the dissipation oI local turbulent kinetic energy, H. A
number oI variations oI the k H model exist. The so-called standard k H model is widely
used. One oI the main drawbacks oI this model is that the eddy viscosity is assumed to be
identical Ior all oI the Reynolds stresses, so that the turbulence has no preIerence direction.
Several modiIications to the standard k H model have been used to predict the plume
entrainment and jet Ilows. Most Iire scenarios will involve transient Iire growth, Iueled by
radiative Ieedback between the Iire source and conIining boundaries. The mass oI air
entrained into a Iire plume controls, to a considerable degree, the process oI smoke Iilling, the
concentrations and temperature in the hot layer and the combustion in the Ilame. Since
gravitational Iorce is applied only in the vertical direction, the standard k H model does not
model the plume correctly. This has been amended by using a k H model with buoyancy
modiIication.
Another common way oI modeling turbulence is termed 'Large Eddy Simulation (LES),
where the time-dependent Ilow equations are solved not only Ior the mean Ilow but also Ior
the largest eddy structures characteristic oI most Iire plumes. The phrase LES reIers to the
description oI turbulent mixing oI gaseous Iuel and combustion products with the local
atmosphere surrounding the Iire. The basic idea behind the LES technique is that the eddies,
which account Ior most oI the mixing, are large enough to be calculated with reasonable
accuracy Irom the equations oI Iluid dynamics.
The equations describing the transport oI mass, momentum and energy by Iire-induced Ilows
must be simpliIied so that they can be eIIiciently solved Ior the Iire scenarios oI interest. The
general equations oI Iluid dynamics describe a rich variety oI physical processes, many oI
which have little to do with Iires. The simpliIying equations have been widely adopted by the
larger combustion research community, where they are reIerred to as the 'low Mach number
combustion equations. They describe the low speed motion oI a gas driven by chemical heat
release and buoyancy Iorces. The Low Mach number equations are solved numerically by
dividing the physical space where the Iire is to be simulated into a large number oI
rectangular cells, within each cell, the gas velocity, temperature, species concentration, etc.,
are assumed to be uniIorm, only changing with time. The accuracy with which the Iire can be
simulated depends on the number oI cells that can be incorporated into the simulation.
Further work is clearly needed on the turbulence models used in CFD codes Ior Iire
applications. Any progress in such modeling must be based on relevant experimental data Ior
a wide range oI Ilow conditions.



Appendix 6 Design TooIs for TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 63!
3.2.2 Radiation Modeling
The radiative transIer equation is an integro-diIIerential equation, and its solution, even Ior a
two-dimensional, planar, gray medium, is quite diIIicult. In Iires, the multidimensional
combustion system consists oI a highly non-isothermal and non-homogeneous medium where
spectral variation oI radiative properties oI the medium must be accounted Ior. It is necessary
to introduce some simpliIying assumptions and strike a compromise between accuracy and
computational eIIort.
The problem is usually divided into two parts: Iirst, an appropriate solution method must be
chosen Ior the integro-diIIerential equation, and second, an assumption must be made on the
radiative properties oI the medium (i.e., combustion gases and particles). The solution
methods can be divided into the Iollowing categories: exact models, statistical methods, zonal
methods, Ilux methods and hybrid methods.
The properties oI combustion systems are a complicated Iunction oI wavelength, temperature,
pressure, composition and path length. The products oI combustion usually consist oI
combustion gases such as H
2
O, CO
2
, CO, etc, and particles such as soot. The combustion
gases are strong absorbers and emitters oI radiant energy, but these radiative properties are a
strong Iunction oI wavelength. Consequently, the variation oI the radiative properties with the
electromagnetic spectrum must be accounted Ior.
Spectral calculations are perIormed by dividing the entire wavelength (or Irequency) into
several bands and assuming that the absorption/emission characteristics oI each species
remain either uniIorm or change smoothly over these bands. The accuracy oI the predictions
is expected to increase as the width oI these bands become narrower. A number oI approaches
to solve this problem have been suggested. Among them, the total absorptivity-emissivity
models, wide-band models and narrow-band models are the most commonly used. From a
series oI numerical experiments, it has been Iound that six (6) bands are usually enough. II
the absorption oI the Iuel is known to be important, separate bands can be reserved Ior Iuel,
and the total number oI bands is increased to ten (10).
3.2.3 Combustion Modeling
The mechanism by which species are Iormed and destroyed in Iire is extremely complex and
involves chemical and physical processes on a molecular and macroscopic level. Ignition,
combustion and extinction occur at the same time within the microstructure oI a turbulent
Ilame. These events occur at high Irequencies with spatial separation oI only a Iew
millimeters. The mixture oI gases can be diluted by complete or incomplete products oI
combustion at a given location. Thousands oI diIIerent states can thus exist at diIIerent points
within the Ilame, at a given time.
In order to avoid these complications, one can give the heat release rate in a certain control
volume as user input, and thereIore not deal with combustion at all. But Ior Iire applications,
it is important to allow the process oI Iuel and air mixing so that the heat release rate oI the
Ilame can be determined by actual Ilow conditions and oxygen concentration levels. This also
allows the prediction oI species concentration and estimation oI soot concentrations, which
has important signiIicance Ior the radiation calculations.
The range oI models used Ior combustion Iall essentially into two categories: (1) models
based on a conserved scalar approach; and (2) Ilamelet combustion models. One typical
choice Ior the conserved scalar approach is a mixture Iraction. The mixture Iraction model
assumes an inIinite combustion reaction. Neither Iuel nor oxidizer can co-exist in the Iuel or
oxidizer streams. Mixture Iraction is used to represent that local concentration oI Iuel,
oxidizer (i.e., oxygen) and the products. The mass Iractions oI all oI the major reactants and
products can be derived Irom the mixture Iraction by means oI the state relationship Ior the
speciIied Iuels.



Appendix 6 Design TooIs for TriaI AIternative Designs

64 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
An extension oI the conserved scalar approach is achieved by taking into the consideration a
second variable, which concentration is aIIected by the chemical kinetics. A popular and
rather successIul example oI this approach is the Eddy-Break-up model. Based on some
reasonably good correlation with measured data, the eddy break-up model assumes a single,
one-step reaction and inIinitely Iast reactions.
However, the Iast chemistry assumption is not always valid. Under the conditions oI Iuel
ignitions and Iire extinctions, the production oI many important species in turbulent Ilames is
chemically controlled, thus, the Iinite rate kinetics eIIects should be modeled. One way to
incorporate Iinite rate chemistry is based on laminar Ilamelet considerations.
The Iinite reaction rate model assumes that the combustion occurs locally in thin laminar
Ilamelets embedded within the turbulent Ilow Iield. For simple Iuels such as methane and
propane Ior which the chemistry is suIIiciently well known, the relationships between the
instantaneous species concentration and mixture Iraction can be calculated directly. This
requires that laminar Ilamelet libraries be established Irom experiments, where the state
relationships oI species concentrations, temperature, enthalpy, viscosity, density and soot
concentrations are stored as a Iunction oI mixture Iraction.
3.3 Boundary and InitiaI Conditions
In order to Iully speciIy a problem, a set oI boundary and initial conditions must be provided.
Boundary conditions place limits on the physical environment. The limits take the Iorm oI a speciIied
parameter value (e.g., solid wall with zero velocity), a Ilux value (e.g., mass Ilow rate) or a time rate
oI change at a speciIied position. Boundary conditions generally can be in two categories: thermal
boundary conditions and velocity boundary conditions. There are Iour types oI thermal boundary
conditions: adiabatic, constant temperature, thermally thick and thermally thin. Only one can be
chosen in one surIace. II the surIace material is assumed to ignite and burn at certain temperature, the
relationship between the pyrolyzing Iuel and the rate oI energy released should be taken into account.
Initial conditions are important Ior transient problems and speciIy the status oI the physical
environment at the start oI the simulation.
Wall boundary conditions are used to speciIy the Iluid velocities adjacent to the wall surIaces, the
wall shear stresses (related to Iluid viscosity), the velocity oI the wall (iI it is moving) and the heat
transIer characteristics. Adjacent velocity boundary conditions aIIect both the normal and tangential
components oI the velocity vector at boundaries. The tangential boundary condition can be no-slip,
Iree-slip or something in between. The normal velocity conditions can be speciIied velocity proIiles
(e.g., parabolic proIile) or a given volume Ilux. Since a large viscosity gradient can occur next to the
wall, many oI the Iluid properties will vary rapidly in the vicinity oI the walls. Very Iine meshes are
usually required to accurately predict the Iluid properties near the wall.
Doors, windows and other types oI vents are usually speciIied as either inlet or pressure boundaries.
When using inlet boundary conditions, thermal and velocity boundary conditions must be speciIied.
However, a pressure boundary could be more viable Ior a vent considered as part oI Iire analysis. In
this case, the pressure is set equal ambient and the derivatives oI the velocity components normal to
the vent surIace are set equal to zero. This permits the Ilow to enter or leave the computational grid, as
required.
A Iinal type oI boundary condition is a plane or axis oI symmetry. All variables are mathematically
symmetrical, thus with no diIIusion across the boundary. For example, Ior a Iire located in the center
oI a room, symmetry planes can be used to model a quarter oI room. This would allow a Iour-Iold
increase in the number oI cells used to model the problem.



