James G. Quintiere - Principles of Fire Behavior, Second Edition-CRC Press (2016)

Principles of
FIRE
BEHAVIOR
SECOND EDITION
Principles of
FIRE
BEHAVIOR
SECOND EDITION
James G. Quintiere
Boca Raton London New York
CRC Press is an imprint of the

Taylor & Francis Group, an informa business
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2017 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works
Printed on acid-free paper

Version Date: 20160324
International Standard Book Number-13: 978-1-4987-3562-9 (Hardback)
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts
have been made to publish reliable data and information, but the author and publisher cannot assume
responsibility for the validity of all materials or the consequences of their use. The authors and publishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize to
copyright holders if permission to publish in this form has not been obtained. If any copyright material has
not been acknowledged please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmit-
ted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented,
including photocopying, microfilming, and recording, or in any information storage or retrieval system,
without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.
com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and
registration for a variety of users. For organizations that have been granted a photocopy license by the CCC,
a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used
only for identification and explanation without intent to infringe.
Library of Congress Cataloging‑in‑Publication Data
Names: Quintiere, James G., author.

Title: Principles of fire behavior / James G. Quintiere.
Description: Second edition. | Boca Raton, FL : Taylor & Francis, CRC Press,
2017. | Includes bibliographical references and index.
Identifiers: LCCN 2016004478 | ISBN 9781498735629
Subjects: LCSH: Fires. | Fire. | Fire investigation. | Fire prevention. |
Flame spread.
Classification: LCC TH9448 .Q55 2017 | DDC 628.9/2--dc23
LC record available at http://lccn.loc.gov/2016004478
Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com
and the CRC Press Web site at

http://www.crcpress.com
To the memory of Philip H. Thomas, a friend, a mentor,
and the father of fire science. He developed the science for
many aspects of fire before others. He set the path.

Contents
Preface .....................................................................................................................xv
Acknowledgments ............................................................................................. xvii
Acronyms ............................................................................................................. xix
Nomenclature ...................................................................................................... xxi
1. Evolution of Fire Science ...............................................................................1

Learning Objectives .........................................................................................1
1.1 Introduction ...........................................................................................1
1.2 What Is Fire? ..........................................................................................3
1.3 Natural Causes of Fire ..........................................................................4
1.3.1 Lightning...................................................................................4
1.3.2 Earthquake ................................................................................5
1.3.3 Meteors ......................................................................................7
1.3.4 Volcanoes ..................................................................................8
1.3.5 Underground Fires ..................................................................8
1.4 Fire and War ..........................................................................................9
1.5 Fire in the United States and Abroad ............................................... 10
1.5.1 U.S. Statistics ........................................................................... 10
1.5.2 United States and the World ................................................ 13
1.5.3 U.S. Fire Prevention Infrastructure ..................................... 13
1.5.4 Motivation for Improvement ................................................ 15
1.5.5 Flammability Tests ................................................................ 15
1.5.6 Cost of Fire .............................................................................. 17
1.6 Fire Research........................................................................................ 20
1.6.1 Disciplines That Underlie Fire ............................................. 20
1.6.2 Computer Simulations and Physics .................................... 20
1.6.3 Brief History of Fire Science ................................................. 21
1.7 Visualization of Fire Phenomena...................................................... 23
1.8 Scientific Language ............................................................................. 25
1.8.1 Units of Measure .................................................................... 25
1.8.2 Symbols ................................................................................... 28
1.8.3 Scientific Notation.................................................................. 28
1.9 Summary .............................................................................................. 29
Review Questions ..........................................................................................30
Activities .........................................................................................................30
References .......................................................................................................31
vii
viii Contents
2. Combustion in Natural Fires...................................................................... 33

Learning Objectives ....................................................................................... 33
2.1 Introduction ......................................................................................... 33
2.2 Fire and Its Ingredients ......................................................................34
2.2.1 Typical Temperatures and Energy Levels to
Achieve Combustion .............................................................34
2.2.2 Fuel Chemistry ....................................................................... 36
2.2.3 Fire Triangle and Tetrahedron ............................................. 36
2.2.4 Combustion Time and Extent .............................................. 37
2.2.5 Types of Fire............................................................................ 38
2.3 Diffusion Flames ................................................................................. 39
2.3.1 Candle Flame..........................................................................43
2.3.2 Anatomy of a Diffusion Flame ............................................ 48
2.3.3 Turbulent Diffusion Flames .................................................54
2.4 Premixed Flames ................................................................................. 56
2.4.1 Laminar Flame Propagation ................................................ 57
2.4.2 Flame Temperatures .............................................................. 59
2.4.3 Turbulent Propagation to Detonation ................................. 61
2.5 Smoldering ........................................................................................... 62
2.6 Spontaneous Combustion ..................................................................64
2.7 Summary .............................................................................................. 69
Review Questions .......................................................................................... 70
True or False ......................................................................................... 70
Activities ......................................................................................................... 70
References .......................................................................................................71
3. Heat Transfer ................................................................................................. 73

Learning Objectives ....................................................................................... 73
3.1 Introduction ......................................................................................... 73
3.2 Definitions and Concepts................................................................... 74
3.3 Forms of Heat Transfer .......................................................................77
3.3.1 Conduction.............................................................................. 78
3.3.1.1 Steady .......................................................................... 78
3.3.1.2 Thermal Penetration Time ........................................80
3.3.2 Convection .............................................................................. 81
3.3.3 Radiation .................................................................................84
3.4 Heat Flux as an Indication of Damage ............................................. 93
3.5 Heat Flux Due to Smoke in Room Fires ........................................... 94
3.6 Heat Flux from Flames ....................................................................... 98
3.7 Summary .............................................................................................. 99
Review Questions .......................................................................................... 99
True or False ....................................................................................... 100
Activities ....................................................................................................... 100
References .....................................................................................................101
Contents ix
4. Ignition ......................................................................................................... 103

Learning Objectives ..................................................................................... 103
4.1 Introduction ....................................................................................... 103
4.2 Piloted Ignition and Autoignition .................................................. 104
4.3 Evaporation in Liquids ..................................................................... 104
4.4 Liquid Fuels ....................................................................................... 106
4.4.1 Piloted Ignition..................................................................... 107
4.4.2 Autoignition.......................................................................... 108
4.5 Solid Fuels .......................................................................................... 110
4.5.1 Ignition of Wood as an Example........................................ 111
4.5.2 Ignition Temperature and Critical Heat Flux .................. 112
4.6 Time for Flaming Ignition ............................................................... 114
4.7 Predicting the Ignition Time for Solid Fuels ................................. 116
4.7.1 Ignition of Thin Objects ...................................................... 117
4.7.2 Ignition of Thick Materials ................................................. 119
4.8 Solid Properties for Piloted Ignition .............................................. 124
4.9 Summary ............................................................................................ 131
Review Questions ........................................................................................ 132
True or False ....................................................................................... 132
Activities ....................................................................................................... 133
References .....................................................................................................133
5. Flame Spread ............................................................................................... 135

5.1 Introduction ....................................................................................... 135
5.2 Definitions .......................................................................................... 136
5.3 General Flame Spread Theory ........................................................ 138
5.4 Spread on Solid Surfaces .................................................................. 140
5.4.1 Effect of Thickness ............................................................... 141
5.4.2 Downward or Lateral Wall Spread on a
Thick Material ..................................................................142
5.4.3 Upward or Wind-Aided Spread on a Thick Material ........145
5.5 Spread through Porous Solid Arrays ............................................. 147
5.6 Spread on Liquids ............................................................................. 151
5.7 Spread through a Dwelling ............................................................. 155
5.8 Typical Fire Spread Rates ................................................................. 157
5.9 Standard Test Methods .................................................................... 157
5.10 Case Study: Fire Spread in a School Gymnasium ........................ 158
5.11 Summary ............................................................................................ 160
True or False ....................................................................................... 161
Activities ....................................................................................................... 162
References .....................................................................................................162
x Contents
6. Burning Rate................................................................................................ 165

6.1 Introduction ....................................................................................... 165
6.2 Definitions and Theory .................................................................... 165
6.2.1 Burning Mass Flux .............................................................. 167
6.2.2 Heat of Gasification, L (kJ/g) .............................................. 168
6.2.3 Approximate Formula for Steady Burning ...................... 169
6.2.4 Computing the Mass Burning Flux................................... 170
6.2.5 Unsteady Burning ................................................................ 171
6.2.6 Difficulties in Computing Burning Rates ........................ 172
6.2.7 Material Property Values for the Heat of Gasification ... 173
6.3 Estimating Burning Mass Flux ....................................................... 173
6.3.1 Example for a Burning Wall ............................................... 175
6.3.2 Pool Fires ............................................................................... 176
6.3.3 Maximum Burning Rates ................................................... 178
6.4 Energy Release Rate, Q
.................................................................... 180
6.4.1 Heat of Combustion, ΔHc .................................................... 180
6.4.2 Heat of Combustion of Wood............................................. 181
6.4.3 Heats of Combustion of Materials ..................................... 182
6.4.4 Heat Release Parameter ...................................................... 182
6.5 Estimating Energy Release Rate ..................................................... 184
6.6 Experimental Firepower (HRR) Results for Selected Items ........ 187
6.7 Fire Growth Rate ............................................................................... 192
6.7.1 NFPA Design Categories .................................................... 201
6.7.2 Fire HRR of Item Constructed ........................................... 203
6.8 Vehicle Fire Behavior ........................................................................ 206
6.9 Extinction ........................................................................................... 207
6.10 Summary ............................................................................................ 209
True or False ....................................................................................... 210
Activities ....................................................................................................... 210
References ..................................................................................................... 211
7. Fire Plumes .................................................................................................. 213

7.1 Introduction ....................................................................................... 213
7.2 Buoyancy and Fluid Dynamics ....................................................... 214
7.3 Turbulent Fire Plumes and Jets ....................................................... 218
7.4 Buoyant Plumes ................................................................................. 219
7.5 Flame Height ..................................................................................... 221
7.5.1 Jet Flames ..............................................................................222
7.5.2 Pool Fire Flames ................................................................... 224
7.6 Fire Plume Temperatures ................................................................. 229
7.6.1 Analyses to Predict the Plume Temperature ................... 230
Contents xi
7.6.2 Flame Height and Temperature Calculations.................. 232

7.6.3 Nature of Turbulent Flame Temperature .........................234
7.7 Flame Lengths for Other Configurations ...................................... 236
7.8 Whirls and Balls ................................................................................ 237
7.9 Summary ............................................................................................ 239
True or False ....................................................................................... 240
Activities ....................................................................................................... 240
References ..................................................................................................... 241
8. Combustion Products ................................................................................ 243

8.1 Introduction ....................................................................................... 243
8.2 Scope of Combustion Products ....................................................... 244
8.3 Yields................................................................................................... 246
8.4 Concentrations................................................................................... 251
8.5 Hazards ..............................................................................................254
8.5.1 Narcotic Gases ......................................................................254
8.5.2 Additive Fractional Incapacitation Doses ........................ 256
8.5.3 Irritant Gases ........................................................................ 257
8.5.4 Smoke Visibility ................................................................... 258
8.5.5 Heat Effects ........................................................................... 261
8.6 Summary ............................................................................................ 263
True or False ....................................................................................... 264
Activities ....................................................................................................... 264
References ..................................................................................................... 265
9. Compartment Fires ..................................................................................... 267

9.1 Introduction ....................................................................................... 267
9.2 Stages of Fire Development ............................................................. 268
9.2.1 Developing Fire .................................................................... 269
9.2.2 Flashover ............................................................................... 270
9.2.3 Fully Developed ................................................................... 270
9.2.4 Cooling Stage ........................................................................ 272
9.2.5 Example of Measured Conditions in Room Fire ............. 272
9.3 Fire-Induced Flows ........................................................................... 273
9.3.1 Duct Fan Pressures in Fire.................................................. 274
9.3.2 Pressure Level Due to Fire.................................................. 276
9.4 Compartment Flow Dynamics........................................................ 276
9.4.1 Flow in a Room .................................................................... 276
9.4.2 Smoke Filling in a Leaky Compartment .......................... 279
xii Contents
9.4.3 Smoke Movement in a Building......................................... 281

9.4.3.1 Corridors................................................................ 282
9.4.3.2 Vertical Shafts .......................................................284
9.5 Single Room Fire Analyses .............................................................. 286
9.5.1 Smoke Filling ........................................................................ 287
9.5.2 Smoldering Fire in a Closed Space .................................... 288
9.5.3 Vent Flows ............................................................................. 291
9.5.4 Smoke Temperature ............................................................. 295
9.5.5 Flashover ............................................................................... 297
9.5.6 Ventilation-Limited Fires .................................................... 298
9.5.7 Fully Developed Fire Size ...................................................300
9.5.7.1 Model for Fuel Generation in a
Compartment ..................................................... 302
9.5.7.2 Fully Developed Compartment Fire
Behavior .............................................................. 303
9.5.7.3 Fuel Load and Burning Duration ...................... 307
9.6 Anatomy of Fire Growth ..................................................................308
9.7 Summary ............................................................................................ 314
True or False ....................................................................................... 316
Activities ....................................................................................................... 316
References ..................................................................................................... 317
10. Design, Investigation, and Case Studies ............................................... 319

10.1 Introduction ....................................................................................... 319
10.2 Fire Safety Design ............................................................................. 320
10.2.1 Examples in Design ............................................................. 323
10.2.1.1 Example 1: Effect of Shaft Vents in a
Building Fire ......................................................... 323
10.2.1.2 Example 2: Smoke Movement in the World
Trade Center, New York, New York ................... 325
10.2.2 Performance Codes .............................................................. 327
10.3 Fire Investigation............................................................................... 328
10.3.1 Dupont Plaza Fire ................................................................ 329
10.3.2 Example 3: The Case of the Laundry Basket Fire............ 330
10.3.3 Example 4: An Analysis of the Waldbaum Fire,
Brooklyn, New York (August 3, 1978) ............................... 332
10.3.3.1 Early Dawn: Ignition (Approximately 6 a.m.) ...... 334
10.3.3.2 Smoldering Stage .................................................. 336
10.3.3.3 Onset of Flaming: Shortly before 8:30 a.m........ 337
10.3.3.4 Fire Growth in the Cockloft ................................ 338
10.3.3.5 Collapse of the Roof due to Truss Member
Failure ~ 9:15 a.m. ................................................. 339
10.3.3.6 Concluding Remarks ........................................... 339
Contents xiii
10.3.4 Example 5: The Branch Davidian Fire near Waco,

Texas (April 19, 1993)............................................................340
10.3.4.1 Congressional Committee Statement on
the Mount Carmel Branch Davidian Fire
(April 19, 1995) .......................................................342
10.3.4.2 Follow-Up to Waco ............................................... 351
10.3.4.3 Scientific Analyses of Some Aspects of the
Waco Fire ............................................................... 351
10.3.5 Patterns ..................................................................................354
10.3.5.1 Soot Patterns ......................................................... 355
10.3.5.2 Clean Burn Pattern ............................................... 356
10.3.5.3 The Hands of Time in a Fire ............................... 357
10.3.5.4 Gasoline versus Fire Damage ............................. 357
10.3.6 World Trade Center Terrorism and Fire (9/11) ................. 359
10.3.6.1 Investigative Efforts ............................................. 360
10.3.6.2 Role of the Jet Fuel ................................................ 362
10.3.6.3 Fuel Load for the Fire........................................... 362
10.3.6.4 Fire Effect on the Structure ................................. 363
10.3.6.5 Afterthoughts on WTC ........................................ 365
10.4 Computer Fire Models...................................................................... 366
10.4.1 Zone Models ......................................................................... 366
10.4.2 Field Models ......................................................................... 369
10.5 Summary ............................................................................................ 370
Activities ....................................................................................................... 371
References ..................................................................................................... 372
Appendix: Mathematics of Science ................................................................ 375
Glossary ............................................................................................................... 397
Index .....................................................................................................................405
Preface
The genesis of this book is early short courses given to fire investigators of
the Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF) starting in
1992 at the University of Maryland, College Park. The material grew into
a book by Delmar Publishers in 1997 within their series on fire and rescue
books. The book received wide usage among institutions for the education
of firefighters in the United States. Internationally, that use continued with a
Korean translation in 2004 and a Japanese translation in 2009. The book was
maintained as the basis in fire investigation training courses offered by the
ATF for both federal and state investigators.
In early 2012, a new edition was submitted to Cengage (formerly Delmar),
and it received positive reviews. Unfortunately, in July of that year, financial
matters caused Cengage to terminate their fire series and this book. Now out
of print, the book languished until 2015 when a fresh revision was prepared
for Taylor & Francis Group/CRC Press.
The new edition aims to improve the introduction of science and math to
the student. An Appendix has been added to introduce or review principles
of algebra and to explain the need for mathematics in science. The book has
been significantly expanded with new information and improved explana-
tions. More examples are presented in each chapter with many related to
real-world incidents. New figures and graphical results have been added,
and the use of color in some should be a significant enhancement.
The student will gain knowledge of fire behavior through scientific prin-
ciples and will achieve the means to make quantitative estimates for aspects
of fire. The book is intended for those with a secondary education and is not a
fundamental book for scientists and engineers. Introductions to the subjects
of combustion, chemistry, heat transfer, and fluid mechanics will allow the
student to learn what constitutes fire and its effects. The style of presenta-
tion is to offer simple explanations for component fire processes and to dem-
onstrate the use of formulas to make estimations. The formulas are based
on sound developments from fire research and can be tested potentially in
classroom or field demonstrations. Carefully planned experiments can make
much of the material in this book resonate with the students.
The book covers all aspects of fire behavior. Fire is combustion, and the stu-
dent is introduced to premixed flames, the basis of ignition and explosions,
diffusion flames that personify accidental fire, and smoldering, which slows
down potentially deadly process. The rudiments of heat transfer are intro-
duced with simple presentations for conduction, convection, and radiation.
The concept of heat flux is introduced as that effect of fire that causes harm,
damage, and ignition. Buoyant fluid flow caused by fire is explained in terms
of the chimney and then related to flows in rooms and buildings. Fire growth
xv
xvi Preface
involves the trifecta of ignition, flame spread, and continued burning. Each of
these elements is explained in terms of simple ideas and with information on
practical data and results. The student will learn how combustion products
are generated in fire, why they can do harm, and how lack of air will affect
the results. Flame dynamics are explained in the open and in compartments.
The stages of fire development in a room are explained in terms of flashover
and ventilation-limited fires. In all cases, aspects of these subjects are given
in simple but sound mathematical formulas that allow estimations. The stu-
dent will learn the elements of fire and how to express them in quantitative
values in the same way measurements are used to describe how to build a
house. The final chapter discusses case studies related to performance-based
fire safety design and to real fire investigation cases. The material in the book
aims to illustrate the aspects of these cases. Application and agreement with
hypothesis may not establish truth, but the examples are given to illustrate
an approach.
This book is intended for several audiences. It has been used in curriculum
for students in firefighting to learn the behavior of fire. Although some have
found the mathematics a distraction to learning, hopefully this new edition
will ease or eliminate that effect. It is essential that firefighters adopt science
to guide their practices. Short courses to fire investigators have given them
an additional set of tools to defend their conclusions. In addition, the use of
science and its education can support their credentials in expert testimony.
While the book is intended for nonengineers, I have been told that the first
edition has proved a ready reference for practicing fire protection engineers
and to new engineers to the field. Hopefully, the second edition will be of
added benefit. Finally, the average person can learn about fire using this book.
Indeed, Professor Marino diMarzo has taught “Why Do Things Burn?” an
honors seminar class for nonengineers at the University of Maryland College
Park since 2010, using this book. That course is a mix of lectures and hands-
on student experiments to reinforce learning. It cannot be emphasized more
strongly that supportive experiments for the material in this book, safely
conducted, can enhance and enliven the learning process.
Acknowledgments
Before the second edition was aborted at Cengage, reviewers had sent me
their comments. The unfinished form of the manuscript put an extra burden
on their task with corrupted text due to a poor conversion from the original.
Their positive comments and constructive suggestions gave me encourage-
ment to complete the final second edition. I am very grateful for their input
and have tried to meet their suggestions. The reviewers include the following:
Glenn P. Corbett
Associate Professor of Fire Science
John Jay College of Criminal Justice/CUNY
Todd Haines
Planning Chief/Fire Protection Engineer
Dallas/Fort Worth International Airport, New York, New York
Gary D. Kistner
Program Coordinator/Director of Graduate Studies
Fire Service Management
Southern Illinois University Carbondale
Carbondale, Illinois
Robert J. Schaal, IAAI-CFI
Consultant/Forensic Investigator
Gulf Coast Fire Investigation, Research, and Education
Mandeville, Louisiana
Tom Sitz
Lieutenant Painesville Twp Fire Department
Lakeland Community College
Auburn Career Center
Concord, Ohio
Joel Journeay, M.A.
Fire Protection & Administration
California State University
xvii
Acronyms
AIT Autoignition temperature

ASTM American Society of Testing and Materials
BATF Bureau of Alcohol, Tobacco, Firearms and Explosives
BTU British thermal unit
CDT Central Daylight Time
CFD Computational fluid dynamics
FLETC Federal Law Enforcement Training Center
FLIR Forward-looking infrared
GDP Gross domestic product
LFL Lower flammable limit
NBS National Bureau of Standards
NFPA National Fire Protection Association
NIST National Institute for Standards and Technology
PMMA Polymethyl methacrylate
PRC Product Research Committee
PS Polystyrene
PU Polyurethane
PVC Polyvinyl chloride
RH Relative humidity
RMV Respiration minute volume
SI International System of Units
UFL Upper flammable limit
xix
Nomenclature
a Constant associated with t-squared fires, Q = at 2

A Area
AF,b Area of fuel covered by the flame
AIT Autoignition temperature
Ao H o Ventilation factor
c Distance in radiation fraction equation
c Specific heat
c Speed of light
CHF Critical heat flux, threshold heat flux for piloted ignition
d Diameter
d Thickness
D Diameter
Db Maximum fireball diameter
Dm Mass optical density, pertains to visibility
Do Optical density per unit path length
E Energy
f Frequency
f Friction factor
F12 View factor
Fr Froude number, ratio of momentum to buoyant force
g Earth’s gravitational force per unit mass, 9.81 N/kg
h Convective heat transfer coefficient
H Height
HL Smoke layer height
HN Neutral plane height
HRP Heat release parameter, ΔHc/L
HRR Heat release rate, same as Q
I Intensity of light
k Thermal conductivity
kρc Thermal inertia
l Length
L Heat of gasification
Lb Maximum height of fireball
Lf Flame length
Lv Visibility, distance able to see through smoke
LFL Lower flammability limit
m Mass
m Mass of fuel burned
m Mass loss or burning rate
m ¢¢ Mass burning flux, or burning rate per unit area
xxi
xxii Nomenclature
m a,max Maximum air flow rate

mCO Mass of CO
m smoke Mass flow rate of smoke
Mspecies Molecular weight of species
p Pressure
q Heat
Q Chemical energy
q Flow rate of heat
q¢¢ Heat flux, or q/A
q¢¢ext External radiant heat flux from hot surroundings
q¢¢flame Incident flame heat flux
q¢¢rr Reradiation heat flux from ignited surface
Q Energy from combustion
Q Combustion energy release rate of fire
Q
Q* Zukoski number, Q* =
racpaTa gDD2
RH Relative humidity
s Stoichiometric air to fuel mass ratio usually for complete chemical
reaction
S Surface area
t Time
tb Duration of burning
t1 Time to reach 1 MW in t-squared fire growth
tig Time to ignite
T Temperature
Ta Air temperature
Tig Ignition temperature
Ts Surface temperature
T∞ Air or initial temperature
TRP Thermal response parameter, Equation 4.4
UFL Upper flammable limit
V Flame spread velocity
Vf Original fuel volume in fireball equations
w Width of the fuel
xƒ Flame height
xp Pyrolysis length
Xr Fraction of chemical energy radiated from flame
Xspecies Mole or volume fraction of species in mixture
yspecies Mass of species produced in combustion per mass of fuel supplied
Yspecies Mass fraction of species in mixture
z Vertical height
α Thermal diffusivity, k/ρc
δƒ Flame forward heat transfer length in spread
Nomenclature xxiii
ε Emissivity
ΔHc Heat of combustion
Δ Means difference
κ Absorption coefficient of the smoke or flame
κs Light extinction coefficient, Equation 8.10
λ Wavelength
ρ Density
ρb Bulk density
σ Stefan–Boltzmann constant, 5.67 × 10 –11 kW/m2-K4
ϕ Parameter in Equation 5.6
Φ Equivalence ratio
1
Evolution of Fire Science
Learning Objectives
Upon completion of this chapter, you should be able to
• Describe what is fire

• Identify fire in history
• Describe the U.S. fire safety infrastructure and statistics
• Describe the history of fire research
• Describe the role of science to predict fire
1.1 Introduction
Before there was life there was fire. It has left its imprint on history in many
ways. We need to understand the role fire has played in history, including
prehistoric events as well. After all, fire has been around from the beginning.
Creation may have begun with nuclear reactions, but fire was an essential
consequence to the development of life. Even the ancients revered fire, as they
imagined Prometheus stole it from Zeus and gave it to us mortals. Perhaps
Zeus knew we would use it and abuse it, and so he punished Prometheus
severely.
Today, fire in the form of controlled combustion is an essential ingredi-
ent of our technology. We could not easily survive without the burning
of coal, gas, and oil. For these reasons, the study of combustion for useful
power is driven by market forces that drive developed world economies. In
contrast, the study of uncontrolled fire is motivated by clear risks to soci-
ety and by societies having the means and desire to invest in such study.
Consequently, we know a lot less about fire than controlled combustion. Fire
events are chronicled and recorded in proportion to the damage rendered
1
2 Principles of Fire Behavior
to people and property. But the study of fire to understand and to improve
humankind is limited. It is a complex area involving many disciplines, and it
is a relatively primitive field compared to other technological areas. Yet, over
the past 50 years, steady progress has been made in its subjects. This book is
a by-product of those studies.
It is important to understand the history of fire science research to
appreciate its contribution to the science. Equally important, it is essential
to appreciate the forces that drive such studies. The impact on people and
property due to fire from recent history and from past history is a driver for
learning. This chapter will try to present some of this information. Of signif-
icance today is an increasing recognition in formal academic study that fire,
as a subject discipline, is important. Today, fire is being studied at numer-
ous universities around the world. The United States has two postgraduate
university programs, while there are several in the United Kingdom, two
in Sweden, several in Japan, and more than a dozen in China. In fact, the
Chinese People’s Armed Police Academy (university) outside of Beijing has
about 7000 students with most studying fire regulations, firefighting, or fire
investigation. Some of their faculty translated my advanced book on fire
into Chinese:
詹姆士 G. 昆棣瑞著 = Fundamentals of fire phenomena/James G.

Quintiere; 杜建科, 王平,.; James G. Quintiere; Jianke Du; Ping Wang;
Yaping Gao Publisher: 化学工, Beijing: Hua xue gong ye chu ban she, 2010.
Indeed, the first edition of the current book has been translated into Korean
and Japanese. Its interest has been primarily from the nonengineering
communities related to firefighter and fire investigator education. Its purpose
was to attempt to translate the research developments into digestible knowl-
edge for these fields. It is couched in terms of descriptions of dissected
aspects of fire growth. In addition, it includes formulas that allow predictive
estimates for these phenomena. In short, it is designed to bring tools from
research to the practitioner.
As a firefighter you need to react to fire with reason and knowledge, not just
training and instinct. As a fire investigator you need to support your opinion
with knowledge and analyses to within a scientific degree of certainty.
In this chapter, the reader is introduced to the subject of fire with its many
features. What is it? Where does it come from? How does it impact us? How
did its research evolve? How can we use that knowledge? In subsequent
chapters, specific aspects of fire are addressed. What is the temperature of
a flame? Why does flashover occur in a room fire? How does ventilation
affect the fire? We will answer these questions and more by knowledge from
research and formulas that allow quantitative estimates. When this informa-
tion is pieced together, you might explain what you saw in a fire or why it
occurred, or you might defend what you think happened.
Now let us start at the beginning.
Evolution of Fire Science 3
1.2 What Is Fire?

To understand fire, we must have a scientific definition of fire consistent
with our perceptions. In scientific terms, fire or combustion is a chemi-
cal reaction involving fuel and an oxidizer—typically, the oxygen (O2) in
the air. But this definition is insufficient, as rusting and the yellowing of
old newsprint fit this definition and they are not fire. The distinction to
be fire or combustion is that significant energy has to be released. Can
we distinguish combustion from fire? Of course we cannot. In scientific
terms, combustion and fire are synonymous. In conventional terms, we
generally treat fire as distinct from combustion, in that fire is combustion
that is not controlled. Firefighters attempt to control it by adding water
or other agents, but the process of fire is not “designed” combustion, as
in a furnace or an engine. Combustion experts who study such systems
may know very little about fire, and those who deal with fire may know
very little about all aspects of combustion. The uncontrolled nature of fire
makes it distinct.
Fire is a chemical reaction that involves the evolution of light and energy
in sufficient amounts to be perceptible. Generally, that energy will emit
light. The color associated with fire might be blue from formative chem-
ical emitters in a flame or yellow to red from light emitted by soot in a
flame or smoldering char. Will there always be light in a flame (fire)? No.
For instance, the burning of hydrogen (H2) with air or oxygen produces
only water vapor from its chemical reaction. Although significant energy
is produced, we would not visibly see flame. But in most other chemical
reactions, we would see the combustion process in terms of light emission
from gases or particulates. A fire or a flame would be energetic enough to
be sensed, particularly with sufficient energy to damage our skin. It may
not be very big, but its chemical energy release rate (per unit volume) would
be sufficient to give us a local burn injury. This is an operational definition
of fire. Fire is a chemical reaction of significant energy release with damag-
ing ability. It need not be only a flame, as a surface reaction on solid can fit
this definition. It is the result of striking a match, the glow of the charcoal
briquette in your grill, the conflagration of the forest, and the spontaneity to
ignition of a large haystack. Fire is combustion, and combustion is a chemi-
cal reaction of significant energy release to cause damage. Operationally,
sufficient damage can be assessed as harm to human tissue or irreversible
damage to materials.
One last word on this chemical reaction called fire. On Earth, it typically
and naturally involves hydrocarbon-based fuels: those composed of atoms
of carbon (C), hydrogen (H), and perhaps some oxygen (O) and nitrogen
(N). Man-made materials have been added to this array of fuels (molecules),
with the addition of chlorine (Cl), bromine (Br), fluorine (F), and other
atoms. For example, wood molecules consist of the atoms involving H, C,
and O; polyvinyl chloride (a plastic) contains H, C, and O plus Cl atoms; and

polyurethane (another plastic) contains H, C, and O plus N atoms. These
additions to the H─C─O base complicate the nature of combustion products
and their potential threat to the environment. It is likely that other elements,
even without the addition of oxygen, could meet our combustion. In short,
fire or combustion is simply a chemical reaction that can suddenly occur to
release significant energy with a relatively high temperature.
All chemical reactions conserve mass, which means all the atoms survive.
In contrast, in a nuclear reaction some atoms are converted into new atoms
with some matter transformed to energy. In a chemical reaction, while atoms
are maintained, molecules are not conserved. Their destruction is the essence
of a chemical reaction. A chemical reaction converts the reacting molecules
into new molecules (generally called products). For combustion or a fire, the
formation of new molecules from the fuel and oxygen molecules gives off
a distinct amount of energy. This energy comes from breaking the binding
forces that held the original molecules together.
This is as close as we get in our discussion to molecular physics. From here
on, our discussion of fire principally concerns what we can see and feel. Of
course we rely on some measurements, but these are at the macroscopic level
in contrast to the microscopic or molecular level. More importantly, we come
to rely on predictive formulae, for the processes, and effects of fire. These
predictive results come from macroscopic measurements and analyses, as
we will see in the forthcoming chapters.
1.3 Natural Causes of Fire

Before humans encountered and generated fire, fire was present in the
universe. Natural high-temperature phenomena that can cause fire are light-
ning and molten rock from volcanic activity. In addition, the energy release
from a high-velocity impact from space can result in fire. These phenomena
date back to the formation of the Earth. As organic matter developed, we
could expect to have fire. Even today, these phenomena are potential causes
of accidental fire, especially lightning.
1.3.1 Lightning
Workers keeping watch over our forests record lightning strikes. These
strikes can cause smoldering to the underbrush on the forest floor. A day
or two after the storm, flames can erupt. The forest crews track these strikes
and attempt to follow up with needed extinguishment. Lightning is even
known to cause house fires. The temperatures in a lightning bolt range to
several thousand of degrees, and it has enormous energy.
1.3.2 Earthquake
Of more concern to modern society is the effect of an earthquake. While there
is no direct fire from an earthquake, the disturbance of combustion devices
and fuel sources provides the means to initiate fire. An earthquake can play
havoc with fire and fuel sources used for heating and cooking. Indeed, in
some earthquake events, the resulting fire has caused more damage than
the ground shock. This was true for the conflagrations resulting from the
San Francisco earthquake of 1906 and the Great Kanto earthquake of 1923
in Tokyo and Yokohama. During the fire following the Kanto earthquake,
38,000 people were killed by a fire whirl—a flaming tornado—that spun
off the main fire column. This unusual phenomenon was likely caused by
winds from a nearby typhoon. People that had taken refuge in a park adja-
cent to the Sumida River in the Hifukusho-ato district of Tokyo were directly
impacted by this large fire whirl. This horror is depicted in the Japanese print
in Figure 1.1a and in the gruesome photograph of corpses resulting from the
fire whirl in Figure 1.1b. A total of 143,000 people overall are reported to have
perished in the Kanto earthquake and fire.
Many refer to the San Francisco earthquake of 1906 as the Great Fire of
San Francisco.1 The fire, caused by broken gas lines and disturbed gas and oil
devices, burned for 74 hours after the quake. It burned over 4.7 square miles
destroying 28,000 buildings. Seven hundred were reported to have died, but
this could be as high as four times this number. The death toll estimate is
(a)
FIGURE 1.1
(a) Japanese painting of the Kanto earthquake fire whirl (1923) from Y. Hasemi. (Continued )
(b)
FIGURE 1.1 (Continued )

(b) Photograph of corpses killed by the fire whirl in Hifukusho-ato, Tokyo (1923) from K. Saito.
not firm, and some place it as high as 6000. Firefighting had limited water
available. It was brought under control by the use of dynamiting.1 In contrast,
the Loma Prieta earthquake of 1989 in San Francisco killed 63 and caused
a fire in the Marina district that destroyed four buildings. Firefighters had
trouble getting water to the fire due to broken water mains.
More recently, earthquakes in Japan produced differing fire damage.
Around Kobe (January 17, 1995) the Great Hanshin Earthquake, in which
6400 died, several fire conflagrations occurred. The fires destroyed over 6000
buildings. In contrast, in March 2011, the Tohoku earthquake and tsunami
did not result in significant fires. However, the effects of the event on the
Fukushima nuclear facility caused explosions in three reactors. These were

explosions due to the release of hydrogen gas from chemical reaction between
zirconium and steam when the nuclear reactor cooling system failed. The
reaction between hydrogen and air caused the explosion, as this sudden
energy release failed the containment vessel of the reactor. Undoubtedly,
fires in the contents resulted as a consequence. Also, the release into the
atmosphere of radioactive products occurred. This dispersion of radioactive
debris is a long-term hazard.
It is surprising that more study is not done on the effects of fire due to
earthquake as opposed to just the structural impact of the jolt. And perhaps
the consequences of fire in a nuclear facility should be studied more. The
event at Three Mile Island was similar to Fukushima without an earthquake
cause. That event was caused by loss of coolant to the reactor and then a sim-
ilar release of hydrogen. A fire in a nuclear waste facility is another potential
disaster, as airborne radioactive waste can be transported very far by the fire
plume in such an accident.
1.3.3 Meteors
There is one more natural cause of fire that humankind has not experienced,
but unfortunately, the dinosaurs did. It has been generally established that
in about 65 million BC a large meteorite smashed into the Earth. The meteor-
ite’s passage through the atmosphere and its impact (equaling many nuclear
bombs) caused frictional heating and pressure that propelled debris and fire
products into the atmosphere. This resulted in a soot cloud that circled the
Earth and affected the atmosphere for at least a year, causing a perpetual
winter and the demise of the dinosaurs by virtually eliminating their food
supply. This cataclysmic event probably occurred more than once over eons,
sealing the fate of the dinosaurs and other creatures and plants. The con-
sequences of this particular fire event are known because remnants of the
meteorite (iridium) have been found in various parts of the world in the
strata of rock from the same geologic time: between the Cretaceous and
Tertiary periods.
The impact scenario causes smoke and dust to enter the atmosphere,
block sunlight, and lower the temperature. The so-called nuclear winter
has the potential to destroy much life on Earth, just as a perpetual winter
destroyed the dinosaurs. This winter effect of a large fire was publicized
in the 1980s when nuclear warfare studies determined that the fire from a
limited nuclear war could produce debris and smoke that would result in
a nuclear winter. Our world would be in jeopardy from a sustained reduc-
tion of light and temperature. Many studies were done to substantiate this
“fallout” from even a survivable nuclear war. The noted science writer, Carl
Sagan, was one of the strongest advocates for studying the consequences
of nuclear winter. Even before the cold war ended, all nuclear parties were
chilled by this possibility.
The recent meteor strike in Russian (2013) was unexpected and caused
much damage from the pressure of its associated shock wave. No fires were
reported as a consequence, but the event has heightened humankind’s alert-
ness to such vulnerabilities.
1.3.4 Volcanoes
Similar to the meteorite consequences, large volcanic activity has been
known to cause catastrophic extinction of the Earth’s species. It might be
of interest to those in the United States to realize that the large crater in
Yellowstone National Park is the remnant of a large volcano. It last cata-
strophically erupted about 640,000 years ago, launching 1,000 km3 of rock
and dust into the atmosphere. Its violent explosion would have produced
tremendous damage for many miles around. In modern times, Krakatoa
(May 1883) erupted with such an effect that barographs recorded it around
the world. The ash was driven 50 miles into the sky, and it settled around
the globe. A subsequent drop in average global temperature of 1.2°C was
found. We, as those in Pompeii (AD 79) buried under 20 ft of ash and
rock, are still vulnerable. This was highlighted when a volcano in Iceland
erupted in 2010. Its ash cloud closed air travel over much of Europe for
about a week.
1.3.5 Underground Fires

There are everyday common examples of fires in subway systems, tunnels,
and mines. In addition, fire can begin and be sustained within the soil.
Consider a couple of examples with historical significance.
Long after the dinosaurs, humankind evolved and eventually cultivated
fire. Aristotle thought it is sufficiently important to classify it as one of
the four elements of matter: fire, earth, air, and water. Fire was used and
abused in many ways. Hazel Rossotti vividly describes the uses, haz-
ards, and spiritual qualities of fire as used by society.2 She describes how
Cherokee Indians learned to preserve fire by burying a smoldering log,
then digging it up, and fanning it into flames. Buried fire is now associ-
ated with waste dumps of wood or recycled plastics. Such fires can start
from spontaneous ignition leading to smoldering and then flames. This
process probably came as a surprise to those who experienced the Stump
Dump Fire near Baltimore, Maryland in 1990. This incident involved more
than 5 acres of buried tree stumps. After many attempts at extinguish-
ment failed, the stumps were more completely buried, probably slowing
the smoldering, but not extinguishing it.
A mystery in an underground fire is associated with the Terracotta
Warriors found outside of Xian in China. The tombs were constructed by
the first Emperor Qin about 221 BC. These tombs contain an army of life-size
terracotta soldiers. On their discovery in 1974, the archaeological findings
displayed evidence of extensive fire debris at the roof level. The tombs were
buried in the ground with a timber roof about 3 m below the surface. As these
were secret, and completely buried, it is not known how the fires occurred.
The determination of the cause—desecration, accidental in robbery, or
natural—can illuminate history.3 Fire forensics can be a tool for archaeolo-
gists that can reveal more than the artifacts themselves.
1.4 Fire and War

The use of fire in war is well known. But before World War II, fire was some-
what limited to ground operations. Aircraft bombing added to its extent and
intensity. The introduction of napalm—a gelled petroleum product—added to
the consequences of aerial bombing. The first systematic and sustained bomb-
ing of a city was the London Blitz lasting from about August 1940 until the
Nazis invaded Russia in June 1941. Incendiary bombing was not used. But by
1943, the Allies established a policy to disturb life in large cities by destroying
large areas of the city and halting normal life activities. This was designed to
target the morale of the population.4
The first city targeted with this policy intention was Dresden. It was
bombed on February 13, 1943 at 10:13 p.m.5 Four thousand pound bombs
were used to splinter the structures to make them more combustible.
Incendiary napalm bombs followed this first attack. The fires were seen from
over 100 miles away. A storm arose; as far as the eye could see, this was
given the name, firestorm. Survivors said it was like being in a burning stove.
Currents of air and fire bombarded the fleeing people. Those that could not
move were asphyxiated where they stopped. Over 75,000 apartment build-
ings were destroyed, and likely 40,000 died, but this number is debated.5
The next firestorm occurred in Hamburg on March 27/28, 1943.4
Lancaster and B-17 bombers were prominent among nearly 800 aircraft.
“The whole yard, the canal, in fact as far as we could see, was just a
whole, great, massive sea of fire.”4 Six thousand apartment-block build-
ings burned in a 1.5 × 3 mile area. People were ignited by the radiation,
some were blown into burning buildings, clothes were torn off, air was
sucked out of shelters, and 40,000 likely died. The massive fire caused
hurricane force winds at ground level, broke branches, and even uprooted
trees. The fire raged for 5 hours.
These large-area fires suck in air from all directions. They pull the oxygen
out of the core and bring high-speed air around the perimeter and into the
center. This is the firestorm. Its name was invented at these fires.
The term firestorm was never applied to incidents in Japan. But General
Curtis LeMay used the same tactics in bombing Japanese cities. These tactics
were refined and studied during bombing experiments in the summer of
1943 at Dugway Proving Grounds in Utah on a simulated city known as

“Little Tokyo.” They demonstrated the vulnerability of the traditional
Japanese wooden houses. During the spring and summer of 1944, most
major Japanese cities were carpet bombed by B-29 bombers. Of these major
cities, from about 25% to 75% of the city’s areas were obliterated and turned
to ash. In Osaka, 16 miles2 were destroyed in contrast to 3 miles2 destroyed
in the Great Chicago Fire of 1871.6 This destruction far exceeded the effects
of the atomic bombs on Hiroshima and Nagasaki.
There is no doubt that henceforth fire will be part of war. Yet with a
smaller “weapon,” a single aircraft filled with liquid fuel was the instru-
ment of destruction for World Trade Center Towers and the Pentagon. Those
were aircraft fuel–initiated fires that ultimately led to sustained fire spread
over ordinary contents and heat-induced failure of the structures. The cause
of the fires is clear, but the reason for the destruction of these fire-resistive
buildings is not so evident.
1.5 Fire in the United States and Abroad

History is punctuated with fire disasters. The Great Fire of London (1666)
and the Chicago Fire (1871) caused the destruction of thousands of build-
ings. These were literally forest fire–like events in that wind was a princi-
pal factor in their spread. Once the wind died, the fire stopped. Mechanical
suppression apparatus could not deal with such fires until calm conditions
prevailed. These fires were conflagrations or mass fires involving large tracts
at a given time. Their lateral flame extent is much greater than their flame
height. It was noted that similar large fires resulted in World War II from
such bombing raids as Hamburg, Dresden, and Tokyo. Today, we begin to
see some return to large conflagrations of dwellings as urban development
has spread into the forest. The term urban wildland fire has now been coined
to refer to these fires.
1.5.1 U.S. Statistics

It is instructive to put the catastrophes of fire into perspective against other
natural and technological disasters. Indeed, many of the greatest disasters in
recent history from natural phenomena are likely to be due to fire. Table 1.1
lists the top 25 U.S. disasters by death toll. The fatality level ranges from 400
to possibly as high as 12,000. More than 40% involved fire. So, fire is a major
contributor to large life loss catastrophes.
Such enormous disasters are relatively infrequent as they span over
more than 100 years. Instead, it is the local effects of fire that have contrib-
uted to our perception of its hazard. According to National Fire Protection
TABLE 1.1
Top 25 Disasters in the United States by Death Toll (Excluding Epidemics
and Heat Waves)
Event Date Fatalities
1. Hurricane, Galveston, TX 1900 6,000–12,000
2. Earthquake and fire, San Francisco 1906 450–6,000
3. Hurricane, Great Okeechobee, FL 1928 1,800–3,000
4. 9/11 Terrorist attacks 2001 3,000
5. Military attack, Pearl Harbor, HI 1941 2,500
6. Flood, Johnstown, PA 1889 2,200–3,000
7. Hurricane, LA 1893 2,000
8. Hurricane Katrina, LA, MS, FL, AL 2005 1,830
9. Ship explosion, Sultana, Mississippi River, TN 1865 1,500–1,700
10. Peshtigo fire, WS, MI 1871 1,200–2,500
11. Ship fire and sinking, East River, NYC 1904 1,000
12. Hurricane, Sea Islands, SC, GA 1893 1,000–2,000
13. Ship sinking, S. S. Eastland, Chicago 1915 800
14. Tri-State Tornadoes, MO, IL, IN 1925 700
15. Fire, Iroquois Theater, Chicago 1903 600
16. Hurricane, Florida Keys 1919 600
17. Flood, St. Francis Dam, CA 1928 600
18. Hurricane, New England 1938 600
19. Explosion (ship), Texas City, TX 1947 550
20. Wildfire, MN 1918 500–800
21. Wildfire, MI 1871 500
22. Fire, Cocoanut Grove, Boston 1942 492
23. Tornadoes, Tupelo (MS)–Gainesville (GA) 1936 430
24. Floods, OH 1913 430
25. Hurricane, Labor Day, FL 1935 400
Note: From various sources with estimated death tolls.
Association (NFPA) over 1.3 million fires were reported in the United States
in 2010, with about 480,000 in structures, 200,000 in vehicles, and 630,000
outside.7 With a U.S. population of 308 million, this suggests a structural
fire each year for at least 1 in 200 households. This is the frequency of fire
occurring possibly in your home. Roughly 3700 civilian deaths occurred
in 2010 in the United States. For a total population of 308 million with a life
expectancy of 70 years, we calculate
308 ´ 106 people

70 years ´ 3100 deaths/year
or roughly 1 in 1400 people will die by fire in the United States over your
lifetime. This fire fatality frequency is roughly 10 times more than for auto-
motive deaths. As an alternative check on these rough estimates, the National
Safety Council estimated for 2007 that the chance of dying by fire in your
lifetime is 1 in 1177, while dying in a motor vehicle is 1 in 88.
So with these current frequency estimates, you may have a neighborhood
fire (over roughly 200 households) about once a year, and you may know at
least one person who will die by fire during your lifetime. These fire statis-
tics relative to other threats to people and property do not cry out for a great
concern about fire.
The causes of fire in dwellings for the United States are displayed in
Figure 1.2. The major causes are cooking and heating equipment. These are
technological causes, except for the human role in cooking fires, and they
extend to areas of electrical appliances, lighting, and clothes washing and
drying appliances. Intentional is still significant, and could even be higher,
as this is not an easy call.
The U.S. overall death rate due to all accidents since about 1900–2010 has
steadily dropped from about 80 per 100,000 of population to 40.8 Figure 1.3
shows the relative significance of fire with other accidental causes of death in
2009. It clearly shows that fire plays a small role compared to other accidental
causes.
Playing with heat source
Exposure to other fire
Candle
Clothes dryer/washer
Smoking materials
Electrical and lighting
Intentional
Heating equipment
Cooking equipment
0 10 20 30 40 50
Portion of fires (%)
FIGURE 1.2
Causes of fire in homes in the United States 2005–2009. (From Accident Facts, National Safety
Council, Itasca, IL, 1995.)
Others
Mechanical suffocation
Categories of accidental death
Fire
Drowning
Choking
Falls
Motor vehicle
Poisoning
0 2 4 6 8 10 12 14
Death rate per 100,000
FIGURE 1.3
Death rate due to unintentional injuries 2009. (From Accident Facts, National Safety Council,
Itasca, IL, 1995.)
1.5.2 United States and the World

It is of interest and curiosity to compare the U.S. fire statistics with the rest
of the world. The Geneva Association (www.genevaassociation.org) main-
tains a program that tracks these statistics, and some additional country data
came from Brushinsky.9,10 In Table 1.2, the death rate statistics are given as
produced in the first edition and amended to more recent times. China is
updated to 2004.11 It is difficult to draw generalities from the data, but there
are some trends.
Cold weather countries tend to have generally high death rate from fire.
China and Russia are at opposite ends, and one might wonder about these
data. Statistical data for South America, the Mid-East, and Africa are not gen-
erally available. It is a benefit to have these statistics, as they can set policy
for a country. Also, if we learn what underlies the statistics in one country,
we might benefit others. Social, culture, technology, climate, and prevention
measures all play a role in these statistics.
1.5.3 U.S. Fire Prevention Infrastructure

Fire events and statistics do move people to action. Throughout the twentieth
century, much effort has been put into the prevention and protection from fire.
In the United States, following the Great Fire of Baltimore (1904) and the fact
that the property loss due to fire was then 10 times that of Europe, a national
research program was initiated at the National Bureau of Standards (NBS)
TABLE 1.2
Annual Fire Deaths per 105 Persons
Country 1992–1994 2004–2008 Country 1992–1994 2004–2008
Russia 10.6a — France 1.26 0.98
Hungary 3.31 1.81 Czech Republic 1.21 1.41
India 2.20a — Germany 1.17 0.68
Finland 2.18 2.08 Australia 0.93 0.48
Union So. Africa 2.00a — New Zealand 0.92 0.75
United States 1.95 1.21 Spain 0.86 0.58
Denmark 1.64 1.28 Poland 0.80a 1.56
Norway 1.60 1.33 Austria 0.74 0.46
Canada 1.58 1.15 Netherlands 0.63 0.52
Japan 1.52 1.62 Switzerland 0.53 0.30
United Kingdom 1.49 0.80 Italy 0.30a 0.46
Belgium 1.47 1.21 China 0.20a 0.19b
Sweden 1.35 1.20 Singapore — 0.11
Source: World Fire Statistics Centre Bulletin 12 & 27, The Geneva Association, Geneva,
Switzerland, 1996/2011 for 1989–1992.
a From Brushlinsky et al.9 for 1994.
b From Guo and Fu.11
to investigate the consequences of fire on building construction materials.