Appendix 6 Design TooIs for TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 65!
3.4 Current AvaiIabIe ModeIs
Over the past decades a number oI general-purpose computer programs have been developed that
permit the solution oI the equations describing Iluid Ilow. A brieI review oI selected Iield models is
given below.
Model Names Countrv Description Availabilitv
ALOFT-FT US Smoke movement Irom large outdoor Iires Freely available
Irom NIST
CFX UK/US General purpose CFD soItware, applicable to Iire and
explosion
May be purchased
FDS (Fire
Dynamics
Simulator)
US A 3-D CFD Iield model speciIied to Iire-driven Ilow oI low
Mach number, predicting the smoke and hot Ilow movement,
and response oI sprinkler links. User-Iriendly. Using LES
technique. Developed by NIST.
Freely available
Irom NIST
FIRE Australia CFD model with water sprays and coupled to solid/liquid
phases Iuel to predict burning rate and extinguishment
May be purchased
FLOW3D General-purpose Iluid dynamic code developed by a Iinite-
diIIerence, transient-solution algorithm solving the convection
equations oI Iluid dynamics.
May be purchased
or leased
FLUENT/
AirPak
US General purpose CFD soItware, applicable to Ilows with heat
transIer and chemical reactions
May be purchased
JASMINE UK Field model Ior analysis oI smoke movement in enclosure
developed Ior Iire simulations (based on PHOENICS).
Restricted
KAMELEON
FireEX
Norway CFD model linked to a Iinite element code Ior thermal response
oI structure
Restricted
KOBRA-3D Germany A 3-D Iield model Ior smoke spread and heat transIer in
complex geometries
May be purchased
MEFE Portugal CFD model Ior one or two compartment, includes time-
response oI thermocouples
Restricted
PHOENICS UK A general purpose, 3-d transient Iluid dynamics code May be purchased
RMFIRE Canada A 2-D Iield model Ior transient calculation oI smoke movement
in room Iires
Restricted
SMARTFIRE UK Field Iire model to investigate the spread oI Iire hazards
through an enclosure
May be purchased
SOFIE
(Simulation oI
Fires in
Enclosure)
US/Sweden A CFD program contains a multitude oI submodels specially
developed Ior applications. Currently not very user-Iriendly,
requires extensive training.
Restricted
SOLVENT US CFD model Ior smoke and heat transport in a tunnel Restricted
SPLASH UK A quasi-Iield model describing the interaction oI sprinkler
sprays with Iire gases
Restricted
STAR-CD UK General purpose CFD soItware May be purchased
UNDSAFE US/Japan Fire Iield model Ior use in open space, or in enclosure, using
3-D Iinite diIIerence scheme
Restricted




Appendix 6 Design TooIs for TriaI AIternative Designs

66 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
3.5 Limitations of FieId ModeIs
While Iield models provide more detail than zone models, they do have limitations. The most
signiIicant limitation oI the Iield model is the cost. Conventional CFD models do not require the
sophistication oI a pyrolysis or combustion model. The chemistry and physics oI the combustion
process is extremely complicated. The Iield models which predict Iire scenarios are thus signiIicantly
more expensive than a conventional CFD model. Although the involved costs continue to limit the
widespread use oI Iield models in Iire protection engineering applications, the Iast advance oI
computer technology and computational techniques will increase the capacity oI the usability oI Iield
models.
Some oI the limitations oI Iield models come Irom the theoretical approximations oI CFD and
combustion chemistry. Field models do not have a direct simulation oI turbulent diIIusion Ilames.
Except Ior some limited cases, the Iire sources must be prescribed by the user. Other major
phenomena that can only be approximated include turbulence, particularly large eddies associated
with strong plumes and Ilames, and thermal radiation interchanges between soot, gases and solid
surIaces. In some cases, the Iuel and oxidizer (air) are initially separate and combustion occurs in the
zone where they mix. Field models do not have a direct simulation oI turbulent diIIusion Ilames.
Some Iield models even yield incorrect results Ior small Iires in a large enclosure or big Iires in a
small enclosure. As the development and application oI Iield models continue, these limitations
should gradually be eliminated.
In addition, the application oI Iield models requires a great deal oI user sophistication to speciIy the
problem and interpret the results. Necessary training is signiIicant to eIIectively implement Iield
models, which require the model users to develop a thorough understanding oI the physics and
chemistry behind the Iire dynamics models.
3.6 Comparisons between FieId ModeIs and Zone ModeIs
Using computer-based Iire models to analyze Iire protection engineering problems is becoming
imperative. Zone models have been successIully applied to a wide range oI these problems. As the
problems grow more complex, zone models will be inadequate to Iully address them. Zone models
provide very limited detail, with bulk average values being predicted in a Iew select locations within
the enclosure. Zone models utilize equations employing empirical relationships and constants
obtained Irom experiments. Such empirical expressions used to describe physical behavior in zone
models could break down as the geometry becomes more complex. ThereIore, the use oI zone models
Ior problems that lie outside the range oI experiments is very limited.
Field models avoid the simpliIications inherent in zone models. In solving the Iundamental equations
oI mass, species, momentum and energy, the compartment is divided into up to millions oI
computational cells. Thus, mass, enthalpy, Ilow velocity, temperature, etc., are calculated Ior each cell
in the grid, and the distribution proIiles are thereaIter provided. In some Iire cases, the geometry oI
the room and its outIitting can have signiIicant eIIects on the nature oI recirculation patterns, thus, the
higher spatial resolution oI Iield models can be important. In many cases, the detailed knowledge oI
the temperature and/or Ilow Iields near sprinklers or smoke detectors is required to accurately predict
the activation.
The use oI Iield modeling to analyze Iire protection engineering problems is growing dramatically. Its
use will become imperative as the complexity oI problems increase. The details about Iluid Ilow and
heat transIer provided by Iield models can prove vital in analyzing problems involving Iar Iield smoke
Ilow, complex geometry (e.g., sprinkler links, Iurniture, etc.), and impact oI Iixed ventilation Ilows.
Fluid dynamics considerations are automatically built into Iield models, rather than being Iorced into
oversimpliIied approximations. Thus, Iield models Iollow the movement oI the plume, rather than
assuming that deposition oI mass and energy Irom combustion/plume zone into the upper layers is
instantaneous. Similarly, they describe the spread oI the ceiling jet into the entire upper layer, rather
than assuming the instantaneous mixing within it.



Appendix 6 Design TooIs for TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 67!
A* 5=L'&*#8';:<9X8N&8%D'*G&%E&<ID*
Over the past Iew decades, a number oI computer Iire models have been developed Ior various Iire
protection applications. While the commonly used zone models and Iield models discussed in
previous Subsections serve Ior multiple applications, there are many other computer models designed
Ior special purposes (e.g., egress, Iire resistance, Iire suppression, smoke control, etc.). Some are only
a collection oI several useIul calculation routines, some programs are available on the Internet Ior Iree
download, some require registration and a small handling Iee, and a Iew programs are commercially
available at relatively high prices. The Iollowing is a collection oI some special-purpose models
available or cited in Iire protection applications:
4.1 Egress ModeIs
Egress models predict the time Ior occupants oI a structure to evacuate. A number oI egress models
are linked to zone models, which will determine the time to the onset oI untenable conditions in a
building or a vessel. Egress models are oIten used in perIormance-based design analysis Ior
alternative design code compliance and Ior determining where congestion areas will develop during
egress.
Many oI these models are quite sophisticated, oIIering computational methods, as well as interesting
Ieatures including the psychological eIIects on occupants due to the eIIects oI smoke toxicity and
decreasing visibility. Many oI these models also have useIul graphic Ieatures so that movement oI
people inside a enclosure can be visualized during a simulation. A brieI review oI selected egress
models is given below.
Model Countrv Description Availabilitv
AllsaIe Norway Egress model including human Iactors N/A
ASERI Germany Movement oI people in complex geometries, including
behavioral response to smoke and Iire spread
N/A
Marinetime
EXODUS
UK A sophisticated evacuation model, taking account oI people-
people, people-Iire and people-structure interactions. It allows
evacuation simulation oI complex vessel structures with many
occupants.
High cost Ior
commercial uses
ESSCAPE Australia Evacuation oI multi-story building via staircases N/A
EGRESS UK Cellular automatic evacuation oI multiple people through
complex geometries. Includes visualization.
N/A
EGRESSPro Australia Egress Program that includes coping times and sprinkler-
detector activations
N/A
ELVAC US Egress program Ior use oI evaluators Ior evacuation N/A
ERM Sweden The Escape and Rescue Model (REM) is based on the same
node and arc method as EVACNET, but is developed
especially Ior hospitals and healthcare Iacilities.
Free
EVACNET Sweden A classic 'network type oI model which determines optional
building evacuation plan. The user deIines a system oI nodes
and arcs, where occupants are positioned at the nodes and
move along the arcs towards the exit.
Free
EVACS Japan Evacuation model Ior determining optimal plan N/A
EXIT89 US Evacuation Irom a high-rise building N/A
EXITT US Node and Arc type egress model with people behavior included N/A



Appendix 6 Design TooIs for TriaI AIternative Designs

68 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
Model Countrv Description Availabilitv
HAZARD I US This multi-component Iire-hazard program calculates the
development oI Iire eIIects (thermal, toxic gases, smoke) and
the interactions and movements oI occupants as well as
detector activation times.
Free
PATHFINDER US Egress model N/A
SEVE-P France Egress model with graphic output that includes obstructions N/A
SIMULEX UK Coordinate-based evacuation model Ior use in geometrically
complex multi-story buildings with many occupants. It allows
input oI CAD drawings Ior a plan oI the building.
Free
STEPS UK Egress Model N/A
WAYOUT Australia Egress part oI the FireWind suite oI programs N/A

4.2 Smoke ControI ModeIs

Model Countrv of
Origin
Comments Availabilitv
ASCOS US (NIST) A program Ior steady airIlow analysis oI smoke control
systems.
Free
FIRE-1.2 Germany/
Norway
Hydrocarbon Iires: eight scenarios
CONTAMW US (NIST) A multi-zone indoor air quality and ventilation analysis
program to predict airIlows, contaminant concentrations and
personal exposures to contaminants.
Free
MFIRE US Mine ventilation network
RISKCOST Canada LiIe and cost in multistory building
SMACS US Smoke in HVAC system
SPREAD US Spreading Iires on wall
UFSG US Upward wall Iire spread
WALLEX Canada Window Fire Plume

4.3 Fire Endurance ModeIs
Fire endurance models simulate the response oI vessel structural elements to Iire exposure. Some oI
these models are stand-alone while others are incorporated into zone or Iield models. The concept oI
Iire endurance models is similar to Iield models. The structural object is divided into smaller volumes,
and the equations Ior thermal heat transIer and mechanical behavior Ior solids are solved to determine
when the structure will Iail. Typically, the material properties are required input Ior the model, as well
as the boundary conditions (i.e., the Iire exposure) Ior the structural element.



Appendix 6 Design TooIs for TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 69!
These models are very useIul Ior determining when a beam or column will deIorm or Iail, and Ior
solving Ior a temperature versus time curve at a certain depth inside the structural element. Since
many structural elements are constructed diIIerently, have diIIerent Ieatures or have diIIerent practical
applications, care must be used in selecting a model that properly characterizes the structural element.
A brieI review oI selected Iire endurance models is given below.
Model Countrv of
Origin
Comments
CEFICOSS Belgium Fire resistance model
CIRCON Canada Fire resistance model oI loaded, reinIorced concrete columns with a circular cross
section
CMPST France Mechanical resistance oI sections at elevated temperatures
COFIL Canada Fire resistance model oI loaded, circular hollow steel columns Iilled with plain
concrete
COMPSL Canada Temperatures oI multi-layer slabs during exposure to Iire
FIRE-T3 US Finite element heat transIer Ior 1-, 2-, or 3-D conduction
HSLAB Sweden Transient temperature development in a heated slab composed oI one or several
materials
INSTAI Canada Fire resistant model oI insulated, circular hollow steel columns
INSTCO Canada Fire resistant model oI insulated, circular concrete-Iilled tabular steel columns
LENAS France Mechanical behavior oI steel columns exposed to Iire
RCCON Canada Fire resistance oI loaded, reinIorced concrete columns with rectangular cross sections
RECTST Canada Fire resistance oI insulated rectangular steel columns
SAFIR Belgium Transient and mechanical analysis oI structures exposed to Iire
SAWTEF US Structural analysis oI metal-plate connected wood trusses exposed to Iire
SISMEF France Mechanical behavior oI steel and concrete composite structures exposed to Iire
SQCON Canada Fire resistance oI square reinIorced concrete columns
STA UK Transient conduction in heated solid objects
TASEF Sweden For 2-3D and axisymmetric shapes, a Iinite element analysis model oI temperature
distribution through a structure exposed to Iire
TCSLBM Canada 2-D temperature distributions Ior Iire-exposed concrete slab/beam assemblies
THELMA UK Finite element code Ior thermal analysis oI building components in Iires
TR8 New
Zealand
Fire resistance oI concrete slabs and Iloor systems
WSHAPS Canada Fire resistance oI loaded, protected W-shape steel columns




Appendix 6 Design TooIs for TriaI AIternative Designs

70 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
4.4 Fire Detection ModeIs

Model Countrv of
Origin
Comments
DETECT-QS US (NIST) Detector actuation-Quasi steady
DETECT-T2 US (NIST) Detector actuation-Time square
LAVENT US (NIST) Response oI sprinklers in enclosure Iires with curtains and ceiling Iires
PALDET Finland UnconIined ceiling
TDISX US UnconIined ceiling, treats Ilow transient

4.5 Fire Suppression ModeIs

Model Countrv of
Origin
Comments
FISCO-3L Norway One-room Iield model, suppression
RADISM UK Zone model, including venting
SPLASH UK Field model, no suppression
FIRDEMND US (NIST) Simulating the suppression oI post Ilashover charring and non-charring solid Iuel Iires
in compartments using water sprays Irom portable hose-nozzle equipment.

4.6 Fire ModeIs from NIST
OI note is that the Building and Fire Research Laboratory at the National Institute oI Standards and
Technology (NIST) has made a large number oI programs Ireely available. In addition to previously
introduced programs and models, the Iollowing programs are also available at NIST`s URL address:
Model Comments
ALOFT-FTTM A Large Outdoor Fire Plume Trajectory model Flat Terrain
ASMET A set oI equations and a zone model Ior analysis oI smoke management system Ior large spaces such as
atria, shopping malls, sport arenas, exhibition halls and airplane hangers, etc.
BREAK1 The program (Berkeley Algorithm Ior Breaking Window Glass in a Compartment Fire) calculates the
temperature history oI a glass exposed to user described Iire conditions.
CCFM
The program (Consolidated Compartment Fire Model version VENTS) is a two-layer zone-
type compartment Iire model computer code, and simulates conditions due to user-speciIied
Iires in a multi-room, multi-level Iacility.
FPETool A set oI engineering equations useIul in engineering potential Iire hazard and the response oI the space
and Iire protection systems to the developing Iires. It also provides the estimation oI the smoke
conditions and human viability resulting Irom exposure to developing conditions within the room.
FASTLite A soItware package that builds on the core routines oI FPETool and computer model CFAST to
provide calculations oI Iire phenomena Ior use.




Appendix 6 Design TooIs for TriaI AIternative Designs

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 71!
B* .'C'&'>;'D*
1. Walton, W.D., 'Zone Computer Fire Models Ior Enclosure, the SFPE Handbook of Fire
Protection Engineers, 2
nd
Edition, Section 3-7, NFPA, Quincy, MA, 1995.
2. McGrattan, K.B., et. al, 'Fire Dynamics Simulator (Version 2) Technical ReIerence Guide,
NISTIR 6783, National Institute oI Standards and Technology, November 2001.
3. Stroup, D.W., 'Using Field Modeling to Simulate Enclosure Fires, the SFPE Handbook of
Fire Protection Engineers, 2
nd
Edition, Section 3-8, NFPA, Quincy, MA, 1995.
4. Quintiere, J.G., 'Compartment Fire Modeling, the SFPE Handbook of Fire Protection
Engineers, 2
nd
Edition, Section 3-5, NFPA, Quincy, MA, 1995.
5. Jones, W.W., 'State oI Art in Zone Modeling oI Fires, Reprinted Irom 9
th
International Fire
Protection Seminar, Proceedings of Engineering Methods for Fire Safetv, pp. A. 4/89-126,
Munich, Germany, May 25-26, 2001
6. Forney, G.P., Moss, W.F., 'Analyzing and Exploiting Numerical Characteristics oI Zone
Models, Fire Science & Technologv, Vol. 14, No. 1 & No. 2, pp 49-62, 1994.
7. Standard Guide for Evaluating the Predictive Capacitv of Deterministic Fire Models, ASTM
E 1355-97, American Society Ior Testing and Materials, 1997.
8. Bukowski, R.W., 'Fire Hazard Assessment, the NFPA Fire Protection Handbook, 18
th

Edition, Section 11/7, NFPA, Quincy, MA, 1996.
9. Peacock, R.D., et al, 'Evaluation oI Complex Fire Models, NISTIR 6030, NIST,
Gaithersburg, MD, 1997.
10. Veynante, D, Verisch, L., 'Turbulent combustion modeling, Progress in Energv and
Combustion Science, Vol. 28, pp. 193-266, Elsevier Science, 2002.
11. Karlsson, B., Quintiere, J.G., Enclosure Fire Dvnamics, CRC Press, 2001.
12. Rehm, R.G., Baum, H.R., 'The Equations oI Motion Ior Thermally Driven, Buoyant Flows,
J. of Research of the NBS, 83:297-308, 1978.
13. Novozhilov, V., 'Computational Iluid dynamics modeling oI compartment Iires, Progress in
Energv and Combustion Science, Vol. 27, pp. 611-666, Elsevier Science, 2001.
14. Cox, G., 'Compartment Fire Modeling, Combustion Fundamentals of Fire, Chapter 6,
Academic Press, London, 1995.
15. Freites, C.J., 'Perspective: Select Benchmarks From Commercial CFD Codes`, Journal of
Fluids Engineering, Vol. 117, pp 208-218, 1995.
16. Olenick, S. M., Carpenter, D. J., 'An Updated International Survey oI Computer Models Ior
Fire and Smoke, Journal of Fire Protection Engineering, Vol. 13, May 2003.


!

This Page Intentionally LeIt Blank


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 73!