Steel and concrete buildings were not necessarily found to be “fireproof.” The
NBS fire program in 1913 was established under Simon H. Ingberg who would
head the Fire Resistance Section of the Heat Division for the next 40 years.
In 1914, due to a Congressional mandate, funding and the challenge of the com-
plex fire problem inspired much work at NBS. NBS history states “so broad
became the scope of the investigation that it soon involved almost every one of
the scientific and engineering laboratories of the Bureau.”12
In addition, the National Fire Protection Association (NFPA) was estab-
lished in 1896 to provide a voluntary basis for technical information and
standards—recommended procedures and practices—concerning fire
safety. In addition, in those early years, the Underwriters’ Laboratory and the
American Society of Testing and Materials began to contribute standard test
methods to assess performance in fire conditions. This led to fire resistance
tests that enable the measurement of the endurance of structural elements of
buildings and structures in fire. Such tests, conducted in standard furnaces
throughout the industrialized world, help to ensure that structures do not
collapse because of fire. But all fires do not match the conditions of the
standard fire resistance furnace tests, and practices are not universal in their
attempts to match the testing results to actual anticipated fire conditions.
More striking is the fact that “standard” furnaces do not always impart the
same fire heating conditions. Standard practices are not universal through-
out the world, and their application to real fire conditions is far from perfect.
1.5.4 Motivation for Improvement

In the early 1970s, it was acknowledged that the United States had the highest
annual death rate from fire of the world’s industrialized nations. It had fire
deaths per year listed at 12,000. Once research commenced, this was quickly
reduced when it was found that an error in compiling vehicle fire deaths had
been listed as 4000 instead of 400!
Congress passed the Fire Prevention and Control Act of 1974. This
act structured the U.S. Fire Administration to enhance the practice of
firefighting, to improve education, to assess national fire statistics, and to
develop research. It focused research on the development of fire in build-
ings: ignition, flame spread, smoke, and flashover (a sudden event in fire
growth that rapidly leads to full involvement of a room). The safety of
people rather than structures became the focus. This attitude of mandated
public safety, coupled with the threat of civil suits due to issues affecting
fire safety, motivated the increasing use of innovative fire safety technolo-
gies. These included residential smoke detectors, smoke control systems in
large buildings, and the extension of sprinklers to public and residential
occupancies. Indeed, many of these technologies have attributed the drop
in annual U.S. fire deaths by more than half of the 8000 in the early 1970s.
Most of this drop is attributed to the widespread use of smoke detectors.
This is remarkable.
Changing technologies bring improvements in fire safety but can also
bring new risks. The high death rate due to fire in the United States and in
other industrialized nations may be associated with the changing technolo-
gies with which we live. Change brings risks, some unexpected. For exam-
ple, in the 1970s, lightweight cellular plastics (such as foam polyurethane
and polystyrene) were seeing new innovative applications, but their fire
hazards were not fully appreciated. Standard tests suggested no problems,
but accident scenarios showed a dramatically fast rate of fire spread. These
incidents brought standard flammability tests into question. The Federal
Trade Commission (FTC) enjoined the plastic industry to take corrective
action. The Product Research Committee (PRC) was established by industry
in 1974 to examine this issue for cellular plastics as a result of a consent order
between the FTC and the industry.13
1.5.5 Flammability Tests

Although the PRC group gained much insight into the issue of foam
plastic flammability, it remains unresolved today. To mitigate concern,
today lightweight cellular plastics are generally required to be covered by
wallboard in building construction applications even after they pass the test
requirements for flammability. Confidence in their flammability ratings by
standard tests is low. But this issue of product flammability is not unique to
cellular plastics; it is widespread.
The disparity among flammability tests for construction materials is

illustrated in Figure 1.4 for six national European tests in the ranking of
24 materials. Professor Howard Emmons publicized this chart after he dis-
covered these data in Europe during a world tour to review the status of
fire science.14 Perfect correlation among all of the six tests should produce
the same ranking for the 24 materials; this would be a 45° straight-line cor-
relation. For example, the number 6 material should be ranked as 6 by all
the tests for a perfect correlation of flammability. The tests decidedly do not
correlate! Each country’s test gives a different measure of flammability for
the same material.
24
22
20
18
16
14
Test rating
12
10
0
2 4 6 8 10 12 14 16 18 20 22 24
Varieties of wallboard
Germany France
Belgium Netherlands
Denmark England
FIGURE 1.4
Disparities among fire tests—rankings of 24 materials by six national flammability test
methods. (From Emmons, H.W., Fire Res. Abstr. Rev., 10(2), 133, 1968.)
The results are not consistent, giving ambiguity to the term flammability.
Perhaps it is appropriate that in a book intended to present accurate prose,
the word flammable is listed among those misused:
Flammable, an oddity, chiefly useful in saving lives. The common word

meaning “combustible” is inflammable. But some people are thrown off
by the in- and think inflammable means “not combustible.” For this rea-
son, trucks carrying gasoline or explosives are now marked flammable.
Unless you are operating such a truck and hence are concerned with the
safety of children and illiterates, use inflammable.15
This change in the word was likely done in the interest of safety, but it can
still cause confusion, especially if you are French. A clear scientific basis is
needed to classify the fire hazard of materials.
We still do not currently have a universal test procedure to establish flam-
mability or, alternatively, inflammability. These current tests are reflections
of our misuse of the word. The world cannot seem to establish a universal
test. In the 1990s, Europe sought to harmonize its flammability tests. First,
they considered the tests of the “Three Sisters”: national tests of Germany,
France, and the United Kingdom. The results gave a chart again like the
Emmons graph. They were using as a benchmark the Room Corner Test.
That test measures the energy release of fire spreading on a wall and ceil-
ing material in a room subject to a corner ignition. That is a realistic test
for lining materials, but it is expensive. So, the desire is instead to have a
small laboratory inexpensive test as the standard. They focused on the Cone
Calorimeter as a candidate test for the standard. That test can measure the
time to ignition and the rate of energy released by a material. This seems like
a good scientific match. But fire standards are political, between the pride of
countries, and the impact on commercial products. Some plastic materials
did not look good in the Cone test, and ultimately, a new test was invented
to “bring harmony” to Europe. This should not be the way progress is made
for fire safety. If only industry and countries would realize that a universal
test based on scientific measurements can be a blessing. A blessing is not just
to safety but also to the cost of doing business and streamlining commerce.
Even if it is not at first a perfect test, a scientific test can be built upon with
future knowledge and made better. Today, an ad hoc test with defects is com-
monly replaced by another arbitrary test.
1.5.6 Cost of Fire

The annual costs of fire are not insignificant, but because of the rela-
tively low frequency of fire compared to other societal threats, they are
not fully appreciated. Moreover, fire safety costs do not increase the pro-
ductivity of the economy; they are a drag on the economy. Costs have
to be expended to make buildings safe from fire and to make products
less likely to contribute to fire. When these costs become too great, more
attention will be made to minimizing them. For now they are viewed as
regulatory costs that are a necessary burden to business. The cost of fire
safety is a societal issue.
It was estimated in 1991 that the annual cost of fire in the United States
was $85 billion.16 The overall cost includes property loss, business interrup-
tion, product liability, insurance administrative costs, fire service (paid),
and fire protection in construction and equipment. Today, this cost is much
higher. Table 1.3 shows the percentage of major country fire costs scaled to
their gross domestic products (GDP) for 2002–2008.9 The costs for the United
States (2008) in actual dollars are as follows: property loss $17 billion, fire
protection in construction $62 billion, fire service $41 billion, and insurance
administrative costs $18 billion for a total of $138 billion.9 The cost continues
to increase, likely due to inflation.
But as can be seen, it is typical for many developed countries to invest
nearly 1% of their GDP in fire-related costs. The United States appears
to spend the most relative to GDP. Unfortunately, not even 1% or 2% of
these fire costs are invested back into improving our knowledge of the
TABLE 1.3
National Annual Fire Costs by Percentage of GDP for 2002–2008
Property Building Fire Fire Insurance
Country Loss Protection Administration Firefighting Total
Canada ND 0.32 ND ND ND
Spain 0.08 ND ND ND ND
Japan 0.12 0.14 0.09 0.30 0.65
Finland 0.17 ND 0.03 0.20 ND
United States 0.11 0.41 0.12 0.27 0.91
Slovenia 0.07 0.16 0.06 0.05 0.34
New Zealand 0.11 0.23 0.08 0.16 0.58
Germany 0.13 ND 0.04 ND ND
Netherlands 0.16 0.31 ND 0.19 ND
Australia 0.08 0.35 ND 0.16 ND
United Kingdom 0.13 0.22 0.10 0.21 0.66
Italy 0.17 0.35 0.05 ND ND
Denmark 0.20 0.26 0.09 0.07 0.62
Sweden 0.17 0.19 0.05 0.13 0.54
France 0.20 0.18 0.07 ND ND
Norway 0.20 0.36 0.10 0.11 0.77
Singapore 0.05 0.39 0.02 0.03 0.49
Source: World Fire Statistics Centre Bulletin 12 & 27, The Geneva Association, Geneva, Switzerland,
1996/2011.
ND, data not available.
technology of fire safety. For the United States, the investment in research
is less than 0.04% of annual cost of fire today by my estimates. Typically,
industry will invest more than 3% of its revenue into research to improve
its products.
In 1975, when the most productive research period for fire research
occurred in the United States, the investment was about 0.06% of total
fire costs. Not a big difference, but a lot more money to stimulate research
and make progress was available back then. Today, this would amount
to an increase in research for fire at about $27 million per year or more.
This is not a lot of money for society to invest, but it would make a big
difference for fire safety. There are few advocates for such an increase.
Lessons are not just learned from history but must also be analyzed
using the science of fire. Research is needed to lay the foundation of the
full knowledge base. Perhaps the future will bring more fire research
as fire investigators and firefighters are becoming more sensitive to its
potential benefits.
Doubtless, future disasters will shape the course of fire safety. In develop-
ing nations, high-rise buildings are being built to heights that dwarf current
levels. It is possible that thousands of occupants will be seriously affected
or killed by a fire disaster in a high-rise building. The previous sentence
was in the first edition published before September 11, 2001. The WTC 9/11
fire event is now classified as the “deadliest large loss fire” in history with
over 2000 dead and a loss of $41 billion (in 2010 $) according to NFPA.7
Since then there has been the Caracas Tower fire (2004) with flames travel-
ing from the 34th floor to 26 stories above; the 38-story Windsor Building
fire in Madrid (2005) that burned for more than 24 hours; and the Beijing
Mandarin Oriental Hotel (550 ft tall), unoccupied, burned out completely in
3 hours. Fortunately, loss of life was not the issue in these three buildings,
but the vulnerability of tall buildings to fire is key. Fire safety in tall build-
ings deserves special attention.
Future living arrangements in outer space under low gravity and in
underground and undersea structures will present unexpected fire
hazards. Many large cities around the world are moving underground,
especially for people in commuting and for shopping. We already see the
issues with moving into the forest environment. Issues related to seeking
new forms of power will bring new problems. Electric and hydrogen-
based vehicles will present new fire issues. With hydrogen it is obvious,
but higher voltage batteries will be the likely cause of fires in the new more
combustible vehicles. Fire affecting radioactive operations and toxic waste
storage sites will continue to present hazards of new dimensions. We have
already seen this at Fukushima. There will be new fire hazard surprises
in future technological advancements and products despite improvements
in fire safety technologies. We are likely not to fully anticipate the fire haz-
ard of new technologies.
1.6 Fire Research

The study of fire is a complex subject that comprises an array of interde-
pendent disciplines. Each of these subjects needs to be developed before the
pieces can be put together to adequately describe fire. Science is the evolution
of many steps and contributions. Eventually, the subject takes shape, and
individuals formulate or unify the subject by quantitative description that
allows for predictions and assessments. This process leads to a formally rec-
ognized scientific discipline for a subject. Let us briefly examine the history
of these component subjects that underlie fire.
1.6.1 Disciplines That Underlie Fire

Fire is uncontrolled combustion involving chemistry, thermodynamics,
fluid mechanics, and heat transfer. Thermodynamics, the study of energy
and states of matter, was principally shaped by Willard Gibbs, a noted
nineteenth-century scientist, who brought unification and clarity to the
subject that is still appreciated today. Heat transfer also had its roots in the
early 1800s. Joseph Fourier, a general in Napoleon’s army, formulated the law
of heat conduction that forms the theoretical basis of the field. But heat trans-
fer in fluids had to await the development of modern fluid mechanics in
the late 1800s when the pioneering work of Osborne Reynolds on turbulent
flow laid the basis for practical engineering analysis in fluid mechanics. In
the 1900s, Theodore von Kármán and others advanced the subject of aero-
dynamics, which paved the way for a more complete framework for heat
transfer. At this point, solutions were based on approximate methods, since
the governing mathematical equations were too complex to be exactly solved.
In combustion, Y.B. Zel’dovitch, a Russian scientist, was able to formulate
solutions for diffusion flames using innovative approximate mathematical
techniques. Although begun in the 1930s, the subject of combustion was not
developed to a mature state until the 1950s.
1.6.2 Computer Simulations and Physics

Today, fast and large capacity computers make it possible to approximately
solve the basic equations with good accuracy. Many facets of combustion
can be rendered by computer simulations, but issues of turbulence, chemi-
cal kinetics, and other small-scale phenomena still cannot be completely
resolved by these computer solutions.
Over the last decade, many more engineers and investigators of fire are
now turning to computer solutions. The Fire Dynamics Simulator, originat-
ing from National Institute of Standards and Technology (NIST), presents
a graphical user-friendly tool. The code is based on the solution of the fun-
damental governing equations but still must rely on underlying models
for combustion and turbulence. FMGlobal has embarked on an alternative

code called FireFOAM. These efforts are advancing the state of the art in fire
computation. There is no doubt that as computers get still bigger and faster,
these computer simulations will be invaluable.
As with all models, computer codes can be misused, as they can present
realistic graphical renditions of fire and smoke in their output. The results
may look right but can be quantitatively inaccurate. Such results must always
be validated for their specific application. This validation must rely on
experimental data that bring generality to the problem. Such experimental
studies have been the backbone of engineering science. They have laid the
foundation of formulas that match the generality of the data and circumvent
the need for fundamental mathematical solutions involving complexities of
turbulence and combustion. This book deals with these backbone formulas.
The use of such formulas relies on understanding the subject. The engineer
must rely on intellectual insight and understanding to use these approxi-
mate formulations correctly. Only when a thorough understanding of a sub-
ject is mastered can simple representations of complex phenomena be made
and appreciated. These formulae and their development form the basis of
education and the transfer of knowledge. Without this basis, we can just turn
on the computer and get an answer. We need both computer simulations
and approximate formula for specific processes. The intelligent use of both
approaches is key for design and analysis. They should provide a counter-
check to each other.
1.6.3 Brief History of Fire Science

Let us return to the history of fire science. The subject of fire needed to build
on all of its component disciplines. These interactions among the disciplines
had to mature before it was even possible to adequately describe and predict
fire. I can remember in the 1970s combustion scientists talking to behavior-
ists, chemists talking to toxicologists, and firefighters talking to engineers.
These discussions are needed today to bring a balance to the science of fire.
Another factor influencing the development of fire science is the moti-
vation to study it in the first place. We can see that fire is a drain on the
economy, and there is no direct market incentive for its study. In Japan, the
consequences of earthquakes led to an extreme sensitivity to fire safety and
its study. Fire science is studied in schools of architecture in Japan as well as
in other scientific fields. It is endemic in Japanese culture and academic dis-
ciplines. Not surprisingly, the first science-based handbook on the quantita-
tive description of fire was published in Japan in the early 1980s. Even today
I am amazed at the infrastructure for fire research in Japan. At their annual
scientific meetings, it is not uncommon to have up to 100 technical presen-
tations on fire. Indeed, the first Society of Fire Protection Engineers (SFPE)
standard issued in May 2011 (SFPE Engineering Standard on Calculating Fire
Exposures to Structures) is derived from Japanese research and standards.
In 1949, England developed one of the most advanced scientific labo-

ratories for fire study in the world (the Fire Research Station, formerly at
Borehamwood). Much of the work through the 1960s was performed under
the leadership of Dennis Lawson, Philip H. Thomas, and David Rasbash.
Rasbash went on to establish the first graduate program in fire engineer-
ing at the University of Edinburgh. P.H. Thomas has been a main force in
disseminating the benefits of fire science throughout the world. Some have
said that he was first to work on many of the subjects of fire and got them
right. Their work has not been fully appreciated today because much of it
was never published in mainstream fire journals. However, today the Fire
Research Notes (1952–1978) can be accessed online through the International
Association for Fire Safety Science (IAFSS, www.iafss.org). The IAFSS was
formed in 1986 with P.H. Thomas as its first head. It was established to facili-
tate the transfer of fire research around the world, and its proceedings con-
tain some of the best of fire research developed.
In the United States, extensive fire research emerged after Japan and the
United Kingdom. A special meeting can mark its start. On November 9, 1959,
many of the pioneers of fire science came together in Washington, DC, for
“The Use of Models in Fire Research.” Walter Berl, the conference organizer,
commented, “The intimate interplay between aerodynamics, heat transfer,
and chemical reaction rates makes the study of fires the intriguing problem
it is.”17 Scientists who had achieved prominence outside the field of fire were
present at this meeting and elected to study fire because of its challenge and
the prospect of its benefits to society.
In the United States, one of the earliest scientists of fire was Hoyt Hottel
of M.I.T. He began the study of fire before World War II but was diverted to
study the effects of fire from weapons during the war. Later, he and Howard
Emmons of Harvard University pursued fundamental research in fire and
lobbied for government research funding. Such research support was real-
ized in the early 1970s by expansions of fire research at the Factory Mutual
Research Corporation (now FMGlobal) and at the NIST (formerly the NBS).
These laboratory expansions and growth into academia were enhanced and
enabled by funding for fire research from the National Science Foundation
and through the National Fire Prevention and Control Act of 1974. The
fruits of that U.S. research effort in the 1970s helped to promote, develop,
and catalyze the disconnected efforts of fire research throughout the world.
Although U.S. funding for basic research in fire has since decreased, the
worldwide activity is expanding in its communication links, and there is
a healthy exchange of knowledge in this small field. For example, proceed-
ings of the symposia sponsored by the IAFSS help to maintain international
exchange in fire research. These proceedings are available online at www.
iafss.org. Also, the SFPE Handbook of Fire Protection Engineering18 is a good
illustration of the current knowledge base of fire science compiled by experts
among the disciplines of fire. It is this synthesis of fire research that makes
possible this current book.
1.7 Visualization of Fire Phenomena

It is crucial to have a visual concept of fire phenomena before one can
establish a framework for study. Many effects are seen (or can be seen if
planned) during the progression of fire and its related smoke movement.
These images must be categorized into scientific terms if we are to learn.
The shape of a flame is influenced by the fluid flow induced by the flame
itself. The nature of smoke movement in buildings can take many forms.
Such visualization must take place in a laboratory for systematic study,
but firefighters and others must be able to articulate their observations to
scientists to promote real study. The scientist cannot allow the size of his
laboratory to limit the scope or relevance of his observations in fire phe-
nomena. However, small-scale studies can be very relevant. We all appreci-
ate the role of wind tunnels in the design of aircraft and in improving the
aerodynamics of motor vehicles. The Wright brothers had a wind tunnel
at their disposal. Such scale modeling techniques pervade many fields of
study; fire is not excluded. Physical scale models based on the laws of sci-
ence can help to design aircraft, ships, oceanic tidal basins, concert halls,
and even analyze vehicle crash dynamics. Scale models can also reason-
ably represent processes in fire. Many of the formulas we will study in
this book have resulted from small-scale experiments and extrapolated to
larger scale by using scientific and mathematical principles. A deep under-
standing of the subject allows a full appreciation of the scale independence
of such formulas.
Figure 1.5 is a schematic illustration of phenomena arising in a room fire.
The recognition of these and other phenomena shown there helped to estab-
lish a framework for the “modeling” of compartment fires. Figures 1.6 and 1.7,
respectively, show the dynamics of smoke movement and fluid flow in a cor-
ridor subjected to a room fire.12 The results were accurately established using
a scale model with glass walls and colored smoke to make the processes
visible. These photographs show the classical representation of an apparent
uniform smoke layer filling the upper half of a compartment (Figure 1.6) and
the complex recirculating flows that occur both in the smoke layer and in the
relatively clear space below (Figure 1.7). Figure 1.6 also illustrates the mix-
ing of smoke into the lower clear space (at the right) as air enters through a
window and entrains smoke.
Figure 1.8 shows fascinating flame patterns as a result of flames imping-
ing on a ceiling.20 The patterns depend on fuel type, fuel flow rate, and
spacing to the ceiling. No computer simulation has yet to predict these
flame shapes.
Today, the availability of small inexpensive video cameras makes it pos-
sible to see fire behavior in room and building fire experiments that were
previously too difficult to see. You cannot model it and confirm it without
solid data, and visualization is an essential key.
Heat
transfer
ts
uc
od
Pr
ct s
Produ
Air
Turbulent Turbulent
mixing mixing
n
tio
a dia
R Air
Fuel Low-speed
buoyant flows
FIGURE 1.5
Phenomena in room fire growth.
WD =
0.109 m
0.057 m
Fire
room
WE 0.35 m
Temp. and
Temp. velocity
Corridor traverse
traverse
0.35 m ℓ = 1.3 m
(a)
FIGURE 1.6
(a) Schematic of a scale model corridor subject to a room fire at the left (doorway width,
WD = 11 cm) and exiting through the right (through a doorway width, WE). (From Quintiere,
J.G. et al., Fire Mater., 2(1), 18, 1978.) (Continued )
(A) No soffit WE/WD = 3.2 (B) WE/WD = 1
(C) WE/WD = 3.2 (D) WE/WD = 1/2
(E) WE/WD = 2 (F) WE/WD = 1/4

(b)

(b) Smoke layer in a scale model corridor subject to a room fire at the left (doorway width,
WD = 11 cm) and exiting through the right (through a doorway width, WE). (From Quintiere,
J.G. et al., Fire Mater., 2(1), 18, 1978.)
1.8 Scientific Language

In order to relate concepts and principles behind fire phenomena, we must
first qualitatively understand the behavior of fire. More importantly, we must
understand the information and methods used to quantitatively describe spe-
cific fire phenomena. Knowledge of physics and algebra is needed. Units for
measure, symbols, and scientific notation provide the language for quantitative
analyses. If the reader is weak in the area of algebra and units of measure, the
section in the Appendix should be read before proceeding further in this book.
1.8.1 Units of Measure

In today’s world, despite the lagging of the United States in the metric
system, the International System of Units (SI) is in widespread use. The
scientific community has universally adopted this system in its publications.
To understand articles in fire science, it is essential to be conversant with the
Room Corridor Exit view
HE = HD
HC
WE
YE
XCS
YD
XCV
XR
Stratified Smoke and air

(a) smoke layer mixing region
(A) No soffit WE/WD = 3.2 (B) WE/WD = 3.2
(C) WE/WD = 2 (D) WE/WD = 1

(b)
FIGURE 1.7
Smoke streaks showing the complex flow pattern in a scale model corridor subject to a room
fire. (From Quintiere, J.G. et al., Fire Mater., 2(1), 18, 1978.)
SI units. Table 1.4 lists the SI units for quantities relevant to the subjects in
this book. The quantities listed arise from the component disciplines of fire.
These quantities will be introduced as we proceed in the book to develop the
scope of fire phenomena.
As an example, let us consider energy. Table 1.3 lists the unit of energy as
a joule (J). The English System of units, common to the United States, would
alternatively have BTU units (British Thermal Units) for energy. 1 BTU is
1055.056 J or about 1 kJ. An alias for a joule is W-s (watt-second). It is common
to name units after scientists from the field and to present aliases. We most
likely realize that we pay our electric bill based on the amount of electrical
energy that we use. The billing is commonly based on kilowatt-hours (kW-h).
This is 3600 kW-s (or kJ) or about 3412 BTUs. Other alternative energy units
are listed in Table 1.5.
Although temperature is more commonly expressed in degrees Fahrenheit
(°F) in the United States, we can readily convert to other units of temperature
as shown in Table 1.5. Celsius or Centigrade (°C) is based on water freezing
and boiling at 0 and 100, respectively, whereas 32 and 212 are, respectively,
assigned on the Fahrenheit scale for these states of water. Both the °C and °F
100
Ceiling height H (mm)
50
The shape of the
propane diffusion
flames as a function
of the heat release
rate Q and the
ceiling–burner
separation
20
10
0 5 10
Heat release rate Q (kW)
FIGURE 1.8
Ceiling flame plume interactions. (From Kokkala, M. and Rinkinen, W.J., Some Observations
on the Shape of Impinging Diffusion Flames, Res. Rep. 461, VTT, Technical Research Centre of
Finland, Espoo, Finland, 1987.)
TABLE 1.4
SI Quantities and Units
Quantity Units
Force N (newton)
Mass kg (kilogram mass)
Time s (second)
Length m (meter)
Temperature °C or K
Energy J (joule)
Power W (watt)
Thermal conductivity W/m-K
Heat transfer coefficient W/m2-K
Specific heat J/kg-K
Heat flux W/m2
scales do not start their zero designation as the lowest possible tempera-
ture. This designation (absolute zero) is where all thermal energy ceases.
Therefore, alternative (absolute) scales of temperature begin with a zero
designation from this point. When the interval is 1°F, this absolute scale is
called the Rankine (°R) scale. Correspondingly, the Kelvin (K) scale is based
TABLE 1.5
Alternative Energy Units
1 BTU will raise 1 lbm of water 1°F at 68°F (lbm is a pound mass).
1 cal will raise 1 g of water 1°C at 20°C.
1 kcal will raise 1 kg of water 1°C at 20°C.
Some conversion factors for the various units of work and energy are as follows:
1 BTU = 778.16 lbf-ft (lbf is a pound force)
1 BTU = 1055 J
1 kcal = 4182 J
1 lbf-ft = 1.356 J
1 BTU = 252 cal
TABLE 1.6
Temperature Conversions
°F—degree Fahrenheit: T (°F) = T (°C) (1.8) + 32
°R—degree Rankine: T (°R) = T (°F) + 459.67
°C—degree Celsius or centigrade: T (°C) = [T (°F) – 32]/1.8
K—Kelvin: T (K) = T (°C) + 273.15
on absolute zero for the Celsius scale. Table 1.6 gives conversion formulas
between these various scales. It is important to appreciation that temper-
atures need to be in “absolute” units for certain formulas. For example,
formulas that pertain to the relationship between electromagnetic energy
(radiation) and temperature require the use of °R or K.
1.8.2 Symbols
Table 1.7 gives other conversion factors that may be useful. In addition,
typically used symbols are assigned to the quantities. These symbols are
used in this text and are fairly common in the fire literature, but they are not
universal. Note that the dot over a symbol implies “rate” or “per unit time,”
and two accents imply “per unit area.” The rate per unit area is commonly
called flux. Greek symbols are also common, for example, ρ (rho) for den-
sity and α (alpha) for thermal diffusively. We will explain the significance of
these qualities as we encounter them in our study of fire.
1.8.3 Scientific Notation

Finally, Table 1.8 lists terminology in scientific notation. This avoids many
zeros in expressing numbers. For example, kW (kilowatts) denotes 1000 W or
103 W. Similarly, 10−3 W is one thousandth of one watt or one mW (milliwatt).
This shorthand in terms of numbers is expressed as powers of 10, and prefix
notation simplifies units of measure.
TABLE 1.7
Common Quantities, Symbols, and Conversion Factors
Length, l 1 m = 3.2808 ft
Area, A 1 m2 = 10.7639 ft2
Density, ρ 1 kg/m3 = 0.06243 lb/ft3
Energy, Q 1 kJ = 0.94783 BTU
Heat, q 1 kJ = 0.94783 BTU
Heat flow rate, q 1 W = 3.4121 BTU/h
Energy release rate, Q 1 W = 3.4121 BTU/h
Heat flux (heat flow rate 1 W/cm2 = 3170 BTU/h-ft2, (1 W/cm2 = 10 kW/m2)
per unit area), q ¢¢
Specific heat, c 1 kJ/kg-°C = 0.23884 BTU/lbm-°F
Thermal conductivity, k 1 W/m-°C = 0.5778 BTU/h-ft-°F
Thermal diffusivity, α 1 m2/s = 10.7639 ft2/s
Pressure, p 1 Pascal (Pa) = 1 N/m2 (1 atm = 14.69595 lbf/in.2 =
1.01325 × 105 N/m2)
TABLE 1.8
Scientific Notation and Prefixes
Multiplier Prefix Abbreviation
1012 Tera T
109 Giga G
106 Mega M
103 Kilo k
102 Hecto h
10−2 Centi c
10−3 Milli m
10−6 Micro μ
10−9 Nano n
10−12 Pico p
10−18 Atto a
Notes: For example, 103 = 1000 and 10−3 = 0.001; 1000 m = 103 m = 1 km =
10−3 Mm = 106 mm.
1.9 Summary
Fire is a chemical reaction, usually involving oxygen from the air, that produces
enough energy to be humanly perceived. Typically, that energy release rate per
unit volume is sufficient to cause a skin burn in seconds. The destruction by fire
throughout history has been dramatic from the cause of the dinosaur extinction
to conflagrations in large cities into the twentieth century. Natural events,
accidents, and war have produced devastating fire consequences. Statistics

give us an appreciation for the loss and costs of fire around the world. The
United States has one of the highest per capita deaths due to fire, as well as in
expenditures for fire per GDP. Fire is rare in society, and consequently, there has
not been a high incentive to study uncontrolled fire. However, research in fire
has progressed throughout the world, and its results can be very useful to fire
safety design practices and to fire investigation determinations. Visualization
of fire phenomena is a first step in understanding and developing tools for fire
protection. To be able to use these tools, the student must understand algebra,
scientific units, and terminology. A goal to achieve by reading this book is to at
least come to think of fire in terms of kilowatts.
Review Questions
1. Explain the difference between combustion and fire.
2. Can volcanoes and meteors cause fires?
3. Which country has the highest death rate due to fire? Do we know why?
4. What is a firestorm?
5. What are NFPA, IAFSS, NIST, and SFPE? What do they do?
6. Convert the following:
85°F _____ °C
400°F _____ °R
695°F _____ K
30,000 BTU/h _____ kW
0.0015 W _____ mW
1385 W _____ kW
6.5 W/cm2 _____ kW/m2
Activities
1. Find examples of the significance of fire in history.
2. Examine the statistics of fire and the attitude of society on fire safety.
3. Test the validity of fire statistics. For example, does the prevalence of elec-
trical fire causes at night suggest something else? What percentage of
accidental fires is actually reported?
4. Examine the SFPE Handbook, the proceedings of the IAFSS (Fire Safety
Science series) on topics of interest to you. Discuss how you can benefit
from this science. Is it understandable?
5. Identify fire organizations and laboratories around the world.
References
1. G. Thomas and M. M. Witts, The San Francisco Earthquake (New York: Stein and
Day, 1971).
2. H. Rossotti, Fire (Oxford, U.K.: Oxford University Press, 1993), p. 22.
3. J. Man, The Terracotta Army (London, U.K.: Bantam Press, 2007).
4. M. Middlebrook, The Battle of Hamburg (London, U.K.: Cassell & Co., 2000).
5. P. Addison and J. A. Crang, eds., Firestorm—The Bombing of Dresden, 1945
(Chicago, IL: Ivan R. Dee, 2006).
6. J. Bradley, Flyboys (New York: Back Bay Books, Little, Brown & Co., 2003).
7. NFPA, Fires in the US during 2010 (Quincy, MA: Fire Analysis and Research
Division, NFPA, 2011).
8. Accident Facts (Itasca, IL: National Safety Council, 1995).
9. World Fire Statistics Centre Bulletin 12 & 27 (Geneva, Switzerland: The Geneva
Association, 1996/2011).
10. N. N. Brushlinsky, A. P. Naumenko, and S. V. Sokolov, Response to query about
fire deaths in Russia (Letters to the Editor), Fire Technology, 31, 3 (August 1995):
279–284.
11. T.-N. Guo and Z.-M. Fu, The fire situation and progress in fire safety science
and technology in China, Fire Safety Journal, 42 (2007): 171–182.
12. R. C. Cochrane, Measures for Progress—A History of the National Bureau of Standards
(Washington, DC: U.S. Department of Commerce, 1974), p. 131.
13. W. E. Becker, Fire research on cellular plastics: The final report of the Products
Research Committee, Library of Congress Cat. No. 80-83306 (1980).
14. H. W. Emmons, Fire research abroad, Fire Research Abstracts and Reviews, 10, 2
(1968): 133.
15. W. Strunk, Jr. and E. B. White, The Elements of Style (New York: MacMillan,
1979), p. 47.
16. W. P. Meade, A first pass at computing the costs of fire safety in a modern soci-
ety, NIST-GCR-91-592 (Gaithersburg, MD: National Institute of Standards and
Technology, March 1991).
17. W. G. Berl, ed., The Use of Models in Fire Research, Publ. 786 (Washington, DC:
National Academy of Sciences, National Research Council, 1961).
18. P. J. DiNenno, ed., The SFPE Handbook of Fire Protection Engineering, 4th ed.
(Quincy, MA: National Fire Protection Association, 2008).
19. J. G. Quintiere, B. J. McCaffrey, and W. Rinkinen, Visualization of room fire induced
smoke movement and flow in a Corridor, Fire and Materials, 2, 1 (1978): 18–24.
20. M. Kokkala, and W. J. Rinkinen, Some Observations on the Shape of Impinging
Diffusion Flames, Res. Rep. 461 (Espoo, Finland: VTT, Technical Research Centre
of Finland, 1987).
2
Combustion in Natural Fires
Learning Objectives
• Identify different forms of natural fire: diffusion flames, smoldering,

and premixed flames
• Explain how a candle flame, a basic diffusion flame, works
• Recite quantitative aspects of natural fire
• Describe the processes leading to ignition and spontaneous ignition
2.1 Introduction
Natural fire processes can take different forms: diffusion flames, smolder-
ing, and premixed flames. These are all chemical reactions that produce
relatively high temperatures. Flames occur as gases. Smoldering involves
the reaction of air with a solid. The smoldering chemical reaction occurs on
the surface of the solid. For a flame, the chemical reaction is solely in the
gas phase. As the reaction occurs at a high temperature, the reaction is fast
and the extent of the reaction is relatively thin. Diffusion, or motion due to
differences in concentration, is common to all flames. A pure diffusion flame
occurs when the fuel and oxygen diffuse from opposite sides of the reac-
tion zone. In the reaction zone, the fuel and oxygen are consumed, so their
concentrations are zero there. In a premixed flame, the fuel and oxygen
(or air) are already well mixed before combustion. The temperature of the
mixture is low so as to not allow the chemical reaction to occur. But with
a hot enough point in this mixture, a flame can emerge as a premixed flame.
Typically, premixed flames are the ingredients of controlled combustion
processes for heating and useful power. However, all diffusion flames are
born from a premixed flame.
33
The term spontaneous combustion might be classed as a distinct fire pro-

cess, but its occurrence results in either smoldering or a flame. This is simply
an ignition process that occurs naturally due to the energy generated by a
chemical reaction.
Let us examine these forms of combustion or fire in more detail.
2.2 Fire and Its Ingredients

A chemical reaction is a process that involves the change in a molecule by
rearranging its atoms into different molecules. Combustion or fire is a chemical
reaction involving the release of energy, some of which can be in the form of
light. Combustion and fire are synonymous. A combustion reaction commonly
involves the fuel molecule combining with oxygen to produce new molecules
as products. A flame is combustion in which a fuel gas reacts with an oxidizer,
commonly oxygen in the air. Smoldering is combustion in which a fuel reacts as
a solid with oxygen in the air. To define a chemical reaction as fire or combus-
tion, sufficient perceptible energy must be released. The rate of energy release
per unit volume of the chemical reaction determines whether that reaction is
fire. The size of the flame is not a factor. For a flame produced by a gaseous fuel
combining with oxygen in normal air, the energy release is strongly dependent
on the temperature of the molecules. Low temperatures produce an impercep-
tible amount of energy. Let us take a deeper look at this process.
2.2.1 Typical Temperatures and Energy Levels to Achieve Combustion

Roughly, the power production rate of a gaseous fuel reacting in air can be
described in terms of how many kilowatts (kW) are produced in a cubic
meter (m3) of the mixture of fuel and air. So, consider such a small volume
region composed of a mixture of fuel and air. The power production depends
on the temperature of the mixture. The speed of the reaction and the energy
released are the study of chemical kinetics, an entire branch of chemistry.
Combustion reactions have high energy, and their speed increases extremely
rapidly with temperature. Based on typical fuel–air mixtures, chemical kinet-
ics allows us to estimate typical production rates according to the mixture
temperature. These are ballpark values.
• At normal room temperature ~25°C, we have a power production

rate of about 10−14 kW/m3. The density of this mixture would be
about 1 kg/m3 and its heat capacity about 1 kJ/kg-°C. If all of this
energy were absorbed by the mixture with no heat loss, the tem-
10 -14 kW/m 3
perature rise can be estimated as = 10 -14 °C/s.
1 kg/m 3 ´ 1 kJ/kg-°C
Combustion in Natural Fires 35
This temperature rise is imperceptible, and any heat losses are likely
to not allow any increase in temperature. So, this mixture would
very slowly react but not at a level to be called combustion.
• Let us raise the temperature of the mixture to say ~300°C, here
chemical kinetics would indicate a faster reaction with an energy
production rate of about 10 kW/m3 (and a temperature rise of about
10°C/s). Now if heat losses are small, the temperature of the mix-
ture would rise, and its energy production rate would rise accord-
ingly. This feedback process between temperature and energy
would make the process go faster and faster. As long as fuel and
air are available, this acceleration process will reach a temperature
level in which the production of energy is balanced by the heat
loss. A higher temperature and energy production rate would be
achieved. This process of moving from about 300°C to the higher
reaction temperature is ignition.
• For combustion in air, a typical flame temperature is about
2000°C. As fuel and air are continuously supplied, we have about
1010 kW/m3 (1010°C/s). Here, the temperature would not continue
to rise because the heat losses would exactly take away all the
energy produced.
There are enormous differences in energy production over these

temperatures. The differences highlight what it takes to initiate ignition
and then to sustain a flame in air. It also shows that to have a “fire,” a
sufficient temperature is needed to set the process in motion. The process
is sometimes called a thermal runaway, as it causes acceleration in tem-
perature. This thermal runaway is the ignition process. It leads to a steady
combustion state (at temperatures approaching 2000°C for reactions in air)
where the flame will be sustained as long as fuel and air are continually
supplied.
The combustion process for smoldering can result in a similar fashion.
Smoldering on paper might been seen as a small (~1 mm) glowing region.
It took some energy to raise the temperature of the paper to sustain this
glowing region. However, with the paper at normal temperature in air, the
same oxidation reaction may take years to even yellow the paper. Minimum
temperatures to sustain smoldering can be as low as about 300°C, but glow-
ing states could be as high as 1000°C. Temperature and the rate of energy
production from the reaction define whether combustion will result for both
a gaseous reaction (flame) and a surface reaction (smoldering). Typically, this
temperature for a flame in air is about 2000°C, and for smoldering, it could
be as low as 300ºC. These temperatures are consistent with the notion that if
the chemical reaction temperature can cause a human skin burn, then it can
be considered as combustion.
2.2.2 Fuel Chemistry

Most fuels are composed of carbon and hydrogen. These are called
hydrocarbons and compose most of the fuels occurring in nature. Synthetic
fuels, particularly plastics, can contain other elements such as oxygen, nitro-
gen, chlorine, and fluorine. When a fuel chemically reacts in air, new mol-
ecules are formed. The resulting molecules are called products. Complete or
ideal reactions are those in which the products are molecules in their most
stable state. This state will not allow their further oxidation. Complete chem-
ical reactions for hydrocarbons would lead to the formation of water vapor
and carbon dioxide. But most flame and all smoldering processes are not
ideal in air. These reactions are interrupted and produce products that are
not fully oxidized. Such reactions are termed incomplete. Incomplete prod-
ucts of combustion for hydrocarbons consist of carbon monoxide, hydrogen,
and soot (mostly carbon). For the synthetic fuels, these products can also
contain hydrogen chloride or fluoride instead of their pure elemental gases,
and nitrogen-bearing fuels are likely to produce hydrogen cyanide instead of
pure nitrogen gas. Factors that promote incomplete reactions are a reduction
in the concentration of oxygen relative to normal air, turbulence in flames,
heat loss, and smoldering in general.
2.2.3 Fire Triangle and Tetrahedron

The fire triangle, as shown in Figure 2.1, is a concept used to describe the com-
ponents needed for fire. The elements of the fire triangle are essential to the
existence of a fire. The triangle consists of (1) fuel combining with (2) oxygen
in a chemical reaction to release (3) energy and other chemical products. The
energy released by the reaction causes heat to be transferred to the solid or
liquid fuel to maintain vaporization into gaseous fuel and to maintain the
temperature to ensure that the chemical reaction can persist. If we take away
sufficient fuel or oxygen, or reduce the energy by extinguishment or retardant
agents, the fire will not survive. This applies to both smoldering and flaming.
Fuel
Heat causes vaporization

Chemical reaction and maintains reaction
temperature
Oxygen Energy
(21% in air)
FIGURE 2.1
The fire triangle.
The fire triangle does not explicitly address the factors needed to make
the combustion reaction occur in the first place. We saw from the subject of
chemical kinetics that temperature is a big factor in initiating the combustion
reaction. But what underlies this dependence on temperature? If the
combustion reaction were put under a chemistry microscope, we would see
much more than a simple transition of fuel and oxygen molecules directly
changing to the resultant products. There would be a series of chemical
reactions with many intermediate species. A sufficient temperature to pro-
duce a chemically active species might initiate the first step in the reaction.
This active species would be unstable and prone to combine with another
species. These unstable species are molecules without their full electron
structure. They are called radicals and are highly reactive. The next step
would produce more active species. This process is called a chain branching
reaction. Specifically, chain branching is an intermediate chemical reaction
that always produces two unstable species. If each radical, in turn, reacts
with any intermediate molecule to form two more radicals (chain branch-
ing), multiple and increasing reactions can occur among the radicals and
intermediates. This is a chain reaction of one reaction always leading to
two. Eventually, a termination reaction is reached that ends the branching.
As long as more active species are continually produced, the process acceler-
ates. All of these chain reactions make the process fast. As energy is released,
overall it would appear that temperature is driving the process. The way
temperature affects the chemical energy production rates is a manifestation
of chain branching. The subject of chemical kinetics studies the full chemis-
try of these chain branching reactions in combustion.
In examining the fire triangle, the factor of chemical kinetics does not
show up. Yet, it is a vital factor in causing the combustion reaction to proceed.
We cannot simply have fuel, oxygen, heat, and energy as the components of
combustion or fire. The chemical kinetics must display sufficient energy and
speed to support combustion. As a consequence, the fire tetrahedron has been
introduced to students. This is a 3D object with four triangular sides. Each
side is now designated as (1) fuel, (2) oxygen, (3) heat and energy, and now
(4) chain branching reactions, or chemical kinetics.
The role of temperature is important to initiating chain branching. How-
ever chain branching could be initiated by an increase in pressure, as would
be the case for some explosives that are pressure sensitive. But most often,
it is the increase in temperature that sets the reaction in motion.
2.2.4 Combustion Time and Extent

Let us make some estimates of how fast the chemical reaction occurs in a
flame, and how big is that reaction region. A temperature above normal air
is needed to get to combustion. The minimum energy needed to achieve
combustion is called the activation energy. For most hydrocarbons, the typical
activation energy is about 160 kJ/mol. A representative molecular weight of a
typical fuel might be taken as 40 g/mol. So, this particular activation energy
is 4 kJ/g. In comparison, it takes about 1.1 kJ/g to vaporize methanol and
2.3 to vaporize water.
Let us use 4 kJ/g as the activation energy to cause combustion. A mixture
of fuel and air that would allow all of the fuel and oxygen to burn completely
is called a stoichiometric mixture. Natural flames operate at this ideal mixture.
A typical stoichiometric mixture for a hydrocarbon fuel is about 12 g air to
1 g fuel. So, we would have 13 g in this mixture. The activation energy for the
mixture is 4 kJ/g fuel/13 g mixture/g fuel or 0.31 kJ/g mixture.
When a flame occurs we have about 2000°C and an energy production
by the chemical reaction of about 1010 kW/m3. At this temperature, the den-
sity of the mixture is about 130 g/m3. The power produced per g mixture
is 1010 kW/m3/130 g/m3 or 7.7 × 107 kW/g. Dividing this into the activation
energy gives a time for a combustion of about 4 × 10−9 seconds. Of course, heat
would be lost and less than 1010 kW/m3 would be retained by the mixture.
But this simple estimate illustrates that the time for combustion to occur is
very, very small. Indeed, one might say that in a flame, when a fuel and oxygen
molecule come together there is almost instant chemistry. Yet underlying all
of this is complex chemical kinetics.
Let us stick with this estimate for the time for combustion as 4 × 10−9
seconds. In this time, oxygen and fuel are being consumed, and diffusion
is the primary manner in which these species are moving. They begin with
their initial concentration and move to a point of zero concentration. The laws
of diffusion govern the distance a species moves in a given time. An approxi-
mate estimate is distance equals the square root of the product diffusivity ×
time. The diffusivity of this mixture at 2000°C is about 8 × 10−4 m2/s. The
estimated distance over which the combustion occurs is about 1.8 × 10−6 m.
This is a very small region.
These estimations were very simplified. A complete calculation must
take into account the actual kinetics, the heat transfer, and the diffusion of
all species. But, in general, combustion occurs quickly (less than a second),
and the combustion region is very thin (less than a millimeter). Flow can
disturb the image of a flame, and in a turbulent flow the thin regions of
the flame become a tangled web. This projects a more global image of the
ensemble that we might call the flame. However, it is composed of small
flame pieces.
2.2.5 Types of Fire

We can categorize fire into three distinct phenomena:
1. Diffusion flames
2. Smoldering
3. Premixed flames
Diffusion flames represent the predominant category. It is the building

fire, the forest fire, or the lit match. Smoldering can initiate or can follow
the death of a diffusion flame. It is the glowing embers of the tragic house
fire, the result of a lightning strike to the forest bed, or the glow of a blown-
out match flame. In both diffusion flames and smoldering, the fuel and
oxygen are initially separated. Some transport process is needed to bring
them together. On the other hand, premixed flames have their fuel oxygen
mixed before the combustion occurs. Premixed flames represent controlled
combustion processes such as in a gasoline internal combustion engine with
spark ignition or in a diesel engine with no sparkplug. It also represents the
incipient flame that occurs in the ignition of solids or liquids. Then a diffu-
sion flame emerges. A fuel and oxidizer can also be premixed as solids. This
state represents propellants and some explosives. As these solid components
create a combustion reaction, gaseous products are likely to be released, and
they must be constructively transported away as in propellants. In explo-
sives, these gases are trapped in the solid explosive mixture and lead to
increases in pressure that accelerates the reaction even further. Let us exam-
ine these forms of combustion.
2.3 Diffusion Flames

A diffusion flame is a combustion process in which the fuel gas and oxygen
are transported into the reaction zone due to concentration differences. This
transport process is called diffusion and is governed by Fick’s Law, which
says that a given species will move from a high to low concentration in the
mixture. In connection with fire, species are the components of the reac-
tion, for example, oxygen, fuel, CO2, and other components comprising the
reactants and products. To understand the nature of diffusion, consider a
drop of blue ink in a glass of still water. Eventually, the dye droplet will
disperse into the water to give a uniform blue tinge. This motion of the dye
into the clear water is diffusion. Oxygen in air will move to the flame zone
where it has a concentration of zero since it is consumed in the reaction. Fuel
is transported into the opposite side of the flame by the same process. The
combustion products diffuse away from the flame in both directions. This
process is illustrated in Figure 2.2.
Most natural flaming fires are diffusion flames. A common example is the
flame of a match or a candle (Figure 2.3). In a candle, the flame melts the
wax, and the melt is transported up the wick by capillary action. The flame
then evaporates the wax, and the gaseous fuel diffuses into the flame where
it meets oxygen. For a wooden match flame, the wood is decomposed by
the heat of the flame into gaseous fuel and char. This thermal decomposi-
tion process of a solid is called pyrolysis. A candle flame is an example of a
Fuel (gas) Oxygen
Reaction
zone
FIGURE 2.2
Schematic of a diffusion flame.
FIGURE 2.3
A candle, an example of a diffusion flame.
laminar diffusion flame governed by pure molecular diffusion. Any flame

higher than approximately 1 ft will naturally possess random fluid mechani-
cal motion, illustrated by visible eddies in the smoke and flame. This is called
turbulence, in contrast to the orderly motion of a laminar flow. So, most fires
due to size are turbulent diffusion flames (Figure 2.4). Smoke from a large
chimney has this turbulent character, and a smoke filament from a cigarette
in a still room, which begins as a laminar flow, then clearly breaks up after
rising about 1 ft to become turbulent (Figure 2.5).
Gravity influences the shape of diffusion flames and profoundly affects
fire processes in general. Because fire creates high temperatures, the hotter
(lighter) gases rise as a result of buoyancy caused by gravity. The ensuing
flow distorts the flame, and eventually, flow instabilities cause turbulence.
Turbulence is due to disturbances, arising naturally, that excite the flow at
its natural frequency. This same phenomenon occurs when an unbalanced
FIGURE 2.4
A turbulent diffusion flame.
tire begins to vibrate erratically at a speed associated with the wheel’s

natural frequency. The mechanical constraints restrict the wheel from rip-
ping off the vehicle. Similarly, fluid friction keeps the erratic fluid’s turbu-
lent motion within limits. The cigarette smoke plume in Figure 2.5 clearly
displays the onset of turbulence at a point where erratic flow vibration
occurred. The size of the turbulent large swirls (eddies) is constrained by
the fluid’s viscosity. Accordingly, buoyancy and turbulence are two factors
that control fire and its associated flows. It is interesting to contemplate the
behavior of diffusion flames in a spaceship in which gravity and therefore
buoyancy are negligible. Only flow from a fan can disturb this flame and
augment the inherent diffusion process. Without the buoyancy of gravity,
the flame would not elongate. Figure 2.6 is a candle flame that was photo-
graphed aboard the International Space Station under nearly zero gravity.
As seen, the flame is nearly symmetrical about the wick, compared to the
flame in normal gravity. In microgravity, diffusion is independent of direc-
tion to give the spherical shape.
FIGURE 2.5
Cigarette smoke.
FIGURE 2.6
Candle flame in normal and microgravity. (From National Aeronautics and Space Administration
Lewis Research Center.)
(a) (b) (c)
FIGURE 2.7
Diffusion flame shapes: (a) jet flame, (b) liquid spill fire, and (c) forest fire.
The general shapes of diffusion flames are illustrated in Figure 2.7. Three
types of diffusion flames are shown: (1) a jet flame, (2) a moderate pool fire,
and (3) a very large fire. Natural fires involving liquid or solid fuels have
very low velocities at the fuel base (~1 cm/s), but a high-pressure gaseous
fuel source can initiate fuel at much higher velocities. For such high-velocity
jet flames, buoyancy effects can be negligible. Jet flames have the charac-
teristic that as the velocity is increased, for a given port diameter, the jet
flame will eventually reach a fixed height. Pool fires with characteristics as
shown in Figure 2.7b are more representative of normal burning commodi-
ties. They achieve heights associated with their fuel supply and continue to
get taller as long as buoyancy controls their air supply. Fires over large areas,
such as forest fires or city conflagrations, have short flames relative to their
base. Their flame height is more controlled by the air flow that is drawn
around its perimeter. As air supply can be limited around the perimeter,
these flames can be drawn down from above as well as drawn in from the
sides. Figure 2.7b will be the principal focus of our attention and represents
diffusion flames associated with commodity and furnishing fires.
2.3.1 Candle Flame

The candle flame provides most of the essential features of natural fires and
a diffusion flame. It will be our learning tool. We are not original in this
approach because we will repeat some of the experiments created by the
nineteenth-century scientist, Michael Faraday (Figure 2.8). These experi-
ments were presented at the Royal Institution in London as a science show
for the public. Known as the Christmas Lectures, they were very popular
The Scientific Discoveries of Michael Faraday

1820 discovered two unknown chlorides of carbon
1821 discovered electromagnetic rotation—set up an
experiment in which a wire carrying an electric
current rotated in the field of a horseshoe magnet.
1823 liquefied chlorine
1825 isolated benzene
1831 demonstrated the principle of electromagnetic
induction, the production of electric current by a
change in magnetic intensity
1844 discovered the rotation of the plane of polarization
of light in a magnetic field
FIGURE 2.8
Michael Faraday and his discoveries. (From Faraday, M., Faraday’s Chemical History of the Candle,
Chicago Review Press, Chicago, IL, 1988 [orig. 1861].)
with children and have been archived in a book that is still available today.1
Some of these lectures are remarkable in their simplicity, yet powerful in
their illustration of the basic principles of flames.
It is astonishing to those who study fire to see how Faraday appreciated the
workings of a candle and its implication to the science of fire. These thoughts are
expressed in his opening remarks at the convening of the Christmas Lectures:
I propose to bring before you, in the course of these lectures, the

Chemical History of a Candle. There is not a law under which any part of
this universe is governed which does not come into play and is touched
upon in these phenomena. There is no better, there is no more open door
by which you can enter into the study of natural philosophy than by
considering the physical phenomena of a candle.1
Light a candle and observe the processes that create the sustained flame.
As illustrated in the sketch in Figure 2.9, you can see several things. The
yellow and blue zones constitute the flame (the former, diffusion; the latter,
premixed). The wick is actually designed to curve so that the flame “clips”
off the wick and limits its height. Why is not the wick destroyed throughout
Luminous yellow zone
Dark zone
“Standoff ” space
Glowing, clipped wick
Melted wax Blue tip
FIGURE 2.9
Characteristics of a candle flame.
the flame? What is the purpose of the wick? With some thought, it can be
deduced that the heat of the flame melts the wax, and the melt soaks the
wick by capillary action, like water feeding tree leaves. The melt keeps the
cotton relatively cool and intact. The heat from the flame evaporates the melt
and diffuses the fuel gas into the luminous flame zone. There it finds oxy-
gen having diffused from the other side. Buoyant flow elongates the candle
flame, but the flow path is small enough so that laminar flow is maintained
within the flame. However, as the hot combustion gases rise from the flame,
they become turbulent. This turbulent phenomenon can be seen by project-
ing collimated light (from a slide projector or strong flashlight) at the flame
and onto a white screen. This is shown in Figure 2.10.
FIGURE 2.10
Image of flame (soot) shadow and shadowgraph white-dark projection of turbulent eddies.
There are several shadow images on the screen. Distinctly, the solid
candle and the wick block the light and form a clear black shadow image.
Looking closely, a shadow-like image of the luminous portion of the flame
also appears on the screen. What might cause this flame shadow? Some
solid ingredients must be blocking the light through the flame. These par-
ticles must form within the flame. Looking more closely about the shadow
of the luminous image on the screen, white curves are seen surrounding
this image. These white lines result from the refraction of light rays as
they bend through the flame. Where the temperature changes the most
quickly (the edge of the hot region), the light is reinforced to give a white
image. This effect of producing the white images is a technique known as
a shadowgraph and marks the edge of the plume around the flame. The
turbulent eddies of the plume region above the flame can also be seen
in the shadowgraph above the flame. These images are very revealing
for the flame. It shows the luminous region blocks light, some distance
lateral from the luminous region it is still hot (due to conduction from the
reaction), and turbulent regions of hot and cold form in the plume above
the flame.
Now place a metal screen (typical window screen mesh, but not alu-
minum) into the flame as shown in Figure 2.11. The metal screen cuts the
flame, extinguishing further combustion. Why? Describe the luminous
flame cut.
Combustion reactions can only be sustained if the flame temperature
is high enough, typically greater than 1300°C. As the height of the screen
is varied in the flame, black smoke is released with the screen high in
the flame and white or gray “smoke” with the screen near the top of the
wick. What do these colored smokes imply? Collect the black smoke on
a piece of white paper, being careful not to insert the paper in the flame.
What is collected on the paper? Where do these black particles originate?
Find that region in Figure 2.9. Blow out the flame and observe the color
and character of the smoke that momentarily emanates from the wick.
How does this compare to the white smoke resulting from the low screen
position?
Relight the candle. After the flame steadies, insert a glass dropper into
the region where you think the white smoke originated. Extract the white
smoke into the dropper tube. Squirt this white gas across the top of the
flame (Figure 2.12). It should ignite and resemble a small flame thrower.
What were these white vapors? It is fuel vapor that has condensed into
small droplets, appearing as white. Clouds in the sky are white for their
condensed water droplets. A cloud of droplets will scatter light, and it will
appear as “white.”
To confirm your conclusion that the white vapor extracted is fuel, blow
out the candle flame again, and quickly try to ignite the white vapors
coming from the wick. You will see a flame propagate down this vapor
Black or white “smoke”
FIGURE 2.11
Sketch and photograph of metal screen cutting the flame.
trail and anchor on the wick. This propagating flame is a premixed flame
(we will explain this further in a bit); the anchored flame is, of course, our
diffusion flame.
Review all of the information deduced about the candle flame and you
should have a good understanding of the diffusion flame and fire. You may
wish to consult Faraday’s text for more informative experiments with the
candle.1 These experiments are simple, and I encourage you to try them for
yourself. You may even invent new procedures to learn more about fire from
the candle flame.
FIGURE 2.12
Extracting interior flame gases.
2.3.2 Anatomy of a Diffusion Flame

More sophisticated experiments with a laminar diffusion flame using meth-
ane (CH4) as the fuel have been conducted by Kermit Smyth and coworkers.2,3
Their flame is formed from controlled flows of CH4 and air, as shown in
Figure 2.13. The burner is a slot burner, so it is planar in character compared
to the cylindrical character of the candle, but the flame dynamics are simi-
lar to the candle. Various point measurement techniques were used in the
flame to quantitatively reveal its anatomy. We can relate these measurements
to what we have deduced from the candle experiments. Now our qualitative
observations can be quantitatively presented. These quantitative results will
be displayed for a height 9 mm above the burner. This position is arbitrary,
representative of examining a “horizontal cut” through the region where a
diffusion flame exists.
Thermocouple measurements of temperature across the flame region are
shown in Figure 2.14 at the 9 mm height. Other heights within the luminous
flame region will show a similar bimodal, two-hump pattern. These 9 mm
measurements are indicative of the two sides of the diffusion flame. The
maximum temperature indicates the location of the chemical reaction. The
combustion needs fuel and oxygen as well as a sufficient temperature to
initiate and to persist. The drop-off in temperature on both sides of the peak
indicates heat loss; however, a high temperature is still achieved to preserve
the flame. Earlier, it was indicated that this required temperature is about
2000°C. The thermocouple probe used here will lose some heat and slightly
cool the flame; therefore, it is difficult to accurately measure the highest
temperatures indicative of the reaction zone. The indicated measured peak
temperature of approximately 1700°C is likely to be more like 1900°C–2000°C.
Nevertheless, this very-high-temperature region measured is indicative of
Stabilizing
screens
Monochromator
Laser
L1 Z
P
L2
Y
Flame zones X
Air
CH4
Air
FIGURE 2.13
Arrangement of a slot burner and point measurement techniques. (From Smyth, K.C. et al.,
Combust. Flame, 62, 157, 1985.)
2000
3 mm
9 mm
15 mm
1600
Temperature, K
1200
800
400
–10 –5 0 5 10
Lateral position, mm
FIGURE 2.14
Temperatures across a laminar diffusion flame. (From Smyth, K.C. et al., Combust. Flame,
62, 157, 1985.)
the chemical reaction zone of the flame. This was the luminous yellow shell
of the candle flame. In Figure 2.14, the width of this peak region is about
1 mm. Recall observing the candle flame through the top of the metal screen
that cuts it horizontally. The luminous region was also about 1 mm.
The rest of the temperature graph shows the regions away from the reaction
or flame zone. The interior temperatures are due to heat conducted inward.
This heat would be what the candlewick would need to sustain the evapora-
tion of the fuel. The outer-wing temperature should eventually become the
temperature of the pure air of about 25°C (298 K). While this is not shown
directly in the graph, it would occur about 5 mm from the flame. The shad-
owgraph image for the candle flame would mark the onset of pure air.
The corresponding vertical flow speed is shown in Figure 2.15. These flow
speeds are principally due to buoyancy. Indeed, the temperature is directly
responsible for these vertical speeds. The peak of 80 cm/s is about 1.8 mph;
a light breeze.
Using special laser-optical techniques, Smyth and coworkers were able to
image the visible flame, OH radicals, and soot. The OH radicals are indicative
of short-lived chemical species in the heart of the combustion reaction. Once
they die out, the flame chain branching reaction dies. They extend to the air
side of the flame. The soot, or black smoke of the candle experiments, comes
from within the (luminous) flame zone. This is clearly seen in their images
80
Vertical velocity, cm/s
40
9 mm
0
–10 –5 0 5 10
FIGURE 2.15
Vertical velocity in a laminar diffusion flame. (From Smyth, K.C. et al., Combust. Flame, 62,
157, 1985.)
128 mm
96 mm
Visible flame
height
(79 mm)
64 mm
Soot
32 mm
OH
0 mm
–10 0 10 mm –10 0 10 mm
Steady-state laminar flame
FIGURE 2.16
Soot and flame reaction zone locations in a laminar diffusion flame. (From Smyth, K.C. et al.,
Combust. Flame, 62, 157, 1985.)
illustrated in Figure 2.16. While that figure is an artistic rendition of their

results, their measurement techniques were quite sophisticated. The shadow
of the soot particles in the candle flame (Figure 2.10) is indicative of their soot
region shown here.
The soot is formed on the fuel side of the diffusion flame. It results from
a complex process as the original fuel (CH4) is heated on diffusing toward
the flame. This intense heating results in cracking the CH4 molecules into
many other hydrocarbon molecules. As acetylene (C2H2) and other precur-
sors form, soot (mainly C atoms) is produced. The soot (a solid formed from
gases) migrates through the reaction (flame) zone, is oxidized (mainly by
OH), and is consumed. We see this oxidation by the yellow (incandescent)
glow of the visible flame. As the soot is being oxidized, it is exposed to the
very high temperatures of the reaction zone. The color of orange-yellow is
indicative of a surface at about 2000°C. This is why flame has that color.
In this flame, as well as the candle flame, nearly all the soot is consumed
before the reaction zone ends. The visible flame height nearly coincides with
the end of the soot height (Figure 2.16), and the OH zone is slightly above
this height.
If we “tear” the flame by “stretching” it (e.g., abruptly pull the candle
flame or disturb it by pulling a small probe through it), soot can escape
through the tear. Again, this illustrates that the soot formation is on the
fuel side of the flame. For a heavily sooting fuel, all the soot may not be
oxidized before the end of the reaction zone; then soot escapes as black
smoke. However, even for fuels with low soot-forming tendencies, as long
as they contain carbon, insufficient air supply to the flame will lead to
escaping soot.
Let us examine the other species that occur in this flame process. Some of
these are shown in Figure 2.17. Smyth et al. took the measurements at the
same 9 mm above the fuel port.2 The concentrations are given in mole frac-
tion, which is equivalent to volume fraction. Volume fraction is the volume of
the species gas to volume of the mixture at the same temperature and pres-
sure. For example, if one extracts the oxygen from air and let it come into bal-
ance with the air pressure, the ratio of its equilibrated oxygen volume to
the volume of the mixture (air) is the volume fraction. Note the oxygen (O2)
1.0
Σx1
N2 CH4
Mole fraction
0.5
O2 H2O
0.0
–1.0 –5 0 5 10
FIGURE 2.17
Major species formed in a laminar diffusion flame (CH4) at 9 mm. (From Smyth, K.C. et al.,
Combust. Flame, 62, 157, 1985.)
in pure air is at 0.21 volume fraction with the nitrogen (N2) making up the
remainder 0.79. These values are measured at the extreme left and right
wings in Figure 2.17. The fuel (methane, CH4) has a volume fraction of nearly
1 (all fuel) at the center. As 100% methane flows from the slot, at 9 mm above
the slot it is still nearly 100%. We see that the fuel and oxygen reach zero at
approximately 6 mm along the lateral position, the peak of the combustion
zone. This zone is about 1 mm in width at a height of 9 mm above the fuel
source. The water vapor (H2O) is also at maximum at 6 mm. Along with
carbon dioxide (CO2), water vapor is a principal compound formed and
released into the atmosphere as a product of combustion. Figure 2.18 shows
the results for CO2 along with other compounds, namely, carbon monox-
ide (CO) and hydrogen (H2), which do not survive to be released into the
atmosphere. As the soot and OH would fall to zero, as the end of the reaction
zone is reached at about the 6 mm lateral position, the other under-oxidized
species, such as CO and H2, form CO2 and H2O.
The concentration of species varies laterally because their relative pro-
portions change due to the production and destruction processes of the
chemical reaction. As their concentrations change, diffusion also occurs
among them. Figure 2.19 qualitatively illustrates the migration of species
taking place in the complex chemical processes of the flame. Superimposed
1.0 2
Φ CO2
1
log (local equivalence ratio)
Mole fraction × 10
0.5 0
H2
CO
–1
0.0 –2
–10 –5 0 5 10
FIGURE 2.18
Additional major species formed in a laminar diffusion flame (CH4) at 9 mm. (From Smyth,
K.C. et al., Combust. Flame, 62, 157, 1985.)
Soot oxidized
CO, H2 oxidized
Maximum temperature
Soot produced
CO produced
H2 produced
(Fuel side)
(Air side)
H2O, CO2 produced
O
OH produced
O2, CH4
5 6 7
Luminous zone
lateral position, mm
FIGURE 2.19
Qualitative chemical processes in a CH4 laminar diffusion flame.
on these chemical processes are the physical processes involving (1) dif-
fusion of species, each migrating toward its lowest concentration; and
(2) motion due to buoyant flow caused by temperature. Due to cracking and
synthesis of the CH4, as the heat from the combustion process is conducted
to the diffusing CH4 molecules, other hydrocarbons, some of which are
shown in Figure 2.20, are formed in reactions preliminary to combustion.
These compounds are “minor species,” achieving short-lived concentrations
of 0.005 mole fraction at most before being completely oxidized. Returning
to Figure 2.19, between about 5 and 7 mm, or over 2 mm, the fuel “cracks,”
soot forms, species come and go, and eventually OH dies out. The entire
process that spans at about 2 mm with the more intense reaction (flame) is
dispersed over about 1 mm.
2.3.3 Turbulent Diffusion Flames

We have seen that in the plume above candle flame becomes turbulent in
less than 1 ft of height. This is typical of fire under normal buoyancy condi-
tions. Therefore, nearly all natural diffusion flames arising due to accidental
fires are mostly likely to become turbulent. The laminar flame processes we
have just reviewed must be viewed as fluctuating erratic processes within
a turbulent flame. The processes depicted in Figures 2.14 through 2.20
1.0
[C2H2]
[C6H6] × 5
[C4H2] × 3
[C4H6] × 10
Mole fraction × 100
0.5
0.0
–10 –5 0 5 10
FIGURE 2.20
Minor chemical species in a CH4 laminar diffusion flame. (From Smyth, K.C. et al., Combust.
Flame, 62, 157, 1985.)
are pushed and pulled in space, blurring the values of the concentrations.
Figure 2.21 is a rendition of a temperature measurement of a turbulent
flame. As the flame moves about the probe, the temperature fluctuates over
the extremes measured for the still laminar flame of Figure 2.14. The tur-
bulent flame can be conceived as an ensemble of laminar flames moving
chaotically in space. An ordinary probe is not likely to keep up with these
extreme and rapid changes. It will tend to experience an average value.
Instead of discerning a clear temperature in the 1 mm reaction zone, the
measurement probe will experience high and low temperatures, with aver-
ages approximately 800°C–1200°C. Thus, peak temperature conditions of a
laminar flame will not be discerned in a turbulent flame. This fluctuating
effect will apply to all of the species as well as temperature. Although the
turbulent flame will not give on average the temperatures and concentra-
tions of its laminar counterpart, the ingredients of the laminar flame can be
used to understand the more complex turbulent flame. A complication is
that due to all of the fluctuating effects some fuel and oxygen may actually
mix before they burn. This is now a premixed flame, not a diffusion flame.
Thus, a turbulent flame is quite complex, and its prediction is still a source
of active research.
Average
Temperature (°C)
0.1 0.2 0.3 Time (s)
FIGURE 2.21
Turbulent temperature fluctuations in a flame.
2.4 Premixed Flames

A premixed flame is the second type of combustion we can call fire. A pre-
mixed flame is a combustion process in which the fuel gas and air
(or oxygen) are first mixed before ignition and propagation occur. In a
diffusion flame, both laminar and turbulent, the fuel is supplied in a
distinct stream, and air is then mixed into it as it burns. In a premixed
flame, the fuel and oxygen, or air, are mixed, then burning occurs within
that mixture. In a confined space, such a combustion process can cause a
rapid pressure increase that manifests itself as an explosion. If sufficient
pressure builds up behind the propagating flame, it will accelerate the
flame. At a higher speed, turbulence will ensue and increase the reaction
rate. Eventually, the flame speed can reach the speed of sound, and with
further increases, a shock wave will form ahead of the flame. That shock
wave introduces now higher temperature and pressure for the fuel and
oxygen supplied into the flame. We know that the chemical reaction rate is
increased as temperature and pressure rise. This leads to a feedback effect.
The speed of the flame increases, and a stronger shock is formed with yet
higher temperature and pressure across it to feed the flame. The end result
will produce an extremely high flame speed with a shock wave preceding it.
This event is then called a detonation. Such an event has great damage
potential; however, even the nondetonating (deflagration) premixed flame
in a confined space can cause significant damage due to pressure rise. Both
events will be perceived as an explosion. While we are considering flames
here, premixed combustion systems can exist as solids (e.g., dynamite, rocket
propellants, etc.). The dynamics of premixed flames can be used to explain
these systems as well.
Fuel
Flame
Air
C2H2
O2
Oxyacetylene torch Methane leak Gasoline engine
FIGURE 2.22
Examples of premixed flames.
As with the diffusion flame, it is sufficient to understand premixed flame

systems in general by studying the simplest system. So, the onset of pre-
mixed flames will be considered. These are laminar flames with relatively
slow speeds. Moreover, premixed flames are common to most useful con-
trolled combustion processes. Figure 2.22 shows common examples of
where premixed flames occur. Acetylene mixed with oxygen forms a very
high-temperature flame for welding and cutting metals. (By the way metals
can burn, too, especially under pure oxygen.) A gasoline internal combus-
tion engine vaporizes the fuel in a chamber with air and an electric arc
ignites the mixture. An accidental scenario is depicted by the methane leak.
The methane mixes with air, and that mixture could find a source of ignition.
The premixed flame will propagate from the ignition source back to the leak.
A diffusion flame can then result at the leak.
2.4.1 Laminar Flame Propagation

Returning to the candle experiments: When the flame is blown out, the white
wax–fuel vapor mixes with air as it rises in its distinct plume. Attempts to
reignite the plume will succeed provided the ignition flame is placed in the
plume stream at the right concentration. Gaseous fuels will ignite within
distinct fuel concentration limits. For every fuel gas in air, there is an upper
(UFL) and lower (LFL) flammable limit concentration in which flame prop-
agation can occur. Flammable here means that the mixture will ignite and
propagate at these concentrations. The speed of propagation for a laminar
flame is relatively low, less than 1 m/s. It is the speed that the flame will
travel along the white vapor of the blown-out candle flame. Another illus-
tration of this speed can be perceived from the candle experiment with the
metal screen cutting the flame. This is shown in Figure 2.23. In this figure, the
flame suspended above the screen results from a plume streak of white fuel
vapor rising up from the center of the diffusion flame around the candlewick
below the screen. When that suspended flame was ignited, white fuel and air
were moving vertically upward toward it. Once ignited, the flame became
suspended because its propagation speed down into the white fuel–air streak
was equal to the upward speed of the plume streak. This is a premixed flame
in the same suspension that would occur on a stovetop premixed burner.
(a) (b)
FIGURE 2.23
Floating premixed flame (a) above screen that clipped the candle flame and (b) premixed flame
above screen and diffusion flame below screen.
A mixture with fuel within the flammability limits will allow flame
propagation once ignited. These values (usually calculated at 25°C in air at
normal pressure) vary slightly with the temperature as shown in Figure 2.24.
The concentration over which flammability is possible will increase as the
mixture temperature increases. At a given mixture temperature, we see that
there is an upper and lower flammability limit. In this range, this means a
small energy source will initiate sustained propagation at a speed that would
be in the range of about 25–100 cm/s. If the temperature of the fuel mixture
Droplet-gas
boundary UFL
Fuel
Fuel concentration
droplets Gaseous flammable Autoignition

mixtures
LFL
TL AIT
Temperature
FIGURE 2.24
Limits of flammability in air showing the effect of temperature from Zabetakis. (From
Zabetakis, M.G., Flammability characteristics of combustible gases and vapors, Bulletin 627,
Bureau of Mines, U.S. Department of the Interior, Washington, DC, 1965.)
is varied, the LFL and UFL change slightly as seen in Figure 2.25. But at very
low temperatures, the fuel may be condensed in the form of droplets. This
droplet mixture also has flammability limits. The temperature (TL) just when
the fuel condenses to a liquid at the LFL is called the flashpoint. We discuss this
phenomenon further when we examine the ignition of liquids in Chapter 4.
Although this ignition process requires an energy source such as a flame or
electric discharge, it is quite small—less than 1 mJ (which is that amount of
energy to raise 1 g of water less than 0.001°C). In contrast, at a high enough
temperature, the fuel–air mixture can ignite without any energy source. The
lowest temperature to cause this spontaneous ignition phenomenon is called
the auto-ignition temperature (AIT), also illustrated in Figure 2.24.
The AIT reported for a fuel is representative of the fuel in the gaseous state
in the presence of air within its flammable limits. It is usually measured in a
specific size vessel and in general depends on size. So, there is not a distinct
AIT for a given fuel. It can vary depending on the circumstances of heat-
ing. But as temperature is critical to initiating the process of combustion,
this initiation temperature is roughly in the range of 300°C–600°C for typical
hydrocarbon fuels mixed with air. At the specific AIT, the mixture will reach
combustion. Table 2.1 gives some typical values for the AIT, along with the
flammability limits needed for propagation.
2.4.2 Flame Temperatures

Once ignited, it is interesting to consider the temperature that the flame will
reach. A fuel gas burning in air must heat up the nitrogen in the air, as it does
not take part in the chemical reaction. If the nitrogen is not present and the fuel
TABLE 2.1
Flammability Limits of Gaseous Fuels in Air at Normal Atmospheric
Temperature and Pressure
Fuel LFL (%) UFL (%) AIT (°C)
Acetylene 2.5 100 305
Benzene 1.3 7.9 560
n-Butane 1.8 8.4 405
Carbon monoxide 12.5 74 609
Ethylene 2.7 36 490
n-Heptane 1.05 6.7 215
Hydrogen 4.0 75 400
Methane 5.0 15.0 540
Propane 2.1 9.5 450
Trichlorethylene 12 40 420
Source: Beyler, C.L., Flammability limits of premixed and diffusion flames, Chap. 2–9, in:
DiNenno, P.J., SFPE Handbook of Fire Protection Engineering, 2nd edn., National Fire
Protection Association, Quincy, MA, June 1995.
burns in pure oxygen, the flame would reach a much higher temperature. The
flame temperature ultimately depends on the energy released in the combustion
reaction, how much gas must be heated, the actual reaction progress, and heat loss.
As a benchmark, consider a flame without heat loss, one that reaches ordinary
atmospheric pressure and achieves an ideal state of chemical balance (equilib-
rium). This is called the adiabatic flame temperature (AFT) and can be regarded as
the maximum possible temperature that can be attained for a given fuel. Now
this AFT depends on the amount of fuel and oxygen present. A mixture of the
two is said to be stoichiometric if none of both the fuel and oxygen remains after
the reaction. As no extra fuel or oxygen is present for a stoichiometric reaction,
this AFT will truly be the ideal maximum flame temperature. It was stated earlier
that this is about 2000°C, without any substantial justification. Table 2.2 shows
some computed AFT values for typical fuels at stoichiometric conditions burn-
ing in air. The prevalent values range from about 1950°C to 1980°C, and these
are representative of the majority of hydrocarbon-based fuels including wood
and plastics. Notable exceptions occur for acetylene and hydrogen. The results
can be explained from the fact that most materials release the same amount of
energy in burning 1 g of oxygen. This observation was brought forth by Huggett6
in recent times, who concluded that 13 kJ/g was a good average value for the
energy released per gram of oxygen burned. A stoichiometric concentration is
about midway between the LFL and the UFL for most fuels, so since this flam-
mability range is typically 1%–10%, a relatively small amount of fuel burns in
proportion to the oxygen in the air. The amount of energy stored in the mixture
of air and fuel is based on how much oxygen burns. This portion of oxygen to the
mass of the original stoichiometric mixture is approximately a fixed number for
all hydrocarbon fuels. Hence, if the energy released per unit mass of oxygen is an
inherent constant, the resulting temperature of the mixture will also be constant.
This explains the relationship between the AFT for stoichiometric burning in air
and the energy release by oxygen burned. Table 2.2 indicates this relationship.
While the AFT is an ideal value, small laminar flames have relatively
little heat loss, so they are a good representation of such real temperatures.
TABLE 2.2
Maximum Flame Temperatures (AFT) in Air and Energy Released
per Oxygen
AFT in Air Energy Release in Burning
Fuel (Stoichiometric) (°C) 1 g of Oxygen (kJ)
Acetylene 2500 15.6
Butane 1970 12.7
Ethane 1955 12.7
Hydrogen 2210 15.1
Methane 1950 12.5
Propane 1980 12.9
Source: From various sources.
The propagation of real flames must be near stoichiometric conditions for

most fuels (with exceptions of acetylene, carbon monoxide, and hydrogen).
Indeed, diffusion flames naturally occur for stoichiometric conditions. So,
these small, candle-like flames most often have temperatures in the range of
about 2000°C burning in air. As a flame becomes turbulent, its local laminar-
like flame will still possess this type of value, but now an average tempera-
ture is more relevant, and it will be substantially lower. But chaotic motion
of turbulent flow has nothing to do with the fuel, so the turbulent flame
temperatures of most hydrocarbon fuels burning in air will be similar.
2.4.3 Turbulent Propagation to Detonation

Once propagation commences in the fuel–air mixtures from the ignition
point, flame propagation occurs under laminar conditions at speeds of
roughly 0.1–1 m/s. As turbulence occurs, the laminar flame will distort. This
“wrinkled” flame will have more surface area and hence more energy, so
its speed will increase. The turbulence can occur naturally or as the flame
passes over obstacles. These effects cause the flame to accelerate. Exceeding
the speed of sound (343 m/s) will cause a shock wave to form ahead of the
flame. The shock will feed higher temperature fuel to the flame and acceler-
ate the process to a higher state. A strong shock will result with extremely
high velocities of over 2000 m/s. If the speed of the flame is below the speed
of sound and no shock wave forms, the process is called a deflagration. Above
the speed of sound, it is called a detonation. Fuel concentration limits needed
for detonations have been determined, and these are listed in Table 2.3 with
the detonation flame velocities. This detonation process is distinctly dif-
ferent from its subsonic counterpart. However, most explosions attributed
to flame propagation in a premixed fuel system are not detonations. They
involve burning in a confined space in which the flames begin as laminar
and progress to turbulent. The propagations at still fast and will cause pres-
sure rise due to a sudden increase in temperature in the confined space. This
can even occur in partially ventilated spaces.
TABLE 2.3
Detonation Limits at Normal Atmospheric Temperature and Pressure
Fuel Lower Limit (%) Upper Limit (%) Detonation Velocity (m/s)
Hydrogen in pure O2 15 90 2821
Hydrogen in air 18.3 59 —
CO in pure moist O2 38 90 1264
Propane in pure O2 3.2 37 2280, 2600
Acetylene in air 4.2 50 —
Acetylene in pure O2 3.5 92 2716
Source: Lewis, B.J. and von Elbe, G., Combustion, in: Flames and Explosions of Gases, 3rd edn.,
Academic Press, Orlando, FL, 1987, pp. 551–555.
2.5 Smoldering
A flame is combustion in a gas, while smoldering is combustion on a solid
surface. Smoldering is a relatively slow combustion process that occurs
between oxygen in the air and the solid fuel. Common examples of smolder-
ing are shown in Figure 2.25. The reaction occurs on the solid surface, and
oxygen diffuses to the surface. The surface can undergo glowing and char-
ring. In fact, when typical polyurethane foam upholstery material smolders,
it will leave a char. But if it burns to form a flame, it will melt. Smoldering
is oxidation that leads to a sufficiently high enough temperature. Rusting
of iron and the yellowing of paper with age are surface oxidation but are
not smoldering or combustion. To class it as combustion, there must be an
acceleration of the initial oxidation to a higher-temperature state. While this
higher temperature can vary depending on conditions, it could be as low as
about 300°C. Again if it can burn your finger, it could be classed as smolder-
ing. Clearly, a glowing oxidizing surface is smoldering, as that is indicative
of a temperature in excess of 1000°C.
Upholstered furniture and mattresses initiated by a discarded cigarette
are prevalent sources of smoldering fires in homes. Coincidently, the ciga-
rette is classic example of smoldering. As air is drawn through the cigarette,
the porous tobacco receives more oxygen and more burns. Smoldering is
controlled by the supply of oxygen. However, as it can occur at relatively
lower rates of combustion than a flame, it can occur at very low levels
of oxygen. Underground fires in abandoned coal mines can smolder for
years. On the other hand, every camper knows that blowing on a glowing
stick of wood will not only increase its degree of glowing but can spring
it into a flame. Airflow is significant to smoldering. Bringing more air to
the oxidizing surface causes a higher reaction rate and higher reaction
temperature.
(a) (b) (c)
FIGURE 2.25
Common examples of smoldering: (a) cigarettes, (b) upholstered chair foam plastic/cotton fabric
system, and (c) charcoal grill.
The fact that smoldering takes place under lower-temperature conditions

than a flame does not allow its intermediate species to burn. The incom-
pleteness of the combustion process leads to high levels of carbon monoxide,
CO, instead of CO2. More than 10% of the fuel mass is converted to CO in
smoldering carbon-based solids. Smoldering fires that persist for a long time
in a confined space as a house can lead to fatal conditions due to CO. Also,
many unburned products can build up, creating a potentially flammable
atmosphere.
Surprisingly, smoldering has received little study compared to flaming
fires. Data can help to illustrate the characteristics of the process and give
one an appreciation of its speed. In general, smoldering can occur in porous
solid fuels, in combinations of fuels, in impervious solids, and in buried
solid fuel waste sites. It requires air, but not much, because the process is so
slow. Air can blow in either direction to the smolder front or can be naturally
induced to flow due to buoyancy. Smoldering is schematically illustrated in
Figure 2.26. The thickness of the smolder region, like a flame reaction zone,
increases as the reaction rate decreases. Its thickness and speed depend on
the reaction rate. This reaction rate is controlled by the supply of oxygen.
In Figure 2.26, air must diffuse through the char from above. If the virgin
material is porous, it might induce air to flow upward and enhance this
process. The process is complex, but relatively slow, generally of the order of
10−3 to 10−2 cm/s or 0.6–6 mm/min. This is very slow but potentially deadly
because it produces carbon monoxide.
Typical smolder velocities are shown in Figure 2.27 and Table 2.4 for common
materials. In Figure 2.27, air was blown through the porous materials, and the
smoldering propagation was opposite to the direction of airflow (reverse smol-
der). It is clearly seen that increasing the air speed increases the smoldering
speed. In Table 2.4, airflow was generally parallel to the solid fuel surface, with
forward smoldering in the same direction as the flow. For wood-fiber board,
the smoldering rates ranged from about 0.8 to 8 mm/min, with the higher
levels at high air flow rates of up to 10 m/s (22 mph). Cardboard and cellulosic
fabric could smolder at about 5 mm/min in relatively still air.
Air Air
Char matrix of spent fuel
Smolder reaction
Unburned fuel
Smolde
r veloc
ity
FIGURE 2.26
Schematic of smoldering.
n
1.5 tio
ns ula
ic i
los
Smolder velocity (10–2 cm/s)
llu m
Ce foa
nurate
ya
m
c
yis o
foa
ol lic foa
m
1.0
ri g id p pheno
ne
lar rigid
anular
tha
nu Gr
Gra
ure
ly
e
e pin
Po
lar whit
0.5 G ranu
0
0 0.2 0.4 0.6 0.8 1.0
Air flow velocity through fuel layer (cm/s)
FIGURE 2.27
Reverse smolder velocities depend on airflow velocity. (From Ohlemiller, T.J., Smoldering
combustion, Chap. 2–11, in: DiNenno, P.J., ed., SFPE Handbook of Fire Protection Engineering,
2nd edn., National Fire Protection Association, Quincy, MA, June 1995.)
Smoldering can revert to flaming combustion particularly due to an

increase in the airflow rate over the fuel. This is implied by the termination
of the graphs in Figure 2.27. A transition to flaming can also occur in wood.
As the smolder rate increases in a block of wood, the process can revert to
flaming after perhaps 1 hour. As the char builds up and the incomplete com-
bustion products move through the hot char, they can reach their AIT and
produce a flame. It should be pointed out, however, that as some wood prod-
ucts like fiberboard might be prone to smolder, the smoldering process on
wood generally takes some heat supply to be sustained. A confined space to
trap the heat could be ideal, once smoldering has begun. Hidden fires gener-
ally begin as smoldering.
2.6 Spontaneous Combustion

In the first edition of this book, spontaneous ignition was classed as a form
of combustion. This is not precisely true. Once ignition occurs, combustion is
in the form of a flame or smoldering. But before that point, we have a chemi-
cal reaction that has not yet made the transition to combustion. To reach
combustion, we must step through ignition. Ignition can be appreciated by
TABLE 2.4
Typical Smolder Velocities Associated with Fuel Configuration
Air Supply Smolder Velocity
Fuel Smolder Configuration Condition (mm/min)
Pressed fiber 1.3 cm thick, horizontal Natural convection/ 0.78–1.3
insulation board, strips, width large diffusion
0.23–0.29 g/cc compared to thickness
“ 1.3 cm × 1.3 cm, strips Natural convection/ 1.6–2.8
varied angle to vertical diffusion
“ 1.3 cm × 5 cm strips, Forced flow,
forward smolder 20 cm/s, 2.1
1500 cm/s 7.8
“ 1.3 cm × 5 cm strips, Forced flow, 1.7–2.1
reverse smolder 80–700 cm/s
Pressed fiberboard 1.3 cm × 30 cm sheets, Forced flow, 0.42
(pine or aspen), horizontal, forward 10–18 cm/s
0.24 g/cc smolder
Cardboard Vertical rolled cardboard Natural convection, 3.0–5.0
cylinder, downward diffusion
propagation, dia.
0.19–0.38 cm
Shredded tobacco 0.8 cm dia. cigarette, Natural convection, 1.8–3.0
horizontal, in open air diffusion
Cellulose fabric + Double fabric layer, 0.2 cm Forced flow, 10 cm/s 6.0
3% NaCl thick, horizontal, forward
smolder
Source: Ohlemiller, T.J., Smoldering combustion, Chap. 2–11, in: DiNenno, P.J., ed., SFPE
Handbook of Fire Protection Engineering, 2nd edn., National Fire Protection Association,
Quincy, MA, June 1995.
considering a coin being pushed toward the edge of the tabletop. It can be
pushed until it is hanging over the edge, yet still on the tabletop. At some
small movement, it becomes out of balance and falls to the floor. There it
stays until some other force comes along. Ignition is the process of falling.
Combustion is sitting on the floor. Before combustion, we have a chemical
reaction in balance that is producing an imperceptible amount of energy.
It is a producer, exothermic, but not a lot at this state. With a little boost to
raise its temperature, the reaction will go out of balance and start to produce
more energy than it can absorb. Eventually, it will come into balance.
The higher-temperature state is combustion. As long as fuel and oxygen are
supplied in the amount it needs, it will stay at this state. Of course, cooling
it to reduce its temperature can send it spiraling down to extinction: its state
before ignition on the table top.
Spontaneous combustion can occur in many media. Figure 2.28 shows
some examples. For the fuel–air mixture, if it reaches its AIT by some
means it will make the transition to a flame. The process requires some heat
Fuel gas
and air
(a) (b) (c)
FIGURE 2.28
Examples of fuel arrays prone to spontaneous combustion: (a) mixture of fuel gas and air
raised to its autoignition temperature, (b) haystack, biological reactions promoted by moisture,
and (c) pile of wood fiberboard or coal waste (slag).
transfer and the availability of oxygen and fuel to be sustained. Haystacks

are porous and can allow air to penetrate. If they are wet, they are a breed-
ing ground for microorganisms. These organisms will use oxygen and pro-
duce energy. The biological process can give the initial push to raise the
temperature of the hay and cascade its chemical oxidation to combustion.
Most typically, the first combustion result in porous solid media will be
smoldering. But as the smoldering continues, air channels into the core will
bring more fresh air to the smoldering and move it to flaming. In the case of
porous media, the process will start in the core for a warm environment, or
at the source of heat. A telltale sign of the process of spontaneous combus-
tion in pile of material will be the degradation of material in the core of a
pile, as this region loses heat least. The bigger the pile, the more difficult it
is to lose heat to surface. So, big piles require a lower AIT. The material will
not have a unique AIT but will depend on the size and heating conditions.
For fuel–air gas mixtures, the AIT is measured in a standard test at a spe-
cific mixture volume. Commonly, the vessel is spherical and placed in an
oven. For solids or liquids, a wide variation of sizes can result in many AITs.
Piles of coal or wood fiberboard can play the same role as the haystack.
Each material has its own peculiar chemical characteristic. Any material
that releases energy in a chemical reaction is prone to spontaneous ignition.
The reaction is usually oxidative, but it can also be simply a decomposition
of the material, such as ammonium nitrate. On April 16, 1947, ammonium
nitrate stored on the French Liberty ship Grandcamp in the harbor at Texas
City, was likely heated by a small fire, ignited, and detonated. About 500
homes were destroyed, and over 3000 people were injured.
Let us consider the spontaneous ignition scenarios of a porous oxidative
solid. Three classic scenarios might be considered: (1) a pile standing in air,
(2) material on a hot surface, and (3) a hot pile exposed to air. In all cases,
the inability of the production of chemical energy to be lost as heat to the
surroundings will cause the ignition. For each of these scenarios, for a given
material, there is a critical configuration that will trigger the ignition process.
While for a gaseous fuel this process is fast (seconds at most), for a porous
solid it can range from minutes to weeks. As oxidation (or decomposition) is
occurring, telltale odors might signal the beginning of the process.
The pile in air might consist of linseed oil absorbed on a bundle of cot-
ton rags. Commonly used to clean and restore furniture, this unsaturated
oil in combination with the cotton makes a potent source. It has been the
cause of many accidental fires, notably the 38-story One Meridian Plaza in
Philadelphia (1991). Rags were used to clean the furniture and left near a
desk in an office. The resulting fire led to the propagation over several floors
with significant structural damage to the building.
The second scenario is a material on a hot surface. A classic old example is
wood in contact with a hot steam pipe. This scenario can show the complex-
ity of the process. It can take many years of heating to prompt this ignition.
Likely, the wood slowly or intermittently heats and char begins to build up.
As the char becomes deeper and more porous, it becomes a good medium for
spontaneous combustion. At a critical char thickness and surface tempera-
ture, ignition of the char can occur. This is likely to be smoldering at first. The
char can smolder away to form a cavity. An increase in airflow and confined
space heating within the cavity can then lead to flaming on the wood.
The third scenario is the hot pile. This scenario is characteristic of mate-
rials undergoing processing, as with wood resin products or plastics. The
formation of plastics involves exothermic reactions. If the reaction is not ter-
minated before the end product is sold, its continuation can lead to sponta-
neous combustion. It might also be associated with commercial laundries
where large hot piles of clothes are removed from a dryer. If the pile cannot
lose the heat fast enough, ignition could be triggered.
Although theories exist for spontaneous combustion, they are difficult
to apply with high accuracy. An advanced treatise on this subject is given
by Bowes, but if one sidesteps his mathematical treatment, he also presents
many practical examples.9 He shows that with the proper material data, a
rationale estimation of spontaneous ignition potential can be determined.
A vivid excerpt from Bowes9 describes the range of results for cotton waste
soaked with linseed oil.
The amounts of material involved in the initiation of fires can be quite
small. For example, Taradoire (1925), investigating the self-ignition of
cleaning rags impregnated with painting materials, stated that experience
indicated the occurrence of self-ignition in as little as 25 g of rags but, in a
series of experiments, found that the best results (sic) were obtained with
75 g of cotton rags impregnated with an equal weight of mixtures of linseed
oil, turpentine, and liquid “driers” (containing manganese resinate) and
exposed to the air in cylindrical containers made of large-mesh wire gauze
(dimensions not stated). Times to ignition generally lie between 1 hour and
6 hour but, for reasons which were not discovered, some samples, which
had not ignited within these times and were considered safe, ignited
after several days. Other examples giving an indication of scale have been
described by Kissling (1895), who obtained self-heating to ignition, from
an initial temperature of 23.5°C, in 50 g of cotton wool soaked with 100 g
of linseed oil and by Gamble (1941), who reports two experiments with
cotton waste soaked in boiled linseed oil (quantities not stated) and lightly
packed into boxes; with the smaller of the two, a cardboard box of dimen-
sions 10 cm × 10 cm × 15 cm, the temperature rose from 21°C to 226°C in
6¼ hour and the cotton was charred.
In short, it demonstrates that not much cotton soaked with linseed oil is
needed to achieve ignition, but the time to occur ranged from 1 hour to sev-
eral days. This propensity to spontaneous combustion can be measured by
subjecting a sample of the substance in question to oven air temperatures
until the lowest air temperature is found to eventually (in hours or days)
cause ignition. The end result can be manifested by smoldering or flaming.
As the size of the fuel increases, reducing the ability of heat to be conducted
from its center, the critical air temperature needed for ignition decreases.
The core temperature at the critical condition will sharply increase to a smol-
dering or flaming state. So, the combustion starts from within. Additives to
the material can enhance this propensity. Bowes9 reports such experimental
results in Table 2.5. As the size of the cube of sawdust increased, a lower oven
temperature (AIT) was needed to set off the ignition. The addition of the oil
clearly made the sawdust more prone to spontaneous ignition. Such testing
and its data can be used to assess the potential for spontaneous ignition in
real and larger scenarios. Indeed, in theory, these data can supply needed
chemical properties to mathematical models to predict ignition different con-
ditions. But because the real conditions may not be duplicated completely in
TABLE 2.5
Sawdust Cubes, with and without Oil, Exposed to Air Temperatures
Cube Size 2r mm Oil Content (%) Ignition Temperature (°C)
25.4 0 212
25.4 11.1 208
51 0 185
51 11.1 167
76 0 173
76 11.1 146
152 0 152
152 11.1 116
303 0 135
303 11.1 99
910 0 109
910 11.1 65
Source: Bowes, P.C., Self-Heating: Evaluating and Controlling the Hazards, Her Majesty’s
Stationery Office, London, U.K., 1984.
the test oven, extrapolations are limited in accuracy. Still such results bring
some quantitative and rational assessment to the potential of spontaneous
ignition. This rationality is especially significant as fires originating due to
spontaneous ignition or smoldering are often misdiagnosed. On the other
hand, electricity as the cause of fire is likely overprescribed, while spontane-
ous ignition or smoldering might be at the root instead.
2.7 Summary
Natural fire can be divided into three types of combustion:
1. Diffusion flames
2. Premixed flames
3. Smoldering
Combustion to a flame or smoldering requires an ignition process. Energy

must be put into the reactants to achieve a thermally runaway process and a
sudden jump in temperature to a combustion state. This jump is caused by the
fact that the energy production rate of the reaction depends on temperature.
Laminar flames in air can achieve temperatures of about 2000°C and a low of
about 1300°C before extinction. Smoldering in air might occur as low as about
300°C. The temperature in a turbulent flame is practically represented by an
average much lower than the laminar fluctuating flames in its structure.
The AFTs in air for many hydrocarbons are about 2000°C with some excep-
tions. This can be explained by the nearly constant energy release of oxygen,
about 13 kJ/g.
A diffusion flame has fuel and oxygen on either side of the flame, while
in a premixed flame the two are mixed before combustion. A candle flame
is representative of a diffusion flame. Soot and other incomplete products of
combustion occur on the fuel side of the flame. Smoldering also requires the
diffusion of air to the solid fuel.
Smoldering is a slow propagation process at speeds of 0.01 cm/s at most.
Laminar premixed flame speeds in air can range from about 20 to 100 cm/s,
with turbulent speeds increasing by up to 100 times. Detonation speeds can
achieve 2000 m/s.
Spontaneous combustion or autoignition occurs because the fuel under-
goes an exothermic reaction (generating chemical energy) that cannot be
sufficiently cooled. As a result, the reaction accelerates and leads to either
flaming or smoldering combustion. Spontaneous ignition can result by
decomposition or by oxidation, as in the drying of linseed oil on cotton.
It can take hours or days to develop ignition.
Review Questions
1. What role does temperature play in combustion?
2. Under what circumstances can a premixed flame ignite and propagate?
3. How does a diffusion flame begin?
4. Explain the role of airflow in smoldering.
5. Why might spontaneous ignition take so long?
True or False
1. Autoignition requires a pilot flame.
2. An oxy-acetylene torch is a diffusion flame.
3. A burning pool of gasoline is a diffusion flame.
4. Charring is not likely during smoldering.
5. Plastics cannot smolder.
6. Smoldering can occur in concealed spaces with very little air.
7. A flame must be 10 ft high to be turbulent.
8. Temperature plays a key role in determining if a flame can occur.
9. A canister of raw plastic material with insufficient antioxidants could
lead to spontaneous ignition.
10. Most hydrocarbon fuels have about the same flame temperature burning
in air.
Activities
1. Conduct and discuss the candle experiments described in this chapter.
2. Measure the velocity of a smoldering process, for example, a cigarette.
See how air speed might change the smoldering rate.
3. Put linseed oil soaked (not too wet) cotton rags in a cardboard box about
1 ft3. Place in a safe place for doing fire tests and let sit for several hours.
Explain the results. Measure the center temperature if possible.
4. Compare the different flames produced by a Bunsen burner by regulat-
ing the air vents. Can you blow the flame off by increasing the flow too
high? Can the premixed flame, with the vents open, be extinguished as
the fuel flow rate is decreased? With the vents closed, as you increase the
fuel flow rate does the flame height level out?
References
1. M. Faraday, Faraday’s Chemical History of the Candle (Chicago, IL: Chicago Review
Press, 1988 [orig. 1861]).
2. K. C. Smyth, J. Houston Miller, R. C. Dorfman, W. G. Mallard, and R. J. Santoro,
Soot inception in a methane/air diffusion flame as characterized by detailed
species profiles, Combustion and Flame, 62 (1985): 157–181.
3. K. Smyth, J. E. Harrington, E. L. Johnson, and W. M. Pitts, Greatly enhanced
soot scattering in flickering CH4/air diffusion flames, Combustion and Flame,
95 (1993): 229–239.
4. M. G. Zabetakis, Flammability characteristics of combustible gases and vapors,
Bulletin 627, Bureau of Mines (Washington, DC: U.S. Department of the Interior,
1965).
5. C. L. Beyler, Flammability limits of premixed and diffusion flames, Chap. 2–9
in SFPE Handbook of Fire Protection Engineering, 2nd edn., edited by P. J. DiNenno
(Quincy, MA: National Fire Protection Association, June 1995).
6. C. Huggett, Estimation of rate of heat release by means of oxygen consumption
measurements, Fire & Materials, 4 (1980): 61–65.
7. B. J. Lewis and G. von Elbe, Combustion, in Flames and Explosions of Gases,
3rd edn. (Orlando, FL: Academic Press, 1987), pp. 551–555.
8. T. J. Ohlemiller, Smoldering combustion, Chap. 2–11 in SFPE Handbook of Fire
Protection Engineering, 2nd edn., edited by P. J. DiNenno (Quincy, MA: National
Fire Protection Association, June 1995).
9. P. C. Bowes, Self-Heating: Evaluating and Controlling the Hazards (London, U.K.:
Her Majesty’s Stationery Office, 1984).
3
Heat Transfer
Learning Objectives
• Describe the difference between energy and heat

• Explain conduction, convection, and radiation heat transfer
• Describe the concept of heat flux and its significance to hazards
from fire
• Perform simple computations in heat transfer
3.1 Introduction
Although chemistry and fluid mechanics are important in the study of fire,
we can forgo their study for elementary quantitative purposes here. Fluid
flow is important to know how air enters a flame and how smoke travels
in a building. Mostly, this flow is due to buoyancy created by the fire and
hot smoke itself. Wind is important for fires in the open, as in forest fires
and conflagrations. In our study of fire plumes (Chapter 7) and compart-
ment fires (Chapter 9), these aspects of fluid flow will be addressed. Of
course, chemistry is important, as it is the basis of combustion. In the previ-
ous chapter, we have learned that chemical reaction takes place relatively
fast, and therefore, as the fuel and oxygen burn, the reaction region is very
thin. Often in even sophisticated analysis of flames, the actual chemical
dynamics and reaction mechanisms can be ignored. An assumption for a
flame of having zero thickness can suffice, and there the fuel and oxygen
immediately become products of combustion. However, the importance of
heat transfer cannot be avoided if we are to develop the ability to make
73
quantitative analyses. Heat transfer causes the solid and liquid fuels to
vaporize to produce the gaseous fuel for the flame. Heat transfer is respon-
sible for making the fire move over surfaces and through the forest. Indeed,
it is the dominant factor in determining the fate of fire growth and the
damage it can produce. Therefore, for fire we must first develop some
understanding and computational ability in heat transfer before we can
appreciate aspects of fire behavior.
3.2 Definitions and Concepts

Before we begin the study of heat transfer, some concepts and definitions
have to be introduced. First, the concept of energy must be introduced. The
study of energy is rooted in the subject of thermodynamics, a very logical
science that carefully defines energy, heat, temperature, and other proper-
ties. As with fluid flow, heat transfer, and chemistry, thermodynamics is at
least a semester course in itself. Therefore, only the essential aspects of ther-
modynamics will be considered.
All matter has energy (E) and mass (m); these are the essential prop-
erties of matter. In the famous equation of Einstein, E = mc 2, mass and
energy are related and can be converted into each other. These, of course,
are nuclear reactions that play no role in fire. The speed of light (c) is so
large that the conversion of mass to energy is enormous. These conver-
sions do not apply here, and for a fixed piece of matter its energy and mass
cannot change. This fact is founded in the conservation laws of mass and
energy. The equation of Einstein that applies only in a nuclear reaction
simply allows the interchange of these two quantities. Indeed, atomic
physicists do not distinguish between the two. The conservation of mass
and energy cannot be proven; it is taken as a general truth. Moreover,
mass and energy cannot be directly defined; they are conceptual. Yet, we
can sense and measure their effects. This is a little deep, so we will try to
get a bit more tangible.
As all matter possesses energy, it can take many forms. Some we can
readily recognize. The other forms of energy are discerned indirectly from
the conservation of the energy. By the way, this conservation law is called
the First Law of Thermodynamics. Let us recognize some of the forms of
energy. When we lift a weight over a distance, the required force times the
distance moved is called work. In the SI system of units, work is force (N for
Newton) times distance (m, meter); or its unit of measure is N-m. An alias
for 1 N-m is a joule, J: a common unit of energy. (In the English system, J
is related to BTU.) The industrial revolution allowed us to burn coal, heat a
boiler to create steam, and move a piston to produce work. In that process,
several other forms of energy show up. Chemical energy results from the
Heat Transfer 75
oxidation of coal to products of combustion. Heat transfer brings energy

to the boiler. The motion of the piston manifests as kinetic energy and is
then converted into work. Energy is a state of matter from which we might
extract work or obtain heat. If there is no motion and no chemical reaction,
does that mean that no more energy exists in the original matter from which
work and heat were extracted? No, there is an intrinsic energy left and per-
sists as long as the atoms are in motion. You might say this is the kinetic
energy of the moving atomic particles. This is called thermal energy (or more
thermodynamically correct, internal energy).
Thermal (or internal) energy is a property of matter directly associated
with the concept of temperature. Something that is “hot” has a relatively
higher internal energy and temperature compared to that which is “cold.”
The level of hot and cold needs to be quantified, and various temperature
metrics have been developed to do this. The scales used to measure tem-
perature are arbitrary and have been established by convenience as well as
by science. The common four scales and their relationship are illustrated
in Figure 3.1 (see Table 1.6). They have been historically based on the freez-
ing and boiling points of water. The “distance” or temperature measure
between them has been arbitrarily assigned +32 and 212 for freezing and
boiling on the Fahrenheit scale and 0 and 100 on the Celsius (or centigrade)
scale. One might begin to see why the Celsius scale is preferred. In addi-
tion, there are two modifications of these scales. These are the absolute
scales that set zero temperature at zero internal energy. They are expressed
in degrees Rankine (°R) for the scale in increments of Fahrenheit degrees
and in Kelvin (K) associated with the Celsius increments. Zero kelvin is
the same as −273°C, and the increments on the two scales are the same.
The absolute temperature scale will be important in heat transfer, as for-
mulas for radiation use this scale. At the absolute zero state, the molecular
°F °R °C K
Boiling point of water 212 672 100 373
Freezing point of water +32 492 0 273
Absolute zero –460 0 –273 0

(No thermal energy)
FIGURE 3.1
Temperature scales.
Hot Cold
High internal Heat Low internal

Energy . Energy
High temperature q Low temperature
(kW)
FIGURE 3.2
Heat transfer as energy exchange between hot and cold.
mechanical activity ceases. Temperature is a direct measure of this activ-

ity, and at this “absolute zero point,” it is the lowest possible temperature
for matter.
Heat is defined as thermal energy in motion that travels from a “hot”
to a “cold” region. This process is symbolically illustrated in Figure 3.2.
There are two laws that govern this heat transfer: one deals with the
energy travel within matter and the other can travel in the absence of
matter (vacuum).
The symbol q represents heat (while Q represents chemical energy
in this chapter). Both are measured in joules (J). It takes 4.182 J (or 1 cal)
to raise 1 gram (g) of water 1°C (see Tables 1.5 and 1.7 for alternative
units and notation). The rate of heat flow or the rate of chemical energy
generated (transformed) is commonly given in J/s or its alias, watt, W. For
example, if we transfer by heat 4.182 J to 1 g of liquid water, its internal
energy will increase by 4.182 J and its temperature by 1°C. If it takes
1 second to accomplish this, we have 4.182 J/s or 4.182 W of heat flow rate.
If that transfer rate continues for 25 seconds, the temperature of water
will increase by 25°C. Hopefully, before the end of this book, you will
relate to fire in terms of its W output in the same way you relate to ratings
for a light bulb, furnace, or toaster. It is common to relate the size of fire
in terms of kW.
An important quantity in heat transfer is heat flux. It is defined as the flow
rate of heat per unit area it passes through. More precisely, imagine heat
flow rate depicted by an ensemble of arrows in the direction of everywhere
it is flowing. The area through which one arrow flows is perpendicular to
this arrow. For a single arrow, consider its flow area shrinking to a point.
The ratio of that heat flow rate associated with the arrow and the small
area reduced to a point gives the heat flux. Its symbol is given as q ¢¢, and
its units are typically kW/m2. Heat flux is a point function, while heat flow
rate applies to sum of all the heat flux passing through the area of the heat
flow. In heat transfer, the heat flux will vary over a region in general. Close
to a hot source, we would experience a higher heat flux than farther away.
Touching a hot plate would give a higher heat flux than touching its insulated
edge. It is the heat flux that we feel by the Sun on a cold day. It is the heat flux
Heat Transfer 77
that makes objects burn and cause damage in fire. It is the most important
quantity that expresses the hazard of fire.
We will examine the forms of heat transfer and how they apply to fire. The
key parameter will be heat flux.
3.3 Forms of Heat Transfer

The “laws” of conduction and radiation govern heat transfer. These “laws”
are not as perfect as the Conservation Laws of Mass and Energy; they are
just very good approximations. In the early 1800s, Joseph Fourier formulated
the law of heat conduction. It described the rate of heat flow through matter
as directly proportional to temperature difference divided by the distance
separating those temperatures in the substance. He was a scientist working in
Napoleon’s army on the cooling of cannon barrels. War motivates discoveries.
It took nearly 100 years later to discover the heat transfer relationship for radi-
ation. Formulated in the early 1900s, Max Planck’s quantum theory led to a
theoretical basis for radiation heat transfer. Such radiation moves at the speed
of light in a vacuum and is very different from the way conduction moves.
Convection is another category of heat transfer but it does not require a
separate “law” to predict it, as it is a subset of conduction. It applies specifi-
cally to conduction in a moving fluid. Figure 3.3 illustrates these forms of
heat transfer in a fire situation of a flame against a wall. Conduction occurs
through the wall, convection occurs at the wall–flame or smoke interface,
and radiation can be felt at a distance and to the wall. Also, radiation comes
from the flame, smoke, and the hot wall.
Flame plume
Wall
Conduction
Radiation
Convection
FIGURE 3.3
Examples of heat transfer in fire.
3.3.1 Conduction
3.3.1.1 Steady
The simplest representation of the law of conduction is given as follows in
steady form (unchanging with time) for the flow rate of heat through a wall.
This is illustrated in Figure 3.4. The equation is given as
kA(T2 - T1 )
q = (3.1)
l
where
k is the thermal conductivity
A is the area through which the heat is transferred
T2 and T1 are the respective temperatures of the wall faces
l is the wall thickness
This equation says that the rate of heat flow (e.g., kJ/s or kW) between two
temperatures, T1 and T2, in a solid is proportional to a property of the solid
known as thermal conductivity (k). Also, the larger the flow area through
This is analogous
which the heat is conducted, the greater is the flow rate, q.
to the increase in the rate of water flowing through a pipe as its diameter
is increased. It should be remarked that this relationship for conduction is
analogous to the flow rate of electrons (current) in a metal wire as being
controlled by the voltage difference and the electrical resistance of the wire.
T2 T1
FIGURE 3.4
Heat conduction through a wall.
Heat Transfer 79
A resistance factor is similarly used for heat conduction. The commonly used
R-value for home siding insulation is the reciprocal of kA/l. This R-value is
called thermal resistance as an analogy to electrical resistance where q takes
on the role of current flow rate and temperature takes the role of voltage. Let
us consider an example.
Example 3.1
Find the heat flow rate per unit area through a wall of polyurethane
foam, 0.05 m thick (l), across a temperature difference of 40°C–20°C. Note
that heat flow rate per unit area is called heat flux (denoted by q¢¢):
q (0.034 W/m-K )( 40 - 20)° C

q ¢¢ = = = 13.6 W/m 2
A 0.05 m
or 0.013 kW/m2
Consider the same conditions of temperature, thickness, and area,

except the solid is now steel. Here, k (from Table 3.1) is 45.8 W/m-K. Note
that whether we use K or °C in the thermal conductivity term (or for
the temperature in the formula), there is no change. This is so because
TABLE 3.1
Table of Thermal Properties
Thermal Specific Thermal Thermal
Conductivity Heat (c) Density Diffusivity Inertia (kρc)
Material (k) (W/m-K) (kJ/kg-K) (ρ) (kg/m3) (α) (m 2 /s) (kW2-s/m4 -K 2)
Copper 387 0.380 8940 1.14 × 10−4 1300.0
Steel (mild) 45.8 0.460 7850 1.26 × 10−5 160.0
Brick (common) 0.69 0.840 1600 5.2 × 10−7 0.93
Concrete 0.8–1.4 0.880 ~2000 5.7 × 10−7 2.0
Glass (plate) 0.76 0.840 2700 3.3 × 10−7 1.7
Gypsum plaster 0.48 0.840 1440 4.1 × 10−7 0.58
PMMA 0.19 1.420 1190 1.1 × 10−7 0.32
Oak 0.17 2.380 800 8.9 × 10−8 0.32
Yellow pine 0.14 2.850 640 8.3 × 10−8 0.25
Asbestos 0.15 1.050 577 2.5 × 10−7 0.091
Fiber insulating 0.041 2.090 229 8.6 × 10−8 0.020
board
Polyurethane 0.034 1.400 20 1.2 × 10−6 9.5 × 10−4
foam
Air 0.026 1.040 1.1 2.2 × 10−5 3.0 × 10−5
Source: Drysdale, D., An Introduction to Fire Dynamics, Wiley, New York, 1985, p. 36.
we are concerned with temperature differences and the degree in the

numerator and denominator cancels out. The calculation for this case is
( 45.8 W/m-K)(20°C)
q ¢¢ = = 18, 320 W/m 2
(0.05 m)
or
q ″̇ = 18.3 kW/m2
Thus, we see the ability of heat flow to conduct through steel is much
greater than through an insulator-like polyurethane.
Under fire conditions, heat conduction can play a role in propagating

the fire. Fires have been known to propagate on ships due to direct heat
conduction through the steel floors and bulkheads. Substances can ignite
due to conduction if they are in contact with a very hot body. Also, cor-
ners will ignite before edges, which in turn will ignite before a flat surface
of the same material for the same heat flux. In those geometries, the one
with the least mass will heat more quickly. In general, the transfer of heat
is more complex than we have depicted it by Equation 3.1. Layers of differ-
ent materials, their shape, and the time for thermal penetration all add to
the complexity and must be considered for an accurate prediction of the
temperature.
In general, heat transfer is unsteady, and it takes some time for the heat to
penetrate through the substance, raising its temperature as the heat reaches
a new depth. A rough estimate of how long it will take the temperature to
be effectively felt at a given depth can be made. If the surface of a material
suddenly has an increase in temperature, the time for the temperature to
increase at a distance (x) within the material is proportional to x2/α.
3.3.1.2 Thermal Penetration Time

The thermal penetration time for a 1% increase in temperature rise at x is
approximately x2/12α or to reach 50% of the increase is approximately x2/α
where α = k/ρc, thermal diffusivity, ρ is the density, and c is the specific heat.
The thermal diffusivity is a measure of how fast it takes a thermal effect to
be felt within a still cup of coffee. The motion of heat in the coffee is analo-
gous to the motion of dissolved sugar from a solid lump at the bottom of
the cup. This phenomenon is known as diffusion of the sugar through the
coffee. The molecules of sugar and coffee liquid are intermingling, and the
concentration of sugar is rising throughout. The heat conduction propaga-
tion within the material is due to thermal diffusion or the diffusion of heat
that excites neighboring molecules. As the internal energy of the molecules
Heat Transfer 81
is raised by atomic vibrations from their hotter neighbors, temperature

increases and heat is propagated, somewhat like a wave effect.
How long would it take the back wall in the polyurethane Example 3.1 to
realize that heat has been applied to the front wall? The thermal diffusivity
can be found from Table 3.1:
(0.05 m)2
Time (1%) = = 174 s
(12 (1.2 ´ 10 -6 m 2 /s))
In contrast, the steel wall has a (1%) penetration time of about 16.4 sec-
onds. Following this transient propagation, steady conduction through
the wall will take place after this penetration time. Then Equation 3.1
can be applied. All of these facets must be considered in analyzing heat
conduction.
In the past, exact sophisticated mathematical solutions were rendered
for conduction problems. Today, modern computers are capable of doing
the same and for more complex problems. However, in the end, the mate-
rial properties have to be known, and conditions in the surroundings of
the conducting material need to be known. If that material is a solid in
a hot fluid such as air, the conduction between the air and the material
needs to be computed. If the air is moving, this conduction will be affected
by the flow.
3.3.2 Convection
In a moving fluid, the heat transfer from the fluid to a solid surface is called
convection, but the governing law is still that of conduction. However, the
temperature difference and distance traversed near the surface are not
readily measured or derived from pure conduction considerations. For
example, as hot air flows over frozen water, the heat transfer by conduction
at the ice surface depends on the temperature difference near the surface, ΔT.
By the law of conduction, the heat transfer at the ice surface from the air as
illustrated in Figure 3.5 is still given by Equation 3.1:
DT
q = kA ×
Dl
where Δl is the distance between the temperatures corresponding to ΔT. The

heat flux to the wall is
q DT
q ¢¢ = =k
A Dl
Temperature
0°C 30°C
Air
∆l
∆T Ice
FIGURE 3.5
Convective heat transfer example.
The ratio ΔT/Δl must be known just at the surface to give a value for this heat
flux. As the temperature in the hot air varies sharply as the air gets closer
to the surface, only very detailed temperature measurements or complex
computations (calculus) can resolve this ratio. Alternatively, the equation is
rearranged in terms of the known temperature difference between the hot
air stream and the ice, as
æ k ö
q¢¢ = ç ÷ (T2 - T1 )
è Dl ø
where
T2 is the air stream temperature (e.g., 30°C)
T1 is the surface temperature (e.g., 0°C)
The flow speed of the air will affect the distance Δl; the higher the flow
speed, the more Δl will be reduced. In physical terms, here Δl is a fraction of
the thickness of the air region between 30°C and 0°C. That region is called
a boundary layer in fluid mechanics. The quantity (k/Δl) is now lumped into
one parameter that is defined as the convective heat transfer coefficient given
by the symbol, h. It depends on the air properties as well as the flow speed.
Therefore, the equation for a flowing fluid that gives the heat flux at the sur-
face is given as
q¢¢ = h(T2 - T1 ) (3.2)
The heat transfer coefficient provides a simple means to evaluate the con-
vective heat flux in a fluid stream flowing over a surface across which the
temperatures are known. It is too complicated for us to explain all the
Heat Transfer 83
TABLE 3.2
Typical Values for Convective Coefficients, h
Fluid Condition h (W/m 2-°C)
Buoyant turbulent flows in air 5–10
Laminar match flame ~30
Turbulent liquid pool fire surface ~20
Fire plume impinging on a ceiling 5–50
2 m/s wind speed in air ~10
35 m/s wind speed in air ~75
methods for determining h. This is the subject of a convective heat transfer

course and involves the prediction of h through measurements or calcula-
tions of the flow over the particular surface.
For fire processes, we do not need much information to make most
estimates for h, as the flow field is primarily natural convection driven
by buoyancy. Table 3.2 gives typical values for h. Fire conditions do not
significantly change the values given for air as suggested in Table 3.2. So, if
we know the temperatures of the flow stream and the flow surface, we can
estimate the heat transfer coefficient from Table 3.2 and use Equation 3.2 to
estimate the convective heat flux to or from the surface.
Let us consider a fire example.
Example 3.2
Find the convective heat flux from a turbulent flame to a cold wall at
20°C. Estimate the h value from Table 3.2 for free convection (a buoyant
flow that is indicative of a flame) as about 5–10 W/m2-°C. Let us select 5
for illustrative purposes here. The maximum time-averaged flame tem-
perature is given as approximately 800°C.
q² = (5 W/m 2 -°C) (800 - 20)°C
= 3900 W/m 2
= 3.9 W/m 2
Had we used 10 W/m2-°C for h instead, our answer would be double.