A P P E N D X R** 2'H'9%8:>E*G'&C%&I<>;'XK<D')*
P&:='&:<*
7* -CC';=D*%C*+:C'*#<C'=F*P&:='&:<*
1.1 Effects of Toxicity
With regard to hazard assessment, the major considerations oI human health when exposed to hazards
are:
i) The time when partially incapacitating eIIects are likely to occur which might delay escape.
ii) The time when incapacitating eIIects are likely to occur which might prevent escape,
compared with the time required Ior escape.
iii) Whether exposure is likely to result in permanent injury or death.
Despite the great complexity in chemical composition oI a smoke atmosphere, the basic toxic eIIects
are caused either by a narcotic (asphyxiant) gas or by irritants.
Narcotic gases cause incapacitation mainly by eIIects on the central nervous system and, to some
extent, the cardiovascular system. The two major narcotic gases in Iires are (1) carbon monoxide (CO)
and (2) hydrogen cyanide (HCN). In addition, low concentrations oI oxygen and very high
concentrations oI carbon dioxide (CO
2
) can also have narcotic eIIects. It is generally recognized that
the vast majority oI deaths associated with accidental enclosure Iires are due to smoke inhalation.
Carbon monoxide combines with hemoglobin in the blood to Iorm carboxyhemoglobin (COHb),
which results in toxic level oI narcotics by reducing the amount oI oxygen supplied to the tissues oI
the body, particularly brain tissue. It has also been shown that the vast majority oI these Iire victims
have COHb levels in their bloodstream suIIicient to induce incapacitation or death. This has led many
researchers to conclude that carbon monoxide (CO) is the dominating toxicant present in Iire cases. It
is also recognized that the elevated carbon dioxide (CO
2
) levels (which result in increased respiration
rates) and depressed oxygen (O
2
) levels associated with the Iire cases act together to increase the
susceptibility oI victims to CO asphyxiation.
Irritant Iire products produce incapacitation during and aIter exposure in Iorms either oI sensory
irritation (consisting oI eye and respiratory tract pain, lacrimation and breathing diIIiculties) or oI
acute pulmonary irritant response.
The degree oI toxicity is determined by Iactors such as the concentration oI toxic product in the target
organ oI the body, and the time period Ior which a toxic concentration is maintained. The
relationships between concentrations inhaled, duration oI exposure and toxicity should be properly
considered based on the characteristics oI Iire scenarios (i.e., material compositions, ventilation, etc.).




Appendix 7 DeveIoping Performance-based Criteria

74 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
1.2 Effects of Smoke
Smoke comprises the total eIIluents Irom a Iire and consists oI two major parts: the invisible vapor
phase and the visible particulate phase. From a toxicological standpoint, all oI the narcotic products
occur in the vapor phase, while irritant products may occur in both phases. The particulate phase
consists oI solid and liquid particles covering a wide range oI particle sizes, depending upon the
nature and age oI the smoke. These particles may contain condensed liquid or solid irritant products;
or irritant products including gaseous ones may be dissolved in liquid particles (as in acid mists), or
may be absorbed on to the surIace oI solid, carbonaceous particles.
Particle size is oI great toxicological importance since it determines how 'deeply particles penetrate
into the respiratory tract and the patterns oI subsequent deposition. For example, particles with a mean
aerodynamic diameter oI less than 5 Pm are capable oI penetrating deep into the lung, while larger
particles tend to deposit in the nasal passages and upper airways.
The other important physiological eIIects oI the particle phase oI smoke is visual obscuration, which
in conjunction with irritant eIIects on the eyes, may impair the ability oI victims to escape Irom Iires.
It is best represented in a hazard model in terms oI a tenability limit concentration. Smoke obscuration
is usually expressed in terms oI smoke density (OD/m) or extinction coeIIicient, K, (K OD/m u 2.3).
1.3 Effects of Radiant Heat
Radiation is important in situations where occupants must pass close to the seat oI Iire in order to
escape or in situations where occupants must pass under a hot eIIluent layer.
Pain occurs when the diIIerence between the rate oI supply oI heat to the skin surIace exceeds the rate
at which heat is conducted away by an amount suIIicient to raise the skin temperature to 44.8C
(111F) at a depth oI 0.1 mm (0.004 in.). The eIIects oI heating the skin are essentially the same,
regardless oI the means oI heat transIer.
The perIormance criterion with respect to thermal radiation is the condition when suIIicient heat is
applied to unprotected or naked skin to induce pain. Pain and damage to skin begin to occur when the
temperature at the basal level (approximately 80 micron (0.02 in) beneath the surIace oI the skin)
exceeds 44C (111F).
A threshold heat Ilux on skin is suggested to be approximately 2.5 kW/m
2
. Below this level, exposure
can be tolerated Ior several minutes and, above which, tolerance time rapidly decreases to a Iew
seconds.
1.4 Egress AnaIysis
A quantitative approach to the egress movement oI people must be balanced by a qualitative
understanding oI the context within which the movement takes place. Egress time can be predicted
either by hand calculations or by available, reliable egress models. When deIining the perIormance
criteria, egress calculations shall be considered as providing only minimum evacuation times.
Tenability issues are important both with respect to the time required Ior escape and the time available
Ior escape.
1.5 Effects of Fire Extinguishing Agents on Occupants
The eIIects oI the by-products oI Iire extinguishing agents shall be properly addressed iI they are used
in the alternative design. In their natural state (the state at rest without being actually discharged into a
Iire), some oI these agents may not be toxic, in Iact most are not toxic iI actually discharged and there
is no heat source. However, iI inhaled, some agents may have toxicological eIIects known as cardiac
sensitization. Cardiac sensitization occurs when a chemical causes an increased sensitivity oI heart to
adrenaline, leading to the sudden onset oI irregular heartbeats and possibly heart attacks.



Appendix 7 DeveIoping Performance-based Criteria

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 75!
Any agent that is to be recognized by these Guidance Notes or proposed Ior inclusion in the
alternative designs should be evaluated in the same manner, Ior instance, as the process used by the
U.S. Environmental Protection Agent`s (EPA) SNAP Program. The SNAP Program was originally
outlined in Federal Register, 59 FR 13044. More details oI SNAP program can also be Iound in the
oIIicial EPA web site listed in the ReIerences. For example, Ior halocarbon clean agents, the toxicity
inIormation (LC
50
or ALC, NOAEL and LOAEL) and the time Ior saIe human exposure should be
evaluated. The LC
50
is the concentration lethal to 50 percent oI a rat population during a 4-hour
exposure. The ALC is the approximate lethal concentration. The NOAEL is the highest concentration
at which no adverse physiological or toxicological eIIect has been observed. The LOAEL is the
lowest concentration at which no adverse physiological or toxicological eIIect has been observed. For
some Iluorocarbon Iire-extinguishing agents, the toxicological eIIects oI hydrogen Iluoride (HF)
should be evaluated. For some inert gas agents, the physiological eIIects (i.e., no eIIect level, low
eIIect level, etc.) should be evaluated.
The design team oI the alternative design and arrangements should be aware oI the characteristics oI
occupancy. It is important to distinguish between normally healthy individuals, Ior example, Iire
Iighter crews, and those with compromised health. Exposure to higher concentration oI HF would be
expected to be tolerated more in healthy individuals, whereas at equal concentrations, escape-
impairing eIIects can occur in those with compromised health.
?* -D=<K9:DL:>E*G'&C%&I<>;'XK<D')*P&:='&:<*C%&*+:C'*#<C'=F*
The purpose oI this Subsection is to provide an example oI establishing the perIormance-based
criteria Ior liIe saIety. Assume that a design team is developing perIormance criteria to prevent the
loss oI liIe outside oI the room or compartment oI Iire origin. The design objective Ior liIe saIety is to
maintain tenable conditions in paths oI egress while occupants outside oI the room or compartment oI
Iire origin escape to a place oI saIety. The design team can set detailed perIormance criteria that
ensure that occupants are not incapacitated by Iire eIIects which include heat, temperature, toxicity
exposure, etc.
2.1 Heat
The tenable condition is to prevent pain and damage to skin (i.e., skin burn) Irom heat radiation and
hot gas exposure. See A7/1.3, 'The EIIects oI Radiant Heat.
2.2 VisibiIity
Beside the toxicological signiIicance oI smoke, the other important psychological eIIect oI the
particulate phase oI smoke is visual obscuration. Depending on whether occupants are Iamiliar with
the escape route, as well as the use oI the space, diIIerent obscuration criteria can be established. For
occupants Iamiliar with the escape route, a criterion oI OD/m 0.5 (approximately 2.5-meter
visibility) is used, and Ior occupants unIamiliar with the escape route, an OD/m 0.008
(approximately 10-meter visibility) criterion is suggested.
2.3 Egress
NFPA 101 A.5.2.2 requires that a design team shall demonstrate that smoke and the toxic gas layer
will not descend to a level lower than 1.8 m (6 It) above the Iloor in any occupied room so that no
occupant is exposed to the eIIects oI Iire. A criterion oI 1.8 m is a conservative value to ensure that no
occupant need be exposed to Iire eIIects, regardless oI where occupants are or where they move.



Appendix 7 DeveIoping Performance-based Criteria

76 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
2.4 Toxicity
Carbon monoxide (CO) represents the most common narcotic Iire product, which accounts Ior over
halI oI all Iire Iatalities due to inhalation. The studies oI the relationship between time to
incapacitation and concentration in active monkeys show that Ior CO concentration oI 1000 ppm, it
takes about 26 minutes, and Ior CO concentration oI 2000 ppm, the time reduces to about 15 minutes.
When CO concentration increases to 8000 ppm, it takes less than Iour minutes Ior incapacitation. A
conservative CO threshold oI 1200 ppm may be chosen based on the revised IDLH concentration
developed by the National Institute Ior Occupational SaIety and Health (NIOSH), which ensures that
occupants will be able to escape without incapacitation due to inhalation.
2.5 Performance-based Criteria
Based on the above analysis, some oI the perIormance-based criteria or threshold values Ior heat,
visibility, egress and toxicity can be summarized as Iollowing:
x Temperature T
skinburn
44C (111F)
x Heat Ilux
skin
q c c 2.5 kW/m
2

x Egress Height oI upper layer smoke ~ 1.8 m (6 It)
x Visibility OD/m 0.5, or about 2 m (6.5 It) oI visibility
x Toxicity CO 1200 ppm
With these perIormance criteria established, the design team can proceed with the evaluation oI the
alternative design(s).
@* /%>X9:C'*#<C'=F*P&:='&:<*
Non-liIe saIety criteria address issues relating to damage thresholds Ior property. Damage thresholds
may relate to thermal energy exposure, resulting in ignition or unacceptable damage. Thresholds
might also consider exposure to smoke aerosols and particulate or corrosive combustion products. In
some cases, unacceptable damage might result Irom small exposure levels.
3.1 ThermaI Effects
Thermal eIIects might include melting, charring, deIormation, or ignition. Considerations include the
source oI energy (e.g., convection, conduction and radiation), the distance oI the target Irom the
source, the geometry oI the source and the target, the material characteristics oI the target (e.g.,
conductivity, density and heat capacity) and the ignition oI the target. The surIace area to mass ratio
oI the Iuels involved is also a Iactor.
3.2 Fire Spreads
The spread oI Iire by progressive ignition should be considered. Factors aIIecting Iire spread include
the geometry and orientation oI the burning surIaces (horizontal versus vertical) as well as the surIace
area to mass ratio oI the Iuels involved. Ventilation and airIlow can increase or decrease Iire spread.
Fire spread can also have an eIIect on liIe saIety. Rapid Iire spread can impair crew and passenger
egress.
3.3 Smoke Damage
Smoke damage includes smoke aerosols and particulate or corrosive combustion products. The
damage thresholds will depend on the sensitivity oI the target to damage. Some works oI art, such as
paintings, have low thresholds, whereas others, such as statuary, might tolerate more smoke. Many
targets, such as electronics, are sensitive to corrosive products at low levels.