This range of 5–10 kW/m2 is representative of most turbulent fire condi-
tions. The convection heat flux to a burning (vaporizing) surface will be
slightly less because the flow of fuel gases will push the hot flame away
from the surface, increasing Δl (see Figure 3.5). Also, for laminar flames,
the flame will be closer to the surface, which corresponds to a smaller Δl.
For this reason, small laminar flames have a higher convective heat flux
than turbulent flames. Near the surface of a pool fire, the flow will be
laminar near its edge and turbulent above.
For the range of convective fire conditions, Table 3.2 suggests that for laminar
flames, the convective coefficient is about 30 W/m2-K. This is an estimate and
decreases from the place where the flow begins on the surface. That is where the
flame is closest. Near the beginning point, this value could be double. Also for
laminar flames, the flame will be thin, and its temperature could be as high
as 2000°C. A computation of the convective laminar flame heat flux from
Equation 3.2 would give heat fluxes ranging from about 60 to 120 kW/m2.
These values are high compared to the more general fire conditions where
turbulent flow prevails. In turbulent flow, the flame is thicker and spread
out due to the fluid fluctuations and mixing in the flow. The temperature
at any fixed point in the reaction region varies rapidly due to these fluc-
tuations. Measurements show that this time-average temperature is about
800°C–1000°C for most fires but could achieve as much as 1200°C for very
large flames. From Table 3.2, the turbulent convective coefficient is about
5–10 W/m2-K for buoyant flames along walls. This gives a peak convective
heat flux of only 10 kW/m2 for turbulent fire conditions. Even in the higher
velocity impingement region from a turbulent flame plume at a ceiling, this
value might only achieve about 50 kW/m2 just at the region of impingement.
We see that the laminar flame is more powerful than its turbulent counter-
part in convective heating. The impact on fire hazard is mitigated as laminar
flames quickly become turbulent after lengths of about 0.3 m (1 ft). But the
turbulent convective heat fluxes alone are not generally sufficient to sustain
fire. We need something else. This comes from the other way heat can be
transferred—radiation.
3.3.3 Radiation
Planck established the theoretical basis for explaining radiation heat transfer.
All radiation is electromagnetic energy consisting of both electric and
magnetic fields. This is distinctly different from the intimate molecular trans-
mission nature of conduction. These radiation rays have both a frequency
and a speed (the speed of light in a vacuum, c = 29,979,245,800 cm/s) and can
traverse within materials and through a vacuum of nothing. The frequency
(f) range determines the descriptive name or phenomena we associate with
the radiation. For example, high-frequency (alternatively low wavelength,
λ = c/f) radiation phenomena, in descending order, include cosmic rays,
gamma rays, and x-rays reaching into ultraviolet, which borders visible
light (a form of radiation). In decreasing order from visible light, we move
to infrared and then radio waves. Thermal radiation is mostly felt across the
infrared and somewhat into the visible. However, it is generated over the
entire frequency spectrum by any piece of matter having a finite tempera-
ture (on the absolute scale). Objects having temperatures of about 1000°C can
appear bright red; we “see” the thermal radiation within the frequency or
wavelength range sensitive to our eye. At about 3000°C, they appear white.
The use of an infrared camera extends our ability to see lower temperatures.
Heat Transfer 85
Night vision cameras do this, making humans visible on a dark night. A

characteristic of smoke is that its transmission of radiation (light) decreases
with increasing wavelength. Such cameras, operating at the higher wave-
length (infrared), enable one to “see through” smoke until its concentration
gets too high to block the light.
Radiation originates in different ways. Radio waves can be electronically devel-
oped for broadcasting. For example, at 100 MHz (Megahertz) or 100 million
cycles per second on our FM dial, we might find station WFIR. Thermal radia-
tion originates solely due to temperature. All matter having a temperature
above absolute zero emits radiation. This radiation has a frequency range and
can be predicted from Planck’s theory; however, we will not examine radiation
in how it varies and is affected by its frequency. Rather than examining how
thermal radiation spans the radio dial we simply examine its overall output.
The maximum possible output of radiation due to temperature is given by
Planck and is expressed in terms of heat flux as
q ¢¢ = sT 4 (3.3)
where
T is the object’s temperature expressed in kelvin (K)
σ is called the Stefan–Boltzmann constant given as 5.67 × 10−11 kW/m2-K4
This formula applies to a perfect radiator or blackbody. It gives the maximum

radiation emitted at a given temperature.
Objects do not necessarily radiate at the maximum output given by
Equation 3.3. Surface effects and absorbing effects reduce this output. The
property that gives the fraction of real output to the maximum value is called
emissivity, symbolized by the Greek letter ε. For solid and liquid surfaces,
ε is typically 0.8 ± 0.2 for fire radiation applications. This means that about
80% of the radiation given by Equation 3.3 is emitted from such an object.
That radiation leaves the surface now as heat and moves to objects of lower
temperature. Here, we are considering a surface that is opaque or nontransmit-
ting. No radiation can pass through an opaque surface. Objects that are not
opaque can have radiation pass through them, and radiation can be emitted
and absorbed throughout its depth at all of the object’s temperatures. A win-
dow glazing does this for the low wavelength radiation contained in the Sun’s
radiation but does not allow the longer wavelength radiation emitted by the
lower-temperature objects in a room to be transmitted out. Radiation emitted
from a transparent object emits radiation from within as well as its surface.
Gases, smoke, and flames can be transparent to radiation. An effective emis-
sivity of a uniform temperature transparent object such as a flame depends
on the thickness of the flame. Flames radiate from gases in the form of dis-
crete pockets of radiation in the wavelength spectrum (fuels and combustion
products comprise these gases). In addition, soot particles radiate over the
entire radiation wavelength spectrum. For turbulent flames, the main source
of radiation is the soot. A single soot particle nearly follows Equation 3.3 for
the radiation emitted from its surface. But for a collection of soot particles
suspended in smoke or flame, each particle radiates and intercepts according
to how much is there. This means that in smoke or flame, the concentration
of soot is important. The amount of radiation that finally reaches the edge of
the smoke or flame from this internal emission must also depend on its over-
all depth. Consequently, it might be approximated that the effective emissiv-
ity of this smoke or flame can be estimated from the formula
ε = 1 − exp(−κl) (3.4)
where
κ is the absorption coefficient of the smoke or flame
l is its thickness
The absorption coefficient is a property of the flame and is a measure of how

easy it is for the radiation to penetrate the flame. It depends on the nature of
the soot and its concentration for the most part. For example, x-rays have a rel-
atively low absorption coefficient for body tissue, implying easy penetration
of the x-rays to see organs. For turbulent flames, κ typically ranges from 0.1 to
1.0 m−1. For common fuel flames having a thickness of 1–2 m or more, their ε
would be nearly 1. This means that such flames emit the maximum radiative
output as an ideal emitter in Equation 3.3. But for very large fires, cold soot
on its edges can actually obscure the flame and reduce the radiant heat flux to
the surroundings from the high-temperature luminous regions of the flame.
This situation is contrary to the notion that large fires emit more radiant heat
flux at their surface than smaller fires. However, as they are bigger, they will
have more surface area, and more radiation will be sent to the surroundings.
Let us consider a calculation for the incident radiant heat flux to a remote
target object caused by a fire or a hot surface such as depicted in Figure 3.6.
We consider two distinct problems here: (1) radiation from the hot surface
and (2) radiation from the flame alone. In both cases, the radiant heat flux
that is received at a target a distance, c, away will be reduced from that emit-
ted by T2. The fraction of energy reduced is called the configuration or view
factor, designated by F12. It represents the fraction of rays that can be seen
from the target relative to the emitting object compared to the full field of
view in the direction of the object (see Figure 3.7). Consider your eye the
target. View an object in the room. The fraction that object subsumes in your
180° field of view is the view factor. If the emitting object were a hemisphere
to a target at the center, the fraction would be 1, as all of the possible rays
leaving the target intercept the hemispherical source. Another interpretation
of F12 is that it is the fraction of radiant heat leaving the source that strikes the
object. F12 depends on distance, c; the size of the source object a × b; and the
surface orientation of both object and target.
Heat Transfer 87
Flame
or
Hot surface
T2
a .
q˝ Target
c
FIGURE 3.6
Radiation to a target from a hot surface or flame at a temperature T2.
Field of view
to object 2
T2
1
Total field of view
FIGURE 3.7
F12 represents the fraction object 2 makes in the total field of view from 1.
For the arrangement depicted in Figure 3.6, F12 can be found from the chart
given in Figure 3.8. In the chart, x = a/c and y = b/c. In that figure, the target
is located on a line that is normal to the bottom corner of the source, repre-
sented as a flat surface. The surface of the source is also normal to this line.
If the face of the target is turned by 90°, it could no longer see the target, and
0.3 X=∞
2
0.2
1
0.7
0.5
0.1
0.4
0.07 0.3
Configuration factor, F12
0.05 0.2
0.15
0.03
0.1
0.02
0.01
0.007
0.005
NACA
0.003
0.05 0.1 0.2 0.5 1.0 2 3 5 7 10 20
y
FIGURE 3.8
Configuration factor, F12. (From Hamilton, D.C. and Morgan, W.R., Radiant-interchange con-
figuration factors, NACA Tech. Note 2836, Washington, DC, December 1952, p. 78.)
F12 would be equal zero. Typical heat transfer references contain many charts
for different configurations.
The heat flux received by the target object is given as
q ¢¢ = esT24 F12 (3.5)
Let us consider an example based on Figure 3.6.
Example 3.3
Find the heat flux due to radiation from a wood fuel flame 1 m tall (a),
0.5 m wide (b), and 0.5 m thick (l) at a distance 3 m away (c). κ is 0.8 m−1 and
the temperature is assumed to be 800°C. (It is very difficult to make precise
radiation calculations because the fuel flame properties are approximate
and these data are scarce. Nevertheless, we can make estimations.) First,
the temperature must be converted to the absolute scale by adding 273:
T2 = 800°C + 273°C = 1073 K
The emissivity of the flame is then computed from Equation 3.4:
ε = 1 − e−κl = 1 − exp[−(0.8 m−1)(0.5 m)] = 0.33

Heat Transfer 89
To determine F12, we need to compute the x and y of Figure 3.8 relative

to Figure 3.6:
x = a/c = 1/3 = 0.333
y = b/c = 0.5/3 = 0.167
From Figure 3.8, locating the x-curve approximately, as there is none for
0.333, and then the horizontal axis y value, F12 = 0.017, is estimated from
the vertical axis. Now Equation 3.5 is used to compute the radiant heat
flux received at the target:
q ¢¢ = esT24 F12 = (0.33)(5.67 ´ 10 -11 kW/m 2 -K 4 )(1073 K )4 (0.017 )
= 0.42 kW / m 2
What if the source in this example were four times as large with the
target on the center axis? We have the same problem four times and the
heat flux received would be 4 × 0.42, or 1.68 kW/m2. This is starting to feel
like that from the Sun on a clear day, as that value is at most 1 kW/m2.
Go back to the original example and compute it again for c = 1 m to see the
difference. In doing so, you would get F12 = 0.09 and 2.2 kW/m 2, or about
five times as much. Some might recall that it is commonly said that “radia-
tion falls off as the distance squared.” From 1 to 3 m, this common rule
would say it falls off by 1/9 or would be reduced to 2.2/9 or 0.24 kW/m2. From
our first computation, the value at 3 m should be 0.42 kW/m 2. So, this com-
mon “rule” is approximate and applies only for the ideal case of radiation
coming from the center point to a distance on a spherical shell. If the radius
of the spherical shell is c, its surface area is 4πc2. We will see in moment that
this ideal case can give us an alternative way of estimating flame radiation
to a target.
The computation of radiant heat flux to a target is very important to assess
potential damage and the possibility of a remote ignition from a fire. Since
the computation depends on flame shape and properties, it can be tedious if
not impractical. Fortunately, there is an alternative formula for a discrete fire
that is accurate as long as it is applied more than two flame diameters from
the center of the flame, that is, c > 2l (see Figure 3.9). The formula is
X rQ
q ¢¢ = (3.6)
4pc 2
where
Q is the combustion energy release rate of fire (kW)
Xr is the fraction of energy radiated relative to the total energy released
.
q˝ Target
FIGURE 3.9
Radiation heat flux to a target from a point source representation of a flame.
Xr is not necessarily a constant for a given fuel and generally varies from
about 0.15 to 0.60 for low sooting fuels such as methane and high sooting
flames, such as polystyrene, respectively. Table 3.3 gives some general guid-
ance for some fuels, and Table 3.4 gives some more specific values according
to the diameter of the fire. In particular, the data for gasoline clearly show
that the radiative fraction drops as the diameter increases. For a 1 m diame-
ter gasoline pool, the radiation fraction is 0.60, but for a 10 m fire, the fraction
is 0.10. Figure 3.10 more clearly shows this trend for a number of fuels. The
drop off is due to the cold soot swirling around the core of the luminous
flame for these large fires, as illustrated in Figure 3.11. The soot blocks the
radiation and traps it with the large turbulent flame.
Equation 3.6 is based on uniform radiant emission from a “point source”
approximating a flame. All of the radiant energy released, X rQ , is uniformly
received over a sphere of radius, c, from the flame. The area of the sphere is
4πc2. The ratio of the rate of energy received to the surface area completes
the deduction of Equation 3.6. The formula only applies to a target looking
directly at the “center” of the flame. This simple result has been shown to
have reasonable accuracy (±50%) as long as c > 2l.
TABLE 3.3
Typical Radiative Energy Fractions, Xr
Fuel (l ≥ 0.5 m) Xr
Methanol, methane 0.15–0.20
Butane, benzene, wood cribs 0.20–0.40
Hexane, gasoline, polystyrene 0.40–0.60
Source: Various.
Heat Transfer 91
TABLE 3.4
Specific Radiation Fraction of Combustion Energy for Hydrocarbon Pool
Fires Showing Dependence on Diameter
Hydrocarbon Pool Size (m) Fraction
Methanol 1.2 0.17
LNG on land 0.18 0.164
0.4–3.05 0.15–0.34a
1.8–6.1 0.20–0.25a
20.0 0.36
LNG on water 8.5–15.0 0.12–31.0a
LPG on land 20.0 0.7
Butane 0.3–0.76 0.20–0.27a
Gasoline 1.22–3.05 0.40–0.13a
1.0–10.0 0.60–0.100a
Benzene 1.22 0.37
Hexane — 0.40
Ethylene — 0.38
Source: Mudan, K.S. and Croce, P.A., Fire hazard calculations for large open hydro-
carbon fires, Chap. 3–11, in: DiNenno, P.J., ed., SFPE Handbook of Fire
Protection Engineering, 2nd edn., National Fire Protection Association,
Quincy, MA, June 1995.
a In these cases, the smaller-diameter fires were associated with higher radiative
fractions.
0.5
0.2
Radiative fraction, Xr
0.1 : Gasoline
: Kerosene
: Heptane
: Crude oil
: Hexane
0.05
: Toluene
: Methanol
: Benzene
: JP-4
0.02
0.2 0.5 1 5 10 50
Pool fire diameter, D (m)
FIGURE 3.10
Radiative fraction dependence on pool diameter. (From Koseki, H., Fire Technol., 25(3), 241, 1989.)
FIGURE 3.11
Illustration of soot blocking the core radiation from a large fire.
Example 3.4
For the previous calculation in Example 3.3 for wood, we can make
an alternative estimate using Equation 3.6. While that example’s
geometry is not cylindrical, we will stretch the application to use
Equation 3.6.
Select Xr = 0.36 (based on Figure 3.13) for a wood flame. (Figure 3.13
shows that the slope of the wood crib data is about 14/400. From
Equation 3.6 this would give Xr = 4π(0.9 m)2 × 14/400 = 0.356.) Based on
an equation for flame height in Chapter 7 the 1 m tall wood flame cor-
responds to a firepower of about 110 kW. Therefore, from Equation 3.6,
(0.36)(110 kW) kW
q ¢¢ = = 0.35
4p(3 m)2 m2
This is very similar to the result of 0.42 kW/m2 previously estimated

for the rectangular wood fuel flame 1 m tall, 0.5 m wide and thick. The
difference is due to the approximate nature of both formulas and the
accuracy of the data we are using to represent the wood flame. But
the differences are within the accuracy we can expect from such a
calculation.
In general, fire calculations such as these could have accuracy as poor
as ±50%, but usually more like ±30%. Their value lies not in perfect
accuracy but in the plausibility of the fire scenario they support. For
example, both of these results give heat fluxes more than that of sunlight.
We would expect them to be potentially harmful. The formulas can
guide us in establishing the plausibility of a hypothesis that we pose for
a fire scenario. Is there a hazard or not?
Heat Transfer 93
3.4 Heat Flux as an Indication of Damage

Heat flux causes objects to get hot and possibly damaged or ignited. As a
benchmark, the radiant heat flux from the Sun at the Earth’s surface is nearly
1 kW/m2, at most. Thresholds (minimum values) of heat flux to cause dam-
age under fire conditions are listed as follows:
Pain to bare skin: ~2 kW/m2

Burn to bare skin: ~4 kW/m2
Ignition of objects: ~10–20 kW/m2
These threshold values will cause the results indicated only after a “long”
exposure of many seconds or minutes. The time to cause damage to bare skin
(Figure 3.12) is illustrated from the work of Stoll and Greene.4 In the figure,
the times to achieve pain and a blister skin burn for heat fluxes higher than
the threshold values are graphically shown. The threshold values appear to
be approached after about 30 seconds. These threshold values give general
guidelines for assessing the potential damage from the fire exposure heat
flux. Therefore, it is very important to be able to estimate fire heat fluxes.
30
25
Threshold levels:
Blister > 4 kW/m2
20
Irradiance (kW/m2)
Pain > 1.0 kW/m2
15
Pa
in
10 5° Blis
ter
(4
C
sk
in
te m
pera
5 ture)
0
0 5 10 15 20 25 30
Time (s)
FIGURE 3.12
Incident radiant heat flux effect on bare skin. (From Stoll, A.M. and Greene, L.C., J. Appl.
Physiol., 14, 373, 1959.)
Target
c
25
c = 0.9 m
Target heat flux (kW/m2)
20
15
10
Wood crib fires
Plastic crib fires
0
0 200 400 600 800 1000 1200
Combustion energy release rate (kW)
FIGURE 3.13
Radiant heat flux to a target facing 0.9 m from the center of wood and plastic crib fires.
(From Quintiere, J.G. and McCaffrey, B.J., The burning of wood and plastic cribs in an
enclosure, Vol. I, NBSIR 80-2054, National Bureau of Standards, Gaithersburg, MD,
September 1980, p. 118.)
Unfortunately, as flame radiation is most responsible, it is the least predict-

able. However, experimental results can give us some empirical means to
make fire estimates.
Figure 3.13 shows measurements of radiant heat flux received by a target
approximately 1 m from wood or plastic cribs. The results are consistent with
Equation 3.6 and yield a value of Xr between 0.30 and 0.40 for both the fuels.
As cribs represent structural furniture, these radiative fractions might be
used accordingly for furniture as a first approximation.
3.5 Heat Flux Due to Smoke in Room Fires

Incident radiative heat fluxes to the floor in a door-vented room fire are
shown in Figure 3.14. These data come from the same tests of the wood and
plastic cribs of Figure 3.13. The heat fluxes are plotted as a function of the
Heat Transfer 95
30
Dark symbols—plastic cribs
Incident radiant heat flux to the floor, q˝floor (kW/m2) Open symbols—wood cribs
Smoke layer
.
20
1m Target
10
0
100 200 300 400 500 600 700 800
Compartment average smoke layer temperature (°C)
FIGURE 3.14
Incident radiative heat flux to the floor in a compartment due to hot smoke. (From Quintiere,
J.G. and McCaffrey, B.J., The burning of wood and plastic cribs in an enclosure, Vol. I, NBSIR
80-2054, National Bureau of Standards, Gaithersburg, MD, September 1980, p. 118.)
upper layer smoke temperatures. Despite the variation in the type of fuel
(wood and plastic) and the variation of ventilation of air through different
doorway sizes, the results are almost solely dependent on the average
temperature of the smoke layer. This dependence also applies to measure-
ments of total (convective plus radiative) heat flux to a ceiling target as
shown in Figure 3.15. The reason for the sole dependence of these results on
the smoke layer temperature can be explained by the radiative view factor of
the smoke layer to the floor or ceiling. As might be inferred by examining the
field of view of both the ceiling and floor targets, the smoke layer dominates
the field of view compared to any radiative contribution from the crib fire.
An observer on the floor looking up sees the bottom of the smoke layer in
about 50% of their forward field of view, as this layer is at about half the room
height. This is a configuration (view) fraction of 0.5. The same observer, still
looking up, only has a very small view angle to the flaming crib. Of course,
just next to the crib, it would be highest. But even that close, it would be less
than 0.5. This suggests that the smoke is the principal source of the heat flux
to the floor. In contrast, the ceiling target has a view factor of 1, as it sees the
entire smoke layer.
Incident total heat flux to the ceiling, qc̋eiling (kW/m2)

Dark symbols—plastic cribs
Open symbols—wood cribs
100 (Shape and size of symbol
pertains to number of cribs
and size of doorway, respectively.)
.
Target
Smoke layer 1m
50
0
100 200 300 400 500 600 700 800 900
Compartment average smoke layer temperature (°C)
FIGURE 3.15
Incident total heat flux to the ceiling in a compartment due to hot smoke. (From Quintiere,
J.G. and McCaffrey, B.J., The burning of wood and plastic cribs in an enclosure, Vol. I, NBSIR
80-2054, National Bureau of Standards, Gaithersburg, MD, September 1980, p. 118.)
Example 3.5: Confirm This Result for the Floor Heat Flux
Take a temperature of 600°C. As the smoke layer for a normal size room
would be about 1 m thick, or more, we will assume it to have an emissivity
of 1. Then from Equation 3.5, q ¢¢ = e s T24 F12
q ¢¢ = (1)(5.67 ´ 10 -11 kW/m 2 -K 4 )(600 + 273 K )4 (0.5) = 16.4 kW/m 2
This compares to about 20 kW/m2 as measured in Figure 3.14. Additional

radiation from the crib fire likely contributes to the difference.
Consider the same smoke layer temperature for estimating the ceiling
heat flux. Here, the configuration factor is 1, as an observer on the ceiling
looking down would only see the top of the smoke in their field of view.
From the calculation given earlier, this readily gives twice the value
to the floor or 32.8 kW/m2. But this compares to about 60 kW/m2 from
Figure 3.14. The difference here is the convection heat transfer that must
be added in. These heat flux measurement sensors are cold at about 25°C.
From Table 3.2, we might estimate the convective heat transfer coefficient
for a plume hitting the ceiling as about 25 W/m2-K. The convective con-
tribution is then estimated from Equation 3.2, q ¢¢ = h(T2 - T1 ):
æ W 10 -3 kW ö
q ¢¢ = ç 25 2 ´ ÷ (600 - 25) K = 14.4 kW/m
2
è m -K W ø
Heat Transfer 97
This gives a total of 47.2 kW/m2, about 20% too low. But again, we have
neglected any radiative contribution from the crib fire. This is within
the range of accuracy we can expect. Thus, we have nearly explained the
experimental results by theoretical calculations.
The threshold of piloted ignition (thin objects at 10 kW/m 2 and thick
objects at 20 kW/m2) suggests that objects begin to have a high propen-
sity to ignite when the smoke layer temperature attains 400°C–600°C.
This is why flashover is often associated with smoke layer temperatures
at about 500°C–600°C. It is a general operational rule for static fires, but
the onset of flashover has a lower critical smoke layer temperature for
spreading fires. The sudden increase in spread will occur at lower smoke
layer temperatures. We will have more to say about this in Chapter 9.
Figures 3.14 and 3.15 are representative of developing room fires in which
there is a distinct smoke layer and distinct central fire. Once more objects get
involved, flames and hot smoke will try to fill the room according to the fuel
and air supply available. The flames and smoke layer can reach 1200°C or
more. As the flames will be thick (>1 m), their emissivity would be near 1. In
this case, the radiative heat flux could achieve 250 kW/m2 in the room. It is not
uncommon to achieve heat flux levels of 100–250 kW/m2 in fully developed
room fires.
Figure 3.16 is presented for illustration to show how radiation and con-
vection heat fluxes can contribute to fire conditions when heating a cold
object. Turbulent convective heat transfer coefficients in room fires can
typically range from 5 to 50 W/m2-K. The radiation is characterized by its
1000
Blackbody radiation
Heat flux to a surface at 25°C, kW/m2
100
Convection at h = 50 W/m2-K
10
Convection at h = 5 W/m2-K
1
0 200 400 600 800 1000 1200 1400 1600
Source temperature, °C
FIGURE 3.16
Overview of heat fluxes from a fire source to a cold object.
maximum potential for a blackbody. This graph puts in perspective the role
of convection and radiation in fire conditions. Even at relatively low source
temperatures, radiation is significant compared to natural convection
(h = 5 W/m2-K). After about 500°C, radiation truly becomes the dominant
factor in heat transfer in fire.
3.6 Heat Flux from Flames

Measurements of the incident heat flux from flames to a surface can be use-
ful in giving us a way to assess their fire hazard. The heat flux will consist
of the total of the convective and radiative components. Thin laminar flame
will have a higher temperature than turbulent flames (about 2000°C over
1000°C, respectively). Convection will dominate these laminar flames. Large
thick flames will be turbulent, and radiation will dominate them. Lattimer7
has assembled an assortment of turbulent flames and their heat flux behav-
ior to surfaces. Only their peak values within the flame region will be listed
here. They are listed in Table 3.5. As they can range from 20 to 200 kW/m2, it
is imperative to understand their associated fire scenarios. Commercial heat
flux sensors have typically been used to obtain these measurements.
It is noteworthy to recognize that small laminar flames, such as candle and
lighter flames, can achieve heat fluxes of 50–100 kW/m2 to a surface at their
tip. From the basic information given in this chapter, it is possible to explain
these data, at least qualitatively.
TABLE 3.5
Turbulent Flame Peak Heat Fluxes to a Surface within the Flame
Description Heat Flux (kW/m 2) Comments
Corridor ceiling burning ~20 Buoyancy suppresses turbulence,
thin flame
Ceiling burning, radial flame 20–30 Buoyancy suppresses turbulence,
thin flame
Wall burning flame 25–40 Increases with flame height
Floor flame impinging on ceiling 80–90 High velocity, thick flame at center
Square floor fire against a wall 50–100 Increases with fire size
(e.g., 50–500 kW)
Window flames from fully 20–200 Increases with room fire size
developed room fires (e.g., 5–10 MW) and size of
(0.5 m above) window
Objects immersed in large pool 100–200 High temperature, thick flames
fires (10–160 m2)
Source: Lattimer, B.J., Heat fluxes from fires to surfaces, Sect. 2, Chap. 14, in: DiNenno, P.J., ed.,
SFPE Handbook of Fire Protection Engineering, 4th edn., National Fire Protection
Association, Quincy, MA, 2008.
Heat Transfer 99
3.7 Summary
Heat is energy transfer due to a temperature difference. Heat transfer is a
significant process in fire. It accounts for the vaporization of the fuel, growth
of the fire, and its damage. Heat flux is the rate of heat (energy) transfer per
unit area normal to its direction. We sense heat flux as we might be exposed
to a fire.
There are three categories of heat transfer: (1) conduction, a molecular phe-
nomenon; (2) convection, conduction in a moving fluid; and (3) radiation, an
electromagnetic phenomenon. The laws of Fourier and Planck provide the
bases for conduction and radiative heat transfer calculations, respectively.
Conduction heat flux is directly proportional to the difference of tempera-
ture, while radiation depends on the fourth power of the absolute temperature.
Absolute zero is the lowest temperature possible.
Heat flux is a key parameter in assessing the potential damage by a fire.
Threshold heat flux levels are approximately 2 kW/m2 for pain to bare
skin, 4 kW/m2 for a burn to bare skin, and 10–20 kW/m2 for the ignitions
of objects. Flames can cause heat fluxes of 20–200 kW/m2, depending on
their size and shape.
The smoke layer temperature in compartment fires of 500°C–600°C is
frequently taken as a benchmark for the initiation of flashover.
Review Questions
1. What is the difference between the absolute temperature scale and the
Celsius scale?
2. What distinguishes heat from other forms of energy?
3. How long will it take heat to penetrate a 3 in. thick concrete wall?
4. Estimate the convective heat flux from a match flame to the surface of
its wood. The flame is roughly 1900°C and the wood pyrolyzes at 350°C.
The convective heat transfer coefficient can be taken as 50 W/m2-K as an
estimate.
5. Estimate the convective heat flux from a gasoline turbulent pool fire to the
evaporating gasoline surface. Gasoline evaporates at roughly 33°C during
burning conditions. Its turbulent flame temperature is 800°C on average.
Take the convective heat transfer coefficient as 20 W/m2-K (see Table 3.2).
6. For the fire in Problem 5, estimate the radiative heat flux to the liquid
surface. Assume a flame emissivity of 0.5. For a target on the surface how
much of the flame does it see?
7. Compute the radiant heat flux 20 m from an 8 MW gasoline pool fire.

Assume the fire is 3 m in diameter. If the fire diameter doubles, the
firepower will increase by a factor of 4. Use Figure 3.10 to make these
estimates.
True or False
1. Convective heat transfer is a type of conduction.
2. Conduction depends on molecules, radiation on electromagnetic energy,
and convection on ether rays.
3. The emissivity of a surface should be less than one, but for a large flame
it can easily be one.
4. Thermal conductivity and thermal diffusivity are properties of a solid
controlling its heat transfer.
5. A threshold for a skin burn could be 4 kW/m2, while at 10 kW/m2 we
could have ignition of clothing.
6. Heat is not a form of energy.
7. A guide for the onset of flashover in a room fire is when the smoke
reaches 500°C–600°C.
8. Laminar small flames always have lower heat fluxes than turbulent
flames.
9. Zero absolute temperature indicates that there is no more energy in the
material.
10. Mass can change into energy in a chemical reaction.
Activities
1. Find out how much time is needed to melt an ice cube. Look up
the energy needed to melt ice (heat of fusion, J/g). Estimate the convec-
tive heat transfer coefficient for a cube sitting in still air. Note that the ice
temperature is always 0°C and record the room temperature. Compare
your computed time to the actual time. Explain the differences.
2. Make a heat flux meter. Connect a thermocouple to a thin (1/8 in.) metal
disk about 1 inch in diameter. Insulate the back of the disk very well.
Roughen the disk and deposit soot on it from a candle flame or paint
it flat black. This will tend to make its emissivity 1. If the disk is thin
enough, when the meter is exposed to a higher heat flux, its surface will
get hot. If no heat is lost through the insulation, all of heat received will
Heat Transfer 101
radiate away. The heat flux radiated away will be exactly equal to that
received. The emitted heat flux is computed by using Equation 3.3 and
the measured temperature. Using the heat flux meter in cold air will
cause a convective heat loss as well, and this would need to be consid-
ered as a source of error. Or this convective heat flux can be added to the
emitted radiant heat flux. The total is the heat flux received.
References
1. D. Drysdale, An Introduction to Fire Dynamics (New York: Wiley, 1985), p. 36.
2. D. C. Hamilton and W. R. Morgan, Radiant-interchange configuration factors,
NACA Tech. Note 2836, U.S. Government (Washington, DC: December 1952), p. 78.
3. K. S. Mudan and P. A. Croce, Fire hazard calculations for large open hydrocar-
bon fires, Chap. 3–11 in SFPE Handbook of Fire Protection Engineering, 2nd edn.,
edited by P. J. DiNenno (Quincy, MA: National Fire Protection Association,
June 1995).
4. H. Koseki, Combustion properties of large liquid pool fires, Fire Technology, 25,
3 (August 1989): 241–255.
5. A. M. Stoll and L. C. Greene, Relationship between pain and tissue damage due
to thermal radiation, Journal of Applied Physiology, 14 (1959): 373–382.
6. J. G. Quintiere and B. J. McCaffrey, The burning of wood and plastic cribs in an
enclosure, Vol. I, NBSIR 80-2054 (Gaithersburg, MD: September 1980, National
Bureau of Standards), p. 118.
7. B. J. Lattimer, Heat fluxes from fires to surfaces, Sect. 2, Chap. 14, in SFPE Handbook
of Fire Protection Engineering, 4th edn., edited by P. J. DiNenno (Quincy, MA:
National Fire Protection Association, 2008).
4
Ignition
Learning Objectives
• Explain the difference between piloted ignition and autoignition

• Define evaporation
• Explain flashpoint and fire point
• Describe the ignition processes in liquids and solids
• Describe the concept of ignition temperature for solids
• Use formulas to predict the ignition time of solids
4.1 Introduction
In Chapters 4 through 6, we discuss the essence of fire growth, which is
composed of ignition, flame spread, and burning rate. These are distinct fire
processes that may have some features in common but must be put together
to establish fire growth. We will clarify their common characteristics and
their differences. This chapter discusses ignition, the start of fire growth.
Ignition is very important, as without it there would be no fire. It is a step
that is easily recognized for flaming ignition and is subtle to discern for
ignition to smoldering. Ignition can occur in different ways, but always heat
should be added. Sometimes, this heat is localized in the form of a small
flame or distributed over a region of the fuel.
103
4.2 Piloted Ignition and Autoignition

Ignition was discussed in Chapter 2 for gaseous fuels. It has two forms:
piloted ignition and autoignition. The former is the process of initiation and
flame propagation in premixed fuel systems. The minimum condition for
piloted ignition occurs at the lower flammable limit (LFL)—that concentra-
tion of fuel that allows propagation with a small sufficient heat source (pilot).
This pilot can be an electric arc that possesses a very high temperature or a
small flame. We saw this process occurs in Chapter 2 when the candle flame
was blown out and the white vapor trail was then ignited. (What happens
when we ignite the candle by applying a flame to the wick?)
The second form of ignition is autoignition that occurs without any pilot.
The fuel must still be within a specific concentration range, typically between
the lower and upper flammability limits. Also, chemical kinetic processes
must generate energy production to exceed the heat loss of the fuel gas–air
mixture. The piloted ignition and autoignition of liquid and solid fuels occur
in an identical fashion for their evaporated or decomposed gas–air mixtures.
This is illustrated in Figure 4.1. The gaseous fuel evolves by heat transfer and
mixes with air. A pilot can cause the ignition, or the hot surface of the solid in
Figure 4.1 might cause ignition. The latter would be autoignition and would
occur in the case of radiant heating of wood. In both cases, a sufficient fuel
gas concentration must be generated.
4.3 Evaporation in Liquids

Liquids generate fuel as a vapor or gas. This process continuously occurs at
the surface. The liquid in Figure 4.1 evaporates at a sufficient rate to form
a concentration near the pilot source at the LFL to allow ignition. The rate
Decomposed fuel Hot surface

Pilot
Radiant
Evaporated fuel heat
Air Air
Air
(a) (b)
FIGURE 4.1
Ignition processes in (a) liquids and (b) solids.
Ignition 105
of evaporation is controlled by the liquid temperature. Let us examine the

process of evaporation for water, as an example. Water will evaporate rela-
tively slowly under normal room temperature. At the liquid temperature, or
energy level, only a relatively small amount of water molecules can break out
of the surface and enter the adjacent air. The escaping vapor creates a pres-
sure that contributes to the total pressure of the atmosphere. Atmospheric
pressure is normally 14.7 psi or about 105 Pa (1 bar). This “partial” pressure
of the water vapor is contributing to the total 1 bar pressure is called vapor
pressure.
The vapor pressure can be more properly defined as the pressure where
a liquid and its vapor are in equilibrium (a state of balance) for a given
temperature. For pure liquid substances, this relationship between vapor
pressure and temperature is a property of the material. In general, the
vapor pressure is a strong function of temperature and increases sharply
as the temperature rises. For example, water at 25°C has a vapor pres-
sure of 0.032 bar (or 0.032 × 14.7 psi = 0.47 psi), which is relatively very
small. But as the liquid temperature is increased, more molecules escape
(evaporation) until 100°C (212°F) is reached. Here, the vapor pressure of
water is now about 1 bar (or precisely 1.014 bar). Now all of the escap-
ing water vapor molecules make up the total pressure of the surrounding
pure air above the liquid. This means that the vapor at the liquid surface
no longer contains any air, or in other words, its concentration is 100%. At
normal atmospheric pressure, this is the maximum temperature that the
liquid water can achieve. In fact, the water vapor concentration (in molec-
ular or molar terms) is equal to the vapor pressure divided by the total
atmospheric pressure (1 bar). At 25°C, water has a concentration of 3.2%.
This concentration is commonly referred to as volume concentration as it
corresponds to the volume of water vapor in a measuring device formed
by extracting it from the air–water mixture and allowing it to expand at a
normal atmospheric pressure (see Figure 4.2).
Air at 1 atm (1 bar, 14.7 psi) and 25°C
Vapor pressure, Vapor pressure,

0.032 bar or 0.47 psi Water vapor in air 1 bar or 0.47 psi
volume volume
at the liquid surface
concentration, 3.2% concentration, 100%
Water at 25°C Water at 100°C
FIGURE 4.2
Water vapor in air.
Dry air at 1 atm and 25°C

0% water vapor, RH = 0%
Diffusion of
water vapor
Water vapor
concentration = 3.2 %
RH = 100%
Water at 25°C
FIGURE 4.3
Relative humidity and diffusion.
For water in air, relative humidity (RH) is a measure of how much water
vapor can be contained in the air. For example, at 25°C if the concentration
of water vapor were 1%, its RH would have been 1/3.2 × 100% or about 31%.
(In other words, 31% of the volume of water vapor that could be contained in
the air is present.) Water vapor will continually leave the surface of the liq-
uid and enter the bulk air at this RH. If the air is initially “dry” (0% RH), the
liquid will continue to evaporate as long as there is a higher concentration
of vapor at the surface than that of the bulk air. The transfer of water vapor
from the surface to the air occurs through diffusion. (Diffusion is the motion
of a species [e.g., vapor] from a high to low concentration.)
The air above the liquid can enhance this transfer process over that of pure
diffusion. This effect due to airflow is called convection with respect now
to mass transfer analogous to convection in heat transfer. If the air reaches
100% RH, there will now be no difference between the concentration of the
water vapor at the liquid surface and in the bulk air, so evaporation will cease
(see Figure 4.3). This is why laundry will not dry in air at 100% RH.
4.4 Liquid Fuels

Pure liquid fuels behave exactly as water. Mixtures of fuels will evaporate
depending on their components. The more volatile components (higher vapor
pressure) will evaporate first in hydrocarbon mixtures such as gasoline.
Ignition 107
Let us stick to a pure liquid model for ease of explanation for ignition.
But remember, in liquid fuel mixtures, it is the more volatile component con-
trolling ignition.
4.4.1 Piloted Ignition

At piloted ignition, the surface concentration needs only to achieve the
LFL. At this concentration, the corresponding surface temperature is called
the flashpoint (TFP). The pilot at the LFL initiates a propagation of flame in
the atmosphere above the liquid surface. This flame is a premixed flame.
It will propagate as long as the mixture is in the flammable range. Hence,
the name flashpoint applies, as this process would appear as a “flash.” The
flash could impart more heat to the liquid surface causing its temperature
to rise. This action would cause an increase in the vapor pressure with more
evaporation. If the evaporation rate continues to increase, it is possible that
a stable diffusion flame will form above the liquid surface. The minimum
liquid temperature to allow a stable diffusion flame is called the fire point
(fire point is the minimum liquid surface temperature needed by the liquid
to sustain a diffusion flame).
For example, n-decane, a liquid at normal room temperature, has a
measured flashpoint between 44°C and 52°C and a fire point of roughly
62°C–66°C. (The variations in these results are due to the way the measure-
ment is made. The location of the pilot and the corresponding concentration
of vapor will affect the results.) The maximum temperature the n-decane can
achieve is 174°C, its boiling point (boiling point—the maximum temperature
a liquid can achieve in air).
The boiling point is the temperature corresponding to when its vapor
pressure equals to the exposed atmospheric pressure. Here, the fuel con-
centration is 100% at the surface. Once a diffusion flame is formed, the heat
transfer from the flame to the surface is sufficient to achieve nearly the boiling
point. This fact allows the boiling point to be the assumed liquid surface
temperature in a fire.
The process of piloted ignition has led from a premixed flame to a diffu-
sion flame. This process takes a little time. Therefore, the time required for
“ignition” to occur is somewhat problematic, as it could be designated at
either the flashpoint or the fire point. However, the time difference between
these events is normally small.
The range of these critical temperatures for some liquid fuels is shown
in Table 4.1.
Because the normal atmospheric temperature is roughly 22°C, everything
above ethanol in Table 4.1 can easily ignite with a pilot under normal condi-
tions. Extra heat should be provided to n-decane, kerosene, and higher flash-
point liquid fuels to bring their surface to the flashpoint.
TABLE 4.1
Critical Temperatures for Liquid Fuels in °Ca
Hot
Fire Boiling Surface
Liquid Formula Flashpoint Point Point AIT Ignition
Propane C3H8 −104 — −42 450 —
Gasoline ~C8H18 ~−45 — ~33 371 >560
Acrolein C3H4O −26 — 53 235 —
Acetone C2H6O −18 — 56 465 —
n-Heptane C7H16 −4 −1 98 233 670
Methanol CH3OH 12 17 64 385 690
Ethanol C2H5OH 13 — 78 363 690,717
n-Decane C10H22 44–52 62–66 174 210 —
Kerosene ~C14H30 49 — 232 260 650
m-Creosol C7H8O 86 — 203 559 —
Formaldehyde CH2O 93 — 97 430 —
Motor oil — 186,216 224 300–560 351 310,740
Sources: Zabetakis, M.G., Flammability characteristics of combustible gases and vapors,
Bulletin 627, Bureau of Mines, U.S. Department of the Interior, Washington, DC, 1965;
Babrauskas, V., Ignition Handbook, Fire Science, Society of Fire Protection Engineers,
Publisher, 2005; Tewarson, A., Thermophysical and fire properties of automobile plastic
parts and engine compartment fluids, Vol. III, Tech Report #0003018009, FM-Global,
Norwood, MA, October 2005.
a Results can vary due to the test used; AIT data commonly from the spherical vessel test.
4.4.2 Autoignition
The process of autoignition is a slightly different mechanism compared to
piloted ignition. As indicated in Figure 4.4, autoignition occurs at a concen-
tration generally between the upper (UFL) and the lower (LFL) flammability
limits. The stoichiometric state in which all the fuel vapor and all the oxygen
Fuel vapor pressure or
concentration in air
Liquid Vapor
region region
UFL
Piloted ignition Autoignition
LFL
TFP AIT
Temperature of fuel–air mixture
FIGURE 4.4
Conditions needed for ignition.
Ignition 109
in air are consumed is most favorable for autoignition. Autoignition is synony-

mous with spontaneous combustion or “self-ignition.” This ignition inherently
occurs due to an imbalance, triggered at the critical temperature, between the
chemical energy production and the local heat loss in the gas. The temperature
at which this occurs is the autoignition temperature (AIT). Once a flame results
in a small region, the process becomes much like piloted ignition, as a pre-
mixed flame is propagated through the remaining flammable mixture.
As seen from Table 4.1, the AIT is mostly higher than the boiling point.
For a liquid to autoignite, some hot surface must heat the vapor–air mixture
to the AIT. But this temperature is not unique and depends on many factors
related to the hot surface heat transfer. The values given in Table 4.1 listed
as AIT undoubtedly came from a standard test in which an open spherical
vessel was placed in an oven. The vessel size and concentration–temperature
distribution of the mixture makes the measurement not so precise. Bigger
vessels give a lower AIT in such a test. ASTM E-659 is an example of a stan-
dard test for determining the AIT for liquid chemicals.
In another configuration of heat, such as that of a hot surface exposed
to the mixture, gives much different results than the spherical vessel oven
test. This scenario can occur when a liquid fuel is sprayed on a hot surface.
Measurements have also been made for ignition by a hot surface. The sur-
face temperature is typically much higher than the AIT (by the vessel test).
The temperature of the surface to cause ignition depends on the nature of the
surface, the way liquid contacts the surface, and the air velocity. Some values
of “hot surface ignition” are given in Table 4.1, but they should be taken as
rough indications, not definitive values.
Note that the AITs of the liquid fuels are much higher than their flash-
points and boiling points. Indeed, all of the liquid would be vaporized at
these AITs. One exception to this difference is asphalt in which the AIT is
nearly the same as its boiling point, about 480°C. An example of a fire involv-
ing asphalt for a roofing application illustrates an interesting scenario. A
heated kettle was used to liquefy the asphalt. As its temperature became too
high and into the boiling range, the liquid ignited when it was tapped from
the kettle. As the liquid came in contact with air, the vapor–air mixture was
now at the AIT. This incident was initially suspected as an arson fire.
Under actual fire conditions, both autoignition and piloted ignition apply
as competent scenarios. However, the piloted ignition case is probably more
predominant once a flame exists in accidental fires. The vapor trail of a liquid
fuel spill may travel some distance before finding a pilot source. This is a
common scenario due to spilling a liquid fuel near a water heater with a pilot
flame as illustrated in Figure 4.5.
In another ignition scenario, a halogen light bulb was suspected of being
the ignition source during paint spraying in a confined space. Indeed, it was
determined that the AIT temperature of the paint (by a vessel test method)
exceeded the light-bulb temperature. However, this scenario would be a hot
surface ignition case. Spraying paint on the bulb and evaporating the paint
Flame propagates back Hot

water
pilot
Heavy fuel or
vapors
migrate
FIGURE 4.5
Evaporating liquid, remote ignition.
within its lens housing indicated no ignition. The hot surface ignition tem-
perature was much higher than that reported by the vessel test.
Another surface not sufficient to trigger ignition is the lit cigarette for
gasoline. This scenario depicting in countless movies has been shown to
not occur in repeated tests. Indeed, several years ago in a student project
this was shown dramatically by evaporating gasoline at the bottom of a
vertical cylindrical tube open at the top. Holes in the vapor region of the
tube allow a competent pilot flame to be inserted to indicate the region of
flammability. When a lit cigarette was placed in the same region, even with
a device to draw on the cigarette, no ignition occurred. It has also been
reported that numerous tests of a lit cigarette tossed into a pool of gasoline
showed no ignition.
4.5 Solid Fuels

Flashpoints for liquids can be measured with reasonably good precision
and can also be computed from theory. But the ignition temperatures of
solids are not precisely determined. They depend on the decomposed
fuel concentration, its components, and the way the solid is heated. The
formation of porous char may contribute to its variation. As a result,
solid ignition temperatures published in the literature can only be taken
as rough indications. Often, these data depend on the experimental con-
ditions or the analysis used to derive them. Moreover, it is very difficult
to directly measure the surface temperature up to the point of ignition.
Despite this limitation on establishing a definitive ignition temperature for
a solid fuel, experimental results will give us some reasonable approximate
values. These and other results can be used to make quantitative estimates
on the prospect of ignition and the time it can take. We will examine some
Ignition 111
predictive models for solid fuel ignition based on the use of effective or
approximate ignition temperatures and other properties.
Let us first examine the processes involved in the ignition of a solid more
closely. Solid fuels tend to be mostly polymers, either synthetic or natural.
Wood is a natural polymer. A polymer is a compound made up of large
molecules consisting of linked chains of a repeated chemical structure.
The chemical structure is called a monomer—a repeated chemical structure
in a polymer.
For example, the chemical structure alone as C2H4 is ethylene, a gas at
normal atmospheric conditions. But the structure linked as the mono-
mer—C2H4 —is now polyethylene, a solid under normal conditions.
In order to get polyethylene to ignite, it must first be vaporized. While
a pure substance, such as ethylene, remains the same chemical while it is
a solid, liquid, or gas, this is not true for polyethylene. When polyethylene
is heated, it will start losing its structure melting at about 120°C–145°C
and will have a reported flashpoint of greater than 220°C. In reality, the
melt may not be a true liquid to a polymer scientist but just a “softened”
solid. Above 220°C where ignition is possible, vapor must be generated.
The vapor is due to thermal decomposition (or pyrolysis—the production
of vapor by heat from a decomposing solid). The polyethylene is ther-
mally broken into other chemical compounds that make up the vapor
produced.
For some polymers, this vapor could ideally be the monomer, but mostly
it is not. The vapor is likely a complex mixture of hydrocarbons. The vapor
produced by the decomposition of the solid is not the same fuel chemically
as in pure liquid evaporation. Whatever it is, it must still achieve the LFL of
that mixture to be ignited in piloted ignition or must be in at a flammable
concentration and at the AIT for autoignition. The flame processes are the
same for the liquid and solid fuels; only, the vaporization process is differ-
ent. The liquid retains its chemical structure, whereas the solid does not as it
decomposes to a vapor. The vapor pyrolysis products for wood will consist
of many compounds. Some will condense in the atmosphere as fine drop-
lets (tars), displayed as white “smoke.” Wood, like some synthetic polymers,
will not totally decompose to a vapor but will leave behind a carbonaceous
porous char.
4.5.1 Ignition of Wood as an Example

Wood can exemplify the full range of factors affecting the ignition of solids.
First, wood can contain absorbed water. When the wood is heated to 100°C,
water vapor will diffuse both into and out of the wood. The released water
vapor will increase the LFL of the air–fuel vapor mixture and retard ignition.
In addition, the water vapor can be trapped in pockets of virgin wood and
the char. This trapped vapor can lead to centers of locally high pressure that
are usually relieved by cracks. The process is much the same as the more
explosive moisture cracking in heated concrete known as spalling (spalling—

explosive breaking of concrete due to the vaporization of trapped water or
other liquids).
Second, the formation of char on the surface can play the role of an insu-
lator to heat conduction into the virgin wood. The char is also a source of
chemical energy as the char oxidizes from air diffusing to the surface.
While this oxidation will release insignificant energy at low temperatures, it
could eventually lead to full-blown ignition to smoldering. This solid-phase
ignition might be termed glowing ignition as the surface will likely turn red
(glowing ignition—start of smoldering at a high enough temperature to emit
red light.)
In theory, the onset of smoldering ignition is an imbalance between the
chemical energy released and the heat losses that result in a sudden jump
in temperature at the surface. The energy released by smoldering can play a
role in enhancing the piloted ignition and autoignition of wood.
Third, the decomposition of wood is not like polyethylene, a melting
solid. Wood thermally decomposes to char and vapor. The vapor will be
composed of a complex array of molecules from low volatility to heavy tars.
The mixture of vapors depends on the rate the wood is heated, so the LFL
of the mixture will not be a precise constant for wood. The LFL is known to
theoretically depend on the heat of combustion. The heat of combustion for
wood volatiles can range from about 10 to 15 kJ/g, so a ±25% variation can
be expected in the LFL for the wood volatiles. In general, the LFL is low—
typically about 1%–3%. So, a small quantity of fuel vapor is only required
for piloted ignition in all cases. In a pure liquid, the transition to vapor is
dependent on a vapor pressure–temperature property relationship. This is
a precise property. For a solid, vaporization depends on the chemistry of
pyrolysis. Yet, both vaporization processes are a strong function of tempera-
ture. As the wood nears a critical temperature, it will produce the needed
vapor concentration for ignition. This temperature at the surface to achieve
the minimum concentration for piloted ignition or autoignition, we term as
the ignition temperature (ignition temperature—effective surface tempera-
ture to cause ignition of the emitted vapors).
4.5.2 Ignition Temperature and Critical Heat Flux

Some experiments on the wood heated by a radiant source in air will serve
to illustrate the conditions critical to ignition. These experiments come from
the work of Spearpoint4 and Boonmee.5 Critical surface temperature to cause
ignition of redwood heated by radiation was measured for the different
modes of ignition: (1) glowing ignition, (2) piloted ignition, and (3) autoigni-
tion. The onset of glowing ignition was designated when the wood surface
temperature exceeded that of an inert insulator exposed to the same radiant
heat flux. This rise showed that the wood surface was now producing sig-
nificant energy due to oxidation. These results are indicative of the variation
Ignition 113
800
Heating parallel to grain
700 Heating perpendicular
to grain
Auto
600
Ignition temperature (°C)
Auto
500 Glowing
400
Piloted
300
Piloted
200
100
0 10 20 30 40 50 60 70 80
Incident radiant heat flux (kW/m2)
FIGURE 4.6
Surface temperatures at ignition for redwood considering flaming (auto, pilot) and smoldering
(glowing), when heated perpendicular and parallel to the wood grain. (From Spearpoint, M.J.
and Quintiere, J.G., Fire Safety J., 36, 391, 2001; Boonmee, N. and Quintiere, J.G., Proc. Combust.
Inst., 29, 289, 2002.)
in ignition temperature for a charring material. Figure 4.6 displays the trend
lines of the temperature data. It is seen that for piloted ignition and autoig-
nition the ignition depends on whether the wood was heated parallel or
perpendicular to its grain. The ignition temperatures are lower for parallel
heating as it is easier for the volatiles to escape along the wood grain paths.
For both heating directions, the AIT is higher than that of piloted ignition.
The glowing ignition temperatures could only be noticed below 40 kW/m2,
as the time to ignite by flaming was too fast above 40 kW/m2 to discriminate
glowing before flaming. The glowing temperature is always between piloted
and AIT. Below 40 kW/m2, the mechanism for autoignition is distinctly a tran-
sition from smoldering to flaming. Thus, at low heat flux, smoldering will
persist for a long time before flaming ensues. As the char layer builds up and
further insulates, the surface temperature can rise to as much as 900°C before
flaming occurs. This hot surface is now the heat source for ignition to flaming.
The times for ignition of this redwood are shown in Figure 4.7. Above
40 kW/m2, it is difficult to distinguish a difference between autoignition and
piloted ignition, as the ignition times are of the order of about 10 seconds
or less. Below 40 kW/m2, glowing ignition can slightly precede piloted igni-
tion, but the difference is not significant. The coincidence of these events
must follow, as the onset of char needed for glowing ignition also denotes
the time of decomposition to fuel vapor. While glowing ignition might occur
104
Flaming ignition from glowing state
1000
Ignition time (s)
100
Autoignition
Piloted and glowing

(approximately coincide)
10
1
0 10 20 30 40 50 60 70
Incident radiant heat flux (kW/m2)
FIGURE 4.7
Ignition times for redwood as a function of incident radiant heat flux. (From Spearpoint, M.J.
and Quintiere, J.G., Fire Safety J., 36, 391, 2001; Boonmee, N. and Quintiere, J.G., Proc. Combust.
Inst., 29, 289, 2002.)
in 100 seconds at about 25 kW/m2, the subsequent transition to flaming or

“autoignition” can take over ½ hour. Below 20 kW/m2, it appears that this
transition will not occur. But an increase in air velocity or simply waiting
longer could discount this limit. The critical limits for piloted and glow-
ing ignition appear to be about 10 kW/m2 for ignition times approaching
about 1 hour. The critical heat flux (CHF) for direct flaming in autoignition
of the redwood, without the transition from smoldering, is about 40 kW/m2.
In general, under radiant or direct flaming ignition, there will be a CHF
that just allows the surface temperature to reach its ignition temperature.
This will depend on the nature of heating and the convective state of the
environment.
4.6 Time for Flaming Ignition

The time for flaming ignition of liquids and solids involves three steps.
The first is predominately the longest. This is the time to heat the liquid or
solid to achieve sufficient vaporization. If the liquid is motionless, conduc-
tion heat transfer solely controls the heating within the liquid, as well as
a solid. The agent for heating could be radiant, convective, or conductive
Ignition 115
heating at the surface and could involve direct flame impingement.

The incident heat flux of this agent is significant as it must be sufficient to
achieve vaporization. Such CHFs are indicated for redwood in Figure 4.7.
The second and third steps are usually small. The second step is the time
for the fuel vapor and air to mix into a flammable mixture. This mixing
process is controlled by diffusion or convection and is roughly a few seconds
after vaporization occurs at the surface. For piloted ignition, it involves the
time to reach the pilot.
The third step is the time it takes for the chemical reaction to release
enough energy to be perceived as a flame. For a piloted condition this is
generally less than a second, as the pilot flame or electric arc provides a
very high local temperature. For autoignition, this reaction time can be
much longer. The AIT (spherical vessel test) of kerosene is listed at 221°C.
This is the minimum temperature to cause ignition of a flammable mixture.
At higher temperatures, autoignition will occur more quickly. Figure 4.8
suggests that at 221°C, ignition of kerosene can take more than 100 seconds.
In general, the autoignition of gases can take much longer than for corre-
sponding piloted ignition.
In considering these steps, it is generally found that the longest and most
significant step, at least in piloted ignition, is the first one. This is the time
to heat the surface to achieve sufficient vaporization capable of ignition.
Flammable concentrations are small and occur quickly. Therefore, pure
conduction heating within the solid or liquid controls most of the process.
100
Kerosene
Ignition time (s)
(minimum) AIT ~ 220°C
10
1
220 240 260 280 300 320 340
Temperature (°C)
FIGURE 4.8
Autoignition time for kerosene vapor to ignite in air. (From Babrauskas, V., Ignition Handbook,
Fire Science, Society of Fire Protection Engineers, Issaquah, WA, 2005.)
Predicting the time to ignite becomes an exercise in heat conduction to

realize the ignition temperature.
4.7 Predicting the Ignition Time for Solid Fuels

From experiments, we can expect piloted ignition temperatures for solids
to range from approximately 250°C to 450°C, with AIT exceeding 500°C.
Fire applications comprise a wide range of materials and commercial
products; it is not possible to have exact thermal data available for each.
However, special fire test data can allow us to obtain effective properties for
items ranging from curtains to countertops. These effective properties can
enable us to estimate ignition times under a range of fire conditions.
As we have seen, the key parameter in determining the ignition of solids
(and liquids) is the surface temperature. If and when that surface attains the
ignition temperature is key to the ignition time. Not only will this depend
on the way the heating occurs and the properties of the material but also on
whether the material is “thin” or “thick.” Thin is not just a physical dimen-
sion but applies under slow heating that produces a temperature is uniform
throughout the thickness. Except for very slow heating conditions, “thin”
is a physical thickness of 1–2 mm, or less than 1/16 in. So, single sheets of
paper, drapes, and garments could be representative of thin objects in a
fire environment. Anything above this dimension is considered to be thick.
Even wallpaper on plasterboard is considered thick, as the paper and
plaster must be considered a composite assembly. In general, it is not just
the thickness that controls whether a material is thermally thin, but it is also
the heat flux. At low heat flux, the physical thickness, for the thermally thin
condition, increases. (Thermally thin refers to a material that has a nearly
uniform temperature during heating.)
A criterion for thermally thin is
2k(Tig - T¥ )
Physical thickness £
q ¢¢i
For example, if the incident heat flux is 10 kW/m2, k is 0.15 W/m2K, and (Tig − T∞)
is about 300°C, then the physical thickness need only be less than 9 mm
or about 3/8th in. For 50 kW/m2, this becomes 1.8 mm. The lower heat flux
produces a large thermally thin criterion.
Our discussion here can also apply to liquids provided the liquid is motion-
less, as would be the case if uniformly heated from the top. But heating from
below or localized heating will cause motion. Motion could make the liquid
act as if it were thermally thin.
Ignition 117
4.7.1 Ignition of Thin Objects

Two cases of thin heating are shown in Figure 4.9. An object of thickness l
heated from one side and perfectly insulated on the other (a) or the same
object of thickness 2l symmetrically heated and cooled (b). The incident heat
flux is given by q¢¢I , and it may be considered radiative or from a flame. There
is a heat loss flux q¢¢L, which can be both radiative and convective. If an object
is to increase in temperature, q¢¢I must be greater than q¢¢L, and this difference
should be great in order to achieve the ignition temperature.
This heat flux is converted into internal energy of the solid material raising
its temperature. How fast that happens depends on the capacity of the solid
to store energy. This energy storage capacity is measured by ρcl, the product
of ρ, the density, c, the specific heat, and l, the thickness. This energy storage
has the following units:
kg kJ kJ
rcl ~ m= 2
m 3 kg-°C m - °C
It tells how many kJ of energy can be stored in a square meter of this thin
material to raise its temperature 1°C. A balance of energy needed to raise the
temperature from T∞ to Tig must be supplied by the heat flux. The heat flux
times the time gives heat added per unit area or kJ/m2 in the units discussed
here. The energy required equals the heat added, or
rcl(T - T¥ ) = (q ¢¢I - q ¢¢L )t
Perfectly .
insulated qI̋ .
qI̋
. or
qL̋
l l
(a) (b)
FIGURE 4.9
Heated thin objects (a) heat from one side and (b) symmetric heating.
This result is only approximate as it has ignored the fact that the temperature
is increasing over time and the surface heat loss (q¢¢L ) is increasing too, as it
depends on temperature. This heat loss can be by radiation and convection to
the surroundings. The formula is only reasonably valid at early time. At later
times, the heat losses bring the temperature to equilibrium or steady state.
As a result, the temperature of the thin solid will increase in early
time (t) as
(q ¢¢I - q ¢¢L )t

T = T¥ + (4.1)
rcl
where T∞ is the initial temperature, and later in time the incident heat flux
will equal the heat loss flux. The CHF needed to just achieve the ignition
temperature is the heat loss flux at that temperature.
The actual temperature response over time is more completely shown in
Figure 4.10 with a labeled ignition temperature (Tig). When Tig is reached, we
have ignition (piloted or auto, depending on our selection of the value for Tig).
The determination of the time to ignite is shown in the figure; but under low
heating conditions, Tig may never be achieved. Note the CHF just causes the
attainment of the Tig value. For the high heating rate cases above the CHF,
the ignition time can be computed using Equation 4.1. A useful approximate
formula for predicting the time to ignite is
rcl(Tig - T¥ )
tig » (4.2)
q ¢¢I
when the incident heat flux is greater than the CHF.
High heating rate

Temperature
Tig Critical heating rate
Low heating rate
T∞
0 Ignition time
Time, t
FIGURE 4.10
Temperature rise.
Ignition 119
Example 4.1
Let us consider a typical material that might represent a drapery or a
thin fabric on an insulating substrate, similar to a cushion or uphol-
stered product. Representative properties for a cotton-like material are
ρ = 0.57 g/cm3, c = 0.34 cal/g-K, and l = 1 mm. For Tig = 300°C, ignition times
range from 5 to 25 seconds for a range of radiant heat fluxes between 40
and 10 kW/m2, respectively. For Tig = 400°C, these times only increase by
about 25%. For this reason, these objects can ignite very quickly regard-
less of their precise ignition temperature. The heat capacity factor, ρcl,
controls the process.
Flashover conditions are sometimes designated with a heat flux to
the floor of 20 kW/m2 (indicative of room smoke layer temperature of
500°C–600°C). So, we see that most thin materials will rapidly ignite
under this condition, consistent with our concept of flashover causing
the fire to suddenly grow.
4.7.2 Ignition of Thick Materials

An equation similar to Equation 4.2 applies for thick materials (usually
l > 2 mm). An approximate solution has been determined for a thick material
(mathematically done for an infinitely thick solid at a high heat flux) giving
(p/4)(krc)(Tig - T¥ )2
tig » , q ¢¢I > CHF (4.3)
q ¢¢I 2
where k is the thermal conductivity of the material. To make a prediction

for a given material, data must be available for the properties in the factor
kρc (called thermal inertia). In addition, the ignition temperature must be
known. In general, k increases in solids as ρ (density) increases. Also k and
c increase with temperature. Because the ignition temperatures might range
from 200°C to 500°C), k and c should increase considerably during the heat-
ing process. In Equation 4.3, a single material value is required for kρc at
some appropriate average temperature. This value of thermal inertia will
typically be higher than that found at ordinary room temperature conditions
for that material.
It is remarkable that Equation 4.3 seems to fit a wide range of material
data. Typically, these data are derived from measuring the time to ignite
under a prescribed constant radiant heat flux. Standard test methods exist
for this purpose (e.g., ASTM E-20586 and ASTM E-13217). The CHF can
be found as the minimum radiant heat flux at which ignition ceases to
occur. In applications to follow, and most common in such testing, only
piloted ignition is considered. An illustration of a typical analysis of such
data in order to obtain the properties needed by Equation 4.3 is shown
in Figure 4.11. The CHF was determined at 8 kW/m2. As the CHF is the
heat loss flux at the ignition temperature, a heat transfer calculation can
Finding t* for POM

1.4
1.2
0.8
qc̋ritical
q˝
y = 0.024x
.
0.6
R2 = 0.9436
.
0.4
0.2
0
0 10 20 30 40 50 60
t*
√tignition
FIGURE 4.11
Ignition data for polyoxymethylene. (From Panagiotou, J. and Quintiere, J.G., Generalizing the
flammability of materials, in: Proceedings of the 10th International Conference on Interflam 2004,
Interscience Communications, London, U.K., 2004, pp. 895–906.)
be made to determine the ignition temperature corresponding to this test

condition. In the results to follow for the standard tests, the heat loss condi-
tions are by both radiation and primarily natural convection. We will not
go into those details. However, the main point of Figure 4.11 is to indicate
how well the data follow the inverse relationship between incident heat
flux and square root of the time to ignite as set forth in the approximate
theory of Equation 4.3. As the data follow this relationship, the slope of
the linear fit to the data can be used to determine the effective kρc for
this material. When the incident heat flux equals CHF, as indicated by the
value 1 in Figure 4.11, the equation no longer will be applicable. It is note-
worthy that just at that condition, the time to ignition (t*) would be 422 or
1764 seconds. This is nearly half an hour or essentially a very long time in
an analysis for ignition hazard.
This illustration is typical of ignition data. Plots like this can be used to
determine property data needed for Equation 4.3. This is a powerful testing
technique to evaluate if a material can ignite and how long it could take.
Some typical piloted ignition times for thick solids are listed in Table 4.2.
We see that at 20 kW/m2, a condition commonly used to imply flashover,
it can take one or more minutes to achieve full flaming for common mate-
rials. But once higher room thermal conditions are achieved sufficient to
ignite other materials, the transition time for full room fire involvement
would be much shorter. During the flashover transition, the heat fluxes
increase, with the heat fluxes in a room at full involvement attaining as
much as 100 kW/m2. As a result, the increasing heat fluxes reduce the times
to ignition or flashover. Our simple formulas, Equations 4.2 and 4.3, do not
Ignition 121
TABLE 4.2
Typical Ignition Times of Thick Solids
Heat Flux (kW/m 2) Time (s) Material
10 300 Plexiglas, polyurethane foam, acrylate carpet
20 70 Wool carpet
20 150 Paper on gypsum board
20 250 Wood particleboard
30 5 Polyisocyanurate foam
30 70 Wool/nylon carpet
30 150 Hardboard
take variable heat flux into account, but we certainly see that increased
heat flux reduces the ignition time.
It is commonly observed in flashover that corner edges will ignite before
the flat surface of an object. This might be explained in terms of less mass
to heat at a corner than a flat surface. In fact, a mathematical solution to heat
conduction at the corner edge would give a similar equation to (4.3) except
with a constant of π/18 instead of π/4. The edge would ignite in about ¼ the
time for the same flat surface material.
Typical ignition data are shown in Figure 4.12 for radiant heating
of wood particleboard. The ignition times can be shown to agree well
with Equation 4.3 (t ~ (1/q¢¢I 2 )), and at low heat fluxes there is no ignition.
The minimum or critical radiant heat flux (CHF) to cause ignition in a
long time (greater than 5 minutes in these experiments) is approximately
18 kW/m2. Also shown in Figure 4.12 are measured ignition temperatures.
These measurements were done with small thermocouples fixed to the
wood surface as in Figure 4.13a. The accuracy of the measurement for
surface temperature is affected by the way the thermocouple is attached.
The scatter shown in the measured ignition temperatures can be expected.
A temperature measurement of 350°C ± 50°C is representative of these
data. Notice that this measured temperature on the particleboard is very
similar to the piloted ignition temperature for redwood heated perpen-
dicular to the grain (see Figure 4.6). The similarity suggests generic data
for wood might be used for other wood products. In general, generic data
should be used with caution and only for guidance.
If the same experiment were done with direct flame heating as shown in
Figure 4.13b,9 the ignition times do not vary with flame energy (see Figure 4.14).
The flame energy, as energy supply rate per width (kW/m) of wood heated,
controls the flame height. In other words, the data in Figure 4.14 were derived
for various flame heights. The results show, despite variations in flame height,
that this flame causes ignition in about 140 seconds (±40 seconds). While there
is significant variation, the trend with flame height or flame energy suggests
300 400
250 350
Tig (°C)
200 300
tig (s)
150 250
100
50
Ignition time
Ignition temperature
0
0 20 40 60
.
Radiant heat, q˝ (kW/m2)
FIGURE 4.12
Measurements of piloted ignition on particleboard. (From Quintiere, J.G., J. Res. Natl. Bur.
Stand., 93(1), 61, January–February 1988.)
Pilot
Radiant
heat
Thin wall flame
.
Thermocouple Flame energy, Q/W (kW/m)
(a) (b)
FIGURE 4.13
Ignition experiments: (a) radiant heating and (b) flame heating.
Ignition 123
300
250
200
tig (s)
150
100
50
0
0 20 40 60 80
.
Flame energy, Q/W (kW/m)
FIGURE 4.14
Measurement of ignition on vertical particleboard by a wall flame. (From Quintiere, J.G., J. Res.
Natl. Bur. Stand., 93(1), 61, January–February 1988.)
a relative constant ignition time. Variations can be due to the perception of

ignition, the local effects of the flame, and uncontrolled ambient airflow. But if
the relative ignition time is agreed to be about a constant, then Equation 4.3
suggests a constant heat flux from the flame. The inferred flame heat flux from
Figure 4.12 at 140 ± 40 seconds is approximately 22 ± 3 kW/m2. This heat
flux is typical of turbulent wall flames less than 1–2 m tall. The flux is found
to be insensitive to the fuel burned because the flame temperature is nearly
independent of the fuel. This experiment with the flame heating related to
the radiant heating tests shows the applicability of Equation 4.3 to different
heating conditions.
Before leaving this section, let us consider an example using Equations 4.2
and 4.3.
Example 4.2
Determine the time for 1/2 in. thick plywood to ignite subject to a flame
heat flux of 25 kW/m2. If a thin delaminated piece, insulated on the back,
with a thickness of 0.5 mm is also heated, determine when it will ignite.
Take the initial temperature as 20°C.
Estimated plywood properties:
Tig = 350°C
k = 0.15 × 10 –3 kW/m-K
ρ = 640 kg/m3
c = 2.9 kJ/kg-K
Thick case (1/2 in.)
(p/4)(krc)(Tig - T¥ )2
tig »
q ¢¢I 2
(p/4)(0.15 ´ 10 -3 kW/m-K ´ 640 kg/m 3 ´ 2.9 kJ/kg-K ´ (350°C - 20°C)2

tig =
(25 kW/m 2 )2
tig = 38.1 s
Notice this time is lower than the data given for a plywood product
shown in Figure 4.17. This difference is an indication of the level of accu-
racy in generic calculations, as all plywoods are not the same.
Thin case (0.5 mm)
rcl(Tig - T¥ )
tig »
q ¢¢I
640 kg/m 3 ´ 2.9 kJ/kg-K ´ 0.5 ´ 10 -3 ´ (350°C - 20°C)

tig =
25 kW/m 2
tig = 12.2 s
We see the time is much smaller in the thin case than in the thick case
due to less mass being heated.
4.8 Solid Properties for Piloted Ignition

A collection of piloted ignition data for thick materials is shown in Figures 4.15
through 4.18. These results were taken by Harkleroad10 in an early appli-
cation of Equation 4.3 to derive properties by the technique illustrated in
Ignition 125
400.0
350.0
Critical flux
300.0
250.0
Time to ignite (s)
200.0
150.0
Equation 4.3
100.0
50.0
0.0
0 10 20 30 40 50 60 70
Heat flux (kW/m2)
FIGURE 4.15
Radiant ignition times of an asphalt shingle. (From Quintiere, J.G. and Harkleroad, M., New
concepts for measuring flame spread properties, NBSIR 84-2943, National Bureau of Standards,
Gaithersburg, MD, November 1984.)
Figure 4.11. In Equation 4.3, the properties kρc and Tig can only be viewed
as effective properties. But by their results, these effective properties give a
very good fit to the data in Figures 4.15 through 4.18. Moreover, the materials
depicted there are complex materials: an asphalt shingle, paper on gypsum
board, plastic foam, and a wood product. The CHF for piloted ignition is
also shown in the figures, and it was found by locating the heat flux
where ignition would just not occur. This value would depend on how much
time was allowed. For practical purposes generally this would not exceed
10–20 minutes, although the experimenter in Figure 4.7 waited for more than
30 minutes. From the CHF, an estimate of the ignition temperature can be
found by a heat transfer analysis. In general, the CHF would depend on the
convective heat loss. If a wind were present, the CFH would have been higher
than the CFH determined here for essentially natural convection conditions.
In the event that there is direct flame heating, the CHF would be lower than
400.0
Critical flux
350.0
300.0
250.0
Time to ignite (s)
200.0
150.0
100.0
Equation 4.3
50.0
0.0
0 10 20 30 40 50 60 70
Heat flux (kW/m2)
FIGURE 4.16
Radiant ignition times of 1.27 cm gypsum board. (From Quintiere, J.G. and Harkleroad, M.,
New concepts for measuring flame spread properties, NBSIR 84-2943, National Bureau of
Standards, Gaithersburg, MD, November 1984.)
that given in Table 4.3, as only surface radiation heat loss applies. In a confined
space, the CHF could be as low as zero.
In Equation 4.3, it is obvious that the numerator involves a set of properties.
Tewarson11 grouped this set into one term: the thermal response parameter
(TRP). The TRP is a material property. In this form, Equation 4.3 is written
more simply as
2
æ TRP ö
tig » ç ÷ , q ¢¢I > CHF (4.4)
è q ¢¢I ø
where TRP = [(π/4)(kρc)]1/2(Tig − T∞).

Table 4.3 gives data for a wide range of materials in terms of TRP, CHF,
and an estimate of the ignition temperature.8,11,12 These have been generally
Ignition 127
400.0
350.0
Critical flux
300.0
250.0
Time to ignite (s)
200.0
150.0
100.0
50.0
Equation 4.3
0.0
0 10 20 30 40 50 60 70
Heat flux (kW/m2)
FIGURE 4.17
Radiant ignition times of 2.54 cm rigid foam plastic, polyurethane. (From Quintiere, J.G. and
Harkleroad, M., New concepts for measuring flame spread properties, NBSIR 84-2943, National
Bureau of Standards, Gaithersburg, MD, November 1984.)
derived from standard test methods such as ASTM E-1321 and ASTM E-2058.
The CHF data by these tests apply to a convective heat loss coefficient of
about 10–15 W/m2-K. Usually, the ignition temperature was estimated from
the CHF. The data show that the CHF can range from about 10 to 45 kW/m2
and that the TRP values can range from 60 to 600 kW/m2-s1/2. As the TRP is
squared in finding the time to ignition by Equation 4.4, the time to ignition
can vary significantly for a given heat flux. The heat flux for flames might
range from 20 to 60 kW/m2 for small to large turbulent flames, respectively,
with small laminar candle-like flames on the high end. In general, precise
knowledge of the incident heat flux is needed for accurate results. Moreover,
specific data are needed for a material considered. The data in Table 4.3 can
only be regarded as generic. Indeed, even for the same polymer, results can
vary. This is most likely due to the fact that a defined polymer can have other
minor ingredients and can vary with processing. Melting materials can also
400.0
Critical flux
350.0
300.0
250.0
Time to ignite (s)
200.0
150.0
100.0
Equation 4.3
50.0
0.0
0 10 20 30 40 50 60 70
Heat flux (kW/m2)
FIGURE 4.18
Radiant ignition times of 1.27 cm plywood. (From Quintiere, J.G. and Harkleroad, M., New
concepts for measuring flame spread properties, NBSIR 84-2943, National Bureau of Standards,
Gaithersburg, MD, November 1984.)
affect the test result, as in the case of foam plastics, because they can with-
draw from the heat source. This can lead results that may not conform to a
material’s real properties. For example, the decomposition temperature of
polystyrene is about 360°C, but there is an entry for polystyrene in Table 4.3
with as estimated ignition temperature of 630°C. This is due to the addi-
tion of fire retardants or the behavior of the melting polymer in the test. So,
some caution needs to be executed in applying the data. Nevertheless, the
test process and the consistency of the data in generating the data of Table 4.3
are sound and offer effective properties that can be used in Equation 4.4 for
making estimates of ignition in other heating applications.
To further facilitate the use of Equation 4.4, Figure 4.19 shows a plot of
the time to ignition for a range of TRP values. The graph is in semiloga-
rithm form to allow ease in discerning ignition times that can range from
1 to 1000 seconds. Note that in applying the graph, the CHF must be known.
Ignition 129
TABLE 4.3
Ignition Properties of Materials
Material CHFa (kW/m 2) Tig (°C) TRP (kW/m2-s1/2)
Ordinary polymers
ABS, acrylonitrile–butadiene–styrene 12 — 340
PE, polyethylene 15 443 454
HDPE, high-density polyethylene 16 — 260
PP, polypropylene 15 443 288
15 443 323
10 274 277
15 443 333
PVC, polyvinylchloride 10 374 215
20 — 390
HIPS, high-impact polystyrene 16 — 320
Nylon 66, polyamide 18 — 250
PMMA, polymethylmethacrylate 10 374 274
8 — 330
15 378 316
9 278 196
POM, polyoxymethylene 10 374 250
8 — 320
SMC ester 20 497 483
High-temperature advanced-engineered polymers
PSF, polysulfone 30 580 469
PEEK, polyetheretherketone 30 580 550
PC, polycarbonate 20 497 357
20 580 434
30 580 455
30 540 455
PEI, polyetherimide 25 540 435
Highly halogenated advanced-engineered polymers
PTFE, polytetrafluoroethylene 50 700 654
FEP, perfluoroethylene–propylene 50 700 680
ETFE, ethylene–tetrafluoroethylene 25 540 481
PCTFE, polychlorotrifluoroethylene 30 580 460
ECTFE, ethylenechlorotrifluoroethylene 38 613 450
PVDF, polyvinylidenefluoride 40 643 506
CPVC, chlorinated PVC 40 643 435
Foams and expanded elastomers
ABS-PVC 19 487 73
PVC 10 374 263
PU, polyurethane foam, rigid 20 435 63
PU, foam, flexible 16 390 183
PS, polystyrene 46 630 331
PS 20 497 146
(Continued)
TABLE 4.3 (Continued)

Ignition Properties of Materials
Material CHFa (kW/m 2) Tig (°C) TRP (kW/m2-s1/2)
Fabrics and woven products
PET, polyethyleneteraphthalate 10 374 174
Nylon 6 20 497 154
Carpet, wool 23 465 129
20 435 182
Carpet, wool, treated 22 455 187
Carpet, wool–nylon blend 18 412 283
Carpet, acrylic 10 300 158
Wood and wood products
Redwood 13 375 145
Red oak — 304 248
Douglas fir 12 384 159
Maple 12 354 239
Massive timber, varnished <20 — 197
Plywood 16 390 219
16 390 238
<20 — 187
Plywood, FR (fire retarded) 44 620 460
22 — 131
Hardboard 10 298 331
14 365 283
Hardboard, gloss paint 17 400 367
Hardboard, nitrocellulose paint 17 400 295
Particleboard 18 412 331
16 382 307
Fiber insulation board 14 355 198
Composites
Paper of gypsum board 35 565 321
Paper of gypsum board, FR 28 510 272
Gypsum board with wall paper 18 412 259
Roof shingle, asphalt 15 378 262
16 390 183
Glass-reinforced polyester
17 400 282
Interior aircraft wall panel, epoxy fiberite 28 505 208
Sources: From Panagiotou, J. and Quintiere, J.G., Generalizing the flammability of materials,
in: Proceedings of the 10th International Conference on Interflam 2004, Interscience
Communications, London, U.K., 2004, pp. 895–906; Tewarson, A., Flammability of poly-
mers, Chap. 11, in: Andrady, A.L., ed., Plastics and the Environment, John Wiley & Sons
Inc., Hoboken, NJ, 2003; Dillon, S.E. et al., Determination of properties and the prediction
of the energy release rate of materials in the ISO 9705 room-corner test, NIST-GCR-98-753,
NIST, Gaithersburg, MD, June 1998.
a CHF based on natural convective, heat transfer coefficient ~ 10–15 W/m2-K.
Ignition 131
1000
100 TRP = 600 kW/m2-s1/2

Time to ignite (s)
450
10
300
150
TRP = 50 kW/m2-s1/2
1
0 20 40 60 80 100 120
Incident heat flux (kW/m2)
FIGURE 4.19
Graph to estimate time to ignition from TRP.
For example, ECTFE has a CHF of 38 kW/m2, so the curve labeled with a
TRP of 450 can only be used above this CHF for ECTFE. Its time to ignite
just above this CHF is more than 100 seconds—a relatively long time for this
high-temperature polymer.
4.9 Summary
Ignition to a flame for solid and liquid fuels begins with a mixture of the
gasified fuel in air. At the correct concentration of fuel vapor, the mixture
can propagate a flame at the LFL with a small energy source (pilot) or at a
sufficient temperature alone (autoignition). With a pilot ignition, the flame
begins as a premixed flame. The temperature of a liquid to cause this propa-
gating premixed flame is called the flashpoint. At a slightly higher tempera-
ture, a diffusion flame can be sustained on the liquid. This is called the fire
point. These temperatures lead to the concept of an ignition temperature for
materials in general. The CHF needed for ignition just allows the ignition
temperature to result from heating.
The vaporization process for liquids is evaporation, while solids must ther-
mally decompose (pyrolysis). Both of these processes increase with tempera-
ture. At a critical temperature, the vapor composition is sufficient to form
a flame. This can occur with a pilot or alone due to the temperature and
concentration. The latter is autoignition. In a solid, ignition to smoldering

likely precedes autoignition.
Two approximate formulas were presented to predict the time to ignition
of solids. These apply to a known heat flux incident on a thin (paper,
fabric, etc.) or thick solid at a heat flux above the critical value. The thin
case involves the product ρcl, while the thick case involves the product kρc.
These comprise the following thermal properties:
ρ = density
c = specific heat
k = thermal conductivity
l = thickness
For thick solids, the TRP was introduced as a convenient grouped prop-
erty. It facilitated a simple formula for the prediction of ignition time in
solids.
Review Questions
1. A laminar match flame imparts roughly 60 kW/m2 to a surface it contacts.
How long would it take Douglas-fir particleboard (Table 4.3) to ignite
under these conditions?
2. A small wastebasket fire in the corner against wood paneling imparts a
heat flux of 40 kW/m2 from the flame. The paneling is painted hardboard
(Table 4.3). How long will it take to ignite the paneling?
3. The AIT of liquids is usually much lower than that needed to ignite the
liquid by a hot surface. Try to explain this difference.
4. Does wood ignite much differently from melting plastics? Explain.
5. What is the CHF needed to ignite redwood to smoldering and to autoig-
nition and piloted ignition?
True or False
1. Thin objects usually ignite more easily than thick objects.
2. Liquid fuels ignite at their boiling point.
3. Thick objects are more difficult to ignite as the material’s density is
increased.
4. Ignition temperature can be precisely measured for solid fuels.
5. Sunlight magnified 15 times could ignite thin paper.
Ignition 133
6. A lit cigarette can ignite gasoline flammable vapors.

7. The flashpoint is the temperature of a solid at autoignition.
8. The boiling point is the maximum temperature of a liquid.
9. When a liquid is fully burning, its surface is likely at the boiling point.
10. Evaporation of water stops at an RH of 100%.
Activities
All of these experiments must be done with safety precautions and with a
fire extinguisher available.
1. Use an electric space heater to ignite pieces of newsprint. Find the dis-
tance from the heater where autoignition will just occur. By locating a lit
match above the paper strip, record the time for piloted ignition to occur
at the same location. Explain the results.
2. Experiment igniting small (very small, half dollar size) dishes of liquid
fuels. Use ethanol or methanol and note its ease of ignition with a match
flame. Cool the alcohol in a bath of ice water and see if it still ignites. Try
igniting kerosene in the same way. Explain the results.
3. Try igniting a thick piece of wood (2 × 4) with a match flame. Does it
ignite? Or, does it not spread a flame? Observe carefully. Are the corners
or edges easier to ignite? Explain the results.
4. Place a thin index card above a candle flame. Investigate ignition without
the flame touching the card. Does the card ignite to a flame or smolder? (This
experiment must be done with caution and not manually holding the card.)
References
1. M. G. Zabetakis, Flammability characteristics of combustible gases and vapors,
Bulletin 627 (Washington, DC: Bureau of Mines, U.S. Department of the Interior,
1965).
2. V. Babrauskas, Ignition Handbook (Issaquah, WA: Fire Science, Publisher/Society
of Fire Protection Engineers, 2005).
3. A. Tewarson, Thermophysical and fire properties of automobile plastic parts
and engine compartment fluids, Vol. III, Tech Report #0003018009 (Norwood,
MA: FM-Global, October 2005).
4. M. J. Spearpoint and J. G. Quintiere, Predicting the piloted ignition of wood in
the cone calorimeter using an integral model—Effect of species, grain orientation
and heat flux, Fire Safety Journal, 36 (2001): 391–415.
5. N. Boonmee and J. G. Quintiere, Glowing and flaming auto-ignition of wood,

Proceedings of the Combustion Institute, 29 (2002): 289–296.
6. ASTM E-2058, Standard test methods for measurement of synthetic polymer mate-
rial flammability using a fire propagation apparatus (FPA) (West Conshhocken,
PA: ASTM International, 2004).
7. ASTM E-1321, Standard test method for determining material ignition and
flame spread properties (Philadelphia, PA: American Society for Testing and
Materials, 1996).
8. J. Panagiotou and J. G. Quintiere, Generalizing the flammability of materials,
Interflam 2004, in Proceedings of the 10th International Conference, Interscience
Communications (London, U.K.: Interscience Communications Ltd., 2004),
pp. 895–906.
9. J. G. Quintiere, The application of flame spread theory to predict material
performance, Journal of Research of the National Bureau of Standards, 93, 1
(January–February 1988): 61–70.
10. J. G. Quintiere and M. Harkleroad, New concepts for measuring flame spread
properties, NBSIR 84-2943 (Gaithersburg, MD: National Bureau of Standards,
November 1984).
11. A. Tewarson, Flammability of polymers, Chap. 11 in Plastics and the Environment,
edited by A. L. Andrady (Hoboken, NJ: John Wiley & Sons Inc., 2003).
12. S. E. Dillon, W. H. Kim, and J. G. Quintiere, Determination of properties and
the prediction of the energy release rate of materials in the ISO 9705 room-
corner test, NIST-GCR-98-753 (Gaithersburg, MD: NIST, June 1998).
5
Flame Spread
Learning Objectives
• Describe the different types of fire spread

• Explain flame spread theory in terms of distance heated and ignition
time
• Compute flame spread speeds on solids for different cases, given
the data
• Describe fire spread rates for many conditions
5.1 Introduction
Ignition occurs in the fuel–gas air mixture. Following ignition, the next step
in fire growth is surface flame spread. Of course, the propagation of flame
through the gas mixture is a form of flame spread, but we are principally
concerned with surface spread in this chapter. Such spread involves the par-
ticipation of the liquid or solid fuel. The surface must reach its ignition tem-
perature to allow its vaporization or pyrolysis region to grow. The perimeter
of its vaporization region defines the flame front.
Flame spread requires certain conditions. Among them, principally, is the
need for sufficient heat transfer. In general, spread does not always occur; so,
the label non-self-propagating has commonly been applied to materials after
testing. However, just because they are nonpropagating in a particular test
does not mean they will not propagate under other initiation conditions.
Such labels have been deceptive and often prove harmful to victims of fire
accidents. Many tests examine the ability of a material to spread or burn.
135
One cannot extrapolate from these tests the way a material might spread in
conditions of your fire scenario.
This chapter examines the theory of flame spread in terms of its significant
phenomena. This information can then be used to explain flame spread in
your scenario and perhaps estimate its speed of growth. We will explain
how a flame can spread over fuel surfaces or through a porous fuel matrix
such as forest brush. Also, fire can propagate through solids as a smoldering
process. Although theory and formulas are presented here, these are diffi-
cult to use in practical applications because the relevant heat flux and needed
material properties are not precisely known. However, we will try to give a
basis to making estimations, and in that process, a quantitative evaluation
might be performed. In the least, we will try to impart a typical level of fire
spread speed to the various types of spread that can occur.
5.2 Definitions
Flame spread is a process in which the perimeter of the fire vaporization
region grows. It could even include the process of remote ignitions, if those
ignition processes are continual. An example of remote ignitions is the
process of firebrands igniting material ahead of a forest fire. Examples of
surface flame spread for a solid or liquid fuel can be seen in many fires.
Specifically, we mean the extension of the burning region. It is not the flame
extent embraced in this definition, but the region volatilizing and supplying
the fuel. In general, fire growth applies to the increase in the combustion pro-
cess including surface flame spread, smoldering, and even the expanding
fire front in premixed flame propagation.
In flame spread, and in fire growth generally, gravitational and wind effects
are important. The flow resulting from the fire’s buoyancy or the natural
wind of the atmosphere can assist (wind-aided) or oppose (opposed-flow)
flame spread. Figure 5.1 depicts these modes of flame spread. Wind-aided
spread occurs in the flame direction of the airflow, while opposed-flow flame
spread is opposite to the flow direction. As seen in the figures, the potential
for the flame to heat the material ahead of it is very dependent on the flow.
This heat transfer is critical.
The flame spread velocity is defined as the rate of motion for the perimeter
position xp in Figure 5.1. The xp denotes the extent of the pyrolysis or vapor-
ization region. Behind the spreading pyrolysis perimeter, there is likely to
be another perimeter in which flaming or combustion has ceased. The region
between these two fronts (spread and burnout) defines the principal flaming
or pyrolysis region. Figure 5.2 shows the burnout and flame fronts for a flame
spreading on a thin horizontal stick. The rate of combustion of the vaporiz-
ing fuel gases from this region we define as the burning rate. We will see in
Flame Spread 137
g
xp
Air flow
δf
Tig
Air flow δf xp
Ts
(a) (b)
x
y
Tig
Ts
g Air flow
δf
x
xp δf
xp Air flow
y
(c) (d)
FIGURE 5.1
Flame spread modes. (a) Natural: flow is induced solely by the buoyancy of the flame. (b) Forced:
flow is caused by the ambient wind or a fan. (c) Opposed-flow refers to the flames spread opposite
to the airflow. (d) Wind aided refers to flame spread in the same direction as the airflow.
FIGURE 5.2
Flame and burnout fronts on a wooden stick.
Chapters 6 through 8 that the burning rate is directly connected to energy

and species produced in a fire. Furthermore, it is significant in establishing
the temperature, visibility, toxicity, and corrosivity of the fire. For this reason,
flame spread controls the extent of the fire and its burning rate. It plays an
important role in establishing the hazard in fire.
5.3 General Flame Spread Theory

The concept of an ignition temperature is key to explaining flame spread in
simple, but physically correct, terms. For the process of flame spread, the pilot
ignition temperature would apply because a pilot (the flame itself) is always
present. If we examine Figure 5.1 again, we can label the position xp at the igni-
tion temperature, Tig. The temperature further ahead of the flame, not affected
by direct heating from the flame, is labeled Ts. The distance along the surface
affected by the flame’s heat transfer is labeled δƒ. In general, the flames can heat
this region ahead in many ways. These depend on the mode of spread—natural,
forced, opposed, or wind-aided—and on the nature of the solid or liquid fuel.
Symbolically, this process is illustrated in Figure 5.3. The illustration depicts
steady flame spread. An observer riding on the flame front, at position xp sees
the new fuel coming toward it at the flame spread velocity, V. In other words,
if the flame front is stationary according to the observer, it appears that the
new material to burn is flowing toward the observer at the flame speed, V, and
into the region burning behind the observer. The general flame spread velocity
formula states that the rate of energy supplied to this newly heated fuel, bring-
ing it to ignition temperature Tig, is equal to the net heat transfer rate from the
burning region, q. Hence, we have an energy balance for spread to occur.
Observer at
Tig
flame front
Ts
Heat transfer
.
rate, q
Flame
spread V
velocity
l w
Cross-sectional Burning region

area, A xp
FIGURE 5.3
Flame spread model.
Flame Spread 139
[Rate of energy required] = [Rate of heat supplied]
rVAc(Tig - Ts ) = q (kW) (5.1)
where
ρ is the fuel density, kg/m3
c is the fuel-specific heat, kJ/kg-K
A is the cross-sectional area, wl, m2
Ts is the fuel temperature beyond the range of the flame’s heat, ºC
The formula is a bit complicated for those new to this type of math
and science. Let us talk our way through it. On the right-hand side is
the heat flow rate needed for the spread in kW. This heat rate must be
capable of heating the new material to its ignition temperature (flash-
point). On the left-hand side, we have the flow rate of a new material
taken up to the ignition temperature from its original value. If we com-
bine all of the quantities together in terms of their units, the left-hand
side has units that emerge as
kg m 2 kJ kJ
m K= º kW
m3 s kg-K s
This is the energy flow rate of the spread material that must be equal to the
heat flow rate from the flame (kW). We have applied a conservation of energy
principle. It must be obeyed if flame spread is to occur. Too little heat, no
spread will occur. Too high an ignition temperature, slow or no spread may
result.
The flame spread velocity is obtained as
q
V= (5.2)
rcA(Tig - Ts )
Many specific cases of flame spread can be derived from this formula by
more carefully describing q and A. We will consider some approximate for-
mulas that follow from Equation 5.2. These formulas are in agreement with
more complex analyses beyond the scope of this text. There are approximate
formulas that should be considered as tools for making estimates, with accu-
rate results subject to assumptions. Again, orientation, wind, and the nature
of the fuel all make a difference.
5.4 Spread on Solid Surfaces

For spread over solid surfaces, it has been shown that the most significant
heat transfer rate is at the surface over a length δƒ, as shown in Figure 5.1.
Heat flow rate is the product of heat flux and the surface area it acts on.
Therefore,
q = q ¢¢d ƒ w (5.3)
where
q¢¢ is the flame forward heat flux
w is the width of the fuel as shown in Figure 5.3
The flow area is normal to the surface area, and it is designated as A or wl,
where l is the fuel’s thickness. Here, we are making a simplification, as we are
implicitly assuming that the heat is uniformly distributed over the thickness,
and a uniform temperature results in depth. This will not be a restrictive
assumption, and other results to follow will be more general.
If we substitute these formulas into Equation 5.2, we obtain
q ¢¢d f
V= (5.4)
rcl(Tig - Ts )
Again, referring to Figure 5.1, V is the rate of movement of position xp.

Orientation and natural flow or forced flow conditions affect the heat-
ing distance δƒ and the associated flame heat flux q¢¢. Measurements
have shown that for downward spread on a wall, the dominant heat flux
extends over a δƒ of approximately 1 mm. It can be as high as 70 kW/m2.
For upward flame spread on a wall, the heat flux is generally 25 kW/m2
(fairly constant with ±5 kW/m2 for flames extending above the vaporizing
region in distances from 0.20 to about 2 m). Here, the flame extension is
the luminous flame region that approximates δƒ. These heat flux results
appear to be independent of the wall fuel; so, whether we have a gasoline
saturated wood wall, or the wood wall itself, the incipient heat flux from
the flame to the wall is up to 70 kW/m2 going down against the flow of
air, or about 25 kW/m2 going up. The downward flame is a laminar and
premixed flame at its front, and the upward flame will shortly become tur-
bulent. The tendency for these heat fluxes to not strongly depend on fuel
type is due to the invariance of flame temperature in air for most common
materials.
Flame Spread 141
5.4.1 Effect of Thickness

Just as for the ignition of solids, we must distinguish between thick and thin
solids in flame spread. In the form of Equation 5.4, the result is strictly for
the thin material, as the temperatures are uniform in depth. In general, for a
thick material, the surface heat transfer from the flames can only penetrate to
a certain depth. This heating in depth takes place during the time to heat the
surface from its original temperature Ts to the ignition temperature Tig. This
heating time is, in fact, the ignition time (tig) caused by the heat flux of the
spreading flame. From our discussion on heat conduction in Chapter 3, the
actual depth heated in this time is related to √(αtig), where α is the thermal
diffusivity, k/ρc. So, if the actual thickness of the material is greater than
this heated depth in flame spread, the thickness of the material is inconse-
quential in the flame spread process. In other words, the material is ther-
mally thick and the spread rate is not dependent on its thickness. While this
heating depth depends also on the heat flux, generally for flame spread heat
fluxes, materials less than 2 mm can be approximated as thin.
For a thin solid (l < 2 mm), the ignition time, tig, is given by Equation 4.2,
which allows us to write Equation 5.4 as
dƒ
V= (5.5)
tig
since the thin solid ignition time is
rcl(Tig - Ts )
tig =
q ¢¢
This is a remarkably simple equation, as it says that the speed of the flame
spread on the surface is the distance heated by the flame divided by the time
to ignite that new material. This is akin to computing your average speed
over a distance traveled.
Indeed, the same Equation 5.5 holds for the thick material (greater than
about 2 mm). For a thick solid, the entire thickness l is not heated during the
ignition time. Recall that the ignition time for a thick material is given by
2
p é Tig - Ts ù
tig = krc ê ú
4 ë q ¢¢ û
The spread velocity for the thick solid is given by Equation 5.5 with this thick
solid ignition time. Most flame spread problems can be put into the form of
Equation 5.5. The solution then depends on determining the heating length
δƒ and the ignition tig caused by the flame heat transfer.
Composite materials, such as plasterboard with a paper face, should be

regarded as a single thick material in determining ignition and spread.
Although the paper on the plasterboard may be physically thin (l < 2 mm),
the combination of the paper and plaster acts as a thick material, because
flame heat transfer from the paper burning will be conducted into the plas-
ter. However, if the paper delaminates, then it can ignite and spread as a
thin material. Such physical effects of delaminating, melting, and so on will
impact the heat flux ahead of the flame, and the distance the heating is felt.
Empirical results from standard tests are likely to contain these complex
physical effects, but the form of their results is usually inadequate to use in
the prediction of spread.
5.4.2 Downward or Lateral Wall Spread on a Thick Material

For downward or lateral flame spread on a wall surface, the combination of
the δƒ and q¢¢, that is, (q¢¢)2 d ƒ, for thick solids, depends on the air flow and on
the way the fuel behaves. In other words, the physical effects in spreading
against the airflow can modify the heat flow and distance in which it acts
ahead of the flame. For such spread in still air, this combination has been
found to depend only on the fuel and can be considered a fuel-flame spread
property. This property is designated by ϕ in the following formula:
f
V= (5.6)
krc(Tig - Ts )2
where
4 2
f= q ¢¢ d f
p
This equation follows by applying Equation 5.5 to a thermally thick material.

Up to now, no restriction has been put on the flame spread formula. Should
it always hold? We know materials do not spread if the heating conditions
are not sufficient. A measure of this is the initial temperature of the mate-
rial. This type of opposed-flow flame spread will only occur if the surface
temperature is above a critical value. Such data can be obtained for materials
from ASTM E-1321: Standard Test Method for Determining Ignition and
Flame Spread Properties.1 The test measures conditions for lateral spread on
a vertical sample. However, similar results apply to downward or horizontal
spread, as the flow conditions into the spreading flame would be similar.
Figure 5.4 shows a typical example of the type of surface flame spread
found in this test. Below the minimum temperature for spread, 120°C, no
spread occurs. Above this temperature, spread occurs at increasing veloc-
ity as the surface temperature is increased. This temperature occurs in the
Flame Spread 143
15
14
13
12
Lateral flame spread velocity, V (mm/s)
11
10
9
8 Surface
No Premixed
7 spread spread flame spread
6
5
4
3
2
1
0 100 200 300 400 500 600

Surface temperature, Ts (°C)
FIGURE 5.4
Lateral flame spread based on the test of plywood in still air.
ASTM E-1321 test by heat from an inclined external radiant panel burner.
The test actually imposes a decreasing surface temperature as the flame
spreads into the material, and accordingly, the flame slows down as Ts
decreases and the ignition time increases. In actual fire conditions, this radi-
ant heat transfer occurs from the hot smoke contained in a room. In that
case, the flame spread can accelerate as the room heats. The flame heats a
very small region (~1 mm) ahead of the lateral or downward spreading front.
The smoke, in contrast, heats the wall to temperatures (Ts). The level of heat-
ing depends on the energy release rate of the fire. As this surface tempera-
ture increases, the surface flame can now heat the wall close to the flame to
the ignition temperature (390°C) more quickly.
Suppose a fire occurs in a room with the plywood wall shown in
Figure 5.4. At some point, the fire has sufficient heat flux (above the criti-
cal heat flux for ignition) to ignite the plywood. In order for lateral or
downward flame spread to occur, the room fire conditions must heat the
plywood to 120°C, its minimum temperature for spread. Here, the spread
rate is only 0.33 mm/s. At Ts = 300°C, it achieves 3 mm/s or about 7 in./min.
This is how room fire conditions can accelerate fire spread over all burn-
ing objects. Equation 5.6 suggests that when the surface temperature of
TABLE 5.1
Lateral Flame Spread Data from ASTM E-1321
Material Tig (°C) krc (kW/m2 K)2 f (kW 2 /m3 ) Ts,min (°C)
Wood fiberboard 355 0.46 2.3 210
Wood hardboard 365 0.88 11.0 40
Plywood 390 0.54 13.0 120
PMMA 380 1.0 14.4 <90
Flexible foam plastic 390 0.32 11.7 120
Rigid foam plastic 435 0.03 4.1 215
Acrylic carpet 300 0.42 9.9 165
Wallpaper on plasterboard 412 0.57 0.8 240
Asphalt shingle 378 0.70 5.4 140
Glass-reinforced plastic 390 0.32 10.0 80
Source: Quintiere, J.G. and Harkleroad, M., New concepts for measuring flame spread properties,
NBSIR 84-2943, National Bureau of Standards, Gaithersburg, MD, November 1984.
the plywood reaches 390°C, its ignition temperature, the speed would be
infinitely fast. This result is not correct because the surface spread formula
no longer applies at this point. As the plywood reaches its ignition tem-
perature, the pyrolyzed gaseous fuel at the surface would be within its
flammable limits. Consequently, the flame would propagate through the
gas mixture near the surface with the appropriate premixed flame speed
(probably about 30–100 mm/s).
Typical data taken from ASTM E-1321 for lateral flame spread in still
air are shown in Table 5.1. These data can be used to construct plots as
depicted in Figure 5.4. It should be emphasized that data should not nec-
essarily be generalized to similar materials. For example, all plywood, or
polycarbonates may not be the same. Material compositions can vary to
a degree; fire retardants may have been included, or composite materials
can vary in construction. However, the data of Table 5.1 are generally rep-
resentative of common construction or interior finish materials. The Tig
and kρc values for these materials in Table 5.1 would apply to other modes
of piloted ignition and flame spread, but the ϕ and Ts,min apply only to
the lateral spread conditions in still air. According to Equation 5.6, ϕ can
be estimated for well-behaved materials. For these, it is known that the
leading premixed flame imparts a level of heat flux of about 60–70 kW/m 2
over about 1 mm of surface. Other materials might delaminate, melt, or
distort to affect these numbers. However, using the well-behaved mate-
rial values, we can compute ϕ:
4 2 4 æ 0.001 m ö
f= q ¢¢ d f = (65 kW/m 2 )2 d f ç 1 mm ´ 2
m3
÷ = 5.38 kW /m
p p è 1 mm ø
Flame Spread 145
We see that real values in Table 5.1 range from about 1 to 15 kW2/m3. These
variations between real and ideal give some indication of the accuracy in
predicting flame spread from first principles. The derived value of ϕ from
the test accounts for real effects not directly modeled. It is an empirical factor.
It should be obvious from Equation 5.6 that a larger ϕ means a higher flame
spread velocity.
The student should try to recognize the general speeds associated with this
type of opposed-flow flame spread in air. It is of the order of about 1–2 mm/s.
It can increase as the surface temperature ahead of the flame increases, and
it needs a critical minimum surface temperature for spread to occur. A wind
into the flame will increase the flame speed because the leading edge convec-
tive heat transfer will increase. However, at a high velocity the fuel may not
have time to fully react and the flame temperature will drop causing the flame
to slow and eventually extinguish. This is called “flame blow-off.”
5.4.3 Upward or Wind-Aided Spread on a Thick Material

Flame spread in the direction of the airflow is generally referred to as wind-
aided spread. Upward flame spread on a wall or flame spread under a ceiling
in still air is also termed wind-aided. In these natural convection cases, the
wind is solely due to the buoyant flow caused by the fire itself. The heat to
make the flame move is mainly caused by the heat transfer from the extended
luminous flame adjacent to the surface. The flame extends the distance δƒ as
shown in Figure 5.1. Unlike opposed-flow spread in which this distance is
primarily a constant that depends on the flow, here this distance depends on
the extent of burning. We will see in Chapter 7 that flame length depends
principally on the energy release rate of the fire driving it. So with increased
burning, a longer flame and a higher speed result.
Let us consider the scenario for upward flame spread on a vertical wall as
shown in Figure 5.5. An igniting fire, placed adjacent to a wall, “insults” it
with a sufficient heat flux to ignite the wall over a heated length xp. At that
point, the combined fuel from the igniting fire and the wall section ignited
produce a flame height xƒ. The flame extension δƒ is xƒ – xp. Our flame spread
velocity at the onset of spread is given by Equation 5.5, where the ignition
time (tig) depends on the flame heat flux at that time. As the wall fire contin-
ues to spread, it will be less and less dependent on the igniting fire. Indeed,
the igniting fire may burn out. Also, the wall may burn out. Both of these
extinctions will reduce the height of the flame, xƒ. Alternatively, the wall
material may contain sufficient fuel (e.g., plywood compared to wallpaper
on plasterboard) to cause xƒ to continue to increase. If xƒ depends directly on
xp, the flame speed will accelerate because xƒ outruns xp.
Suppose xf was always 50% greater than xp. Then δϕ would equal 0.5xp
and the speed would continue to increase as xp increases. Take typical
plywood with a TRP of about 229 kW-s1/2/m2 burning with a turbulent
flame on a wall (see Equation 4.6 and Table 4.3). Turbulent wall flames,
xf
xf xf
xp
xp x
xp
x x x
Igniter On On Off Off
Wall flame Off On On Burning out
FIGURE 5.5
Sequence of events in upward flame spread.
burning with heights of about 0.2–2 m give a heat flux of about 25 kW/m2. This
gives an ignition time in this flame spread of [(229 kW-s1/2/m2)/25 kW/m2]2
or 83.7 seconds. For the 50% flame growth case, according to Equation 5.5
the speed of the wall would be 0.5xp/83.7 seconds or about 6 mm/s for a xp of
1 m. As xp continues to grow, the speed increases accordingly. Here, we have
some appreciation of the magnitude of upward flame spread. More easily
ignitable materials than plywood could have a speed 10 times higher. Thin
materials that burn out quickly could cause the speed to decrease. Materials
with low burning rates may not be able to get a sufficient flame length, and
the flame will not propagate. This type of wind-aided spread is more com-
plex than opposed-flow spread. It is generally unsteady, whereas opposed-
flow spread is generally constant if it occurs for a given material.
In general, upward or wind-aided flame spread depends on the nature
of the igniting fire and the combustion properties of the wall. Of course, if
this upward flame spread scenario occurs in a room, the smoke layer will
increase Ts causing the surface velocity to also increase as in downward or
lateral spread. But unlike downward spread where δƒ is small and nearly
constant (~1 mm), for upward spread δƒ can sharply increase or decrease due
to combustion and flow properties that affect flame length, xƒ. As a result of
different flow features, flame spread up a wall, under a ceiling, or driven by
airflow in a ventilation duct is all different. Our ability to write perfect for-
mulas for each case is not yet possible, so we are left with only the framework
of Equation 5.5 at this time. Unlike the case of opposed-flow spread where δƒ
is constant, these wind-aided cases will be unsteady, because δƒ will change
as the flame propagates. Spread will accelerate or stop. Typical wind-aided
spread rates might range from roughly 1 to 100 cm/s.
Flame Spread 147
5.5 Spread through Porous Solid Arrays

Flame spread can take place through porous solid fuels such as forest brush
and debris. Even wind-aided flame spread through the arrays of urban dwell-
ings can be considered in this class of flame spread. Figure 5.6 illustrates this
type of spread in still air through a wood crib. In this case, the forward heat
flux to cause spread is mainly within the array, not just at the outer surface.
There will be a flame above the porous material, but the principal factor
controlling the spread is the heat transfer within. A wind can increase both
heating levels. The spread rate is controlled by the fuel mass in the array
characterized by the bulk density, ρb. The bulk density is the mass of the
solid fuel divided by the volume of the entire array—the fuel and the air.
Assume again that the fuel elements are thin so that they heat uniformly.
From Equation 5.2, for the array we obtain
q ¢¢
V= (5.7)
rbc(Tig - Ts )
where q¢¢ represents the overall heat flux within and on top.
FIGURE 5.6
Flame spread through a porous array—a wood crib.
Most often, this porous array is considered to be composed of cellulosic

fuel matter, so that fixes the fuel properties. However, moisture content is a
big factor. Equation 5.7 is only symbolic as to implement it requires discrimi-
nating the heat fluxes and other effects, such as wind, terrain, and moisture.
The forward heat flux in the array is a complex combination of radiation
from the fuel elements and convection due to wind effects.
In wildland fires, the U.S. Forest Service uses a model formulated by
Rothermel that is qualitatively explained in Reference 4 in which these com-
plexities are discussed. Equation 5.7 forms the basis of the Rothermel model
in which more detail is contained with specific data for forest fuels. It is used
as a tool in combating forest fires.
Wildland fires are very unique. In many cases, they begin with a lightning
strike that leads to smoldering of the forest ground covering dead brush.
This can lead to flaming much later, very likely due to a wind. This natural
process of burning out the ground forest material is a good thing for a for-
est. Sometimes, the process is intentional; it is called “prescribed burning.”
In any case, whether intentional or accidental, the combination of dry forest
material and a sustained wind can cause the fire to grow. The wind can push
this early ground brush fire to a higher level, and with severe conditions
it can lead to burning the live forest materials. It can even get to the top of
trees; this is called a “crown fire.” Such a fire condition is very bad and nearly
impossible to defend against. Now the embers generated as firebrands can
travel well ahead of the fire front and cause new fires. With the development
of homes built into the forest, the phenomenon of the “urban wildland fire”
has become a common event.
Let us address some specific analyses for forest and porous fuel fires
that give some quantitative estimate for their spread speed. Roughly, the
heat flux and the thermal properties for forest materials are nearly con-
stant. Thomas5,6 reports results for porous arrays where the flame speed,
V, inversely depends primarily on ρb. This has been found to hold true
for woody elements over a wide range of thicknesses. He reports that for
spread in still air:
C (5.8)
V=
rb
where
V is in m/s
ρb is in kg/m3
C is a constant of roughly 0.07 kg/m2-s for typical wildland fuels and
0.05 kg/m2-s for wood cribs of sticks up to 3 cm in diameter
In general, C might be regarded to vary as 0.06 ± 0.02 kg/m2-s.5,6 This con-

stant contains complex heat flux and fuel property effects. It is remarkable
that it varies relatively little between wildland fuels and composed array of
sticks.
Flame Spread 149
Wind is a serious factor in forest and urban conflagration fire spread.