Appendix 7 DeveIoping Performance-based Criteria

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 77!
3.4 Fire Barrier Damage and StructuraI Integrity
The loss oI Iire barriers can result in damage Irom extension oI heat and smoke. Opening protection
operations and penetrations are Iactors. Minimal acceptable perIormance in terms oI amount oI
potential Ior extension will depend on the sensitivity oI the target to heat and smoke. Structural
collapse is an issue in both liIe saIety and property protection. The stability oI a structure is important
Ior crews and passengers during the time necessary Ior egress and Ior emergency responders during
rescue and suppression activities.
3.5 Damage to Exposed Properties
PerIormance criteria might need to be developed to prevent or limit damage or Iire spread to exposed
properties. The mechanism oI damage can be heat or smoke. Separate distance, material Ilammability
characteristics and geometry are important considerations.
3.6 Damage to the Environment
PerIormance criteria, by limiting the eIIluent associated with Iire suppression systems and Iire-
Iighting operations, limiting the release oI contaminants Irom combustion and extinguishing media,
might need to be developed to protect the vessel environment.
A* .'C'&'>;'D*
1. The SFPE Engineering Guide to PerIormance-Based Fire Protection Analysis and Design oI
Buildings, Society oI Fire Protection Engineers and National Fire Protection Association,
1999.
2. ISO/TR 13387-1 through 8, 'Fire SaIety Engineering, International Organization Ior
Standardization (ISO), 1999.
3. SOLAS, Amendments 2000, IMO, London, 2001.
4. SOLAS, Consolidated Edition 2001, IMO, London, 2001.
5. Guide to Establish Equivalency to Fire SaIety Regulations Ior Small Passenger Vessels (46
CFR Subchapter), United States Coast Guard, 2001.
6. Purser, D., 'Toxicity Assessment oI Combustion Products, the SFPE Handbook oI Fire
Protection Engineers, 2
nd
Edition, Section 3-7, NFPA, Quincy, MA, 1995.
7. Engineering Guide Ior Predicting 1
st
and 2
nd
Degree Skin Burns Irom Thermal Radiation,
Society oI Fire Protection Engineers, 2000.
8. Pauls, J., 'Movement oI People, the SFPE Handbook oI Fire Protection Engineers, 2
nd

Edition, Section 3-7, NFPA, Quincy, MA, 1995.
9. NFPA 101: LiIe SaIety Code, 2000 Edition.
10 NFPA 301: Code Ior SaIety to LiIe Irom Fire on Merchant Vessels, 2001 Edition.
11. Ahlers, H., NIOSH documentation Ior Immediately Dangerous to LiIe or Health
Concentrations (IDLHs), National Institute Ior Occupational SaIety and Health, CDC, 1995.
12. NFPA 2001: Standard on Clean Agent Fire Extinguishing Systems, 2000 Edition.
13. U.S. Federal Register, Volume 59, Page 13044, EPA SNAP Program.
14. U.S. Environmental Protection Agent (EPA), 'Substitutes Ior Ozone-Depleting Substances,
October 9, 2003, http://www.epa.gov/ozone/snap/~.


!

This Page Intentionally LeIt Blank


!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 79!

A P P E N D X Y** !9='&><=:H'*2'D:E>*<>)*-J<I89'*
!><9FD:D*Z*!&&<>E'I'>=D*C%&*
P%>=<:>'&DL:8*P<&E%*#8<;'D*
0>=&%)N;=:%>*
SOLAS Chapter II-2 regulation 10.7.1.3 requires that cargo vessels oI 2000 gross tonnage and
upwards shall be protected by a Iixed carbon dioxide or inert gas Iire extinguishing system complying
with the provisions oI Fire SaIety Systems Code (FSS Code). FSS Code Chapter 5, section 2.2
stipulates that the quantity oI carbon dioxide available shall be suIIicient to give a minimum volume
oI Iree gas equal to 30 oI the gross volume oI the largest space to be protected.
During the discharge oI carbon dioxide in non-weathertight cargoes, the loss oI a certain amount oI
carbon dioxide is inevitable through the gaps and the labyrinths oI hatch covers. Although not
mandatory, FP 47/WP 6.2 suggests about 10 increase in CO
2
by taking into account the leakage oI
CO
2
Iire extinguishing media through clear gaps between hatchway covers.
Due to the complexity oI Iire scenarios, loading conditions oI cargoes and toxicity oI carbon dioxide,
very Iew data sources are available to correctly predict the leakage. In this study, the nature oI the
distribution oI carbon dioxide is investigated inside cargo holds during its discharge. The object oI
this project is to provide a clear picture oI carbon dioxide discharge and propose a perIormance-based
alternative Ior Iire engineering design. In this project, CFD models are used to simulate the transient
distributions oI carbon dioxide during the discharge.
3%<9D*%C*2'D:E>*
An important consideration in the design oI carbon dioxide total Ilooding systems in cargo holds and
machinery spaces is the successIul completion oI a Iull discharge test to veriIy, as required by SOLAS
or per NFPA 12, that carbon dioxide is indeed discharged through the system piping, and that the
design concentration is achieved and maintained Ior the required holding time.
For non-tight cargo holds, it is also important to estimate the leakage oI carbon dioxide and the
duration oI discharge to achieve the requirements oI CO
2
percent concentrations stipulated in SOLAS.
The narrow gaps or passages among containers make it diIIicult Ior CO
2
gas to penetrate through,
resulting in CO
2
escaping Irom the openings on the top oI the cargo hold.
AIter several meetings with ABS Iire protection specialists, it is decided that the primary Iire saIety
goals Ior this study should be limited to the Iollowing:
x Estimate the amount oI CO
2
required to control Iires in a cargo space;
x Provide Iire extinguishing system to minimize the loss oI carbon dioxide through the openings oI
the non-tight hatch covers;
x Maximize the distribution oI carbon dioxide throughout the narrow passages among containers.


Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

80 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
5K[';=:H'D*%C*2'D:E>*
The design objectives are developed aIter Iurther reIining the accepted goals and are as Iollows:
x Calculate the minimum amount oI CO
2
Ior eIIective Iire suppression by taking leakage into
consideration.
x Propose an alternative design oI CO
2
discharge by simulating Iire scenarios inside a non-tight
cargo hold
.N9'*.'MN:&'I'>=D*%C*P<&K%>*2:%J:)'*#FD='ID**
The amount oI carbon dioxide required in SOLAS regulations is speciIied in terms oI a volumetric
concentration. The rules associated with carbon dioxide systems are identiIied below:
x For cargo spaces, the quantity oI carbon carbonate available shall, unless otherwise provided, be
suIIicient to give a minimum volume oI Iree gas equal to 30 oI the gross volume oI the largest
cargo space to be protected in the vessel (FSS Code 2.2.1.1).
x The volume oI Iree carbon dioxide shall be calculated at 0.56 m
3
/kg (FSS Code 2.2.1.4).
x The piping Ior the distribution oI Iire-extinguishing medium shall be arranged and discharge
nozzles so positioned that a uniIorm distribution oI the medium is sought (FSS Code 2.1.2.1).
x IMO FP 47/WP.6.2 proposes the increase oI carbon dioxide by taking into consideration the
leakage oI carbon dioxide Iire extinguishing media through clear gaps between hatchway covers.
The amount oI increase should be in accordance with oI the Iollowing Iormulae, as appropriate:
CO
2
INC
30
60 A
T
2 / B
CO
2
INC
45
4 A
T
2 / B
where
CO
2
INC
30
increase oI carbon dioxide Ior cargo spaces not intended Ior carriage oI
motor vehicles with Iuel in their tanks Ior their own propulsion, in kg
CO
2
INC
45
increase oI carbon dioxide Iro cargo spaces intended Ior carriage oI
motor vehicles with Iuel in their tanks Ior their own propulsion, in kg
A
T
total maximum area oI clear gaps, in m
2

B breadth oI cargo space protected by the carbon dioxide Iire extinguishing
systems, in m
(Note that FP 47/WP. 6.2 is not mandatory Ior IMO members.)
No discharge time is speciIically required Ior cargo spaces in the FSS Code, presuming that carbon
dioxide can be discharged in a series oI stages and more settling time is required Ior the
extinguishments oI Iires inside cargo spaces.
For reIerence, land-based NFPA 12 (2000 Edition), 'Standard on Carbon Dioxide Extinguishing
Systems, requires that Ior surIace Iires, the design concentration shall be achieved within one minute
(2-5.2.1), and Ior deep-seated Iires within seven minutes (2-5.2.3). For marine systems, NFPA 12
stipulates that cargo spaces other than vehicle spaces shall be supplied with carbon dioxide based on
1 lb/30 It
3
based on the gross volume (6-2.6).


Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 81!
NFPA 12 also suggests that multiple discharges can be applied to control Iires. The initial quantity oI
carbon dioxide discharged shall be based on the net volume oI the space as determined by the amount
oI cargo in the cargo space. Additional carbon dioxide shall be released as needed to maintain control
oI the Iire |6-2.6 (b)|.
The major diIIerence concerning required CO
2
quantities between the marine (IMO, CFR) and the
land-based (NFPA 12) is that the land-based requirements are Iuel-speciIic and dependent on whether
the design basis Iire is a surIace Iire or deep-seated Iire. NFPA 12 requires the determination oI
proper concentration oI CO
2
required Ior the type oI Ilammable materials involved in the hazard.
Deep-seated Iires and certain Ilammable liquids and vapors require substantially higher CO
2

concentrations Ior extinguishments. For some Iires involving either Iuel oil or lubricating oil, the
concentration requirements Ior these Iires are identical in the land-based and marine standards. Some
Iires involving electrical equipment and cables have the potential oI becoming deep-seated. ThereIore,
NFPA requires higher CO
2
concentrations Ior these Iires than called Ior in the CFR and IMO
regulations. Table 2-3.2.1, NFPA 12, lists the minimum design CO
2
concentration percent Ior various
materials, ranging Irom 34 to 72, depending on materials. In no case shall a concentration be less
than 34, compared to 30 required by IMO.
G'&C%&I<>;'*P&:='&:<*
The perIormance criteria were developed through a Iurther reIinement oI the design objectives. They
represent numerical values to which the trial design will be compared. The perIormance criteria used
Ior this study are as Iollows.
x The volumetric concentration oI carbon dioxide shall be at least 30 oI the gross volume oI the
cargo space in this study.
x The above concentration shall be achieved within seven (7) minutes aIter the discharge oI carbon
dioxide.
,&:<9*!9='&><=:H'*2'D:E>*<>)*!&&<>E'I'>=D*
The top oI the cargo hold is covered by three partially weathertight hatch covers which are separated
by a gap up to 50 mm. Such gaps provide the openings oI ventilation Ior the cargo hold. Labyrinths,
gutters and other equivalent means Iitted close to the edges oI each hatch cover also can cause
leakages oI carbon dioxide. During the discharge, the cargo hold is partially pressurized which causes
the mixture oI air and carbon dioxide to leak out through hatch coaming labyrinth bars. The passages
oI openings are equivalent to 50 mm in width in this study. Such openings are parts oI the vessel`s
structure and cannot be closed or sealed during the CO
2
discharge.
Typical designs oI CO
2
extinguishing system place Iour (4) discharge nozzles either underneath the
hatch covers or at the middle oI the transverse bulkheads. The latter design seeks to improve the
distribution oI carbon dioxide throughout the cargo hold.
The alternative design aims to reduce the amount oI carbon dioxide Iire extinguishing media required
by FSS Code 2.2.1.1. Such reduction is based on the numerical calculations oI CO
2
concentrations oI
carbon dioxide Ior suIIiciently controlling Iires required by NFPA 12.


Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

82 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
0)'>=:C:;<=:%>*%C*$:&'*\<]<&)D*
A Hazard IdentiIication (HAZID) technique is applied Ior the risk assessment Ior Iires in cargo holds.
A HAZID should look at three Iactors: sources including likelihood, consequences and mitigation.
1. Sources of hazards:
The potential hazards inside a container are various in types, ranging Irom non-combustible materials
to highly Ilammable, dangerous goods. The sources oI Iire can come Irom the Iollowing types oI
scenarios:
x Improper loading: Improper loading allows movement and Iriction oI goods inside a container,
and could cause an ignition oI goods such as wetted matches, leather gloves, rubber gloves,
plastic toys, outdoor nylon tenting materials, sewing notions, etc.
x Spontaneous combustion. Spontaneous combustion is thought to have contributed to a large
claims-related Iire in 46 percent oI cases. Cargoes suIIering Irom spontaneous combustion and
smoking included Iishmeal, tobacco, cotton, wood pulp and coal.
x Self-ignition of chemicals: Most oI the Iires experienced in recent years have been associated with
the chemical calcium hypochlorite in its hydrated Iorm, used amongst others Ior the sterilization
oI swimming pools. Containers loaded with this chemical, but not clearly marked, were
occasionally loaded in locations exposed to elevated temperatures, such as in a hold adjacent to
the engine room. As the selI-igniting temperature oI this cargo, when stowed in larger quantities
may be as low as in the thirties C, Iires did start in containers with mostly devastating results.
The inIormation on the contents oI the containers was Irequently received only when the vessel
was already out at sea. The crew was either not aware oI any such risk or could not do anything
about it.
2. Consequence
The Iires inside cargo holds can be catastrophic both to the vessel and the environment. Heat-
damaged containers, along with distorted container cells and loose cargo, cannot be removed by
normal oIIloading systems. The container cargo in the Iire area essentially becomes 'bulk in eIIect
a pile oI junk trapped by remaining shells and distorted cell structures. II the Iire cannot be
extinguishing promptly, all containers in the cargo hold can be damaged; extensive heat can also
cause the vessel hull structure to become deIormed.
3. Mitigation
For Iire suppression systems that are conventionally located at the top oI cargoes, gaseous Iorms oI
Iire extinguishing agents should penetrate through the narrow gaps oI cell guiding racks to reach the
Iire sites. ThereIore, the most demanding conditions can be that oI Iires originating near the bottom oI
a cargo hold where adequate length oI time and suIIicient amount oI Iire extinguishing agents are
required.
#8';:C:;<=:%>*%C*2'D:E>*$:&'*#;'><&:%D*
Characteristics of fires
The purpose oI this study is to demonstrate the equivalency oI alternative design and arrangements Ior
carbon dioxide systems. Due to the lack oI statistical data oI Iires in containers, only the most
demanding requirement Ior such systems should be selected to bound all potential Iire scenarios. A
container near the bottom oI the cargo hold is chosen Ior the analysis oI the eIIectiveness oI CO
2

systems.


Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 83!
The heat release rates ranges Irom 1 MW Ior goods such as wood, clothes, etc., to 40 MW Ior
explosive chemicals such as calcium hypochlorite, polystyrene, etc. It is oIten diIIicult to predict the
Iuel quantities Ior an individual container. In order to accommodate Ior this uncertainty, this study has
chosen a standard t-squared Iast Iire to represent Iire growth in all possible scenarios.
While it may be considered adequate Ior the t
2
Iire to represent the growth rate oI Iires up to
Ilashover, it is not realistic to assume that the growth can remain parabolic Ior long periods. This is
particularly relevant in situations where Ilashover either does not occur or requires a high rate oI heat
release rate. For the cases oI Iires inside a cargo space with small ventilation opening, the actual heat
release rate is nearly proportional to the size oI the ventilation size.
For this study, the heat release rates are estimated up to the order oI 8.2 MW within seven minutes.
The Iire will be conIined to one container; heat transIer Irom Iire source to surrounding will be
calculated by solving the diIIerential equations oI continuity, momentum and energy.
Geometry
Containers operating in the marine mode are oIten stowed in vertical stacks within cells in a hold.
When stowed in this manner, containers will be restrained at the end Irames against longitudinal and
transverse movement by the cell structure. The reactions oI entire stack oI containers are taken
through the Iour bottom corner Iittings oI the lowest container.
A typical containership cargo hold measuring 12.561 m length by 38.164 m width by 23.740 m height
is chosen, with a total oI 15 stacks oI containers in the hold. Appendix 8, Figure 1 demonstrates the
cross sectional view oI the cargo hold. The lower leIt and right corner cutouts are the spaces Ior
piping and cables which are not considered in the domain oI numerical simulation.
By taking into the consideration the Ieet height at each corner, the actual size oI a container is
modiIied as 12.180 by 2.426 by 2.878 in meter. Due to the complexity oI the corner Iittings and cell
structures, the net volume oI such structures are neglected in this study. There are three types oI gaps
by locations with respect to the longitudinal axis oI the vessel: vertical, horizontal, and longitudinal,
as shown in Appendix 8, Figure 2.
In this study, all containers are assumed to be made oI steel, except the container with Iire. Since large
volume oI air is expected inside a packed container, the interior space oI a container is assumed to be
void space with air.



Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

84 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
FIGURE 1
Cargo HoId FuIIy Loaded with Containers
(not to actual scale)
0.04855
0.0305
0.458
Hatch Cover
Inner bottom
38.164
23.740
Longitudinal enclosures
excluded Irom cargo hold
0.0305
0.062 0.062 0.062
0.184 0.184
0.331 0.331
2.426
12.180
2.878
Container
Y
X
Z
1.772
12.180
2.910
unit: meter


FIGURE 2
Hatch Cover Opening and CO
2
NozzIes at Tops
(not to actual scale)
Hatch Covers
Inner bottom
Hatch openings
(up to 0.050 m)
Port side Starboard side
12.561
CO
2
Injection
nozzles at tops
38.164
23.740
Units: meters



Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 85!
ON<>=:=F*%C*P<&K%>*2:%J:)'*-J=:>EN:DL:>E*4'):<*
a. SOLAS Requirement
Dimension oI Cargo hold, in meters: 12.561 by 38.164 by 23.740
Total gloss volume oI cargo hold, in m
3
: 11,253
(excluding longitudinal enclosures
in Appendix 8, Figure 1)
Carbon dioxide by volume, in m
3
: 3,376
Carbon dioxide by weight, in kg: 6,028
II CO
2
is discharged through a 45-kg CO
2
cylinder, then 134 cylinders shall be present on board in
accordance with SOLAS regulations.
b. IMO FP 47/WP.6.2
Area oI clear gaps, A
T
, in m
2
: 2.271
Breadth oI cargo space, B, in m: 38.164
CO
2
INC
30
by weight, in kg: 595
CO
2
INC
45
by weight, in kg: 40
2:D;L<&E'*#:IN9<=:%>*
Presumably, when liquid carbon dioxide Ilows Irom storage tank through pipelines to the discharge
nozzles, the pressure drop in the pipeline signiIicantly reduces the temperature inside oI the pipes, and
thereIore, part oI the liquid carbon dioxide becomes a solid Iorm and is discharged as a solid. As a
result, carbon dioxide is discharged into the cargo space in both gaseous and solid Iorms. Some oI the
solid phase CO
2
vaporizes as it settles, and the rest Iorms an evaporating layer on the surIaces oI the
containers and the cargo hold. As the solid phase oI carbon dioxide evaporates, its gaseous volume
expands dramatically. The solid volume oI carbon dioxide is negligible, but the gaseous volume can
be estimated by Iree carbon dioxide density as 0.56 m
3
/kg.
Several theoretical attempts have been made to calculate the dependent pressure, temperature and gas
concentrations developed during the discharge and post-discharge oI carbon dioxide into an enclosure.
In this project, the homogeneous gaseous discharge oI carbon dioxide is assumed.
2:D;L<&E'*.<='D*
Carbon dioxide is injected into cargo hold in both gaseous and solid Iorms, with the solid Iraction
determined by treating the discharge processes as an isentropic expansion Irom storage temperature
and pressure to atmospheric pressure. The actual discharge rate varies by storage pressure, design
nozzle sizes and pressure loss in the piping system.
SOLAS Chapter II-2 does not provide any discharge rate or nozzle size speciIications. The FSS Code
(2.2.1.1) only requires that the piping Ior distribution shall be arranged and discharge nozzles so
positioned that a uniIorm distribution oI the medium is obtained.


Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

86 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
According to NFPA 12 (2000 edition), the design concentration shall be achieved within one (1)
minute Irom the start oI discharge Ior surIace Iires, and within seven (7) minutes Ior deep-seated Iires.
However, the discharge rate shall be no less than that required to develop a concentration oI 30
percent in two minutes. Since the gaseous Iorm oI carbon dioxide is assumed during discharge in this
study, Ior deep-seated Iires, the minimum discharge rate is calculated as 482 m
3
/min Ior seven (7)
minutes. Note that an average rate oI discharge is about 204 m
3
/min, according to a series oI tests
conducted by FM Research Corporation in 1990.
In this study, the discharge rate Ior each oI Iour (4) nozzles at the end oI the carbon dioxide piping
system is set at 121 m
3
/min. The eIIective cross-sectional area oI each discharge nozzle is estimated
to be 1 m
2
.
P%I8N=<=:%><9*,%%9*
In this case study, the concentrations and placement oI carbon dioxide as a Iunction oI time are
predicted by a deterministic method in a cargo hold. The passages among the containers provide a
unique challenge to Iield models because they are very narrow compared to the size oI a container and
require non-structure meshing in model generation.
A commercially available computational Iluid dynamics (CFD) program, FLUENT/AirPak, is used in
this study to predict the distribution oI carbon dioxide inside the cargo spaces. Widely used Ior HVAC
applications, AirPak is capable oI multi-species calculations with non-structure meshing. It is also
capable oI quickly and accurately simulating temperature distribution and air, smoke and suppressant
Ilow patterns. Many reviews and cases oI applications have been published to justiIy the assumptions
and approximations over a variety oI applications.
0>8N=*G<&<I'='&D*
AirPak requires the descriptions oI the cargo space, discharge rates oI carbon dioxide, ambient and
initial temperatures and heat release rates oI Iiring container. The description oI the cargo space
includes the geometry oI cargo hold, the locations oI containers and the dimensions oI openings and
vents (used here in lieu oI discharge nozzles). The input data oI heat release rates include timeline
rates oI heat release. The walls oI the cargo hold and containers are prescribed to be adiabatic, and the
only heat loss is through convective heat transIer in the openings.
ON<>=:=<=:H'*.'DN9=D*
The transient variables oI Ilow, temperature and species concentrations oI CO
2
injection are solved by
AirPak to demonstrate the distribution oI CO
2
in the narrow passages among containers. The transient
solutions oI continuity, velocity, temperature and Iractions oI air and CO
2
are obtained simultaneously
by solving the equations oI continuity, momentum energy and transport. Discrete cells up to 1.5
millions oI hexahedral meshes were adopted Ior simulations. The mixing and transport oI species are
modeled by solving conservation equations describing convection and diIIusion Ior each component
species. For a turbulent mixing, the diIIusion, turbulent viscosity and Schmidt numbers are
incorporated to the diIIusion terms. In this project, the transport equations were solved Ior the
components oI carbon dioxide and air. Due to the inherently non-linear nature oI this problem, the
solution procedures are highly iterative. The details oI the Iormulation oI cell elements and
diIIerential equations used in the modeling are explained in many reIerences.
The loading conditions with Iully loaded, halI loaded and empty cargoes, respectively, are used in this
study. Carbon dioxide is injected either Irom the top oI the hatch covers or Irom the sides oI the
vertical bulkheads.


Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 87!
!H'&<E'*1%9NI'=&:;*P%>;'>=&<=:%>D*%C*P<&K%>*2:%J:)'*
The eIIectiveness oI CO
2
discharge is measured by the transient volumetric concentrations oI carbon
dioxide that are averaged over the open spaces inside the cargo hold. The higher the value, the more
eIIective the discharge system. The discharge nozzles are positioned either at the top oI the cargo hold
or underneath the hatch covers, or at nearly halI depth oI the transverse bulkheads. The results oI the
averaged CO
2
percent concentrations are shown in Appendix 8, Figure 3 Ior a Iully loaded cargo.
FIGURE 3
CO
2
Percent Concentrations in a FuIIy Loaded Cargo
Average CO
2
Percent in FuIIy Loaded Cargo
0.00
0.20
0.40
0.60
0.80
1.00
0 60 120 180 240 300 360 420 480
Discharge Time, (sec)
C
O
2

P
e
r
c
e
n
t

C
o
n
c
e
n
t
r
a
t
i
o
n
s
Nozzles at top
Nozzles at half


As shown in Appendix 8, Figure 3, the volumetric concentrations oI carbon dioxide reach 30 in
about one minute Ior both discharge modes, and the perIormance criteria are achieved. Appendix 8,
Figure 3 also shows that the discharge is more eIIective iI the nozzles are positioned at the halI depth
oI the bulkheads, and CO
2
is more easily dispersed into the narrow spaces between containers. Within
three minutes aIter initial discharge, the CO
2
percent reaches 69, which is close to the highest
minimum requirement in Table 2-3.2.1, NFPA 12.
For various loading conditions, only the concentrations in the eIIective spaces are comparable and
meaningIul. In this study, the eIIective spaces are deIined by the spaces separating containers. For
Iully loaded cargo, the eIIective spaces include whole open spaces inside the cargo. Similarly, Ior halI
loaded cargo, the eIIective spaces only include the spaces oI the lower halI oI the cargo where
containers sit. Appendix 8, Figure 4 shows the CO
2
percent in eIIective spaces Ior the cases with the
nozzles at the halI depth oI the bulkheads. In a partially loaded cargo hold, air is pushed out by CO
2

injection Irom the bottom up, resulting in a high CO
2
percent in the eIIective spaces.


Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

88 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
FIGURE 4
CO
2
Percent Concentrations in Effective Spaces
Average CO
2
Percent in Effective Spaces
0.00
0.20
0.40
0.60
0.80
1.00
0 60 120 180 240 300 360 420 480
Discharge Time, (sec)
C
O
2

P
e
r
c
e
n
t
Fully loaded
Half loaded


Instant CO
2
percent on the surIaces adjacent to the transverse bulkhead, inner skin bulkhead and hatch
covers at 120 seconds aIter discharge is shown in Appendix 8, Figure 5. The blue squares in the Iigure
represent discharge nozzles. Appendix 8, Figure 6 shows the cut-through proIiles oI CO
2
percent in
aIt view. Detailed Ilow at the corners oI containers is shown in Appendix 8, Figure 7 with the vectors
originating Irom corresponding meshing points.
FIGURE 5
CO
2
Percent at 120 Seconds in Isometric View
in Performance-based Design (test # FuIIbb)



Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 89!
FIGURE 6
CO
2
Percent at 120 Seconds in Aft View
in Performance-based Design (test # FuIIbb)


FIGURE 7
DetaiI FIow at the Corners of Containers




Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

90 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
-H<9N<=:%>*(:=L*$:&'D*0>D:)'*=L'*P<&E%*\%9)*
The distributions oI carbon dioxide and temperature will be aIIected by the heat and smoke generated
by Iires. The temperature around the corners oI containers is shown in Appendix 8, Figure 8. The
white arrows indicate the velocity vectors, the pink solid blocks are non-permeable, conductive
containers, and the block in the lower right corner is the container with the design Iast Iire at 420
seconds. Heat transIer Irom Iire heats up the air around containers, which Iorms a hot plume and
ascends through the gaps among the containers to the top oI the cargo hold. Meanwhile, cool air with
CO
2
is entrained to Iire region, creating a plume oI a mixture oI air and CO
2
gas, shown as a green
vertical plume on the right side oI Appendix 8, Figure 9. The color bar at upper right corner indicates
the volumetric molar Iraction (same as volumetric concentration) oI CO
2
, Irom 0.0 in blue to 1.0 in
red. Only halI oI the cargo hold is shown in Appendix 8, Figure 9. The attraction oI CO
2
to the Iire
region helps control Iire spread and eventually extinguishes the Iire. The inIluence oI Iire on the
distribution oI CO
2
is still under study.
Also shown in Appendix 8, Figure 8, the temperature inside the container could reach up to 1500C iI
the Iire is not controlled. Such high temperature and intense heat will weaken the properties oI steel
and could cause the steel structure oI the container to collapse.