Indeed, if the wind dies, the ability to contain these fires becomes more real-
istic. For wind-aided flame spread through porous arrays, Thomas6 reports a
modification to Equation 5.8 as
æCö
V = ç ÷ (1 + aV¥ ) (5.9)
è rb ø
where
V∞ is the wind speed
a is a constant ≈ 1 s/m
The wind speed effect on the flame speed is approximated here as linear.
In reality, it can be more complex. This formula represents a good first
approximation to the effect of wind.
Figures 5.7 and 5.8 show the accuracy of these simple formulas for differ-
ent porous wood-based fuel arrays. Their comparison is against controlled
laboratory data and field data. Figure 5.7 shows that a wide array of data
varying in bulk densities from about 20 to 200 kg/m3 gives an average for
C of 0.077 kg/m2-s. Figure 5.8 shows a trend curve for wildland fuels and
two straight trend lines for wood crib data. Equation 5.9 is a good approxi-
mation up to a wind speed of about 5 m/s for these fuels, after which the
wildland spread rate increases more sharply with wind speed. Urban fire
data for conflagrations of Japanese cities are also shown in Figure 5.8 where
the bulk density of the dwellings has been taken as 10 kg/m3 in order to
plot the data. The scatter in these real field data can be expected, but their
trend is clear. Figure 5.9 shows more recent results of fire spread through
the part of Kobe following the Hyogoken Nanbu earthquake of 1995. These
data also show scatter as can be expected from the actual organization of
the houses. They show leeward speed (in the wind direction, wind-aided),
windward (into the wind, opposed), and wind side (normal to the wind
direction). We expect the latter two to be similar as they are in the category
of opposed-flow spread. If we take the Thomas estimation to be the same for
these wooden houses of Kobe (10 kg/m3), then Equation 5.9 can be used to
estimate the earthquake results as
C 0.06 kg/m 2s 3600 s m

V= (1 + V¥ ) = 3
´ (1 + V¥ m/s) = 21.6 (1 + V¥ m/s)
rb 10 kg/m 1h h
Figure 5.9 shows a fire spread of about 21 m/h upwind of for no wind, sur-
prisingly similar to the estimate. However, at a wind speed of 4 m/s, the fire
spread in the wind direction (leeward) is estimated by the formula as 108 m/h
compared to the data ranging from about 60 to 80 m/h. Still, not so bad.
0.20
0.15
V (cm/s)
0.10
0.05
The slope of the line is
Vρb = 7.7 mg/cm2/s
0.00
5 10 15 20 25
1/ρb (cm3/g)
FIGURE 5.7
The effect of bulk density of fire spread in a porous fuel array. (From Thomas, P.H., Forestry,
20(2), 139, 1967.)
Recently, Tanaka and coworkers have presented a model to predict the

spread in the Kobe postearthquake fires.8 Their model is based on the mecha-
nism depicted in Figure 5.10.
The model is based on a computer zone model for building compartments.
The model was extended to include building-to-building spread. The heat
transfer mechanisms that promote spread include
1. Thermal radiation from the building fires

2. Convection from the wind-blown plumes
3. Conduction from the ejected and transported fire brands
The model gave reasonable prediction regarding the number of burnt build-
ings in a given conflagration area of Kobe. While quite complex in its detail
Flame Spread 151
Cribs
20 Gorse and heather
Urban fires in Japan
15
Spread velocity (m/min)
Vρb (kg/m2-s)
2
10
6 1
0
0 5 10 15 20 25 30
Wind velocity (m/s)
FIGURE 5.8
Comparison of spread rates in porous arrays with the effect of wind. (From Thomas, P.H.,
Forestry, 44(2), 155, 1971.)
and execution, the spread framework follows the basic principle: sufficient
heat transfer has to cause the next region to reach its ignition temperature.
5.6 Spread on Liquids

Surface flame spread on liquids is similar to the mechanisms discussed
for surface flame on solids. However, the liquid differs in that motion can be
induced in the liquid due to the spreading flame. The fact that the surface of
the liquid is hotter under the flame than the cooler liquid upstream leads to
a buoyant flow that can influence the flame spread process. Also, motion can
be induced in the liquid by viscous or surface tension effects. Surface tension
increases as the temperature decreases. So, as one moves ahead of the flame
on a liquid surface toward the cooler fluid, the surface tension increases and
80
m/h Leeward
area
Leeward area
60 Windside area
Windward area
Fire spread speed
Windside
area
40
Windward
20
area
0
0.0 1.0 2.0 3.0 4.0 4.5
Wind velocity m/s
FIGURE 5.9
Wind velocity affecting the flame spread among urban dwellings following the Kobe earth-
quake. (From Nagano, Y., Fires, in: Kaji, H., ed., Comprehensive study of the Great Hanshin
earthquake, UNCRD Research Report Series No. 12, United Nations Centre for Regional
Development, Nagoya, Japan, 1995, p. 117.)
pulls the flame forward. This is a distinct difference to that of a solid, and
it adds to the flame spread speed caused by heat transfer alone. The surface
tension variation with temperature is the chief mechanism that makes liquid
surface flame spread different from solids. Here, we are talking about flame
spread on a horizontal liquid pool unaided by wind. This surface tension
effect would not be a significant mechanism in wind-aided spread on liquid
pools because surface flame heat flux would dominate. For flame spread on
a liquid pool in still air, the surface tension effect causes significant energy
transfer to the liquid ahead of the flame. This liquid energy transfer raises
the temperature (Ts) ahead of the flame front, reducing the energy needed for
spread. Therefore, from Equation 5.2, the liquid flame spread speed is higher
over what we would expect for a solid with comparable properties.
Figure 5.11 shows experimental results of Akita9 for a 2.6 × 1.0 cm deep
pool of methanol. The ignition temperature for methanol is the same as the
flashpoint, which is 11°C. At that temperature, a fuel–air mixture exists at
the surface that can be ignited by the surface flame. The flame speed results
are plotted against the original liquid temperature Ts of the liquid similar
to Figure 5.4 for plywood. The results between the plywood and the liquid
fuel are distinctly different. The general enhancement of the surface ten-
sion motion of the liquid produces a flame speed of greater than 1 cm/s,
while for the plywood it is generally about 1 mm/s before Ts nears its ignition
temperature. Also, there is a pulsating region that applies to liquid spread.
Here, the spread rate can oscillate between the maximum and minimum
Flame Spread
Factors of building-to-building fire spread

Heat transfer from fire plumes Spotting of firebrands Heat transfer from fire involved building
∆T∞
U∞
.
. ΣQL .
. . QB m
Q m .
qR̋
. .
qR̋
. ΣQL
. QB .
m m
II. Building—collapsed I. Building—not collapsed A. Damage on structural frames

(Wood crib fire) (Group of compartment fires) B. Deficit on walls and windows
C. Falling of exterior material
Structural damage and mode of combustion
Type of structural damage
FIGURE 5.10
Mechanisms of post-earthquake spread. (From Himoto, K. et al., A post-earthquake fire spread model considering damage of building components due
to seismic motion and heating of fire, in: Proceedings of the 10th International Symposium on Fire Safety Science, 2011, pp. 1319–1330.)
153
1 cm
200
Methanol
100
Preheat-type spread
60
Uniform
40 region
Pulsating
Rate of flame spread (cm/s)
region
20
Premixed-type
Pseudo- spread
uniform
10 region
6
Max
Tstoich = 20.5°C
rate
4
Tflash = 11°C
2
Min
rate
0.6
–20 –10 0 10 20 30
Liquid temperature (°C)
FIGURE 5.11
Rate of spread on a liquid pool of methanol as a function of its temperature. (From Akita, K.,
Some problems of flame spread along a liquid surface, in: 14th Symposium (International) on
Combustion, The Combustion Institute, Pittsburgh, PA, 1973, p. 1075.)
values shown. For both the plywood and the methanol, as Ts reach their igni-
tion temperature a flammable mixture is reached. In Figure 5.11, the pre-
mixed gas-phase velocities are explicitly shown for temperatures above the
flashpoint. These range from about 30 to 180 cm/s. Such premixed results
were not explicitly shown for the plywood in Figure 5.4.
The discussion here has been for pools of liquid that have a considerable
depth. A liquid spilled on a solid like a carpet, wood, or concrete floor would
Flame Spread 155
0.14
0.12
To
30°C
20°C
0.10
0.08
Vf (cm/s)
0.06 Stable Unstable

spread spread
0.04
0.02
0 50 100 150 200 250 300 350

U (cm/s)
FIGURE 5.12
Flame spread on a kerosene soaked sand bed as a function of opposed air speed. (From
Suzuki, T. et al., Flame spread over soaked sand in an opposed air stream, in: Proceedings of the
Second International Symposium on Fire Safety Science, 1989, pp. 199–208.)
not possess the same flow effects. The spread will be slower than a pool of
liquid. The liquid and porous solid would act like a composite, and the heat
loss into the solid would mitigate the spread. Figure 5.12 shows data for kero-
sene soaked into a bed of sand from Hirano and coworkers.10 In the figure,
the effect of opposed-flow velocity is shown on the flame speed. In still air,
the radiation from the vertical flame leads to the highest spread rate in these
experiments. As the air speed increases, the flame is blown over into a thin
boundary layer flame that heats the fuel ahead by convection mostly. Over
an air speed of about 300 m/s, the flame is blown off.
5.7 Spread through a Dwelling

Distinct spread in a particular medium has been discussed. It is far more
complex to determine the fire spread rate through a series of rooms.
However, experiments were done some time ago, and the overall fire
growth rate was measured in terms of the volume of the fire and the con-
sistency and nature of the fuel in dwellings. The results are quite sim-
plistic, yet their predictability from basic principles is explainable. Apply
Equation 5.1 to this case. The rate of the volumetric fire growth can be
represented as VA (speed times area, and note the units: m/s × m2 is m3/s).
The heat transfer put out by the fire would increase with the volume of the
fire, so heat transfer could be taken as proportional to the fire volume as a
first approximation. This indicates that the rate of growth for the fire vol-
ume is proportional to the volume of the fire itself. The constant of propor-
tionality depends on the fuel and its configuration. We would expect the
density, specific heat, and ignition temperature of the fuel in the rooms to
affect this constant. This simple result yields that, for a given consistency
of fuel in a series of rooms, the fire volume will double in a fixed time. This
fire doubling time for normal dwellings might range from 2 to 10 minutes.
Figure 5.13 shows data from a dwelling with similarly distributed fuel
among its rooms.11 It has a fire volume doubling time of 7.5 minutes. When
the ratio is 2, the time is 7.5 minutes, and at 4 it is 15 minutes, and so on.
So, this complex fire growth through a building can be classified into a
quantitative format.
20
Ratio of fire volume to room of origin volume
10
9
8
7
6
5
4
1
0 5 10 15 20 25
Time after room of origin flashover (min)
FIGURE 5.13
Fire growth in a dwelling. (From Labes, W.G., Fire Technol., 2(4), 287, May 1966.)
Flame Spread 157
5.8 Typical Fire Spread Rates

As we have seen, flame spread is a complicated phenomenon of fire growth
that does not easily lend itself to quantitative analysis. Formulas do exist, and
they fit a simple framework (e.g., Equation 5.2). But they are usually lacking in
information for the specific heat flux and fuel properties. However, with some
estimations, data and formulas can be used to arrive at representative results.
To put fire spread rates in perspective, Table 5.2 lists typical values that
may be useful in characterizing different spread phenomena. Even without the
application of formulas, the table data can usefully characterize fire spread
phenomena. Smoldering spread is likely below 0.01 cm/s, opposed-flow flame
spread normally is of the order of magnitude 0.1 cm/s, and forest spread
through fine needles, liquid spread and wind-aided spread could range from
roughly 1 to 100 cm/s at most. Anything greater is gas-phase premixed flame
propagation.
5.9 Standard Test Methods

A number of flame spread–based test methods exist for materials, some of
which form the basis for fire safety regulations in the United States. Most of
these tests produce relative indexes for assessing spread and do not support
basic data for the computation of flame spread. Some can even be misleading
because the test performance may not reflect realistic fire scenarios. Other
tests have played a traditional role in regulating fire safety and have been
supported on an empirical bed of data (see Table 5.3). The Steiner tunnel
test (ASTM E-84) is a wind-aided flame spread test with the material on the
ceiling of a duct. It gives a flame spread rating that is normalized at a rating
of 100 for the rate associated with burning red oak. Although ASTM E-162
TABLE 5.2
Typical Flame Spread Rates
Spread Rate (cm/s)
Smoldering 0.001–0.01
Lateral or downward spread on thick solids 0.1–1
Wind-driven spread through forest debris or brush 1–30
Upward spread on thick solids 1–100
Horizontal spread on liquids 1–100
Premixed flames (laminar) 10–100
Premixed flames (detonations) ≈105
TABLE 5.3
Empirical Standard Flame Spread Tests
ASTM Designation: E-84
Standard test method for surface burning characteristics of building materials
Standard test method for surface flammability of materials using a radiant heat energy
source
Standard test method for critical radiant flux of floor-covering systems using a radiant heat
energy source
Standard test method for determining material ignition and flame spread properties
is a downward flame spread test, its measurements are processed to give

similar index results to ASTM E-84. ASTM E-648 is one of the few tests that
actually give results in engineering terms. It gives the minimum heat flux
required for horizontal flame spread on a floor material. In contrast, ASTM
E-1321 is not used for regulatory purposes but generates engineering data for
flame spread computations. Such data have been listed in Table 5.1. Suffice
it to say that no standard test has been invented that directly gives a full set
of engineering data needed to predict flame spread. This deficiency is a seri-
ous problem in assessing the fire growth hazard of new materials and their
applications. Also, the reliance on test results without a consideration of the
fire scenario is problematic.
5.10 Case Study: Fire Spread in a School Gymnasium

Several years ago, I investigated a fire in a high school gymnasium
(Figure 5.14). I usually take my job as trying to explain what happened. But
no one can exactly know how any fire progressed. In this case, the issue
was what product was significant in spreading the fire beyond the upper
gym. The fire was clearly seen on a pile of head-high mats against a wall. An
attempt was made to suppress it with available portable fire extinguishers.
These untrained observers failed to suppress the mat fire and vacated the
gym while the fire was still on the mats. They proceeded by moving away
down the corridor. As they progressed, they looked back and saw, through
the corridor wall glazing, that the fire had just started to bridge the upper
corner of the wood partition wall about 20 ft above the upper gym floor. This
occurred in less than a minute.
Flame Spread 159
North Central High school fire

Fire on polyurethane mats
Next seen on upper corner of wood partition 1 minute later
Complete burnout of upper gym occurs
Some blame spread to seats as responsible.
Firefighter sees paint burning on wall after fire controlled.
X observers of fire X
PU mats
Wood partition
PS seats
Folded
Plan view
Upper gym Main gym
Elevation
FIGURE 5.14
Gymnasium fire.
Some investigators argued that the fire progressed from the mats, along
possibly some floor mats, to the bleacher seats made of polystyrene.
They suggested that flames from the bleachers attacked the wood parti-
tion wall. If we consider about 10 m to the bleachers, a floor fire would
propagate at about 0.1 cm/s, and it can be computed then to take more
than 2 hours to move that distance. Even if one might say the polyurethane
mats might melt to form a liquid pool, bumping up the speed to 1 cm/s,
still this gives an inconsistent time (12 minutes) relative to the time taken
by the observers. Now the flame height on the pile of mats could barely
reach the ceiling. So, how could the fire move from the mats to involve the
wood partition at the ceiling level? After reading much material, I came
across an observation of a firefighter that was mopping up in the upper
gym after the fire had ravished the school. He said some of the paint on
the concrete block wall was still burning. Realizing now that the paint, and
likely many coats, was on the wall, we now have the possibility of upward
fire spread. Considering a propagation rate in this wind-aided direction
to the wood partition, we can estimate a speed of at least 1 to say 10 cm/s
for a travel distance of about 10 m. We now get about 100–1000 seconds
(or about 1–10 minutes) to reach the partition. This scenario is much more
plausible, especially recognizing that such a wind-aided fire will accelerate.
Hence, this scenario is likely what happened, and it is consistent with the
time period taken by the witnesses. We have not precisely predicted this
fire growth, but we have used the science of fire to make a rational analysis
and found the most consistent scenario.
In general, paint on inert concrete can ignite and burn. In NYC, paint
on concrete block walls in stairwells of public housing was found to be
responsible for several fires that propagated up many floors. The city
sought a remedy to either cover or remove the paint. These two painted
wall examples indicate the hazard of upward flame spread. Of course in
both cases, a significant fire had to get it going.
5.11 Summary
Fire spread can occur through gaseous fuel at incipient speeds of 10–100 cm/s
reaching detonations in confined regions of 105 cm/s. On the other hand,
surface flame spread can range from 1 to 100 cm/s on liquids and solids,
especially if it is wind driven. Downward or lateral spread can have much
lower speeds, and smoldering almost stands still at 0.001–0.01 cm/s. All
of these speeds contribute to the fire growth and the extent of the burning
region. The extent of the burning region is directly related to hazard and
damage due to fire.
Theories have been described to illustrate the factors important in
fire spread and to provide a framework for computations. Fire spread
depends on many factors: the fuel, its orientation, the wind, the direction
of spread, and heat transfer. Fire spread velocity can be described as the
ratio of the flame heated distance to ignition time. The heated distance
depends on the reach of the fire’s direct heat flux. It has been shown how
such a simple theory can be applied to the many situations encountered
in fire spread.
This chapter has dealt with the nature of flame spread and how to esti-
mate its speed. But for most common materials, flame spread is not always
possible under normal atmospheric conditions. In opposed-flow spread,
the leading edge flame is a premixed flame. If its heat losses are too great
to sustain a flame temperature in the range of 1300°C, the flame will extin-
guish. If the opposed wind speed is too great, the fuel will not have enough
time to completely react, and the flame will die—blown out. Under wind-
aided flame spread, if the flame does not extend sufficiently in the forward
direction to heat the new material to its ignition temperature, the flame will
not advance. The issue of extinction is complex, and simple formulas do
Flame Spread 161
not generally apply. But one must always be aware that flame spread is not
automatic. Yet if it occurs, the formulas presented here can give reasonable
estimations and explanations of flame spread behavior.
Review Questions
1. Data for a cardboard match are as follows: Treat the cardboard as thick.
k = 0.20 W/m-K
ρ = 550 kg/m3
c = 2.5 J/g-°C
ϕ = 12 kW2/m3
Tig = 350°C
a. Compute the TRP of the cardboard.
Assume the flame imparts a heat flux of 50 kW/m2. At normal room
temperature, T∞ = 20°C, compute the flame spread speed in b and c.
b. For downward spread, the heating distance (δƒ) is 1.5 mm.
c. For upward spread when the flame extends from the pyrolysis front
by 2 cm.
2. A rigid foam plastic wall lining is ignited. The flame heat flux is
25 kW/m2, and after several seconds, it extends 1.5 m from the ignition
region. Compute the upward spread speed using data in Table 5.1.
3. Dry grassland has a bulk density of 25 kg/m3. On flat ground with
a wind speed of 20 mph, estimate the possible flame speed through the
grass. What if there was no wind?
4. List all the possible factors that could control flame spread.
5. Could flame spread on a steel clad foam plastic occur if there are joint
seams in the steel?
True or False
1. Thickness is not a factor in flame spread.
2. As materials are heated by hot smoke in a fire, they will spread faster
than at normal temperatures.
3. Flame spread factors on liquid fuels are the same as that on horizontal
solids.
4. The bulk density or porosity of forest brush is an important variable in
flame spread.
5. Winds do not change the spread rate but can blow out the fire.
Activities
1. Tape the edges of strips of paper so as to suspend them from a wire
clothes hanger. The tape is used for mounting and holding the paper in
place. Ignite the vertical strip to observe upward and downward spread.
See if you can observe the leading edge premixed flame in the downward
propagation. Try to clock the time for the pyrolysis front to spread over
the strip. Moisten the paper with alcohol (ethyl or methyl) and repeat the
tests. Explain your results. Caution: Such an experiment needs to be done
in a safe place, not in your home.
2. Place two 2 × 4’s about 1 foot high together. Find the gap thickness that
will allow flame to spread up or down the gap by using a match flame
size ignitor. Explain your results.
3. Obtain a shallow metal or glass tray about 6–10 in. long and 1 in. wide.
Separately pour kerosene, then methyl or ethyl alcohol into the tray about
1/4 in. in depth. Ignite one end and record the spread time. Do not let the
alcohol sit long. Explain the results. Again be cautious in doing this. It
should be done in an appropriate setting with safety precautions taken.
References
1. Standard test method for determining material ignition and flame spread prop-
erties, ASTM E-1321, in 1996 Annual Book of ASTM Standards, Sec. 4, Construction
(West Conshohocken, PA: American Society for Testing and Materials, 1996).
2. J. G. Quintiere and M. Harkleroad, New concepts for measuring flame spread
properties, NBSIR 84-2943 (Gaithersburg, MD: National Bureau of Standards,
November 1984).
3. J. G. Quintiere, A simulation model for fire growth on materials subject to a
room-corner test, Fire Safety Journal, 20 (1993): 313–339.
4. C. Chandler, P. Cheney, P. H. Thomas, L. Trabaud, and D. Williams, Fire in
Forestry, Vol. 1 (New York: John Wiley & Sons, 1983), p. 23.
5. P. H. Thomas, Some aspects of the growth and spread of fire in the open,
Forestry, 20, 2 (1967): 139–164.
6. P. H. Thomas, Rates of spread of some wind-driven fires, Forestry, 44, 2 (1971):
155–175.
7. Y. Nagano, Fires, in Comprehensive study of the Great Hanshin earthquake,
UNCRD Research Report Series No. 12, edited by H. Kaji (Nagoya, Japan:
United Nations Centre for Regional Development, 1995), p. 117.
8. K. Himoto, K. Mukaibo, R. Kurada, Y. Akimoto, A. Hokugo, and T. Tanaka,
A post-earthquake fire spread model considering damage of building com-
ponents due to seismic motion and heating of fire, in Proceedings of the
10th International Symposium on Fire Safety Science (2011), pp. 1319–1330.
Flame Spread 163
9. K. Akita, Some problems of flame spread along a liquid surface, in Fourteenth

Symposium (International) on Combustion (Pittsburgh, PA: The Combustion
Institute, 1973), pp. 1075–1084.
10. T. Suzuki, M. Kawamata, and T. Hirano, Flame spread over soaked sand in
an opposed air stream, in Proceedings of the Second International Symposium on
Fire Safety Science (1989), 199–208.
11. W. G. Labes, The Ellis Parkway and gray dwelling burns, Fire Technology, 2, 4
(May 1966), 287.
6
Burning Rate
Learning Objectives
Upon completion of this chapter, you should be able to accomplish the following:
• Describe the factors influencing energy release rate.

• Execute formulas to predict burning rate and energy release rate.
• Define heat of gasification and heat of combustion.
• Select or construct energy release rate signatures of real items.
6.1 Introduction
The fire triangle (fuel, oxygen, and energy) helps to explain the nature of fire,
but ignition, flame spread, and burning rate, as illustrated in Figure 6.1, com-
prise the elements of fire growth. You need this trifecta to achieve growth.
Ignition tells us when fire growth begins, flame spread theory allows us to
define the extent of the fire’s boundaries, and burning rate gives us the con-
sumption of fuel within those boundaries. In this chapter, we describe how
objects burn, gasifying their liquid or solid fuel from the heat of their own
flame. A burning rate theory is presented, but real objects demand real data.
We demonstrate how to use these data.
6.2 Definitions and Theory

Burning rate is defined as the mass of solid or liquid fuel consumed per unit
time. By consumed, we mean it chemically reacts with oxygen. In large struc-
tural fires, fuel may vaporize due to heating, but not immediately burn due
165
Burning
Ignition rate Flame spread
FIGURE 6.1
The trifecta of fire growth.
to limited oxygen. Burning is combustion; fuel supply is that flow of burnable

gases that potentially can make a flame. Therefore, one must distinguish
between the mass loss rate of fuel in fire as fuel supplied and as the mass burned
(fuel reacted with oxygen). In this chapter, we are considering the burning
of single objects primarily in air. Here, these two distinct quantities—mass
loss rate or supply rate and burning rate—are considered synonymous. We
consider all of the supplied fuel to be burned here. So, if we burn a pan of
alcohol on a scale (Figure 6.2), the heat from the flame will cause the liquid to
evaporate and its mass loss can be recorded over time. The mass lost divided
by the time to lose this mass is the mass loss rate, m, usually conveniently
expressed in grams per second (g/s). Here it is synonymous with burning
rate as it is very likely that all of the alcohol evaporated will be consumed
in the flame, leaving products of combustion obviously behind. Some of the
products of combustion can still be compounds of that might be capable of
still reacting. If the products do not go to their normal stable states (e.g., water
vapor and carbon dioxide for hydrocarbons), the chemical reaction with oxy-
gen is termed “incomplete.”
If we were to simply let the alcohol evaporate without ignition, the mass
loss rate would not be synonymous with burning. However, in this chapter,
for a material or object with free exchange of normal air, once ignited, the
fuel supply rate and burning rate will be considered to be one and the same.
In order to ignite the alcohol in the pan in Figure 6.2, it must at least be
at its flashpoint temperature. Alcohols typically have their flashpoints less
than normal air temperature, so a small flame near its surface will cause
ignition. Once a diffusion flame forms over its surface, the surface tempera-
ture increases to nearly its boiling point. The boiling point is the maximum
temperature a pure liquid can reach when heated in normal atmospheric
conditions. If we boil water at sea level (1 bar), it would reach 100°C, but
on Mt. Everest (0.260 bar), it boils at 69°C. So as a liquid fuel is heated,
its temperature rises, and its evaporation rate increases. But the tempera-
ture at the surface of a pure liquid cannot exceed its boiling point. Hence,
the alcohol will start to burn at its flashpoint as a short duration premixed
flame, then as the diffusion flame forms (fire point) and gets bigger, the
liquid temperature will reach its boiling point and stay there. During this
increase, the rate of evaporation or the burning rate also increases accord-
ingly. This leads to an initial burning rate once ignited and then an increase
Burning Rate 167
FIGURE 6.2
Measuring the burning rate.
until the rate becomes relatively steady. This changing or unsteady burn-
ing is typical of the way fuels burn. They will start low, then increase, and
try to reach a steady state. But complex fuels, like wood that char, will first
increase then decrease as the insulating char layer prevents the heat to get
to the virgin wood.
6.2.1 Burning Mass Flux

In general, the rate of burning depends on the fuel properties, its orientation
or configuration, and the area involved. Wood logs burn well together,
but not alone. A surface may burn easily on its underside, but not its top-
side. The area of burning is controlled by the ignition and flame spread pro-
cesses. The rate of burning over that area may not be steady and may not
be uniform. The character of uniformity can be expressed by the intensity
of burning over the surface. This quantity is expressed as burning rate per
unit area or mass burning flux, m ¢¢. It describes how each point on the surface
burns. If we divide the fuel supply area into small surface regions, associat-
ing each surface with an m ¢¢, then the sum of m ¢¢ times each corresponding
surface area gives us the total rate of burning. In the calculus, this would be
refined as an integration of the mass flux over its area.
Flame heat flux

Flame
0 R
r
Vaporizing 0 R
surface
FIGURE 6.3
Illustration of nonuniform burning.
This intensity of burning is analogous to the characterization of a sprin-

kler spray hitting the floor below. Putting buckets on the floor to catch the
individual water amounts can be used to find the local intensity of the spray
in terms of typically gpm/ft2. In sprinkler nomenclature, this intensity is
commonly called the “water flux density” or just “water density.”
Figure 6.3 shows an illustration of how the local burning flux can vary.
It depicts a small laminar flame, like a candle size. The flame at the edges
will be near the surface and have a high heat flux. As the flame is farthest
away from the surface at its center, the heat flux will drop. But in the cen-
ter region, the flame could be thicker and may possess some radiation as
well as convection. This added radiation could make the heat flux increase.
In this way, the heat flux can have this bimodal distribution about the center
as shown in Figure 6.3. The heat flux distribution is directly responsible for
the evaporation rate or burning rate of the fuel at each point. Therefore, each
point on the surface will have a corresponding mass flux and will vary in
accordance with this surface heat flux.
6.2.2 Heat of Gasification, L (kJ/g)

The heat of gasification can be regarded as a property of a liquid or solid
fuel. It is the energy needed to make the solid or liquid produce a vapor
and is defined on a unit mass basis of the vapor, commonly given in kJ/g. It is
an exact thermodynamic property for liquid fuels and can be expressed
with great accuracy. For a liquid, it represents the sum of energy needed
to evaporate the liquid at the surface, plus the energy needed to bring the
liquid from its original temperature to its boiling point. For a pure liquid,
the vapor produced will be the same chemical as the liquid. For a solid, the
solid might melt then gasify, or simply pyrolyze (thermal decomposition)
to a vapor. Both the melt and the vapor are rarely the same compound as
Burning Rate 169
the solid. Indeed, for charring solids, the original solid will transform in
thermal decomposition to both char and vapor, simultaneously. The heat of
gasification is not an exact property for the solid because of its decomposi-
tion to new compounds.
The utility of the heat of gasification is that it can be used to predict the
burning rate. Although imprecise and approximate for solids, its measure-
ment process shows that it can be determined with reasonable consistency.
Its subsequent use then depends on this measurement process. Typically
that process adds a known amount of energy to the solid or liquid by heat
transfer. The solid or liquid is placed on a scale, and its mass loss due to
gasification is recorded. Repeating this measurement with varying amounts
of energy or heat transfer, and averaging the results over time, leads to a
fairly constant value for the energy added per mass lost. This output is the
heat of gasification. It can depend on how the process is averaged over time.
For example, for a charring solid, it might be averaged only over the period
of high mass loss rate. Alternatively, it might be averaged over the entire
gasification time. Usually the peak burning condition is used for solids.
In any case, the heat of gasification is a precise property for a liquid, but an
approximate property for a solid.
6.2.3 Approximate Formula for Steady Burning

A general predictive formula for the mass burning rate per unit area, or mass
burning flux, is given in terms of the heat of gasification as
q ¢¢
¢¢ =
m (6.1)
L
where q¢¢ is the net heat flux to the fuel surface.

This formula applies exactly to the steady burning of a deep liquid pool.
It gives the final steady burning flux for the pool. For solids, it gives only an
average result over the flaming period. As typical data on L for solids are
based on the average peak burning rate, Equation 6.1 attempts to predict the
peak burning conditions.
For shallow liquid pools or liquid fuels saturated into solids, the formula
overestimates the mass burning flux because of heat lost into the substrate of
the liquid. Also, for something like gasoline poured on wood, the gasoline
would burn off first. The wood would assume the boiling point of the gaso-
line fuel components. As the less volatile components of the gasoline begin
to vaporize, the wood temperature would rise. If sufficient heat transfer
remains after the gasoline burns off, the wood might begin to vaporize and
continue to burn. A burning liquid fuel with a low boiling point absorbed
into the wood is likely to only cool a horizontal piece of wood (except at the
edge of the flame where the wood is dry).
6.2.4 Computing the Mass Burning Flux

The heat flux in the equation is the sum of heat from all possible sources. In a
fire, the heat comes from the flame and the heated surroundings. Because the
surface of the material burning is relatively hot, it will radiate heat to the sur-
roundings. In the formula of Equation 6.1, the net heat flux for a material in a
fire is composed of the flame above the surface (the incident flame heat flux,
q¢¢flame) and external radiant heat flux from hot surroundings (q¢¢ext ). In addition,
there is a reradiation term due to the radiant energy emitted from the surface.
The reradiation flux q¢¢rr can be estimated as approximately the blackbody
radiation from the vaporizing surface. It is always greater than the critical
flux for piloted ignition (CHF). In the event that the reradiation flux is not
known, one might use CHF as a first low estimate.
The net heat flux for a fire condition is the algebraic sum of the heat flux
components at the surface as shown in Figure 6.4. So the net heat flux is
given as
q ¢¢ = q ¢¢flame + q ¢¢ext - q ¢¢rr (6.2)
This formula would apply for a material burning in a fire that is heated by
both its flame and radiation from the surroundings. For other scenarios, it
would have to be modified. For example, a material burning on its own in
normal ambient air would only have the incident flame heat flux and the
radiant exchange between the surface and the ambient surroundings. On the
other hand, in a test apparatus to measure L, the process might take place in
nitrogen with heat transfer from a radiant heater. For this case, there would
be no flame heat flux in the formula. The measurement or estimation of these
heat fluxes in different scenarios is crucial to using Equation 6.1 to estimate
steady burning.
.
qe̋xt
.
qf̋ lame
σT 4B (reradiation)
Fuel (gas)
Fuel (solid or liquid)
FIGURE 6.4
Heat flux components causing burning on a horizontal surface.
Burning Rate 171
6.2.5 Unsteady Burning

In general, materials will not burn in a steady fashion even under fixed heat-
ing conditions. At ignition, a material will have a minimal mass flux. Once
the flame occurs, additional heat will be applied, and the burning rate will
increase. Equation 6.1 for a deep pool liquid will give the final steady burning
rate. Typically it will take some seconds to reach this value. For melting solids,
it could take minutes. For charring solids, the process is more complex because
of the increasing char layer that builds up. Initially, the charring material will
burn like a melting solid and try to achieve a steady value. However, as the
char builds up, it forms an increasing insulating layer. The char is porous,
as the vaporizing component has been depleted from the original material.
Also the char tends to be nearly all carbon and is effectively inert. This inert
increasing porous layer allows vapor from the virgin material to easily pass
through it but diminishes the heat transfer by conduction from the flame
to the virgin material. This reduced heat transfer causes the burning rate to
decrease. As the char layer increases, less and less heat can be transferred to
the virgin solid material for vaporization. The flame will be sustained until the
burning flux gets to a critically low value and extinguishes. The flame elimi-
nation is followed by direct oxidation of the char. Smoldering then ensues,
and the char slowing burns away. Once the char is depleted, flaming could
resume if sufficient heating persists at the surface. Charring materials, such as
heavy timber, likely burn in this repeated fashion in building fires.
Figure 6.5 shows how wood behaves in horizontal burning at an external heat
flux of 50 kW/m2. Up to about 350 seconds flaming is sustained, then smolder-
ing ensues. During both periods, the mass loss rates appear almost constant at
about 40 g/440 s = 1/11 g/s for flaming and about 5 g/800 s = 1/160 g/s for
smoldering. On closer inspection there is a sharp decrease in mass at about
Specimen mass
50
45
40
35
30
Mass (g)
25
20
15
10
dm/dt = –0.0918
5
0
0 200 400 600 800 1000 1200
Time (s)
FIGURE 6.5
Burning of wood with an external heat flux of 50 kW/m 2. (From Dillon, S.E. et al., Determination
of properties and the prediction of the energy release rate of materials in the ISO 9705 room-
corner test, NIST-GCR-98-753, NIST, Gaithersburg, MD, June 1998.)
20 seconds. This is the peak burning rate before the char interferes with the heat
transfer into the wood. Later at about 200 seconds, there is a slight increase in
the rate of mass loss until about 350 seconds where it slows to smoldering. This
slight increase in the mass loss rate is likely due to the noncombustible insula-
tion on the back face of the wood that traps more energy into the virgin wood.
The L value for the wood in this experiment depends on the period of burn-
ing over which it is averaged in Equation 6.1. The surface of the burning wood
increases in temperature as the char builds up, and therefore, the reradiation
heat flux increases. The flame heat flux changes as the mass loss decreases, and
the incident radiant heat flux of 50 kW/m2 remains the same at the surface.
However, the heat flux arriving at the virgin wood will decrease. The virgin
wood has inherently a reasonably well-defined heat of gasification; however,
the heat of gasification derived from such an experiment will use the known
heat flux from the radiant source, a measured heat flux from the flame at peak
burning conditions and an associated reradiation heat flux. L will likely be
computed from the measured mass loss rate of the peak burning condition.
6.2.6 Difficulties in Computing Burning Rates

In the context of the simple formula in Equation 6.1, L is only a very approx-
imate property for char formers. Nevertheless, Equation 6.1 gives us a
simple, but proper, quantitative tool for estimating the mass burning flux.
Unfortunately, the net heat flux q¢¢, or more specifically, the flame heat flux,
is not readily available from the literature. This is a quantity not easily pre-
dicted. While L is a property of the liquid or solid, the net heat flux depends
on the flame: its size, orientation, soot concentration, and flow conditions.
We must rely on data and insight to achieve practical estimates for it.
Typical mass burning fluxes for materials can range from about 1 to
100 g/m2-s. Values below 1–2 g/m2-s usually lead to extinguishment. As
shown directly by Equation 6.1, external heating added to the flame heat flux
will increase the mass burning flux. Such external heating can arise from a
fire in a room resulting in a feedback loop that can accelerate burning. The
room gets hot giving more heat flux; more heat flux makes the burning rate
increase. This added burning makes the room get hotter. This is the feedback
loop that increases the burning mass flux as the room gets hotter. However,
the oxygen level in the room can eventually mitigate this process since low
oxygen will drop the flame temperature. In general, atmospheric oxygen is
directly related to flame temperature. For that reason, burning in elevated
oxygen atmospheres will give a higher flame temperature and burning rate
than burning in vitiated (depleted oxygen) atmospheres.
Suppressing a fire by cooling with water can lead to extinguishment of
the flame as the critical mass flux level is reached. This cooling effect can be
taken as an effective heat loss flux due to evaporation and might be included
in Equation 6.2. Qualitatively, the effect of suppression by water can be
explained, but quantifying this effect has not been generally developed.
Burning Rate 173
Accordingly, Equation 6.1 can help explain a range of burning rate effects.
It can explain the increase in burning in a room, it can explain a decrease in
burning if the ambient oxygen drops, and it can explain the suppression by
water to a critical value for extinction. However, methods to predict the vari-
ous heat fluxes needed in Equation 6.1 are nonexistent or empirical at best.
6.2.7 Material Property Values for the Heat of Gasification

Typical values of heats of gasification are listed in Table 6.1. In addition, the
reradiation heat is listed, if known. These data have been assembled from vari-
ous sources and are only generically listed. A manufactured material would
only be representative of its type. For pure liquids, the heat of vaporization and
boiling temperatures are distinct and precise. For charring materials and com-
mercial products, these quantities are approximate. Chemical additives and the
physical form of the material (granular, foam, solid) can affect the value of L
for solid fuels. Indeed, a chemical retardant could increase L for the solid fuel,
making it more difficult to vaporize. As was stated, values of L for pure liquid
fuels are precise properties, while for a material such as wood, it is approxi-
mate because it includes the effect of char and changing conditions over time.
Moreover, the use of such approximate L values from Table 6.1 attempts to give
an estimate of the peak burning conditions. For steady or near peak burning
condition, the L includes the energy needed to vaporize, plus the energy needed
to bring the material to its vaporization temperature. A pot of water in full boil
will be at 100°C throughout, not just at the surface. So for pure liquids, burn-
ing at a full boil, L would not include the energy needed to bring the rest of the
liquid to its boiling point; it would already be at that value. So under full boil
conditions, L for a liquid in Table 6.1 will be lower than that given.
As the degradation process becomes more complex, as for char formers,
L is an approximate representation of the process. For solid fuels, approxi-
mate values of L have been deduced by special heating experiments based
on using Equation 6.1. In general, it can be seen from Table 6.1 that values
of L tend to increase as we move from liquids (<1 kJ/g), to melting plastics
(1–3 kJ/g), and then to char formers (3 kJ/g and higher). This is generally
true. Liquids are easiest to vaporize and char formers require a lot more
energy to get the virgin material to vaporize.
6.3 Estimating Burning Mass Flux

Although Equation 6.1 is a useful quantitative tool in explaining burning
rate, it has limited practical use. The principal problem is how to specify
the flame heat flux. This quantity can depend on the fuel, its orientation,
its configuration, and its environment.
TABLE 6.1
Burning Properties of Materials
Reradiation Heat Flux, q ¢¢rr Heat of Gasification,
Materials (kW/m 2) L (kJ/g)
Liquids
Gasoline ~0.5 ~0.33
Hexane 0.88 0.45
Heptane 1.07 0.50
Kerosene ~3.7 ~0.67
Ethanol 0.86 1.00
Methanol 0.74 0.74
Ordinary polymers
ABS, acrylonitrile–butadiene–styrene 10 1.4, 3.2
PE, polyethylene 15 1.9
HDPE, high-density polyethylene 15 2.0, 2.2
PP, polypropylene 15 2.0
PVC, polyvinylchloride 15 1.5, 2.5
PS, polystyrene foams 10–13 1.3–1.9
Nylon 6,6, polyamide 15 1.5, 2.4
PMMA, polymethylmethacrylate 11 1.6, 2.3
POM, polyoxymethylene 13 2.0, 2.4
PU, polyurethane rigid foams 14–22 1.2–5.3
PIU, polyisocyanurate foams 13–37 1.2–6.4
PE foams 12 1.4–1.7
High-temperature polymers
Phenolic foam 20 1.6
Phenolic/glass fibers 20 7.3
PC, polycarbonate 11 2.1, 4.8
Phenolic–aromatic polyamide 15 7.8
Halogenated polymers
FEP, perfluoroethylene propylene 38 2.4
ETFE, ethylene tetrafluoroethylene 48 0.8–1.8
CPVC, chlorinated PVC (25%–48% Cl) 10–12 2.1–3.1
Wood and wood productsa
Redwood — 4.6, 9.4b
Red oak — 7.9, 9.4b
Douglas fir — 6.8, 12.5b
Maple — 4.7, 6.3b
Plywood — 7.3
Plywood, FR (fire retarded) — 9.3
Chipboard, FR — 10.0
(Continued)
Burning Rate 175
TABLE 6.1 (Continued )

Burning Properties of Materials
Reradiation Heat Flux, q ¢¢rr Heat of Gasification,
Materials (kW/m 2) L (kJ/g)
Compositesa
Paper of gypsum board — 4.8
Wallpaper on rigid PU — 5.5
Sources: Dillon, S.E. et al., Determination of properties and the prediction of the energy release
rate of materials in the ISO 9705 room-corner test, NIST-GCR-98-753, NIST,
Gaithersburg, MD, June 1998; Tewarson, A., Flammability of polymers, Chap. 11, in:
Andrady, A.L., ed., Plastics and the Environment, John Wiley & Sons, Inc., Hoboken, NJ,
2003; Panagiotou, J. and Quintiere, J. G., Generalizing the flammability of materials,
in: Proceedings of the 10th International Conference on Interflam 2004, Interscience
Communication, London, U.K., 2004, pp. 895–906; Spearpoint, M.J., Predicting the
ignition and burning rate of wood in the Cone Calorimeter using an integral method,
Masters thesis, Department of Fire Protection Engineering, University of Maryland,
College Park, MD, 1999.
a Based on peak average burning.
b Heating parallel, perpendicular to wood grain.
6.3.1 Example for a Burning Wall

For burning walls, not exceeding about 2 m tall, the wall flame heat flux to
the wall appears to have an approximate constant value of 25 ± 5 kW/m2
independent of the material. This gives us a strategy for estimating the burn-
ing rate of such a wall fire.
Example 6.1
Consider a burning wall of nylon 6/6 having a vaporization temperature
of 380°C. Estimate its burning mass flux.
Assume for the flame an incident heat flux = 30 kW/m2 as a typical
wall flame.
Compute the reradiation surface heat flux from blackbody theory of
radiation for a surface temperature of 380°C.
sT 4 = 5.67 ´ 10 -11 kW/m 2 -K 4 (380 + 273)4 = 10.3 kW/m 2
This value of 10.3 kW/m2 for the reradiation flux of nylon 6/6 computed
here is in contrast to 15 kW/m2 given in Table 6.1. The difference is in the
vaporization temperature of that particular nylon or in the measurement
accuracy of vaporization temperature of the nylon.
Since this reradiation heat flux is lost from the surface, the net heat
flux is computed as
q ¢¢ = 30 – 10.3 = 19.7 kW/m 2

From Table 6.1, the heat of gasification, L, is taken as 2.4 kJ/g. The burning
mass flux is now computed as
q ¢¢ 19.7
¢¢ =
m = = 8.2 g/m 2 -s
L 2.4
If we had selected 1.5 for the heat of gasification from Table 6.1, the result
would be higher.
This example illustrates how, for this wall configuration, we can develop
an estimate for the burning mass flux. It also illustrates the variations in
the estimation process due to the uncertainty of generic properties. An
accurate value could only be obtained from precise values for the par-
ticular nylon on the wall. The fact that the computed value is above the
critical value for extinction of about 2 g/m2-s says that the nylon wall will
sustain burning. As nylon melts, no consideration of melting is included
in this estimate. The nylon could drip, the wall could fall onto the floor, or
simply distort in other ways due to melting. This aspect is not included
in the formula. It assumes the nylon stays in place on the wall with the
flame heat flux specified.
6.3.2 Pool Fires

For other configurations other than a vertical wall, we cannot easily settle
on estimates of q¢¢ without data from specific experimental measurements.
Consider the horizontal burning configuration of Figure 6.4. For liquid fuels,
a variety of experiments have been conducted. From these experiments, it
was found that the diameter of the liquid pool has a significant effect on the
heat flux, and therefore the burning flux. A typical burning mass flux curve
is shown in Figure 6.6 for methanol. For very small diameters, d < 5 cm, the
flame is laminar in character, possessing a very thin flame of nearly 2000°C.
As the diameter increases in this range, the flame sheet moves farther from
the surface, reducing the heat flux and consequently the mass flux as shown.
For the range of 5–20 cm diameter, the flame is more turbulent with a lower
heat transfer coefficient and with the average turbulent flame temperature
becoming approximately 800°C–1000°C. As we saw in Chapter 3, the con-
vective heat transfer coefficient is roughly 5–10 W/m2-°C in this turbulent
region and invariant with diameter. With temperature differences of up to
1000°C between the flame and the methanol, the convective heat flux from
the flame would be about 10 kW/m2 at most. So, for this turbulent diameter
range, the burning mass flux is nearly constant. It can be estimated using the
data in Table 6.1 as
10 kW/m 2 - 0.74
¢¢ =
m = 7.5 g/m 2 -s
1.23 kJ/g
Burning Rate 177
60
Methanol
50
Corlett and Fu5
Kung and Stravrianidis6
Burning rate (g/m2-s)
40
30
20
10
0 100 200 300
Diameter (cm)
FIGURE 6.6
Burning mass flux for methanol pool fires as a function of diameter. (From Corlett, R.C. and Fu,
T.M., Pyrodynamics, 4, 253, 1966; Kung, H.C. and Stravrianidis, P., Buoyant plumes of large-scale
pool fires, in: Nineteenth Symposium (International) on Combustion, The Combustion Institute,
Pittsburgh, PA, 1982, pp. 905–912.)
This computed value is lower than the experimental value in Figure 6.6.
(At about a diameter of 20 cm, the burning flux is about 12 g/m2-s.) It shows
the accuracy of trying to predict the flame heat flux. Perhaps adding some
radiation heat flux from the flame might make the estimate better.
As the diameter further increases, the size and thickness of the turbu-
lent flame increases, providing an increasing source of additional radiant
heat flux from the thick flame. However, as we saw in Chapter 3, the flame
radiation reaches a maximum value when the flame emissivity becomes
equal to 1. As a result, the burning mass flux attains a maximum value
when the flame emits at this maximum. The occurrence of this maximum
value depends on the sooting tendency of the fuel but generally occurs
for a fire diameter of about 1–2 m. A similar result is shown in Figure 6.7
for gasoline pool fires with diameter values greater than 20 cm. The maxi-
mum value is seen to reach 55 g/m2-s. All fuels capable of burning in a
horizontal pool configuration will attain this type of maximum burning
flux. As it occurs around diameters of 1 m or above, it can provide a very
good upper limit estimate in computing burning rates in fire for fuels
over or near this size.
0.200
0.100
m˝ (kg/m2-s)
0.050
.
0.020
0.010
0.2 0.5 1.0 2.0 5.0 10 20
Pool diameter (m)
FIGURE 6.7
Mass burning flux for gasoline pools. (From Babrauskas, V., Burning rates, Chap. 3-1, in:
DiNenno, P.J., ed., SFPE Handbook of Fire Protection Engineering, 2nd edn., National Fire
Protection Association, Quincy, MA, June 1995.)
6.3.3 Maximum Burning Rates

These maximum pool fire burning mass flux values have been experimen-
tally determined for other liquid fuels, as well as for solid fuel materials,
and even arrays of objects. They have been tabulated from various sources
in Table 6.2. Although these values must be taken as very rough estimates
for the solid fuels, they do offer a framework for computing worst-case
burning conditions. Most of the materials in Table 6.2 burn in pool-like con-
figuration with a circular area surface. The last two items—cartons and cribs
(array of wood sticks)—should be regarded as low piles with the burning
area primarily their top surface. For some of the materials, the flame heat
flux has been roughly estimated from using the information in Table 6.1
and Equation 6.1. The maximum burning flux is highest for the liquefied
gases, drops off to the liquid fuels, melting solids, and finally, the charring
materials have the lowest. The incident flame heat fluxes might be surpris-
ing for these big fires. The flames are very turbulent and tall compared to
their diameters. This will encourage high radiation from the flames, but
the large body of raw fuel near the pool surface and the possibility of soot
there as well can block the radiation like a cloud of smoke. So the process of
flame heating is very complex. The flame heat fluxes have been estimated in
Table 6.2 by Equation 6.1 and data from Table 6.1. These incident flame heat
fluxes range from about 20–80 kW/m2. The order of magnitude is likely cor-
rect, but the individual accuracies cannot be assessed. Some of the variation
is likely due to the absorption of the flame radiation by the fuel vapor over
the surface of the pool.
Burning Rate 179
TABLE 6.2
Maximum Burning Flux Values and Associated Flame Heat Flux Levels
Estimated Flame Heat
Fuel ¢¢ (g/m 2-s)
Mass Flux, m Flux (kW/m 2)
Liquified propane 100–130 —
Liquified natural gas 80–100 —
Benzene 90 —
Butane 80 —
Hexane 70–80 35
Xylene 70 —
JP-4 50–70 —
Heptane 65–75 36
Gasoline 50–60 19
Acetone 40 —
Methanol 22 28
Polystyrene (granular) 38 70
Polymethyl methacrylate (granular) 28 78
Polyethylene (granular) 26 80
Polypropylene (granular) 24 67
Rigid polyurethane foam 22–25 —
Flexible polyurethane foam 21–27 64
Polyvinyl chloride (granular) 16 37
Corrugated paper cartons 14 38
Wood crib 11 66
Source: Tewarson, A., Generation of heat and chemical compounds in fires, Chap. 3–4, in:
DiNenno, P.J., ed., SFPE Handbook of Fire Protection Engineering, 2nd edn., National
Fire Protection Association, Quincy, MA, June 1995.
The high surface heat flux for the wood crib is an artifact of the stick array.
There is a relative high vaporization rate within the pile due to radiation
exchange from the sticks, and this gives a relatively high mass flux at the
top plane surface and associated heat flux. It is very likely that a 1 m diam-
eter single flat piece of wood in horizontal burning would barely sustain
burning if at all. An empirical, but accurate formula to estimate the average
burning rate of cylindrical or square sticks of wood depends inversely on
the thickness dimension, d (cm), as
0.9 ± 0.1
¢¢ =
m mg/cm 2 -s
d1/2
For example, the average burning rate of thin match flame from a stick diam-
eter of 2 mm would be 0.9/(0.2)1/2 = 2.01 or 20.1 g/m2-s. Thin objects can
easily store energy and burn at higher levels than thick objects. Note this thin
matchstick flame can burn at a higher mass flux than the wood crib array in
Table 6.2.
6.4 Energy Release Rate, Q

Perhaps the most important quantity related to fire is the energy release rate,
or the firepower measured in kilowatts, kW, and denoted here by the symbol,
In much of the literature in fire, this has been called “heat release rate,”
Q.
denoted by HRR. It, more than any other factor, represents the size of the fire
and its potential for damage. We shall see that it is directly related to flame
height, and from Chapter 3, we saw that it is directly related to the radiant heat
flux surrounding the fire. The entire potential for fire growth and flashover in
a room can be related to Q.
So HRR in kilowatts is the power of the fire. It is a key number in assess-
ing the “size” of the fire. It is synonymous to the power of a light bulb, your
furnace, and even your car (as 1 kW = 1.34 hp). Most average persons do not
consider kilowatt as a measure of fire. You should.
Once I was interviewed on national early morning television about a
notable fire that had just occurred. I was a novice and seduced by the
reporters into speaking. They promised I would not be asked specific
questions about the fire, only about the investigative process. Well, such
an agreement is void as soon as you as go on the air. After many fire ques-
tions that I fielded and dodged, since we had not had any time to really
bear into the investigation, I was asked, “Just what do you mean by a
large ignition source?” When I said about “200–300 kW,” the questions
took another path. The response indicated the interviewer did not know
how to digest the size of the fire in kilowatts. This is the power of knowing
what kilowatts mean in fire.
6.4.1 Heat of Combustion, ΔHc

The energy release rate is found through knowledge of the burning mass
flux. This burning flux times the area over which it acts give us the total
burning rate (g/s). The heat of combustion is a property of the fuel that tells
us how many kilojoules (kJ) of energy we get for each gram (g) of the fuel
gases burned. Consequently, we can write the formula
Q = m
¢¢ADH c (6.3)
where
A is the area involved in vaporization
ΔHc is the effective heat of combustion
The term “effective” heat of combustion is used as opposed to the theoreti-

cal or complete heat of combustion. The former applies during the flaming
portion of the fire, while the latter assumes all of the fuel burns to chemical
Burning Rate 181
species that can no longer still burn. This is “complete” combustion. The heat
of combustion represents the chemical energy released per unit mass of vaporized
fuel that is reacted. It is possible to measure the theoretical complete value for
a solid such as wood in a device called an “oxygen bomb.” In this device, all
of the combustible content of the wood would be reacted and the resultant
energy and mass consumed would be measured. Only inorganic ash would
remain. Typically this heat of combustion for a wood (theoretical) would be
approximately 19 kJ/g. If the same measurements were conducted under
typical fire conditions for wood, we would measure ΔHc of approximately
12–15 kJ/g for the flaming period and ΔHc of approximately 30 kJ/g for the
smoldering phase of the char.
6.4.2 Heat of Combustion of Wood

It should be appreciated that the g (grams) in kJ/g is the mass of the fuel
being reacted. For example, if about 70% of the wood turns to fuel gas, and
the remaining 30% burns as char, then as energy must be conserved, using
14 kJ/g for the heat of combustion of the fuel gas, and 30 kJ/g for that in
smoldering:
(14 kJ/g gas)(0.7 gas/wood) + (30 kJ/g char)(0.3 char/wood) = 18.8 kJ/g wood
This shows how wood releases its energy in two stages: flaming and smol-
dering. The process is illustrated in Figure 6.8 for a typical wood-like mate-
rial. It corresponds to the mass loss in Figure 6.5. In the figure, the power
per unit area reaches a peak at about 20 seconds just after ignition. This is
followed by surface charring, insulating the heat flux, and the power. A sec-
ond peak occurs as the virgin wood becomes thermally thin and its backside
insulation reduces heat conduction. So more energy stays in the thin wood
and its power rises again, but eventually it falls again as the virgin wood is
expended. But now the remaining char structure smolders beginning at 380
seconds. The heat of combustion measured as well during the periods of
flaming and smoldering clearly shows an average heat of combustion dur-
ing flaming of about 12–14 kJ/g and about 30–35 kJ/g during smoldering.
This heat of combustion was determined by dividing the measured energy
by the mass loss in a controlled experiment.
As the char layer builds, flaming will eventually cease at a mass loss flux
of about 2 g/m2-s or less. In the data of Figure 6.8, flaming ceases at about
50 kW/m2 and the heat of combustion at that time is about 20 kJ/g. This
gives a mass flux at extinction of 2.5 g/m2-s. Smoldering then consumes the
char by oxidation has a decreasing burning flux of 1.5 to about 0.5 g/m2-s.
At the peak burning condition in Figure 6.8, the wood achieves about
17 g/m2-s (260 kW/m2/15 kJ/g).
Heat release rate

400
Heat release rate (kW/m2)
350
300
250
200
150
100
50
0
0 200 400 600 800 1000 1200
Time (s)
Heat of combustion
40
Heat of combustion (kJ/g)
35
30
25
20
15
10
5
0
0 200 400 600 800 1000 1200
Time (s)
FIGURE 6.8
Heat release rate and corresponding heat of combustion for wood burning horizontally under
and external radiant heat flux of 50 kW/m 2. (From Dillon, S.E. et al., Determination of proper-
ties and the prediction of the energy release rate of materials in the ISO 9705 room-corner test,
NIST-GCR-98-753, NIST, Gaithersburg, MD, June 1998.)
6.4.3 Heats of Combustion of Materials

In this chapter, we are primarily concerned about the effective heat of com-
bustion during the flaming period. In general, both the production of char
and the formation of incomplete products of combustion such as soot and
carbon monoxide reduce the heat of combustion during flaming. Chemical
additives—retardants—can also play a role here to reduce the heat of combus-
tion. That is why ΔHc must be measured for a specific material to be accurate.
The effective heats of combustion during flaming tend to be highest for gas
and liquid fuels, and least for char formers. Table 6.3 lists typical values.
6.4.4 Heat Release Parameter

An important parameter in assessing fire hazard is the ratio of ΔHc/L. This
has been termed the Heat Release Parameter, HRP. It is important because HRP
expresses the combustion energy released per energy required to vaporize
Burning Rate 183
TABLE 6.3
Energy Release Properties of Materials
Heat of Combustion,
Polymers ΔHc (kJ/g) HRP, ΔHc /L (—)
Gases
Methane 50.0 —
Ethane 47.5 —
Ethene 50.4 —
Propane 46.5 —
Carbon monoxide 10.1 —
Liquids
n-Butane 45.7 118
n-Hexane 43.8 97
Heptane 44.6 89
Gasoline 43.7 135
Kerosene 43.2 65
Benzene 40.0 75
Acetone 30.8 49
Ethanol 26.8 27
Methanol 19.8 27
Ordinary polymers
High-density polyethylene, HDPE 40.0 21
Polyethylene, PE 43.4 37
Polypropylene, PP 44.0 32
Polystyrene, PS 35.8 17
Nylon 27.9 30
Nylon 6 28.8 21
Polyoxymethylene, POM 13.4 6, 13
Polymethylmethacrylate, PMMA 24.2 12
Polybutylene terephthalate, PBT 20.9 23
Acrylonitrile–butadiene–styrene, ABS 30.0 14, 21
ABS-FR 11.7 4
Polyurethane, PU foam 18.4 19
Polyvinylchloride, PVC 9–11 3, 6
Chlorinated PVC 5.8 1
Polyester/glass fibers (30%) 16.0 6
Polyvinyl ester 22.0 13
Epoxy 25.0 11
Epoxy/glass fibers (69%) 27.5 2
High-temperature polymers and composites
Polycarbonate, PC 21.9 5, 21
Cross linked polyethylene (XLPE) 23.8 5
Polyetheretherketone, PEEK/glass fibers (30%) 20.5 7
Phenolic/glass fibers (45%) 22.0 5
(Continued )
TABLE 6.3 (Continued )

Energy Release Properties of Materials
Heat of Combustion,
Polymers ΔHc (kJ/g) HRP, ΔHc /L (—)
Wood
Douglas fir 14.7 1–2
Hemlock 13.3 1.5–3
Plywood 11.9 1.6
Plywood, FR 11.2 1.2
Chipboard, FR 9.2 0.9
Textiles
Wool 19.5 5
Acrylic fiber 27.5 6
Composites
Paper on gypsum 6.4 1.3
Wallpaper on rigid PU 18.9 3.4
Sources: Dillon, S.E. et al., Determination of properties and the prediction of the energy release
rate of materials in the ISO 9705 room-corner test, NIST-GCR-98-753, NIST,
Gaithersburg, MD, June 1998; Tewarson, A., Flammability of polymers, Chap. 11 in:
Andrady, A.L., ed., Plastics and the Environment, John Wiley & Sons, Inc., Hoboken,
NJ, 2003; Panagiotou, J. and Quintiere, J. G., Generalizing the flammability of materi-
als, in: Proceedings of the 10th International Conference on Interflam 2004, Interscience
Communication, London, U.K., 2004, pp. 895–906.
the fuel. The HRP values are also listed in Table 6.3. As seen from the table,
the HRP values range from about 1 to 100, as materials move from char form-
ers to liquid fuels. This is a tremendous difference. It can be seen that liquid
fuels are comparably more dangerous than solid fuels, as they will produce
much more HRR for a given area burning. The HRP is an indicator of haz-
ard for a material, as a high value implies a lot of energy is released or little
energy is required to cause burning. Materials that have a HRP lower than
about 3 are not likely to burn since only a portion of the energy released is
transferred back to the fuel. Most of the energy is transported away in the
buoyant flow, and some is lost to the surroundings by radiation.
6.5 Estimating Energy Release Rate

Example 6.2
Compute the mass loss rate per unit area and the energy release rate for
a 1 m diameter pool fire of wood, polystyrene, heptane, and gasoline.
Treat the wood as a crib, not a flat slab. Assume the maximum burning
rate per unit area applies for these relatively large pool fires. Figure 6.9
depicts the dynamic feature of this pool fire example.
Burning Rate 185
.
Q, Energy release
rate (kW)
.
q˝, Heat flux
. to surface
m˝
Air
Fuel
1 m Diameter
FIGURE 6.9
Example configuration as a pool fire and its dynamics of burning. (Note: Fuel vapor near
surface can block heat and reduce q¢¢.)
1. Wood. From Table 6.2 for wood cribs, the maximum burning
flux is m ¢¢ = 11 g/m 2 -s. (Note: If this were actual wood cribs or
stacked pallets, we would have to consider the exposed sur-
face. Instead, here we consider only a top surface representing
a shallow array.)
p 2 p
A= D = (l m)2 = 0.785 m 2
4 4
Q = m
¢¢ADH c
= (11 g/m 2 -s)(0.785 m 2 )(15.0 kJ/g )
= 130 kJ/s = 130 kW
2. Polystyrene. Table 6.2 gives for a granular fuel
¢¢ = 38 g/m 2 -s
m
Q = m
¢¢ADH c
= (38 g/m 2 -s)(0.785 m 2 )(39.85 kJ/g )
= 1189 kW
3. Heptane. Table 6.2 gives 65–75 g/m2-s for the maximum mass
flux. Therefore, we see some differences that must depend on
the source of the data. We will use the higher value, 75 g/m2-s.
Q = (75 g/m 2 -s)(0.785 m 2 )( 44.6 kJ/g)

= 2626 kW
4. Gasoline. Again, using Table 6.2 we find 55 g/m2-s.
Q = m
¢¢ADH c
= (55 g/m 2 -s)(0.785 m 2 )( 43.7 kJ/g)
= 1887 kW
Summary for 1 m diameter pool fires.
¢¢ (g/m2 -s)
m Q (kW) HRP (−)
Wood 11 130 1–2
Polystyrene 38 1189 17
Heptane 75 2650 89
Gasoline 55 1887 135
These results are very illustrative of the relative potential for damage of
these four fuels. Despite their relatively small area (1 m diameter), the
liquid fuels would present fires that can rapidly bring typical residential
rooms to full involvement. The polystyrene would also pose a serious
threat, but the wood at this size would be a manageable fire.
A relatively low net heat flux of 19 kW/m2 is causing the gasoline fire com-
pared to 70 kW/m2 for the polystyrene fire. These heat flux values can
be computed from Equation 6.1 and Table 6.1 and are listed in Table 6.2.
The difference in these heat fluxes accentuates our lack of understanding in
how to predict them from theory. Their difference is almost counterintuitive,
because the gasoline fire is larger in power, Q . One explanation suggests
these differences are due to radiation absorption. The large amount of vapor-
ized gasoline by-products near the fuel surface acts like a cloud in blocking
the flame radiation.
Tables 6.2 and 6.3 provide a very practical way to estimate burning rate
and firepower for a number of common fuels in a horizontal orientation.
It gives an upper bound and is likely very accurate for fires above 1 m in
diameter. More data can be obtained in the literature. For a vertical orienta-
tion, using a flame heat flux of about 25 kW/m2 can be a starting point with
Equation 6.1.
Burning Rate 187
The method of computing the firepower and burning rate for materials is
limited. It is for a static fire. It does not address a real object. In principle, it
is possible to synthesize the actions of ignition, flame spread, and burning
on horizontal and vertical surfaces to represent a real object. In practice,
this is impractical. An object ignited and spreading to its full involvement is
difficult to compute from first principles.
6.6 Experimental Firepower (HRR) Results for Selected Items

The only practical way to determine the burning rate or energy release rate of
an item is by direct measurement. This measurement depends on the man-
ner of ignition, particularly for the initial fire growth over time. Weighing
devices or load cells can be used to determine mass loss while an item is
burning. If the products of combustion are collected in an exhaust duct and
the oxygen content of these gases is measured, the energy release rate can also
be determined. Such a device is called an “oxygen consumption calorimeter”
(Figure 6.10). The device works on the principle that the heat of combustion
per unit mass of oxygen consumed is nearly a constant (13 kJ/g) for a wide
range of ordinary fuel compounds. The energy release rate is found directly
from recording the rate of oxygen consumed. This type of device is available
in many forms. Some can measure small material samples, others are very
large devices that might contain a small house. Figure 6.11 shows a large item
being burned at the SP Technical Research Institute of Sweden. The prod-
ucts of combustion are fully collected in the hood, and they are measured
for oxygen consumption in the exhaust. The firepower (kW) is computed as
13 kJ/g times the consumption rate of oxygen (g/s).
Typical examples of results for upholstered furniture are shown in
Figure 6.12. Typically, the early part of the growth curve depends on the igni-
tion process. The steep increase in Q is due to the overall fire growth—flame
spread and burning rate—for the furniture. The decay from the peak energy
release rate reflects the burnout of various portions of the furniture. The width
of the curves, or more precisely, the area under them, represents the amount
of burnable fuel (in joules, J). The amount of burnable fuel or available energy
is responsible for the duration of the fire. Duration is just as important as the
peak power of the item. Indeed, duration of building fires is responsible for
the response of structure to the fire. The longer a fire burns, the more the
structure can be heated. At a critical temperature, the structure can fail.
The simplest way to estimate energy release is to measure mass loss rate
by burning an object on a scale. From Equation 6.2, the Q is found by mul-
tiplying this mass loss rate by an appropriate heat of combustion. Table 6.4
gives some typical peak values of burning rate for real items along with very
rough estimates of their peak energy release rates. These are presented to
primarily illustrate the spectrum of firepower, not for precise generic use.
Smoke
transmission path
Exhaust fan
(1.5 m3/s)
Gas sampling ring probe

Thermocouple
H Bidirectional velocity probe
Water-cooled skirt
Gardon gage Wastebasket

simulation burner
Weighing platform
(0–50 kg live load)
FIGURE 6.10
The furniture calorimeter. (Reprinted with permission from Babrauskas, V., Burning rates,
Chap. 3-1 , in: DiNenno, P.J., ed., SFPE Handbook of Fire Protection Engineering, 2nd edn., National
Fire Protection Association, Quincy, MA, June 1995.)
Figures 6.12 through Figure 6.22 give experimental results for a range
of items, including trash, televisions, Christmas trees, upholstered chairs,
and pallets. Again their peak value and their duration are both important.
Such data are important for making estimates of how typical furnishing
can affect fire growth. Starting with one object, its interaction with another
object can be addressed. For example, we know it takes time to ignite. If one
of these items were to ignite another object, it would have to impose a suf-
ficient heat flux for a sufficient time. From Chapter 3, knowing the firepower
can give the radiant heat flux to the surroundings. Knowing the ignition
properties of the next object can provide a means to assess the potential for
ignition. Such a calculation can tell if the first fire is capable of spreading
to other furnishings. Of course, it must be emphasized that these data in
Figures 6.12 through 6.22 are specific. They may be used as generic rep-
resentations, but each object is specific. The data depend not just on the
properties of the object but also on how it was ignited, as that ignition likely
controlled the early rate of spread.
Burning Rate 189
FIGURE 6.11
Sofa burning in a large calorimeter. (Courtesy of SP Technical Research Institute of Sweden,
Borås, Sweden.)
3000 Large hood tests

Rate of energy release (kW)
F32 (sofa)
2000
F31 (loveseat)
F21 (single chair)
1000
0 200 400 600 800 1000

Time (s)
FIGURE 6.12
Typical upholstered chair energy release rates. (Reprinted from Babrauskas, V., Burning rates,
Chap. 3-1, in: DiNenno, P.J., ed., SFPE Handbook of Fire Protection Engineering, 2nd edn., National
Fire Protection Association, Quincy, MA, June 1995. With permission.)
TABLE 6.4
Typical Peak Burning Rate and Firepower Values
Generic Item Estimated Burn Rate (g/s) Estimated Firepower (kW)
Small waste containers (18–40 L) 3–6 100
Large waste containers (70–120 L) 5–10 300
Chairs, wood and upholstered 10–60 1200
Sofas 20–100 2000
Beds 20–140 2500
Closet ~40 1000
Office ~90 2300
Bedroom ~130 3300
Kitchen ~190 4800
House ~30,000 750,000
Source: Based on data from Quintiere, J., Growth of fire in building compartments, in: Robertson,
A.F., ed., Fire Standards and Safety, ASTM STP 614, American Society for Testing and
Materials, West Conshohocken, PA, 1977, pp. 131–167.
600
Rubbish
500
One bag, straw, grass, duff,
4.1 kg total
400 Three bags, paper trash,

3.51 kg total
300 Two bags, paper trash,

2.34 kg total
One bag, paper trash,

200
1.17 kg total
100
0
0 1 2 3 4 5 6 7 8 9 10
Time (min)
FIGURE 6.13
Energy release rate of representative rubbish. (Reprinted from Babrauskas, V., Burning rates,
These results (Figures 6.12 through 6.22) show the variation in burning
behavior within the category of a single generic item. This variation can be
due to many factors: size, materials, nature of the ignition source, and the
effect of fire retardants. As with Table 6.4, they represent a range of typical
values. To accurately know how an item will burn, it must be tested in a
device similar to that shown in Figure 6.10.
Burning Rate 191
300
Test 3
200
Energy release rate (kW)
100
Test 1
0
0 300 600 900 1200 1500
Time (s)
FIGURE 6.14
Energy release rate results for television sets. (Reprinted from Babrauskas, V., Burning rates,
Table 6.5 gives typical experimental values for warehouse commodities in

terms of their floor area (A) covered. These firepower (HRR) values per unit
area, Q ¢¢ should be interpreted as peak average values for a fixed floor area
covered. Their duration of burning (tb) can be found by knowing the total
mass (m) available to burn:
mDH
tb = c (6.4)
Q¢¢A
By putting the average peak power together with the burn time, we can
construct a very approximate rendition of the power curves for a particular
object. In the next section, we will see how this construction can be perhaps
improved.
800
Test 17
600
Test 18
Energy release rate (kW)
400
200
Test 16
0
0 300 600 900
Time (s)
FIGURE 6.15
Energy release rate results for Christmas trees. (Reprinted from Babrauskas, V., Burning rates,
6.7 Fire Growth Rate

As we have seen, the fire growth rate depends on the ignition process; flame
spread, which defines its perimeter; and the mass burning flux over the area
involved. For items of furniture and commodities, this complex process can-
not be predicted by simple formulas. However, each item, due to its composi-
tion and configuration, must have a characteristic growth time. In other words,
a given item, once ignited, may achieve 1 MW (megawatt, a million watts)
in say 130 seconds. For another object, it might take 80 seconds. Such data
have been compiled and are useful in fire analyses. Also it has been found
that the growth rate approximately follows a relationship proportional to time
squared. Considering the flame spread velocity as constant over time can pro-
vide an explanation for this time-squared behavior. Then, the fire area would
grow as t2 (length is velocity times time, and area is related to length squared).
Burning Rate 193
500
450
400
350
300
250
200
150
100
50
0
0 300 600 900 1200 1500 1800 2100 2400 2700 3000
Time (s)
Furniture calorimeter
FIGURE 6.16
Upholstered chair. (After Gross, D., Data sources for parameters used in predictive modeling
of fire growth and smoke spread, NBSIR 85-3223, National Bureau of Standards, Gaithersburg,
MD, September 1985.)
Alternatively, if the rate of fire growth depends on the fire itself, the growth
rate would be exponential (et), not t2. Both of these growth rates have been sug-
gested for the early fire growth. But the t-squared relationship is simplest and
it seems to work well. The empirical growth rate formula commonly used is
Q = at 2 (6.5)
1000
900
800
700
600
500
400
300
200
100
0
0 300 600 800 1200 1500 1800 2100 2400 2700 3000
Time (s)
FIGURE 6.17
where a is a constant associated with the item. If t1 is the characteristic time

to reach 1 MW, then a = 1 MW/t12
and
2
ætö
Q = ç ÷
è t1 ø (6.6)
This growth rate expression has been fitted to experimental data by ignor-
ing the early incubation period where the growth process is establishing a
Burning Rate 195
3000
2700
2400
2100
1800
1500
1200
900
600
300
0
0 300 600 900 1200 1500 1800 2100 2400 2700 3000
Time (s)
FIGURE 6.18
foothold. This incubation period is most likely due to spurious flame spread
and the nature of the ignition process. The t2 fit is illustrated in Figure 6.23,
along with an illustration of the incubation time. The precise incubation time
is impossible to predict. However, it can be of significance if one is using the
t-squared growth rate to estimate the time for detection. Detection may have
occurred during the incubation growth period.
Table 6.6 lists fire growth times to 1 MW (t1) corresponding to t2 fires
for commodities that have unlimited lateral extent for growth. In reality,
at some point, the decay process must occur as indicated by the burning
curves shown in Figures 6.12 through 6.22. In other words, Table 6.6 gives
4000
3600
3200
2800
2400
2000
1600
1200
800
400
0
0 300 600 900 1200 1500 1800 2100 2400 2700 3000
Time (s)
FIGURE 6.19
Sofa. (From Gross, D., Data sources for parameters used in predictive modeling of fire growth and
smoke spread, NBSIR 85-3223, National Bureau of Standards, Gaithersburg, MD, September 1985.)
the growth time; then for a given area of that fuel item, Table 6.5 would give
its peak power. In principle, the process of estimating the growth rate of an
actual item using the t-squared behavior can be done from Equations 6.3
and 6.4 and Tables 6.5 and 6.6.
Example 6.3
Consider 18 ft2 of wood pallets stacked 5 ft high. Let us construct a fire-
power graph. From Table 6.6, t1 is 97–187 seconds. Take t1 = 100 s as a
representative value for illustration. Using Equation 6.5:
2
æ t(s) ö
Q = ç ÷ MW
è 100 ø
Burning Rate 197
SIDE B
1000
900
800
700
600
500
400
300
200
100
0
0 300 600 900 1200 1500 1800 2100 2400 2700 3000
Time (s)
FIGURE 6.20
Mattress. (From Gross, D., Data sources for parameters used in predictive modeling of fire
growth and smoke spread, NBSIR 85-3223, National Bureau of Standards, Gaithersburg,
Typical pallet stack

4 Height = 1.22 m
3
Energy release rate (MW)
0
0 100 200 300 400 500 600 700 800 900
Time (s)
FIGURE 6.21
Typical energy release rate for a wood pallet stack. (Reprinted from Babrauskas, V., Burning
rates, Chap. 3-1, in: DiNenno, P.J., ed., SFPE Handbook of Fire Protection Engineering, 2nd edn.,
National Fire Protection Association, Quincy, MA, June 1995. With permission.)
From Table 6.5, the peak energy release rate is computed as
Q = (3.7 MW/m 2 ) ´ (18 ft 2 ´ 0.0929 m 2 /ft 2 ) = 6.19 MW
It will take a time to reach this level. It can be computed by equating the
firepower:
t grow = 100 ´ (6.19)1/2 = 249 s
The duration of this fire will depend on how much wood is avail-
able in the pallet stack. Suppose the pallets weighed 200 kg, and the
wood heat of combustion is 15 kJ/g. The available energy to consume
would be 200 kg × 15 MJ/kg = 3000 MJ. This entire pile would burn
Burning Rate 199
14
Wood pallet stacks
1.22 m × 1.22 m × 0.14 m each pallet
12 Assumed: ΔHc = 12 MJ/kg
Peak energy release rate (MW)
10
4 points
8
4
Mean of 8 points
0
0 1 2 3 4 5 6 7
Pile height (m)
FIGURE 6.22
Dependence of pallet burning rate on stack height. (Reprinted from Babrauskas, V., Burning
rates, Chap. 3-1, in: DiNenno, P.J., ed., SFPE Handbook of Fire Protection Engineering, 2nd edn.,
National Fire Protection Association, Quincy, MA, June 1995. With permission.)
2 at2
Q
.
1 MW
Incubation time
0 Time
0 t1
FIGURE 6.23
Fire growth rate curve of t2 fitted to data. (From Heskestad, G., Venting practices, Sect. 6,
Chap. 10 in: Cote, A.E. and Linville, J.L., eds., Fire Protection Handbook, 17th edn., National Fire
Protection Association, Quincy, MA, 1991, pp. 6.104–6.116.)
TABLE 6.5
Fire Behavior of Warehouse Commodities: Fully Involved Energy Release
Rates for a Fixed Floor Area
Commodity
Q/Floor Area Covered (MW/m2 )
Methanol 0.72
Diesel oil 1.9
Kerosene 2.2
Gasoline 2.2
Wood pallets, stacked 1½ ft high 1.3
Wood pallets, stacked 5 ft high 3.7
Mail bags, filled, stored 5 ft high 0.39
PE letter trays, filled, stacked 5 ft high 8.2
PS insulation board, rigid foam, stacked 3.1
14 ft high
PU insulation board, rigid foam, stacked 1.9
15 ft high
PS tubs rested in cartons, stacked 14 ft high 5.1
FRP shower stalls in cartons, stacked 15 ft high 1.2
PE bottles in cartons, stacked 15 ft high 1.9
PS toy parts in cartons, stacked 15 ft high 2.0
PE trash barrels in cartons, stacked 15 ft high 2.9
Cartons, compartmented, stacked 15 ft high 2.2
PVC bottles packed in cartons, 3.3
compartmented, stacked 15 ft high
PP tubs packed in cartons, compartmented, 4.2
stacked 15 ft high
PE bottles packed in cartons, compartmented, 6.1
stacked 15 ft high
PS jars packed in cartons, compartmented, 14.0
stacked 15 ft high
Source: Based on data from Heskestad, G., Venting practices, Sect. 6, Chap. 10 in:
Cote, A.E. and Linville, J.L., eds., Fire Protection Handbook, 17th edn.,
National Fire Protection Association, Quincy, MA, 1991, pp. 6.104–6.116.
Notes: PE, polyethylene; PU, polyurethane; PVC, polyvinyl chloride; PS, poly-
styrene; PP, polypropylene; FRP, fiberglass-reinforced polyester.
in about 485 seconds (3000 MJ/6.19 MJ/s). Here, no account of the

energy consumed during the t-squared growth period was included
in the previous time estimate. To do so would involve some calculus.
If that were done the actual burnout time would be about 650 seconds
following ignition. The constructed graph is shown in Figure 6.24. It
starts as t-squared, reaches the maximum of 6.19 MW at 249 seconds,
and burns out at 650 seconds.
Burning Rate 201
TABLE 6.6
Characteristic Times to Reach 1 MW for t2 Fires
Commodity t 1 (s)
Wood pallets, stacked 1½ ft high 155–310
Wood pallets, stacked 5 ft high 92–187
Mail bags, filled, stored 5 ft high 187
Cartons, compartmented, stacked 15 ft high 58
Paper, vertical rolls, stacked 20 ft high 16–26
Cotton, polyester garments in 12 ft high rack 21–42
“Ordinary combustibles” rack storage, 15–30 ft high 39–262
Paper products, densely packed in cartons, rack storage, 20 ft high 461
PE letter trays, filled, stacked 5 ft high on cart 189
PE trash barrels in cartons, stacked 15 ft high 53
PE bottles packed in compartmented cartons, 15 ft high 82
PE bottles in cartons, stacked 15 ft high 72
PE pallets, stacked 3 ft high 145
PE pallets, stacked 6–8 ft high 31–55
PU mattress, single, horizontal 115
PU insulation board, rigid foam, stacked 15 ft high 7
PS jars packed in compartmented cartons, 15 ft high 53
PS tubs nested in cartons, stacked 15 ft high 115
PS insulation board, rigid foam, stacked 14 ft high 6
PUS bottles packed in compartmented cartons, 15 ft high 8
PP tubs packed in compartmented cartons, 15 ft high 9
PP and PE film in rolls, stacked 14 ft high 38
Distilled spirits in barrels, stacked 20 ft high 24–39
Source: Based on data from Heskestad, G., Venting practices, Sect. 6, Chap. 10 in: Cote, A.E.
and Linville, J.L., eds., Fire Protection Handbook, 17th edn., National Fire Protection
Association, Quincy, MA, 1991, pp. 6.104–6.116.
Note: PE, polyethylene; PU, polyurethane; PVC, polyvinyl chloride; PS, polystyrene;
PP, polypropylene; FRP, fiberglass-reinforced polyester.
6.7.1 NFPA Design Categories

The t2 relationship has proved useful and has been adopted into NFPA 92B12
for design of smoke control systems. These NFPA standards classify the
growth times as follows and as illustrated in Figure 6.25:
Slow t1 = 600 s
Medium t1 = 300 s
Fast t1 = 150 s
Ultrafast t1 = 75 s
5
Firepower (MW)
0
0 100 200 300 400 500 600 700
Time (s)
FIGURE 6.24
Example of fire growth rate constructed for wood pallets.
Cartons 15 ft high, various contents,

fastest if empty or containing
plastic foam
Thin plywood wardrobe Full mail bags, 3 ft high
Wood pallets pallet stack
Fastest burning 5 ft high Cotton/polyester
upholstered furniture innerspring
Ultrafast Fast mattress Medium
6000
5000
4000
Q (kW)
3000
2000 Slow
1000
0
0 100 200 300 400 500 600 700
Time from ignition (s)
FIGURE 6.25
Categories of t2 fires. (Reprinted from Guide for smoke management systems in malls, atria,
and large areas, National Fire Codes, Vol. 10, NFPA 92B, 1991 edn., National Fire Protection
Association, Quincy, MA, p. 92B-30, 1995. With permission.)
Burning Rate 203
These categories are only for rough guidance and perhaps standardiza-
tion in design. They should not be used to hypothesize fire growth without
a suitable basis. The growth rate must be established by relevant data as
in Table 6.6. As the previous example illustrated, a practical application to
construct the entire growth and decay process must use data such as that
in Tables 6.5 and 6.6. The early growth can be considered as t2 up to the
full involvement, and then burning ceases when the available mass of
fuel is depleted. However, the fire does not necessarily stop abruptly as in
Figure 6.24. This abrupt ending can be refined.
6.7.2 Fire HRR of Item Constructed

Recently, Natori13 has compiled and correlated burning items in a systematic
form. These data include furnishings, appliances, pine trees (dry), and oth-
ers items tabulated completely in a working group document of Japan. Some
selected results are summarized here in Table 6.7. The burning of the items
can be represented in the form of a t-squared growth to a steady maximum
and followed by a t-squared decay. Mathematically this is expressed in terms
of a, for growth, and ad for decay, to best represent the experimental data
on energy release rate of the item. This format is shown in Figure 6.26. The
decay is shown to follow a decreasing t-squared behavior. The form of this
power curve has been successfully fitted to a large assortment of data. Table
6.7 contains some of these data. The values of the parameters from Figure
6.26 are listed in the table. Because up to 25 items of a given class have been
included in the data, there is a distribution in the results of a given class. For
example, the peak power of a sofa could reach 1826 kW on average with a
standard deviation of 676 kW. All the parameters are given by a mean and
deviation. This shows the variability in a given class, and that one cannot
generalize data for a single sofa as precise for a class of sofas. (In Table 6.7 for
the dry [Christmas] trees, the total heat release [THR] is given in MJ instead
of the time for peak burning [tmax].)
It may be somewhat more sophisticated to use these detailed data of
Table 6.7, but they represent the burning process more completely. By the
variations in Table 6.7, it should be obvious that such data can only be used as
representative, and not definitive. For the HRR of a specific item, a calorimeter
test must be made. Then, it is very likely the results can be tabulated as in the
form of Table 6.7. For rough approximations of any item, Natori13 recommends
some simple formulas that have been correlated in terms of mass and area.
She defines a cylindrical surface area (Ab) encompassing the item’s overall
diameter (D) and height (H). Also its encompassing volume is designated as
Vb. Then a is given in terms of bulk density (ρb = Mass/Vb in kg/m3) as
7.5
a (kW/s 2 ) » for wooden items
r1b.2 (6.7)
204
TABLE 6.7
HRR Parameters for Selected Items, Mean and Range
Item Mass (kg) HRR max (kW) a (kW/s2) a d (kW/s2) tmax (s)
Chair with cushioning 9.8 (8.1–15.5) 362 ± 82 0.040 ± 0.03 0.0054 ± 0.0028 118 ± 94
Chair, rigid plastic 13.1 (10.8–17.7) 594 ± 131 0.019 ± 0.009 0.0029 ± 0.0013 83 ± 81
Single chair upholstered 19.9 (11.5–28.3) 1185 ± 56 0.11 ± 0.058 0.066 ± 0.0055 56 ± 48
Sofa, upholstered 25.5 (3.8–54.6) 1826 ± 676 0.12 ± 0.078 0.078 ± 0.018 46 ± 33
THR (MJ)
Christmas trees (dry) ~5–10 800–1600 1–4 0.2–1.6 20–80
~12–20 1600–5000 0.4–9 0.4–2.4 90–200
Source: Natori, A., Fire Sci. Technol., 27(3), 310, 2008.
Principles of Fire Behavior
Burning Rate 205
.
Q max
HRR (kW)
2
at
2
a(tdecay – t)
tgrow tmax tdecay

Time (t)
FIGURE 6.26
Natori firepower diagram. (From Natori, A., Fire Sci. Technol., 27(3), 310, 2008.)
and
0.15
a (kW/s 2 ) » for plastic items
r0b.5 (6.8)
The associated maximum HRR (firepower), HRRmax, is given as
HRR max = (148 ± 90 kW/m 2 )Ab (m 2 ) for wooden items (6.9)
and
HRR max = (272 ± 204 kW/m 2 )Ab (m 2 ) for plastic items (6.10)
For estimations in HRR using Equations 6.7 through and 6.10, the portrayal
of the fire growth curve would follow that of Figure 6.24 with no gradual
t-squared decay. The duration of the fire is found by setting the area under
the power–time curve equal to the THR based on the mass and heat of com-
bustion of the item. For wood, recall the heat of combustion can range from
about 11 to 15 kJ/g, while for a plastic, it can vary from 5 to 50 kJ/g.
Many other items can be categorized. The method of ignition and the
incubation period before the t2 rate ensues are important factors in deter-
mining the precise growth rate. The incubation period could be due to
smoldering or simply the effect of a very slow initial spurious flame spread.
Recently, I investigated a fire that began from a candle burning down in
a small glass cup holder filled with wax. This issue was the response of a
smoke detector to the early fire growth. Another investigator hypothesized
the fire growth as starting directly as a “medium” fire growth in the NFPA
category of Figure 6.25. I considered the glass cup breaking and its wax
burning before an adjacent item got involved. Experiments showed this is
possible, and a surveillance video actually recorded such a start to fire in
a closed restaurant bar. The growth rates for these two scenarios are very
different.
6.8 Vehicle Fire Behavior

A very different type of growth rate compared to commodities and furnish-
ing is that in an automobile. Figure 6.27 shows the rate of growth in a typical
vintage sedan-type automobile passenger compartment of the 1990s due to a
gasoline fire beneath, following collision damage. Yet in Figure 6.28, a post-
collision fire originating in the engine compartment has a spurious growth
for the various vehicles depicted. However, the shifted curves eliminating the
incubation times show a remarkable similarity (with some typical variations)
2000
t-squared fire, at2 a = 0.18 kW/s2

Experiment
Vehicle cabin fire growth (kW)
1500
1000
500
0
0 20 40 60 80 100 120
Time after fire enters cabin (s)
FIGURE 6.27
Fire growth in a vehicle passenger compartment due to gasoline spill following collision dam-
age. (From Tewarson, A. et al., Comprehensive analysis of post collision motor vehicle fires,
FM Global Technology Report 3018009, October 2004.)
Burning Rate 207
2000
1500
Vehicle cabin fire growth (kW)
1000
500
0
0 5 10 15 20 25 30
Time after fire enters cabin (min)
#119 Battery fire #190 data shifted to origin
#190 HVAC (FR&Std) #178 data shifted to origin
#178 HVAC #203 data shifted to origin
#203 Washer fluid t-squared, 0.057 kW/s2
#119 data shifted to origin
FIGURE 6.28
Growth rate in a passenger vehicle cabin from a fire coming from the engine, postcrash. (From
Tewarson, A. et al., Comprehensive analysis of post collision motor vehicle fires, FM Global
Technology Report 3018009, October 2004.)
as a t2 fire growth in the passenger cabin.14 Again, while a t-squared fit elimi-
nating the incubation time appears reasonable, the incubation time is critical
to a fire growth originating in the engine compartment and progressing into
the cabin.
6.9 Extinction
Extinction of burning occurs if there is insufficient heat transfer to keep the
gaseous fuel supplied, and the flame is cooled to a temperature that limits
its survival. It shuts down. Recently, analysis of experiments in the burn-
ing of horizontal specimens has shown a tendency to extinguish at a criti-
cal (minimum) energy release flux (kW/m2).15 (Do not confuse this with heat
flux. This is the energy release rate [HRR] per unit area of burning surface.)
PAI
PI
CE
EP
PESU
PEEK
PEN
PPZ
PPS
PC
PBT
Material
PA6
PUR
PS
PS
PP
PP
PE Average = 64 kW/m2
PE
PMMA
PMMA
POM
Benzyl Al RQ
Xylene RQ
Decane RQ
0 20 40 60 80 100 120
Qcrit fire point (kW/m2)
FIGURE 6.29
Minimum energy flux required for burning in the cone calorimeter. (From Lyon, R.E. and
Quintiere, J.G., Combust. Flame, 151(4), December 2007.)
Figure 6.29 shows data taken for burning in air with natural convection
(associated with a convective heat transfer coefficient of about 15 W/m2-K)
for polymers burning in the 10 × 10 cm configuration. While there is varia-
tion for the different materials, it is believed that most of the variation is
inaccuracy caused by the inability to precisely measure the energy flux
just before extinction. Unsteady burning and the lag time of scale response
are responsible for such inaccuracies. It is readily seen that even for repeat
results of the same polymer (PS, PP, PE, and PMMA), the variation is great.
On average, the critical value is about 64 kW/m 2 with a standard deviation
of about 22.
It can be shown from15 that the critical energy flux for burning is depen-
dent on the heat transfer coefficient and on the oxygen concentration.
As a flame goes to extinction on a solid or liquid fuel, it becomes small
and pure convection heat transfer from the flame might be expected.
An approximate theoretical formula for the critical value of the energy flux
when burning in air is
Q b¢¢, crit » 5.4 hc (W/m 2 -K ) in kW/m 2 (6.11)

Burning Rate 209
The data shown in Figure 6.29 were taken in the cone calorimeter in which
the heat transfer coefficient is about 12 W/m2-K. The computed critical
value is 5.4(12) = 64.3 kW/m2… not bad! It should be recalled that earlier we
indicated a critical mass flux at extinction of about 2 g/m2-s. Now we see
this value depends on the heat of combustion. The heat of combustion cor-
responding to 2 g/m2-s would be 64.3/2 = 32.1 kJ/g.
6.10 Summary
After ignition a fire may grow over an area by flame spread. The burning
rate (or fuel supply rate) can be determined for a material or item by knowing
its mass burning flux (m ¢¢) , usually given in g/m2-s. Values less than 2 usu-
ally lead to extinguishment, and typical values range from 1 to 100 g/m2-s
for surfaces, and higher for stacked commodities. The property called “heat
of gasification” measures how much energy it takes to produce this gasified
fuel; the property called “heat of combustion” gives how much energy is
released when it burns. The ratio of these two properties, HRP, is a measure
of the energy produced to energy required. The bigger it is, the greater the
burning rate potential for a material.
A simple theory was presented to compute the mass burning flux from
surfaces from its heat of gasification provided the flame heat flux is known.
Despite some data for typical flame heat flux (see Chapter 3), a prediction of
flame heat flux for a given fuel is beyond current science. However, much
data exist on the maximum mass burning flux for fuels and for actual items
burned. Energy release rate data, derived from large-scale calorimeters,
provide a source for making estimates. A large amount of practical data
has been expressed in terms of t-squared fires, and these data can provide a
generic first estimate. The energy release rate (or alternatively, the burning
rate) is the most significant factor in the indication of fire hazard. It relates
to the potential for ignition of nearby items, to flashover potential in a room,
and to the rate of water needed to extinguish the fire. The critical value of the
energy flux at extinction for burning in air under natural convection is about
64 kW/m2 and appears to be independent of fuel type on average.
Review Questions
1. Plexiglas, polymethyl methacrylate (PMMA), is often used as display
windows and cases in art galleries and museums. If its flame imparts
30 kW/m2 to the surface when it burns, estimate the energy release rate of
a 2 m2 square sheet (one side). You will need Tables 6.1 and 6.3, respectively.
2. Wood pallets 5 ft high and 6 × 6 ft in their plan dimension are

ignited and burnt. Using Tables 6.5 and 6.6, respectively, estimate the
following:
a. Maximum energy release rate
b. Time to reach 1 MW
c. Its t2 energy release rate curve
3. Graph the energy release rate of a generic sofa over time.
4. What is the peak energy release rate of a dry Christmas tree weighing
about 7 kg?
5. A small wood stick match (2 mm diameter) flame has a heat transfer
coefficient of 50 W/m2-K. What is its energy flux when burning in air?
Compare this to critical energy flux for burning.
True or False
1. All the gaseous fuel produced in a fully involved room fire burns in the
room.
2. Weighing the object as it burns cannot determine its burning rate.
3. The heat of gasification and the heat of combustion are usually equal.
4. Fires on objects tend to grow proportional to time squared.
5. The heat flux from a turbulent flame, like its temperature, is nearly the
same for all fuels.
6. Wood generally burns at a lower mass flux than plastics because it chars.
7. It is impossible to collect data for burning objects and make any sense out
of it.
8. Objects burn at a higher rate in atmospheres of higher oxygen concen-
tration as contemplated for space travel.
9. Water can work for extinguishment because it can lower the flame tem-
perature and wet the surface.
10. Solid and liquid fuels can burn if their flame heat flux is high enough.
Activities
All fire experiments must be done with care in a safe place with protection
and extinguishers available.
1. Obtain small shallow glass dishes ranging in size from 1 to 10 cm in
diameter. Pour about 1/4 in. depth of a liquid fuel of high flashpoint
(>25°C) into a dish. Ignite it and time its burning period. How does
burning time vary with fuel diameter? Explain your results.
Burning Rate 211
2. Obtain a collection of small wood sticks, roughly 4 × 1/4 × 1/4 in. Make
a pile about 4 × 4 × 2 high, first putting all the sticks in contact, then
leaving 1/4 in. spacing between the sticks in alternating layers. Try to
ignite both with a match flame. Which pile burns? Why?
3. Examine the items contained in a room. From the data of this chapter,
try to estimate the potential energy release rate of each item that could
burn. Add to find their sum. Recognize that in a typical residential room,
1 MW could jeopardize it into flashover.
4. Do this with caution and safety glasses. Place a candle flame under a
clear inverted glass bowl and observe extinction. Do not allow the flame
to get near or touch the glass, and protect against cracking glass. Repeat
this with a plant that remained in the inverted bowl overnight in an
attempt to increase the oxygen.
References
1. S. E. Dillon, W. H. Kim, and J. G. Quintiere, Determination of properties and
the prediction of the energy release rate of materials in the ISO 9705 room-
corner test, NIST-GCR-98-753, NIST, Gaithersburg, MD, June 1998.
2. A. Tewarson, Flammability of polymers, Chap. 11 in Plastics and the Environment,
edited by A. L. Andrady (Hoboken, NJ: John Wiley & Sons, Inc., 2003).
3. J. Panagiotou and J. G. Quintiere, Generalizing the flammability of materials,
in Proceedings of the 10th International Conference on Interflam 2004 (London, U.K.:
Interscience Communication, 2004), pp. 895–906.
4. M. J. Spearpoint, Predicting the ignition and burning rate of wood in the Cone
Calorimeter using an integral method, Masters thesis, Department of Fire
Protection Engineering, University of Maryland, College Park, MD, 1999.
5. R. C. Corlett and T. M. Fu, Some recent experiments with pool fires, Pyrodynamics,
4 (1966): 253–269.
6. H. C. Kung and P. Stravrianidis, Buoyant plumes of large-scale pool fires,
in Nineteenth Symposium (International) on Combustion (Pittsburgh, PA: The
Combustion Institute, 1982), pp. 905–912.
7. V. Babrauskas, Burning rates, Chap. 3-1 in SFPE Handbook of Fire Protection
Engineering, 2nd edn., edited by P. J. DiNenno (Quincy, MA: National Fire
Protection Association, June 1995).
8. A. Tewarson, Generation of heat and chemical compounds in fires, Chap. 3-4 in
SFPE Handbook of Fire Protection Engineering, 2nd edn., edited by P. J. DiNenno
9. J. Quintiere, Growth of fire in building compartments, in Fire Standards and
Safety, ASTM STP 614, edited by A. F. Robertson (West Conshohocken, PA:
American Society for Testing and Materials, 1977), pp. 131–167.
10. D. Gross, Data sources for parameters used in predictive modeling of fire
growth and smoke spread, NBSIR 85-3223 (Gaithersburg, MD: National Bureau
of Standards, September 1985).
11. G. Heskestad, Venting practices, Sect. 6, Chap. 10 in Fire Protection Handbook,

17th edn., edited by A. E. Cote and J. L. Linville (Quincy, MA: National Fire
Protection Association, 1991), pp. 6.104–6.116.
12. Guide for smoke management systems in malls, atria, and large areas, National
Fire Codes, Vol. 10, NFPA 92B, 1991 edn. (Quincy, MA: National Fire Protection
Association, 1995), p. 92B-30.
13. A. Natori, Research regarding analysis of the target safety standard for fire
resistance design and its mode of expression—Estimation for the safety standard
targeted by the verification method for fire resistance performance, Fire Science
and Technology, 27, 3 (2008): 310–489.
14. A. Tewarson, J. G. Quintiere, and D. Purser, Comprehensive analysis of post
collision motor vehicle fires. FM Global Technology Report 3018009, FM Global,
Norwood, MA October 2004.
15. R. E. Lyon and J. G. Quintiere, Criteria for piloted ignition of combustible
solids, Combustion and Flame, 151, 4 (December 2007) 551–559.
7
Fire Plumes
Learning Objectives
Upon completion of this chapter, you should be able to accomplish the
following:
• Describe the meaning of buoyancy.

• Describe the behavior of fire plumes.
• Calculate flame height.
• Estimate temperature above a fire.
7.1 Introduction
All fires involve buoyancy, which sets the flow pattern and the nature of the
flame. This flow pattern directs the growth of the fire and sucks in the air
supply. It is also responsible for the characteristics of the smoke that departs
the fire. Even under wind conditions, buoyancy plays a significant role in
the fire’s behavior. In this chapter, we consider in some detail the sym-
metric vertical fire that comes from the ignition of a release of gaseous fuel.
When the fuel release is distributed on a horizontal surface (floor), this is
commonly called a “pool” fire. It is representative of a liquid fuel spill or
even the horizontal portion of cushion on a chair. Such fires are mostly tur-
bulent, and the release of fuel gas at the surface is at a relatively low veloc-
ity. In a quiet atmosphere, the gases of the flame and hot smoke will rise
vertically. This rising flow stream is called a “plume”—more specifically, a
buoyant plume. However, in a wind, the plume will bend according to its
213
interaction with the wind, and eventually will follow the wind direction as
the plume loses its buoyancy. Just what is buoyancy? And how do we dif-
ferentiate between a flame due to fuel flowing at a high velocity and a flame
that emerges from a burning liquid spill on a floor?
In this chapter, we shall examine the origin of buoyant flow, discriminate
between jets and plumes, and develop an appreciation of how to compute
the length of flames and their temperatures. For the most part, we shall
consider fire plumes driven by natural convection alone. Equations will be
developed in this chapter with the use of symbols to represent properties
and flow features. The student having trouble with this should try to under-
stand the process and not necessarily all the details. The main point is to
focus on the result.
7.2 Buoyancy and Fluid Dynamics

The mixing of fuel gases and air to form a flame cannot be appreciated with-
out some knowledge of fluid dynamics. The motion of a fluid (gas or liq-
uid) is complex, and the equations that govern this flow are impossible to
exactly solve. Computer models strive to do this and are further enlisted
to solve the combustion that occurs in a flame. They are referred to as com-
putational fluid dynamics (CFD) models. Any CFD model has approxima-
tions and cannot accurately represent all aspects of flows. Simplified models
sometimes apply that can give us a first approximation to a complex fluid
problem. Experimental data can be organized into general formulas that
can apply over a wide range of problem variables. This latter process of
equation development will be emphasized here. While the use of comput-
ers can be fun, the user may lack an understanding of what is happening.
This is not good. In contrast, equation development requires an understand-
ing of the fluid processes. The physics of the problem must be translated
into mathematics. But often theory alone cannot complete the equation, and
the strategic use of data can supply the ingredient to complete the process.
This process will be illustrated.
The equations of fluid flow began with the work of Daniel Bernoulli in
about 1733. He found that the kinetic energy (KE) of a fluid was nearly
balanced by its potential energy (PE) due to gravity. However, there was
a missing factor and that turned out to be pressure. You might say that
Bernoulli discovered fluid pressure. Today, we know his relationship
between velocity, gravity, and pressure as Bernoulli’s equation:
rV 2
+ rgz + p = constant along a streamline (7.1)
2
Fire Plumes 215
We now know that Bernoulli ignored the effects of friction in the fluid
(or viscosity), and his equation only best applies away from surfaces. Here,
ρ is the fluid density

V is its velocity
g is the symbol for the Earth’s gravitational force per unit fluid mass,
9.81 N/kg
z is the vertical height dimension
p is pressure (force per unit area—all fluids exert this pressure equally
in all directions)
The first term in Equation 7.1 is the KE per unit volume of fluid, and the
second term is the PE per unit volume. The third term, pressure, could be
considered as work done on the fluid (force × distance) per unit volume.
Pressure supplies the work needed to push the fluid.
A streamline is the path; a particle attached to the fluid would follow in
a steady flow in or around an object. A photograph of white streamlines in
Figure 7.1 illustrates the flow over a cylinder. Notice the smooth orderly flow
(laminar) at first is followed by the chaotic flow (turbulent) downstream. After
a bit, all flows suffer from instability that leads to turbulence. Bernoulli’s
equation will hold provided fluid friction is not significant and will even
hold along an average streamline in turbulent flow. Fluid friction is not so
important away from surfaces, and Bernoulli will work very well there.
This is a big approximation to the complex equations derived by Navier and
Stokes in the early 1800s and now solved by CFD models today.
With this simple background of Bernoulli in fluid flow, jets and plumes can
be examined as long as they are free of wall interference. A jet is a flow where
the initial velocity is so large that it dominates the streamline pattern of the
flow. A plume is a more quiet flow due to the motion of a light (or hot) fluid in a
Streamlines
Laminar
Onset of turbulence
FIGURE 7.1
Streamlines in white for flow around a cylinder from an Album of Fluid Motion by van Dyke (1982).
heavier fluid (cold). This difference in density (or temperature) gives rise to an
effective force called “buoyancy.” A flow solely, or predominately, driven by its
buoyancy, is called a “plume” if it is in a free atmosphere, not confined by walls.
A chimney flow has confined walls. But it is a bit easier to precisely
describe the meaning and origin of buoyancy for the chimney flow. Let us
apply Bernoulli’s equation to a chimney flow and see how the buoyant force
emerges. It will form the basis of plume flow.
Consider a vertical chimney with a horizontal intake at the bottom illus-
trated in Figure 7.2. Apply Equation 7.1 along the streamline from point 1 to
2. Then the values of the three terms, KE + PE + pressure work, are the same
at 1 and 2. As the flow starts at rest in a big room, its KE or velocity is zero in
the room at 1. If we take the datum for the elevation, z, at point 1, then z is zero
there. So at point 1, we only have the pressure term. We label it p1. At point 2,
we have a velocity and an elevation giving PE. We also have a pressure that we
label p2 different in value from p1. As this is a chimney, the temperature inside
(TH) is higher than the still air outside (TC). From experience, we know hot air
rises; we will now see why. Bernoulli’s equation can explain this motion.
Apply Equation 7.1 between 1 and 2, recognizing that the constant in
Equation 7.1 is the same at 1 and 2:
rHV22
p1 = + rH gH + p2 (7.2)
2
2
Streamline
H
TC
TH
1
Streamline
FIGURE 7.2
A streamline through a chimney.
Fire Plumes 217
The difference between the pressure at 1 and 2 outside the chimney is due to
the weight of the cold, heavy, stationary air outside the chimney. The pres-
sures at 1 and 2 are the same inside and outside the chimney, respectively.
One can apply Equation 7.1 to the cold fluid at rest.
p1 - p2 = rC gH (7.3)
Combining Equations 7.2 and 7.3 to eliminate the pressures gives a result for
the motion of hot gases out of the chimney:
rHV22
= (rC - rH ) gH (7.4)
2
The term on the right-hand side is the buoyant force per unit volume. It could be
interpreted as the PE difference between the cold and hot air. This is buoy-
ancy. The difference in density has given rise to the velocity (V2) out of the
chimney.
An alternative means to express this answer for the outflow is to put it in
terms of temperatures instead of densities. If this chimney is at sea level, the
atmospheric pressure is 101.325 kPa (kilopascal) (or 14.696 psi), and at 20°C,
the density is 1.20 kg/m3. Then the pressure difference between 1 and 2 from
Equation 7.4 would be
p1 - p2 = rC gH = 1.20 kg/m 3 ´ 9.81 N/kg ´ H (m)
For H = 1 m, this is about 11.8 Pa (1 N/m2 = 1 Pa). Even if H is 1000 m, the

pressure difference is still about 1/10th of the total atmospheric pressure
(11.8 kPa compared to 101 kPa). Therefore, the pressure difference over prac-
tical heights in the Earth’s atmosphere is small. So the pressure throughout
the height might be considered almost constant. This is the case for fire flows
in general, even throughout a tall building.
However, while pressure overall will be nearly constant, differences in
pressure will arise due to buoyancy. More appropriately for fire-induced
flows, these differences are better described in terms of temperature.
For example, hot smoke properties can be related to cold smoke (mostly
nitrogen for air and combustion products) by a relationship that requires
ρCTC = ρHTH. (In general, this is called an “equation of state.”) From this result,
Equation 7.4 can be rearranged as
V22
=2 (7.5)
((TH - TC )/ TC ) gH
The term in the denominator represents the buoyancy effect. The quantity
on the left-hand side of Equation 7.5 has no dimensions since it is equal to
a number 2. It is a “dimensionless number,” and it has the name Froude
(Fr) number. Physically, it can be considered as the ratio of momentum to
buoyant force. It is seen in this case for the chimney that Fr = 2.
Removing the walls from this chimney example can then be considered
as an approximation to a vertical plume. For vertical turbulent plumes, in a
static atmosphere, with and without flames, this Froude number is found to
be a constant of about 1.5 ± 0.1 at all heights.1 Recall that Bernoulli’s equation
holds for ideal frictionless flow and real plumes dissipate some energy by
friction. This can explain the difference between 1.5 and ideally 2. Still the
result of 1.5 compared to 2 suggests very little is lost due to friction in a buoy-
ant plume. Momentum appears to lose about 25% (2–1.5)/2, to friction at any
location. So for a turbulent fire plume Equation 7.5 applies with a constant
of 1.5.
7.3 Turbulent Fire Plumes and Jets

Let us just consider here just turbulent conditions, as laminar flames are very
small and their plume quickly becomes turbulent above the flame. A fire
plume is the buoyant column of flame and hot combustion products rising
above the fuel source. Buoyancy is the force that arises due to density differ-
ences. Because density is inversely proportional to temperature for gases,
plume temperatures cause an upward force on the hot gases relative to the
surrounding air (see Figure 7.3). If the hot plume gases cool to the local air
Buoyant
force
Eddies
Air
entrainment
Flame
FIGURE 7.3
Buoyancy and fire plumes.
Fire Plumes 219
temperature, the buoyant force would become zero, causing the plume to
cease rising. This condition is responsible for cigarette smoke stratifying in
a room with a common slight rise in temperature as it moves up. Also the
plume from a forest fire will eventually even out in the atmosphere, and
move only horizontally with the wind, as do clouds of water vapor. If buoy-
ancy is the only force responsible for the motion of the plume, we expect its
Froude number to be about 1.5. However, the character of the flow is very
different if the initial velocity of the hot gases or the initial velocity of the
fuel is large. Such a flow can arise from a high-pressure tank. In this case,
the initial “jet” velocity controls the flow, not the buoyancy. In quantitative
terms, this will occur if the initial Froude number is greater than 1.5. A jet
can be distinguished from a hot plume by how far it will travel horizontally
before rising due to its buoyancy. The greater the horizontal distance, the
stronger is the jet.
7.4 Buoyant Plumes

As the hot gases rise in a buoyant plume, the colder air is induced to flow
into the fire plume. This flow process is called “entrainment” (see Figure 7.3).
In fact, all flows in a free atmosphere at rest will entrain the surrounding
flow. This occurs for two reasons. In purely laminar flows, the fluid friction
drags the neighboring fluid along, and it becomes part of the jet or plume as
it continues. The rate of this entrainment flow is responsible for the flame
height and the characteristics of the fire plume. Consequently, the fire is
intertwined with the airflow, each depending on the other.
In general, flow can be laminar or turbulent. Laminar flow is smooth
and orderly. The candle flame is laminar, but its buoyant plume of heated
gases becomes turbulent at roughly 1 ft (or less) above the flame. This char-
acteristic can be seen by directing a collimated beam of light through the
candle flame and its plume and projecting the resultant image on a screen.
The fluctuating light and dark images of the plume illustrate high-frequency
turbulence. Within a turbulent flame, this high-frequency motion of the reac-
tion zone can produce oscillations of hundreds of degrees at a fixed point.
Superimposed on these high-frequency fluctuations, as high as 10 cycles per
second (10 Hertz [Hz]), are large-scale fluctuations as a direct result of buoy-
ancy and fluid friction. As the buoyancy causes the plume fluid to rise, its
cooler edges are slowed and fall back down. This results in recirculating
regions or eddies comparable in size to the fuel diameter, as illustrated in
Figure 7.4. These eddies rotate as well as rise along the edges of the plume.
They undoubtedly influence entrainment.
The large eddy turbulent effects present in all buoyant plumes can have
a profound effect on flame height in combusting plumes. Although these
Combusting
rising eddy
Fuel
Air
FIGURE 7.4
Turbulent burning eddy.
FIGURE 7.5
A sequence of flame images over a 0.3 second period for a 0.3 m diameter fire. (After McCaffrey,
B.J., Purely buoyant diffusion flames: Some experimental results, NBSIR 79-1910, National
Bureau of Standards, Gaithersburg, MD, 1979.)
turbulent flames are diffusion flames, the diffusion mechanism of bring-

ing the fuel and oxygen together by concentration differences alone is now
embedded within the turbulent flow. The outer eddies produce a rising
vortex between them. The high-frequency turbulence distorts this image,
as illustrated in Figure 7.5. As this burning eddy rises, it extends the flame
until the fuel burns out. A new eddy then carries the flame to this former
height. As a result, the flame height can fluctuate over a significant distance.
Figure 7.5 shows that the maximum height can be roughly twice its mini-
mum. Although these sketches are shown for flames of a 0.3 m diameter
burner with an eddy frequency of 3 Hz, this difference in heights is typi-
cal of buoyant turbulent flames. When we address flame height and plume
temperatures, we should keep in mind that the flame is unsteady, undergo-
ing high frequency as well as slower eddy fluctuations. Formulas for flame
height usually only give average results and do not account for the periodic
“ripping off” of the flame shown in Figure 7.5.
The frequency of formation and departure of the outer eddies from the
base of the plume (called vortex shedding) depends on the base diameter of
the fire, D. This vortex shedding frequency in Hertz for a fire of diameter D
in meters is approximately
1.5
f = (7.6)
D
Fire Plumes 221
FIGURE 7.6
Photograph of a 10 m diameter liquid fire. (Courtesy of L.A. Gritzo, Sandia National
Laboratories, Albuquerque, NM.)
A 10 cm fire would shed eddies about five times per second, but a large wild-
land fire of 100 m would form a vortex every 10 seconds. Figure 7.6 shows
a photograph of a large liquid fuel fire (approximately 10 m diameter) that
illustrates the eddies. Note that the eddies are of size comparable to the diam-
eter, whereas the high-frequency turbulence has much finer length scales. This
photograph also illustrates the engulfment of the flame by black soot carried
by these eddies to the colder fringes of the flame where the soot does not burn.
7.5 Flame Height

Here, we restrict our attention to symmetrical fires from circular horizontal
fuel sources. These are called “pool fires,” representative of liquid fuel spills.
However, the results can approximate other symmetric fuel shapes burning
on a floor. We will discuss this later. Line fires and wall fires have somewhat
different quantitative characteristics but are qualitatively similar in behavior
of pool fires. We also will address principally turbulent fires, with no wind
effects, dominated by buoyant effects. However, let us first consider a range
of conditions to give a quantitative perspective.
7.5.1 Jet Flames

Consider gaseous fuel issuing from a pipe of diameter D at an exit velocity
Ve. For a given fuel and a fixed pipe diameter, the ignited gases will result
in an increasing laminar flame as Ve is increased, as shown in Figure 7.7.
As the exit velocity is increased, the flame is maintained as laminar up to a
maximum length of nearly a distance of 200 diameters. At higher velocities,
the flow becomes turbulent and the flame length becomes fixed for a given
fuel and pipe diameter. This phenomenon of a fixed turbulent flame height
is a direct result of the entrainment process. The basis for the prediction of
flame length is that the visible flame ends when all of the fuel is burned.
Because the fuel is supplied at the base, the end of the combustion process
is controlled by how fast the air is entrained over the length of the fire
plume. This entrainment process is illustrated in Figure 7.8. The chemical
reaction rate is fast compared to how fast air is supplied to meet the fuel
Laminar Turbulent
Flame height envelope

Flame length, Lf
0 Exit velocity, Ve
FIGURE 7.7
Flame lengths for gaseous jets issuing from a pipe of fixed diameter. (From Hottel, H.C. and
Hawthorne, W.R., Diffusion in laminar flame jets, in: Third Symposium on Combustion and Flames
and Explosion Phenomena, Williams & Wilkins, Baltimore, MD, 1949, pp. 254–266.)
Fire Plumes 223
Sufficient air is entrained
All fuel is combusted
Air entrainment
Lf
Fuel supplied
FIGURE 7.8
Air entrainment and flame length.
so the fuel reacts almost immediately when it encounters air. This is why
the entrainment of air controls the combustion process, and therefore the
height of the flame.
Let us examine the turbulent fuel jet of Figure 7.7. The fuel supplied must
completely react by the end of the flame and combine with the air entrained.
Let us see if we can explain why the flame stays at a fixed length. The rate of
air supplied (by mass) is equal to air density (ρa) times entrainment velocity
(Ve) times the area through which the air was entrained. This area can be
approximated as πDLf since the jet does not expand by much. Then propor-
tionally, we represent
Rate of air mass entrained = ra pDL f Ve (7.7)
where
ρa is the density of the surrounding air
D is the pipe exit flow diameter
Lf is the flame length
The mass rate of fuel supplied through the pipe is the density of the fuel
(ρf) times the velocity of the fuel supplied (Vf) times the area of the pipe exit
(πD2/4). In symbols, this is
p
f =
m r f Vf D2 (7.8)
4
These two mass flow rates must be at least in stoichiometric proportion

in order for the air to burn all of the fuel. Usually, for turbulent condi-
tions, this process may not be so efficient and it requires more air than
needed for stoichiometric combustion. Turbulence controls the excess air,
and this is not a function of the fuel type. We will have more air than
needed; the process is said to be “fuel lean.” Stoichiometric here implies
that the stoichiometric air-to-fuel mass ratio (s) is that determined for a com-
plete chemical reaction.
Before relating these two mass flows by their stoichiometric ratio, we
need one more element. The jet velocity controls the entrainment velocity
into the jet, where the fuel and air mix. This is a result that holds up from
measurements in the turbulent flow of jets and plumes. As a consequence,
the entrainment velocity is found to be about 1/10th of the jet velocity. For
the jet, as the pipe exit velocity controls the flow in the combustion region,
we take the jet velocity to be the fuel velocity at the exit. Therefore, from
Equations 7.7 and 7.8, by replacing the fuel velocity with the entrainment
velocity, we have the following proportionality:
raDL f Ve
µs
r f (p/4) D2Ve
or solving,
L f (r f )s
~ (7.9)
D ra
This result explains why the turbulent jet flame length becomes constant for
a given fuel (having density ρf and stoichiometric ratio s) at a given diam-
eter. Such turbulent jet flames commence at exit velocities much higher than
those due to purely buoyancy-induced flame velocities. Jet flames would
be relevant for ruptures of high-pressure gas lines. We now examine more
closely purely buoyant flames.
7.5.2 Pool Fire Flames

If we examine the “exit” velocity of fuel gases leaving the vaporizing solid
or liquid fuel as they burn in still air, we would find these velocities are
very low, not in the range to be considered jet velocities. For example, in
Chapter 6, we saw that the maximum burning mass flux for a gasoline pool
fire is 55 g/m2-s. Because the vaporized gasoline is about twice as dense as
air, the velocity of these vapors would be approximately
55 g/m 2 -s
Ve » » 3 cm/s
2000 g/m 3
Fire Plumes 225
Buoyant plume
V Unit volume at plume gas at density, ρ

and temperature, T
Unit volume air density, ρa

Z and temperature, Ta
FIGURE 7.9
Exchange of buoyant potential energy into kinetic energy of the fire plume.
This velocity is not responsible for the subsequent entrainment of air.