FIGURE 8
Temperature at Corners of Container Fire

Container with Iire




Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 91!
FIGURE 9
CO
2
Concentration in a Design Fire Scenario


+'<^<E'*%C*P5
2
*=L&%NEL*58'>:>ED*)N&:>E*2:D;L<&E'*
Appendix 8, Figure 3 also shows that the percentage oI CO
2
concentration is Iar exceeding the
required 30 oI open volume in Iully loaded cargoes two minutes aIter discharge. While CO
2
is
dispersed and quickly Iills up the gaps, a great amount oI CO
2
leaks out oI the cargo hold through the
openings at the hatch covers. The loss oI CO
2
due to leakage is shown in Appendix 8, Figure 10, and
the rate oI total leakage to total discharge is shown in Appendix 8, Figure 11.
From Appendix 8, Figures 10 and 11, it can be seen that a signiIicant amount oI CO
2

is lost Irom
openings oI hatch covers. II all oI the 138 cylinders oI CO
2
are Iully discharged into the cargo hold
within seven minutes, over 60 (or 3650 kg) oI the CO
2
is actually lost due to the leakage Irom the
openings oI hatch covers, as shown in Appendix 8, Figure 11. This amount oI leakage is six times
more than the amount suggested by IMO FP 47/WP 6.2.

Container
with Iire


Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

92 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
FIGURE 10
Loss of CO
2


Leakage of CO
2
from Openings
in a FuIIy Loaded Cargo
0
1000
2000
3000
4000
5000
0 60 120 180 240 300 360 420 480
Discharge Time, (sec)
L
e
a
k
a
g
e

o
f

C
O
2
,

(
k
g
)
Nozzles at top
Nozzles at half


FIGURE 11
Loss Rates of CO
2

Leakage Rate: totaI Ieakage / totaI discharged
into a fuIIy-Ioaded cargo
0.00
0.20
0.40
0.60
0.80
1.00
0 60 120 180 240 300 360 420 480
Discharge Time, (sec)
L
e
a
k
a
g
e

R
a
t
e
Nozzles at top
Nozzles at half




Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 93!
4:>:INI*!I%N>=*%C*P<&K%>*2:%J:)'*:>*4N9=:89'*2:D;L<&E:>E*
#FD='ID*
The results oI the CO
2
percent and loss calculations in Appendix 8, Figures 3, 4, 10 and 11 clearly
demonstrate that the average CO
2
concentration oI 30 can be achieved within two minutes oI initial
discharge (the assumption oI non-penetration container block is applied in this calculation). However,
what really matters to eIIectively control or extinguish Iire is to assure that 30 oI CO
2
percent is
achieved in the every part oI open spaces inside oI the cargo hold. ThereIore, it is essential to
calculate the time at which 30 is achieved, and subsequently, the corresponding amount oI carbon
dioxide. Appendix 8, Figure 12 provides the minimum CO
2
percent concentrations in a Iully loaded
cargo.
FIGURE 12
Minimum CO
2
Percent Concentrations
Minimum CO
2
Percent in a FuIIy Loaded Cargo
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0 60 120 180 240 300
Discharge time, (sec)
C
O
2

P
e
r
c
e
n
t
Nozzles at top
Nozzles at half


Appendix 8, Figure 12 shows that the discharge time to achieve minimum concentration oI 30 in
every open space inside a cargo hold is about three minutes. Note that the average CO
2
concentration
is above 70 at three minutes, as shown in Appendix 8, Figures 3 and 4. ThereIore, three minutes oI
discharge can provide the required concentrations stipulated in SOLAS regulations. Accordingly, the
needed amount oI carbon dioxide is decided by:
3 min 482 m
3
/min 1446 m
3

or 2582 kg oI CO
2
, which is equivalent to total oI 58 cylinders oI 45 kg CO
2
.
Prescriptive requirements by SOLAS require that 6026 kg or 134 cylinders oI CO
2
shall be provided
to control and extinguish Iires in the cargo hold discussed in this study, with 595 kg (or 14 cylinders)
more suggested by IMO FP 47/WP 6.2. The perIormance-based design, supported by numerical
calculations and modeling, clearly shows that without the compromise oI Iire saIety level, not only
can the required 134 cylinders be cut down to 58 cylinders, but the increase in CO
2
suggested by IMO
FP 47/WP 6.2 is unnecessary.


Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

94 !"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445!
II the dispersion oI CO
2
into non-sealed containers and the continuous leakage oI CO
2
through clear
gaps between hatch covers aIter initial discharge are taken into consideration, more cylinders may be
needed in addition to 58 to account Ior the loss and reduction in CO
2
concentrations. The actual
number oI cylinders should be subject to engineers` judgment.
#'>D:=:H:=FT*_>;'&=<:>=F*<>)*+:I:=<=:%>D*%C*/NI'&:;<9*4%)'9:>E*
The results oI sensitivity analysis identiIy the input parameters that have the greatest eIIect on the
output variables. Among input parameters oI model geometry, the small breath oI the gaps (Irom 0.05
to 0.03 m) between containers is very sensitive compared to most other dimensions such as the sizes
oI containers. The small open spaces inside a cargo hold determine the transport characteristics oI
Ilow that dominates the dissipation oI heat and CO
2
.
The uncertainty oI numerical simulations can be evaluated iI the Iull-scale test data are available.
Empirical data or general observations Irom industrial practices may also be used iI the test data is not
available. While multiple stages oI CO
2
discharges are commonly adopted to contain or control a Iire
inside a cargo hold, no Iull-scale test data Ior CO
2
discharge in a cargo hold has been available Ior
comparison. ThereIore, the analysis oI uncertainty is very limited. The major limitations oI numerical
simulations came Irom the Iollowing three areas:
x Modeling oI cargo hold and containers
x Selection oI design Iire scenarios
x Algorithms oI numerical calculations and discretion
Once the dimensions oI the cargo hold and containers are decided, the quality oI meshing is critical to
the accuracy oI numerical results. Even though the nodes in this study range up to 1.2 million Ior the
Iully loaded cargo hold, it is still desirable to make Iiner meshes to account Ior the thermal-driven
Ilows in narrow gaps among containers.
The selection oI design Iire scenarios is crucial to accurately predicting the Iire and, subsequently, the
distribution oI carbon dioxide. The results oI calculation show that heat transIer Irom Iire heats up air
around the Iire source, pushing hot air upward to the hatch covers. The upliIting hot air creates
negative pressure zones around the Iire source while cooler air with injected carbon dioxide is sucked
to the Iire source. ThereIore, the Iire actually enhances the movement oI air inside the cargo hold,
causing quick dissipation oI CO
2
to the Iire sources.
The inherent algorithms oI numerical schemes also determine the accuracy oI numerical calculations.
Appropriate turbulent model, radiation modeling, species transport, etc., all contribute to the Iinal
solutions. The validation oI the soItware is available elsewhere.
P%>;9ND:%>D*
The main purpose oI this study is to provide perIormance design Ior carbon dioxide extinguishing
systems in a cargo hold. A CFD model is applied to evaluate the transient distribution oI gaseous
carbon dioxide inside a cargo hold. Numerical results oI multiple loading scenarios with two
discharge modes clearly demonstrate that with perIormance-based design, the 30 oI CO
2

concentration can be achieved within three minutes oI discharge. Thus, the actual amount oI CO
2

needed to get this 30 is signiIicantly less that the amount prescribed by current SOLAS regulations.
A reduction in CO
2
cylinders or storage tanks can be implied Ior the perIormance-based design.
ThereIore, the increase in CO
2
suggested in FP 47/WP.6.2 is unnecessary Ior partially weathertight
hatchways covers onboard containerships.


Appendix 8 ExampIe AnaIysis - AIternative Design and Arrangements for Containership Cargo
Spaces

!"#
!
"#$%&'()!'*+),!*'!&-+).'&+$/)!%),$"'!&'%!&..&'")0)'+,!1*.!1$.)!,&1)+2!
!
!3445! 95!
The numerical calculations also show that a signiIicant amount oI carbon dioxide is leaking through
the clear openings oI non-weathertight hatch covers. The actual leakage is more than six times the
amount in increase suggested in FP 47/WP.6.2.
.'C'&'>;'D*
1. SOLAS 2000 Amendments, IMO, London, 2001.
2. SOLAS, Consolidated Edition, IMO, London, 2001.
3. International Code Ior Fire SaIety Systems (FSS Code), IMO, London, 2001.
4. NFPA 12, Standard on Carbon Dioxide Extinguishing Systems, 2000 Edition.
5. Odense L185 General Arrangement, Building No. L185 Series, Odense Steel Shipyard Ltd.,
2001.
6. Rules for Certification of Cargo Containers, American Bureau oI Shipping, 1998.
7. IMO FP 47/WP.6, 'Any Other Business DraIt MSC circular on Guidelines on partially
weathertight hatchway covers on board containerships, IMO, London, 2003.
8. Zalosh, R., Huang, C. W., 'Carbon Dioxide Discharge Test Modeling, Fire SaIety Science-
Proceeding oI the Fourth International Symposium, pp. 889-900.
9. AirPak Training Notes, FLUENT Inc., 2001.


!

This Page Intentionally LeIt Blank