That entrainment results from the substantial increase in fire plume veloci-
ties due to combustion and its resulting buoyancy.
Let us estimate this buoyant flow velocity. A simple explanation for this
velocity is that the relative buoyant PE is converted into KE. For a unit
volume of gas in the fire plume, Figure 7.9 illustrates this energy transfer.
We reason that the relative PE per unit volume at height z under the gravita-
tional acceleration g is
(ra - r) gz
and the KE per unit volume (considering negligible starting KE) is
V2
r
2
Equating these energies gives the plume velocity as
2(ra - r) gz
V=
r
or in terms of temperatures,
2(T - Ta ) gz
V= (7.10)
Ta
where T/Ta = ρa/ρ, that is, the temperature is inversely dependent on den-
sity. Because the gravitational acceleration, g, is 9.81 m/s2, this velocity is
approximately 4.5 m/s for z = 1 m and (T – Ta)/Ta = 1. So, for our gasoline
fire example with a flame height of say 1 m, the fire plume velocity is more
than 100 times the vaporization velocity (i.e., 3 cm/s). This fire plume veloc-
ity controls the entrainment of air into the fire. By a process similar to that
leading to Equation 7.9, we can obtain a formula for the height of this buoyant
flame. We shall arrive at a formula that will illustrate the nature of the factors
that control flame height for a fire plume. Unlike the jet where the exiting
fuel velocity controls entrainment, here the plume velocity controls. So in
Equation 7.4, Ve is proportional to V of Equation 7.10. The fuel mass flow rate
(m f ) times the stoichiometric ratio of air to fuel is related to the mass flow rate
of air entrained up to the end of the flame. Writing this as a proportionality:
raDL f Ve
µs
mf
we can now solve for the flame height. But first recognize from Equation 7.10
that the maximum plume velocity and therefore the entrainment velocity
is proportional to gL f where Lf denotes the flame length, z. This assumes
the flame temperature is constant, and we shall prove this shortly. Now sub-
stituting and rearranging, we find that the ratio of flame height to fire base
diameter is proportional to
2/3
Lf æ fs ö
m
µç ÷
D çè ra gDD2 ÷ø
We can stop here, but it is useful to introduce the ratio of combustion energy
per unit mass (the heat of combustion, ∆Hc) to the thermal energy per unit
mass of air (cpaTa). Then we get
2/3
Lf æ ö 2/3
m f DH c æ scpaTa ö
µç ÷ ç DH ÷
D çè racpaTa gDD ÷ø2
è c ø
The second quantity scpaTa/ΔHc is a constant as the specific heat of air and its
temperature is fixed, and ∆Hc/s is the heat of combustion per unit mass of
air, a near constant of about 2.9–3.0 kJ/g for most hydrocarbons. The first
quantity is a very important factor that shows up in many fire formulas.
It represents the combustion energy release rate (HHR or Q) to the ther-
mal flow energy of the air. It is given the symbol Q* as first introduced by
Professor Edward Zukoski of the California Institute of Technology. Ed was
Fire Plumes 227
a pioneer of fire science and contributed to many aspects of fire research

including flame height, plume entrainment, and compartment fires. He often
showed the presence of Q* in formulas related to these phenomena. Many
now call Q* the Zukoski number.
This quantity Q* is a dimensionless property of the turbulent fire and is
computed as
Q
Q* = (7.11)
racpaTa gDD2
where
ρa is the air density (1.2 kg/m3)
cpa is the air specific heat (1.0 kJ/kg-K)
Ta is the air temperature (usually at 20ºC or 293 K)
Q is the energy release rate of the fire (kW)
D is the pool diameter (m)
g is the gravitational acceleration (9.81 m/s2)
In these units
Q (kW)
Q* = (7.12)
1101[D (m)]5/2
Our 1 m diameter, gasoline fire of Chapter 6 where Q = 1887 kW has a Q*

of 1.7. Typical natural fires may range from Q* values of 0.5 to 100 for very
large diameter fires to stacked wooden pallets, respectively.
While Q* is the ratio of combustion energy to a nominal flow energy of the
plume, it can also be interpreted as a length ratio. By taking it to the 2/5th power:
(Q /racpaTa g )2/5

(Q* )2/5 = (7.13)
D
The numerator now has dimensions of length. This numerator represents a

characteristic combustion length of the fire and is directly related to the flame
length. It also can be characteristic of the size of the large vortices associated
with buoyant plumes.
Measurements of flame height by various investigators have all shown
that the flame length is related to Q*. Figure 7.10 shows a range of experimen-
tal results that illustrates the full spectrum from low Q* buoyant plume fires
(conflagrations) to jet flames where buoyancy is unimportant (~Q* = 106).
An approximate demarcation between purely buoyant pool fire behavior and
where jet flames commence is approximately Q* ~ 104. Over a range of most
Q*
10–2 10 –0 102 104 106

1000
Propane
Hydrogen
100
Pool Jet
fires flames
(Approx.)
Lf /D
10
1 (Range of data)
0.1
0.1 1 10 100 1000
Q*2/5
FIGURE 7.10
Flame length for a range of symmetric fire conditions in terms of Q*. (After McCaffrey, B.J.,
Flame height, Chap. 2-1, in: DiNenno, P.J., ed., SFPE Handbook of Fire Protection Engineering,
2nd edn., National Fire Protection Association, Quincy, MA, June, 1995.)
natural fires, 0.5 ≤ Q* < 100, a relatively simple formula (from Heskestad4)
gives a good estimate of the average flame height.
L f = 0.23 Q 2/5 - 1.02D (7.14)
where
Q is in kilowatts
Lƒ and D are in meters
Recall that the flame height may fluctuate in time by a factor of 2 from its
minimum to its maximum height, as depicted in Figure 7.5, so Equation 7.14
represents an average flame height. The most uncertainty exists below Q* of
0.5. We shall see from the next section that Equation 7.14 favors the maximum
flame height. As the flame fluctuates, there is a region where combustion
Fire Plumes 229
always occurs (a continuous flame region), and above it there is an intermit-

tent region where flames reach their maximum.
Let us investigate the temperatures in this intermittent flame and in the
above noncombusting plume.
7.6 Fire Plume Temperatures

The temperatures within a turbulent fire plume vary across the width of
the plume from a maximum at the center to the air temperature at the edge.
We shall consider the centerline maximum temperature along the height
(z) of a turbulent buoyant fire plume. The variation with height depends on
whether z is within the combustion zone (z ≤ Lƒ) or the noncombusting zone
(z > Lƒ). We can consider a simple theory to explain these vertical tempera-
ture variations. Over the combusting region, the flame generates Q energy
release rate and radiates a fraction (Xr) of this energy away from the flame
to the surroundings. The energy that stays in the flame at the end of the
combustion region at z = Lf is (1 - X r )Q . The energy release rate is distributed
over the flame length as the entrained air reacts with the fuel present. Once
all the fuel is reacted at z = Lf, the energy rate (1 – Xr)Q remains constant after
that throughout the noncombusting plume region. These processes are illus-
trated in Figure 7.11. Let us see if our background of fire plumes can give us
some information on how the temperature will behave.
z z
Noncombusting zone
TCL
.
Plume
(1 – Xr)Q
Ta
Height
Intermittent
Lf Z = Lf
Combusting
Flame
D 300 800
Fire plume energy TCL (°C)
FIGURE 7.11
Fire plume energetics.
7.6.1 Analyses to Predict the Plume Temperature

Consider the combustion zone where z ≤ Lƒ. The energy by combustion that
stays in the fire plume by conservation of energy must equal the thermal
flow energy of the plume. The flow energy rate of the plume is represented
by the mass flow rate times the thermal energy of the plume. In symbols,
this is
Q = mc
p (T - Ta ) (kW) (7.15)
where
cp is the specific heat of the plume gases
m is the mass flow rate of the plume gases
The mass flow rate in the plume is nearly equal to the air entrainment rate,
as the amount of fuel supplied is relatively small. We can write, in detail, the
energy released due to the combustion process as
Q = m
f DH c (1 - X r ) (kW) (7.16)
where
m f is the mass rate of fuel reacted up to height, z, g/s;
ΔHc is the heat of combustion, (kJ/g)
It has been stated that more air than stoichiometrically required is entrained
over the flame. In other words, the turbulent mixing does not allow the
available fuel to readily react with the entrained air. Let us say n-times stoi-
chiometric air is entrained and is needed to complete this fuel-lean combus-
tion. We can therefore write that the mass air entrainment rate is
= nsm
m f (7.17)
where
s is the stoichiometric mass air-to-fuel ratio
n is the multiple of stoichiometric air required due to inefficient mixing
By equating Equations 7.15 and 7.16 and substituting Equation 7.17, it follows
that the temperature rise in the fire plume is
DH c (1 - X r )
T - Ta = (7.18)
cp ns
ΔHc/s is approximately constant (2.9 kJ/g air) for most all hydrocar-
bon fuels, cp ≈ 10 –3 kJ/g-K for air, and it is experimentally found that n is
Fire Plumes 231
approximately 9.6.1 The average turbulent flame temperature rise along the
centerline of a fire plume is approximately (for air initially at 20°C)
T - Ta » 1450(1 - X r ) (°C) (7.19)
This result demonstrates that the temperature within the combusting zone
is not dependent on fuel type except for variations in Xr. We saw that Xr can
range from 0.15 to 0.60 for different fuels and may also depend on the scale
of the pool fire (Figure 3.10 and Tables 3.3 and 3.4). Large pool fires or very
sooty fuels may have the tendency to shroud the luminous flame in cooler
black soot as seen in Figure 7.6. This shroud of soot can block the outgoing
flame radiation, making Xr smaller and T larger. This result has been found
for large pool fires and indicates a maximum possible fire plume turbulent
temperature of nearly 1500°C.
How about the temperature along the plume centerline as we move above
the flame. Continued entrainment of air will drop the temperature. It is not
surprising that this temperature rise depends on Q*. But here Q* is based on
the dimension z, the height along the plume, rather than the base diameter.
Although there is a small effect of the base diameter, we will ignore it here.
We represent this behavior as
T - Ta z
~ function (7.20)
Ta (Q / racpaTa g )2/5
for a wide range of pool fire data.

Figure 7.12 represents this correlation shown for methane fires of 15 kW to
60 kW. Although the data are scattered to some degree, the line represents
the correlation distinctly in three regions: (1) combusting, (2) intermittent,
and (3) noncombusting plume. The end of the intermittent zone—the maxi-
mum flame tip—is at z/Q 2/5 = 0.20 m/kW 2/5, which is consistent with the
coefficient 0.23 of Equation 7.14. The correlation shows a maximum (time-
averaged) turbulent flame temperature of approximately 800°C for these data
in which Xr was about 0.3. Applying Equation 7.19 would give a centerline
temperature rise in the combustion zone of 1015°C. Actually, this is more cor-
rect than the data in Figure 7.12 because the thermocouples used to measure
the fire temperatures were not corrected for their radiation loss. So by losing
heat, they read too low. As the temperature drops, the data are more accu-
rate. It should be noted that the centerline temperature of slightly over 300°C
occurs at the tip of the flame. Heskestad4 commonly associated a tempera-
ture rise of 500°C as his measure of the average flame height. Importantly,
turbulent flame plumes have a similarity in that, despite the value of the
flame height, the fire plume’s temperature rise at the flame’s maximum tip at
the end of the intermittent zone can always be taken as about 300°C.
900
800
700
600
500
400
.
Q (kW)
300 14.4
21.7
33.0
TCL – Ta (°C)
44.9
200 57.5
Flame Intermittent Plume

100
80
60
50
0.008 0.01 0.02 0.03 0.04 0.05 0.06 0.08 0.1 0.2 0.3 0.4 0.5
.
z/Q2/5 (m/kW2/5)
FIGURE 7.12
Centerline fire plume temperatures. (After McCaffrey, B.J., Purely buoyant diffusion flames: Some
experimental results, NBSIR 79-1910, National Bureau of Standards, Gaithersburg, MD, 1979.)
Although Figure 7.12 only contains data from one fuel (methane), other
experimental results for hydrocarbon fuels have shown the approximate
universality of the graph. However, as the pool fire diameter increases,
sooting can reduce Xr causing maximum centerline turbulent flame temper-
atures as high as 1200°C or more. Except for this higher plateau in the flame
region of Figure 7.12 as the fire gets larger in diameter, the rest of the correla-
tion still applies. More important, for hydrocarbon fuels, the turbulent flame
temperature is independent of fuel and can generally be taken as approxi-
mately 800°C for moderate fire sizes. Again, this 800°C rise is not corrected
for radiation error, and more accurately it is more like 1000°C. But commonly
thermocouple data are not corrected in fire experiments.
Gasoline and wood fires should show no temperature differences at the flame
tip, but their flames lengths may be significantly different. These flames have
a similarity. They can be stretched and squeezed, but their temperature dis-
tributions are the same relative to the combusting and noncombusting zones.
7.6.2 Flame Height and Temperature Calculations

Compute the average flame height, and the centerline fire plume tempera-
ture at this flame height for four fuels, 1 m in diameter. These fires were
Fire Plumes 233
TCL
Ta = 25°C . Lf
Q
D=1m
FIGURE 7.13
Example computing the average flame height and temperature at that location.
previously used in Chapter 6. The four fuels are wood, polystyrene, heptane,
and gasoline. See Figure 7.13 as an illustration of the example.
Using Equation 7.14,
L f = 0.23Q 2/5 - 1.02D
where Q is in kilowatts, and D = 1 m. Use the values of Q computed in

Chapter 6. The flame height results follow:
Wood:
Lf = 0.23(130 kW)2/5 – 1.02(1 m) = 0.59 m
Polystyrene:
Lf = 0.23(1189 kW)2/5 – 1.02(1 m) = 2.89 m
Heptane:
Lf = 0.23(2661 kW)2/5 – 1.02(1 m) = 4.37 m
Gasoline:
Lf = 0.23(1887 kW)2/5 – 1.02(1 m) = 3.68 m

The temperature is found from Figure 7.12. Although flame radiation var-
ies for the fuels, the graph does not account for the effect of Xr. This is not
perfectly correct, but we do not have better results to work with here. More
sophisticated analyses would not change our results from the graph. So let
us proceed. The following fire plume temperatures at z = Lf on the centerline
are obtained by computing the horizontal axis of the graph and obtaining
the temperature rise from the vertical axis.
Fuel T(°C)
Wood 740
Polystyrene 400
Heptane 330
Gasoline 340
The wood result is indicative of a temperature in the continuous flame

region, while the others are indicative of the peak flame height. The wood
result is not correct. Recall that the wood here is a wood crib that has some
height to it. Indeed, Heskestad5 has shown that Equation 7.14 holds for arrays
of commodities as well as pool (flat-based) fires. His arrays were stacked
wood pallets 0.3–3.4 m high and rack storage commodities of 3–6 m high.
The equation applies provided the flame height is measured from the base of
the array, and the diameter has an effective value taken as D = ( 4Q /pQ ¢¢)1/2.
The energy flux for the array can be taken from Table 6.5. If the wood were
represented as a crib with some height to it, say 1.3 ft, by Table 6.5 this wood
pallet stack (crib) would have a firepower of 1021 kW for a 1 m diameter.
This firepower is more than that estimated for the wood by using Table 6.2.
The value in Table 6.2 must depend on the height of the crib and cannot be
general. With this adjustment for the wood, the temperature rise is more like
400°C. Recall that no effect of the radiation fraction was taken into account,
and that the temperatures of Figure 7.12 were not corrected. Roughly, the
results indicate that the temperature rise at the flame height was always about
the same, 330°C–400°C, with some inconsistency due to using a formula for
flame height with a general correlation for fire temperature. In principle, the
temperatures should be the same at the flame with some variation due to
differences in radiation fraction of the fuel.
7.6.3 Nature of Turbulent Flame Temperature

We have just seen that the turbulent flame temperature rise for moder-
ate size pool fires (and it could be shown for other fire configurations) is
nearly a constant maximum of approximately 800°C and independent of
fuel, particularly hydrocarbons. Literature values of adiabatic flame tempera-
tures for hydrocarbon fuels burning in air typically range from 2000°C to
Fire Plumes 235
2300°C with measured laminar flame temperature of 1800°C–2000°C. Why

are turbulent fires 800°C or 1200°C at most? To understand this, we must
understand what is being measured and therefore sensed by a thermometer
or thermocouple.
The adiabatic flame temperature is the maximum possible temperature that
can be achieved for that particular combustion process of a fuel. Adiabatic
means perfectly insulated; there is no heat loss from the reaction zone.
As shown in Chapter 2, the candle flame is representative of a laminar dif-
fusion flame. It is thin and reasonably steady in its position. Because it is
thin, not much heat is lost by radiation. Primarily, heat is lost by conduction
through the gas. Accordingly, we might expect its temperature to be slightly
below the adiabatic flame temperature. Values of 1800°C–2000°C are typical
of such laminar flames. If by cooling or dilution, its temperature drops to
about 1300°C it is likely to extinguish. Then, how can a turbulent flame at
800°C–1200°C exist?
Figure 7.14 illustrates the measurement process for laminar and turbu-
lent flames. A probe at a fixed point attempts to measure the flame tem-
perature. The relatively stationary laminar flame is an easy target for the
temperature probe. The fluctuating turbulent flame is not. The probe can
be in and out of the flame on either the air side or fuel side. A typical
trace of the probe output over time is illustrated. Even a small probe may
Fluctuating turbulent flame

Laminar flame
Fuel Air
Tf
Tf
1500
Tf (°C)
1000
500
100 200 300 400 500

Time (s)
FIGURE 7.14
Measuring flame temperatures.
not respond fast enough in the fluctuating flame so the true maximum
and minimum temperatures may not even be realized. The mean time or
average temperature is what would be reported in Figure 7.14 for a turbu-
lent flame. The 800°C represents a time-average temperature. If we were
capable of “riding” on the fluctuating flame, we would measure tempera-
tures in excess of 1300°C and closer to the laminar flame temperatures.
The thermal threat of the turbulent flame is represented by the time-average
temperature. That is the temperature an object within or exposed to the
turbulent flame would sense from the turbulent flame. So the time-average
temperature is very important and is a measure of the damage potential
of turbulent flame.
7.7 Flame Lengths for Other Configurations

Up to now, we have primarily focused on pool fire-like flames. Heskestad has
said that his flame height formula can also apply to arrays. Others have con-
sidered different configurations. These include a fire in the corner, against
at wall, on a wall, and impinging on the ceiling. Not surprisingly, the factor
Q* enters into all of these formulas. Some results are couched in terms of
the average flame length, and others differentiate between the continuous
flame height and the peak height. Hasemi and Tokunaga6 give results for a
square fire in a corner or against a wall; Ahmad and Faeth7 give results for
a fire burning on a wall, and recent Ding8 give an approximate result for the
flame extension along a ceiling from a pool fire below. These results are in
terms of Q* as
Lf
= C f Q*n - b (7.21)
D
or alternatively in specific units for Q in kW and D in m:
n
Lf æ Q ö
= C¢f ç 5/2 ÷ - b (7.22)
D èD ø
Table 7.1 gives the values for Cf, C′f, and n; and Figure 7.15 describes the con-
figurations. Figure 7.16 gives a graph for all the configurations. The ceiling
case should be used when the pool flame is computed as being above the
ceiling height.
Fire Plumes 237
TABLE 7.1
Flame Length Parameters for Equations 7.21 and 7.22
Configuration Lf D N Cf C'f a b
Pool Height Diameter 2/5 3.8 0.23 1.02
1/2
æ 4Q ö
Array Height çç ÷÷ 2/5 3.8 0.23 1.02
è pQ¢¢ ø
Fire in corner Height Side of square 2/3 Continuous 3.0 0.028 0
Maximum 4.3 0.040 0
Fire against Height Side of square 2/3 Continuous 2.2 0.021 0
wall Maximum 3.5 0.033 0
Fire on wall Height Width 2/3 4.9 0.046 0
Ceiling flame Radius Diameter of 2/5 1.6 0.015 0
pool fire
a C′f in units according to kW and m.
Lf
Lf
Lf
D D D D D D
D
FIGURE 7.15
Flame configurations.
7.8 Whirls and Balls

There are other phenomena related to fire in the free atmosphere. A plume in
the wind can lean over, and its length and angle can be computed as a func-
tion of Q* and the Froude number. A wind can also induce a rotation in a
plume. This forms a fire whirl—a spinning plume. It can happen at the edges
of large fire such as forest fires. It can also happen in urban conflagrations,
as occurred in 1923 during the fire following the Great Kanto earthquake in
Hifukusho-Ato. As the whirl moved off the main fire, it killed 38,000 who
were taking refuge near the river. Fire whirls are still the subject of research.
According to Saito, Dobashi, and coworkers,9 the fire whirl achieves a much
taller flame than its originating plume due to an increase in burning rate.
The swirling motion at the base of the fuel increases the convective heat
transfer. Anyone that has seen a fire whirl is amazed at how fast is spins and
100
In corner
max.
cont.
Pool and array
10 Against wall
max.
cont.
Ceiling
Lf /D
Wall
0.1
10 100
Q (kW)2/5/D (m)
FIGURE 7.16
Flame lengths for various configurations (see Figure 7.15).
its jump in height. A large cylinder, cut in half along vertical lines, and dis-
placed slightly around a centered liquid pool fire, admits a circulating flow
to cause a whirl.
Another atmospheric fire phenomenon is a fireball. This is often seen in
movies after a car careens off a cliff or following an “explosion.” In actuality,
the car incident is mostly Hollywood. The fireball occurs due to a sudden
release of gaseous or aerosol fuel that is ignited. The core of fuel cannot burn
all at once; it burns on its edges due to turbulent diffusion. The increase in
temperature of the combustion process gives buoyancy, and this carries the
raw fuel and burning fuel upward like a burning sphere. Buoyancy distorts
the sphere, and turbulence sets in. Just as the flame ends in a plume when
the fuel burns out, the fireball stops in the same way. As it rises, it gets big-
ger and higher; then it is gone. The maximum diameter, its height, and its
burning time can all be computed. They are based on the original volume
of fuel released. This usually happens when there is a crash or breakage of
a tank of gaseous or liquid fuel. The fireballs at the WTC towers on 911 are
indicative. Fay and Lewis10 give simple formulas for the fireball in terms of
the original fuel volume. The fuel volume (Vf) should be computed from the
mass of fuel released divided by it density at the temperature and pressure
of the air.
Height of fireball:
Lb (cm) = 12.7[Vf (cm3)]1/3 (7.23)

Fire Plumes 239
Maximum diameter:
Db (cm) = 7.7[Vf (cm3)]1/3 (7.24)
Burn time:
tb (s) = 0.28[Vf (cm3)]1/6 (7.25)
These simple formulas were derived from theory and laboratory experiments.
They work quite well in large-scale events as they are based on sound physics.
7.9 Summary
Buoyancy plays a significant role in natural fire plumes. Buoyancy is a force
on the fluid due to gravity and a density or temperature difference. A fire
plume is driven by it buoyancy, and a jet flame is driven by the container’s
exit velocity of the fuel. Turbulent jet flames reach a fixed length according to
their pipe diameter. Fire plumes reach a flame height based on the Zukoski
number, Q*. The Zukoski number is the ratio of combustion energy to flow
energy in the plume. Turbulent flame temperatures are lower than their
laminar counterparts and commonly do not exceed 800°C in moderate size
pool fires as measured by thermocouples. But this temperature can increase
with fire size, achieving a theoretical maximum rise in air of about 1450°C.
These temperatures primarily depend on the radiative fractional loss of the
fuel. Because temperatures decrease very rapidly after combustion due to air
entrainment, flame height is a critical measure of hazard. Objects above the
turbulent flame are not likely to ignite; flame contact is generally required
except for remote ignition by radiant heat of very large fires. Formulas were
presented, and their origin was guided by the theory of fluids, mainly fol-
lowing that of Bernoulli. Flame height formulas for various configurations
were presented, along with formulas for fireball phenomena.
Review Questions
1. For a 500 kW flame compute the flame height if its base is 10 cm and its
diameter is 100 cm.
2. A ceiling detector will alarm at a gas temperature of 150°C. For a fire
directly under it, at 3 m distance, find the smallest energy release rate fire
source that can cause the alarm.
3. Describe how buoyancy occurs and what it depends on.

4. A 0.5 m diameter pool fire has a flame 1 m above a ceiling of 3 m tall.
What is the energy output of this fire?
5. For the problem in 4, compute the radial extent of the flame under the
ceiling.
True or False
1. Although a turbulent flame temperature is roughly 800°C, the temper-
ature at the fluctuating flame tip is no more than 350°C.
2. Eddies or vortexes are rotating regions of flow.
3. Gasoline fires are much hotter than wood fires.
4. The velocity of gasoline vapors leaving the surface of a liquid pool fire
originates at 3 m/s or more.
5. Theoretically, a buoyant plume can rise forever in a cold atmosphere.
6. A fire whirl is a ride in an amusement park.
7. A fireball can occur if a dump of gaseous fuel is suddenly ignited.
8. Buoyancy can only occur in a container such as a balloon.
9. A Froude number for a plume is about 1.5.
10. The parameter Q* has some strange units.
Activities
All these experiments need to be done in a safe laboratory site and with
attention to fire spread and its extinguishment.
1. Observe buoyant plumes in the atmosphere from smoke stacks, fires, or
other sources. Note their nature, eddy characteristics, and how they are
affected by wind.
2. Make a cylindrical shape about 2 ft in diameter and 2 ft high, and cut
it vertically to make two equal parts. Put a small pan (about 2–3 in.
in diameter), ignite it, and surround the flame with the two cylin-
der pieces, displaced slightly to form two vertical gaps. As the air is
drawn into the gaps to be entrained into the fire plume it will create a
fire whirl.
3. In a laboratory setting under a hood, ignite a small pan of heptane or
similar flashpoint liquid fuel. Use a pan no larger than 10 in. in diameter.
Observe the fluctuations of the flame, its necking in at its base, and the
eddy vortices.
Fire Plumes 241
4. Fill a balloon with natural gas or methane. Attach a string to the bal-
loon. Wet the string with a flammable liquid to form a wick. Tie the bal-
loon to safe point and ignite the string. As the flame moves to the balloon
and bursts it, a fireball will occur.
5. Use a square pan about 2 in. on a side. Ignite it in an open space, then
against a wall, then in a corner. Observe the flame height and check if it
follows the formulas.
References
1. B. J. McCaffrey, Purely buoyant diffusion flames: Some experimental results,
NBSIR 79-1910 (Gaithersburg, MD: National Bureau of Standards, 1979).
2. H. C. Hottel and W. R. Hawthorne, Diffusion in laminar flame jets, in Third
Symposium on Combustion and Flames and Explosion Phenomena (Baltimore, MD:
Williams & Wilkins, 1949), pp. 254–266.
3. B. J. McCaffrey, Flame height, Chap. 2-1 in SFPE Handbook of Fire Protection
Engineering, 2nd edn., edited by P. J. DiNenno (Quincy, MA: National Fire
Protection Association, June 1995).
4. G. Heskestad, Luminous heights of turbulent diffusion flames, Fire Safety
Journal, 5 (1983), 103–108.
5. G. Heskestad, Flame heights of fuel arrays with combustion in depth, in
Proceedings of the Fifth International Symposium on Fire Safety Science (1997),
pp. 427–438.
6. Y. Hasemi and T. Tokunaga, Flame geometry effects on the buoyant plumes
from turbulent diffusion flames, Combustion Science and Technology, 40 (1984),
1–17.
7. T. Ahmad and G. M. Faeth, Turbulent wall fires, Proceedings of the Combustion
Institute, 17 (1978), 1149–1162.
8. H. Ding, An integral model for turbulent flame radial lengths under a ceiling,
Master of Science, Department of Fire Protection Engineering, University of
Maryland, College Park, MD, 2010.
9. K. Kuwana, S. Morishita, R. Dobashi, K. H. Chuah, and K. Saito, The Burning
rate’s effect on the flame length of weak fire whirls, Proceedings of the Combustion
Institute, 33 (2011), 2425–2432.
10. J. A. Fay and D. H. Lewis, Unsteady burning of unconfined fuel vapor clouds,
Proceedings of the Combustion Institute, 16 (1976), 1397–1405.
8
Combustion Products
Learning Objectives
• Explain the nature and levels of combustion products for various

fire conditions
• Explain the property called yield and how it relates to concentration
of that product in smoke
• Calculate the hazards of combustion products in smoke
• Explain the role of equivalence ratio in producing combustion
products
8.1 Introduction
We have seen that the mass burning rate is a significant quantity that results
from the burning rate flux (m ¢¢) and the area (A) involved in burning due to
ignition and spread: m =m ¢¢A. From this burning rate (m ), we can derive all
the products that result from the fire’s chemical reaction. We have seen that
energy release rate (Q ) is derived as m DH c, where DH c is the heat of com-
bustion. Q controls the smoke temperature and the surrounding heat flux.
It causes an immediate effect on its surroundings. All of the other products of
the fire’s chemical reaction are frozen in the smoke emanated in the buoyant
plume leaving the fire. (Smoke here is defined as the gases and particulates
that flow in stream from the flame.) As with Q proportional to m by DH c,
the rates of production of the other products are proportional to m accord-
ing to the fire’s chemical reaction. This proportionality is called the yield for
each product species. These products of combustion involve carbon dioxide
and water vapor for the complete burning of hydrocarbon fuels; however, other
products are possible due to incomplete combustion or due to the presence of
243
other elements, besides carbon, hydrogen, and oxygen, in more complex fuels.
These incomplete combustion products can have a variety of harmful effects
on people, on commodities, and on equipment. Usually, the harm or damage
is related to the concentration of combustion product and its exposure over
time. In this chapter, we present quantitative criteria for several products of
combustion. An elementary background in chemistry is required.
8.2 Scope of Combustion Products

The nature of the combustion products depends on the fuel composition
and on the fire process. We have seen that fire can be characterized as (1) a
premixed flame, (2) a diffusion flame, and (3) smoldering or surface oxida-
tion. Only diffusion flames and smoldering, to some degree, are consid-
ered in this chapter. Ultimately, we need to represent the chemical reaction
occurring in the fire process. This reaction depends on the completeness of
combustion. For example, if we burned methane in air (recall 1 mol of air
is 0.21 mol of O2 and 0.79 mol of N2), we can write the following chemical
equation:
( 4) ( 4)(0.79)
CH 4 + (0.21 O 2 + 0.79N 2 ) ® CO 2 + 2H 2O + N2
0.42 0.42
For complete or ideal combustion, the products, by definition, are the most
stable compounds: CO2 (carbon dioxide) and H2O (water). At combustion
temperatures, these species are gases, although the water could condense
if it finds a cool surface. The coefficients of each term in the chemical equa-
tion have been deduced so that atoms are not destroyed (e.g., 1 C atom and
4 H atoms in CH4). From this balance of atoms (or mass), we can compute for
every 16 g of CH4 (recall that 1 mol of CH4 has a mass of 16 g, 12 g for C and
1 g for each H) that we must have (4)/0.42[0.21(32) + 0.79(28)] or 274.7 g of air.
(Note that 1 g of air has about 0.23 g of O2 and 0.77 g of N2.) These propor-
tions insure that all of the methane and oxygen are completely consumed
in the reaction. The ratio of this mass of air to mass of fuel is called the
(ideal) stoichiometric air-to-fuel mass ratio (s), that is,
274.7 g air
s= = 17.2 g air/g fuel
16 g CH 4
If more air is available, the condition is said to be overventilated or the process

is fuel lean. For less air than required for stoichiometric burning, the process
is termed underventilated or fuel rich. All of the oxygen in that air is reacted
Combustion Products 245
by the time the smoke leaves the flame. In all cases, the nitrogen in the air is
passive and remains constant. However, it will be heated as a result of the
chemical reaction’s energy release.
Typical fuels involved in fire processes are hydrocarbons, that is, they are
composed of hydrogen (H) and carbon (C). Other fuels may also include oxy-
gen (O), nitrogen (N), chlorine (Cl), fluorine (F), and bromine (Br). The resul-
tant combustion products depend on the process, as illustrated in Figure 8.1.
In an ideal complete reaction, the N, Cl, F, and Br would yield their respec-
tive gases. However, in any fire process, the reaction is not complete and the
resultant gases are normally hydrogen cyanide (HCN), hydrogen chloride
(HCl), hydrogen fluoride (HF), and hydrogen bromide (HBr), respectively.
In addition, incomplete combustion leads to carbon monoxide (CO) in place
of carbon dioxide (CO2), soot (principally carbon), and many hydrocarbons
(HCs) as a result of the thermal decomposition of the original fuel. Gaseous
hydrogen can also be found in incomplete combustion processes. Figure 8.1
shows the various fire processes and the associated products of combustion.
It is very important to recognize the particular fire process because the prod-
ucts of combustion depend on the process as well as on the fuel. Smoldering
requires very little air to maintain. Its combustion is incomplete resulting in
a significant yield of CO.
Initially, there is sufficient air available in a room fire to support flaming
combustion. If fire occurs in a closed room, the oxygen in the air will be
expended and the process becomes underventilated. A small opening to the
Fuel (F) Process in air (A) Products
C, Carbon F + A (Ideal) : CO2, H2O, N2, Cl2, F2, Br2
A Flame
H, Hydrogen
F Overventilated : CO2, H2O, some CO, soot
HCN, HCl, HF, HBr
O, Oxygen
A
N, Nitrogen Underventilated : Same as above, but

F
more CO, soot,
Cl, Chlorine HC compounds
A
F, Fluorine F Smoldering : Same as above, but
predominately CO
Br, Bromine
Heat
Evaporation, : Same as above, but
F
Pyrolysis more HC compounds
FIGURE 8.1
Typical products of combustion in fire.
room will bring in some air, but if the fuel generated is too great, the fire will
remain underventilated. Extinction will generally occur when the oxygen
feeding the fire is about 11% or less (less for higher room temperatures).
The parameter that quantitatively represents the over- and underventila-
tion states is called the equivalence ratio, Φ, defined as the mass ratio of fuel
(gas) available to air available times the stoichiometric ratio, s.
æ mass of fuel (gas) ö

F=ç ÷´ s (8.1)
è mass of air ø
That is, if Φ > 1, the fire is underventilated. For Φ > 1, these decomposition
products are not burned to completion. For Φ < 1, there is sufficient air and
the process will be complete: This case is said to be overventilated. All fires
start with Φ < 1 and with continued growth can reach Φ > 1 states. The
chemistry of these processes is complex, and consequently, empirical data
must be used to quantify the nature of the combustion products for each
fuel. These predictions by fundamental modeling are difficult and still the
subject of much research.
8.3 Yields
As the heat of combustion gives us the energy release per unit mass of fuel
burned, yields give us the mass production of each product species per unit
mass of fuel burned. Just as the heat of combustion for a fire reaction is not
the complete ideal value and must be measured, so too is the yield of a spe-
cies. Calorimetry devices serve to measure both the actual heat of combus-
tion and the yields. For example, the yield of CO is defined as
mCO
y CO = (8.2)
m
where mCO is the mass of CO produced and m is the mass of fuel burned.
Alternatively, we could consider the rates of mass, that is, y CO = m CO /m
,
where m is the rate of burning and m CO is the rate of CO mass produced.
It is generally accepted that the yields are fairly constant for a given fuel as
long as Φ < 1. But as the fire condition becomes underventilated, the yields
change.
Tewarson1 of FMGlobal has nearly single-handedly established a wealth
of consistent data on the yields of various fuels. Although mainly derived
from relatively small-scale tests but over a realistic range of oxygen and heat-
ing for fire, it is generally accepted that these results apply to realistic fire
TABLE 8.1
Fuel Properties as a Function of Ventilation
Conditions Overventilated Conditions Underventilated
Combustion Products
Fuel yCO2 (g/g) yCO (g/g) ysoot (g/g) DH c (kJ/g) DH c ,ideal (kJ/g) D m (m2/g) yCO (g/g) yH2 (g/g) yHCl (g/g)
Gases
Propane 2.85 0.005 0.024 43.7 46.4 0.155 0.229 0.0110 0
Acetylene 2.60 0.042 0.096 36.7 48.2 0.315 NAa NA 0
Liquids
Ethyl alcohol 1.77 0.001 0.008 25.6 26.8 NA 0.219 0.0098 0
Heptane 2.85 0.010 0.037 41.2 44.6 0.190 NA NA 0
Solids
Wood (red oak, pine) 1.27 0.004 0.015 12.4 17.7 0.037 0.138 0.0024 0
Polymethyl methacrylate (PMMA) 2.12 0.010 0.022 24.2 25.2 0.109 0.189 0.0032 0
Polystyrene (PS) 2.33 0.060 0.164 27.0 39.2 0.335 NA NA 0
Nylon 2.06 0.038 0.075 27.1 30.8 0.230 NA NA 0
Polyurethane (PU) flexible foam 1.51 0.031 0.227 19.0 27.2 0.326 NA NA 0
Polyvinyl chloride (PVC) 0.46 0.063 0.172 5.70 16.4 0.400 0.360 NA 0.4
Source: Based on data from Tewarson, A., Generation of heat and chemical compounds in fires, Chap. 3-4, in: DiNenno, P.J., ed., SFPE Handbook of
Fire Protection Engineering, 2nd edn., National Fire Protection Association, Quincy, MA, June 1995.
a NA, not available.
247
conditions. Table 8.1 lists a sample of his data for representative fuel classes:
gases, liquids, and solids. The yields are shown for CO2, CO, and soot along
with the measured heat of combustion, DH c. These are experimentally
derived values representative of over- and underventilated fire process. The DH c
is less than its ideal value that would occur for the ideal complete reaction,
as indicated in Table 8.1. The table also gives the yields of hydrogen (H2) and
hydrogen chloride (HCl) for PVC plastic containing chlorine. The quantity
Dm relates to visibility, and we will come back to it later in the chapter.
In most cases, the yields (or negative yields, as in the consumption of
oxygen) are slightly less than their ideal chemical reaction values for CO2,
H2O, and O2. The yields associated with the products of incomplete combus-
tion (e.g., CO, soot) are relatively small. But as we shall see, these are the cul-
prits that do the damage. Carbon monoxide can kill, HCl can corrode metal,
and soot can damage.
Some typical underventilated values of yields for CO and H2 are listed in
Table 8.1, but their precise values depend on Φ. For Φ < 1, the overventilated
values in Table 8.1 apply. Figure 8.2 displays how the underventilated yields
vary with Φ. They are represented as a ratio of the yield to its overventilated
1000
CO: Ethyl alcohol

100 CO: Propane
CO: PMMA
Ratio of yield to yield in open air
CO: Nylon
CO: Wood
10
CO: Polystyrene CO: Polypropylene

Soot
1
O2
0.1
CO2
0.01
0.1 1 10 100
Equivalence ratio, Φ
FIGURE 8.2
Effect of underventilation on yields for many materials—approximate representation of data.
(After Tewarson, A., Generation of heat and chemical compounds in fires, Chap. 3-4, in: DiNenno,
P.J., ed., SFPE Handbook of Fire Protection Engineering, 2nd edn., National Fire Protection Association,
Quincy, MA, June 1995.)
value. This graph comes from Tewarson1 and is very valuable. For exam-
ple, it can explain how carbon monoxide is increased as the fire becomes
underventilated (a state sometimes called ventilation limited in compartment
fires). Also, the results show that CO yields depend on fuel, but the yields
(±) of O2, soot, and CO2 are essentially fuel independent when plotted as
yunderventilated /y overventilated versus Φ.
Example 8.1
Let us consider an example. In Chapter 6, we computed the burning flux
of wood and polystyrene (PS) as 11 and 38 g/m2-s. Suppose we have the
same size fire for each fuel: say 2000 kW. Since (from Equation 6.2)
Q = m
¢¢ADH c
the fuel areas involved can be computed as
Q
A=
¢¢DH c
m
For wood
2000 kW
A= = 15.2 m 2
(11 g/m 2 -s)(12 kJ/g)
and for PS
2000 kW
A= = 1.9 m 2
(38 g/m 2 -s)(27 kJ/g)
Fire in open air: overventilated. For a fire in the open air (free burning and
overventilated as ample air is available) or in a large room with sufficient
open windows, the overventilation yields apply. The production rate of
CO is then computed by
m ¢¢A,
CO = y CO m (8.3)
or for wood
CO = (0.004)(11 g/m 2 -s)(15.2 m 2 ) = 0.67 g/s

m
and for PS
CO = (0.060)(38 g/m 2 -s)(1.9 m 2 ) = 4.3 g/s

m
As a result, we see more CO is produced by PS.
Fire in a room that is underventilated. What if these same fires are in a

room where the air supply through windows is 500 g/s, typical of a
small broken window? Let us calculate Φ from the room to examine the
degree of underventilation. For this computation, we need the stoichio-
metric ratio of air to fuel required for burning, s. Let us estimate s from
the ideal reaction using ΔHc, ideal, which is derived as follows:
DH c,ideal
s» (8.4)
3 kJ/g air
where the approximation of a constant energy release for burning the

oxygen in air is used. For most ordinary combustibles, approximately
3 kJ is released for each gram of air used by the fire reaction. Then, for
wood
17.7
s= = 5.9 g air/g wood
3
and for PS
39.2
s= = 13.1 g air/g PS
3
The corresponding Φ values are (fuel to air available times s)
(11 g/m 2 -s)(15.2 m 2 )(5.9)

F wood = = 1.97
(500 g/s)
and
(38 g/m 2 -s)(1.9 m 2 )(13.1)

F PS = = 1.9
(500 g/s)
From Figure 8.2 and the overventilated values from Table 8.1, we can esti-
mate that for wood at Φ = 1.97: (Table 8.1 gives the overventilated yield
as 0.004, and Figure 8.2 reads approximately y underventilated/y overventilated ~ 50.)
yCO ≈ 50 (0.004) ≈ 0.20
and for PS at Φ = 1.9:
yCO ≈ 2.0 (0.060) ≈ 0.12
The amount of CO produced has only doubled for the PS but has increased
by 50 times for the wood for this particular room fire condition. This example
should illustrate that CO production depends on the fuel as well as the air
supply. For the wood CO yield of 0.2, this state would give a CO production
rate of 33.4 g/s provided the burning rate of the wood can be maintained at
11 g/m2-s at this underventilated state. Similarly, the production rate of the
PS is 8.7 g/s for this unventilated case. The wood fire in the underventilated
state has become more hazardous than the PS with this higher production
rate of CO. Notice a lot more wood is burning here than PS.
Finally, it should be pointed out that smoldering, especially the notorious
fire scenario of a cotton or polyurethane foam mattress or sofa, produces CO
yields of as high as 0.2–0.3. These yield levels are comparable to the under-
ventilation values in flaming for fuels. But the production rate in smolder
is much less than generally that of underventilated fires. This is why they
take a long time to do harm. A growing fire going through flashover to a
ventilation-limited state can suddenly produce toxic smoke.
8.4 Concentrations
Yields measure what is produced at the fire source. The principal hazard of
fire is the composition and associated concentrations of harmful species in
the smoke. By smoke, we mean the gas stream that flows away from the fire
(as illustrated in Figure 8.3) continuing to mix with air and with having no
further chemical reactions. The amount of mixing with air determines the
concentration of a particular species yielded in the flame. Associated with
each species yield, we have a corresponding concentration at different points
.
msmoke
Air Air
Fuel
FIGURE 8.3
Illustration of smoke movement from a fire.
in the smoke. Moreover, associated with the energy yield, DH c we have the
concentration-like variable, temperature at each point in the smoke. The spe-
cies concentration will decay as air is entrained into the smoke. Some species,
like temperature, decay as well due to losses of mass as the smoke encoun-
ters clean, cold surfaces. For example, soot is deposited, and HCl (gas) can
condense on surfaces. This surface transfer is increased as the local smoke
velocity increases; that is why heavy depositions of soot are seen around
door cracks as the smoke squirts through at high velocity. In addition, soot is
transported by both a concentration and temperature difference. The latter
is called thermophoresis and mainly responsible for soot deposition early in
a fire onto cold surfaces. Objects adjacent to a flame can have these deposits
subsequently burn off.
Concentration can be expressed in several ways. Mass concentration is
defined as
species y species m
m
Yspecies = = g species/g sm
moke (8.5)
smoke
m m smoke
where m smoke is the mass flow rate of the smoke. Usually, concentrations are
represented as volume fractions, that is, the volume occupied by the spe-
cies at normal atmospheric temperature and pressure to the corresponding
smoke volume. For example, if we could extract an individual species and
let it expand in normal atmospheric pressure, then its volume compared to
the volume resulting from extracting the entire smoke mixture is volume
fraction. This volume fraction can be estimated as
(29 mol of smoke per g of smoke)Yspecies (8.6)

X species =
( Mspecies mol of species per g of species)
in moles (or volume) of species per moles (volume) of smoke

where Mspecies is the molecular weight of the species, and 29 is an approximate
molecular weight of the smoke (since it mostly contains N2, 28 g/mol).
Example 8.2
In the previous example for production of CO from the PS fire, the flow
rate of the smoke flow leaving the room must balance the fuel and air
smoke = 500 g air entering + (38 g/m2-s)(1.9 m2) fuel sup-
flow rate, so that m
plied ≈ 572 g/s.
Therefore, from Equations 8.3 and 8.5, for the underventilated case
(0.12)(38 g/m 2 -s)(1.9 m 2 )

YCO, under = = 0.015 or 1.5%.
572 g/s
The corresponding volume concentrations is computed as
(1.5%)(29)
X CO, under = = 1.56%
(28)
If an overventilated yield for PS were used in error, the overventilated

case would give concentrations of
4.3 g CO/s
YCO, over = = 0.0075 or 0.75%
572 g/s
and
(0.75%)(29)
X CO, over = = 0.78%.
(12 + 16)
Let us examine the hazard these concentrations of CO present. It is common

to express low concentrations (by volume) in terms of parts per million (ppm).
For example, the CO volume fraction given above as 0.0075 is 7500 ppm
because 0.0075 = 7500/106 .
a is the mass flow rate of air into a room and m
In general, if m is the rate of
fuel supplied, it can be shown that the mass concentration is directly related
to the yield from Equation 8.5 as
species y species m
m y species m (s/m
a)
m
Yspecies = = =
smoke
m m smoke (ma +m )(s/m a)
æ F ö
Yspecies = y species ç ÷. (8.7)
è s+F ø
Here, Φ is based on air and fuel supplied, Equation 8.1, and the stoichio-
metric ratio, s, of the fuel. So for a given fuel, there is a one-to-one corre-
spondence between yield and concentration at a given equivalence ratio.
As the equivalence ratio increases toward 1 and more, the concentration
increases sharply. As the hazard is related to concentration and exposure
time, survivability diminishes.
8.5 Hazards
Each concentration of species can present a specific hazard. The hazard can
be to humans, to equipment, and to property. Here, we consider only the
hazard to humans. However, essentially in each case, the hazard is measured
by the duration of the concentration exposed. Below a threshold concentra-
tion, nothing is damaged. Here, we use concentration in a general sense.
It can include temperature, the reduction in oxygen (vitiation, 0.21 — X O2 ), as
well as specific harmful combustion product concentrations such as carbon
monoxide, XCO.
Combustion products such as soot and odor-bearing hydrocarbons can
destroy significant quantities of goods, even when the actual fire damage
by heat is small. Soot ladened with HCl or other acid gases (e.g., HBr, HF)
can deposit on equipment. If not cleaned, this acidity can cause considerable
damage later through corrosion. Such effects on goods and equipment, we
do not pursue further.
We draw from the review by Purser2 to examine toxicity effects.
8.5.1 Narcotic Gases

Narcosis is the state of induced sleep, which reduces the capability of escape
and can lead to death. It can be caused by the combustion gases CO (carbon
monoxide), HCN (hydrogen cyanide), and even CO2 (carbon dioxide) at high
concentrations. It can also be induced by the reduction of oxygen to low con-
centrations. All animals require oxygen to be transported in the body by
oxyhemoglobin (O2Hb) in the blood. The introduction of CO into the lungs
causes the hemoglobin to attach instead to the CO as carboxyhemoglobin
(COHb), which prevents oxygen in the blood from attaching. The introduc-
tion of HCN affects the cells and also prevents oxygen from being used by
the body. Roughly, all of these effects are additive, so that if you are half dead
from CO, 3/8 dead from HCN, and 1/8 dead from the reduction of oxygen,
you are dead in accordance with the additivity of narcosis effects.
Figure 8.4 shows the time to reach loss of consciousness for exposure
to HCN and CO in primates for a specific activity level. We see from this
figure that HCN is more toxic: It requires much less concentration to cause
unconsciousness. Also the threshold levels below which no loss of con-
sciousness occurs is about 90 ppm for HCN and 900 ppm (or 0.09%) for CO.
Below these levels, incapacitation is still possible, but not unconscious-
ness. Table 8.2 shows the effect of oxygen loss due to oxygen or carbon
monoxide on humans. Incapacitation can occur at as low as 8%–12% O2
and 0.015%–0.040% CO corresponding to 20%–40% COHb. Usually, 40%
COHb indicates an incapacitation level and 60% COHb a lethal state.
These levels are not precise and can vary in humans depending on health
and other factors.
ppm
HCN CO
8000
7000
6000
CO
5000
4000
300 3000
200 2000
HCN
100 1000
10 20 30 min
Time to loss of consciousness
FIGURE 8.4
Time to unconsciousness in primates by exposures to CO or HCN. (After Purser, D.A.,
Toxicity assessment of combustion products, Sect. 8, Chap. 2, in: DiNenno, P.J., ed., SFPE
Handbook of Fire Protection Engineering, 2nd edn., National Fire Protection Association,
Quincy, MA, June 1995.)
TABLE 8.2
Effects of O2 Loss in Blood
Loss due to Decrease Loss due to Increase in
in Oxygen Carbon Monoxide
Concentration Concentration
O2Hb (%) X O2 (%) COHb (%) X CO (%) Effect
90–100 15–21 0–10 <0.008 None
80–90 12–15 10–20 0.008–0.015 Fatigue
60–80 8–12 20–40 0.015–0.04 Dizziness, nausea, possible paralysis
50–60 6–8 40–50 0.04–0.06 Prostration, asphyxiation, collapse
30–50 3–6 50–70 0.06–0.3 Unconscious in minutes, death
0–30 0–3 70–100 >0.3 Unconscious in seconds, death likely
Source: Based on data from Purser, D.A., Toxicity assessment of combustion products, Sect. 8,
Chap. 2, in: DiNenno, P.J., ed., SFPE Handbook of Fire Protection Engineering, 2nd edn.,
National Fire Protection Association, Quincy, MA, June 1995.
TABLE 8.3
RMV (Liters/Minute) for a 70 kg Man
RMV Activity
8.5 Resting
25 Light work
50 Heavy work, slow running
Source: Based on data from Purser, D.A., Toxicity assessment of com-
bustion products, Sect. 8, Chap. 2, in: DiNenno, P.J., ed., SFPE
Handbook of Fire Protection Engineering, 2nd edn., National Fire
Protection Association, Quincy, MA, June 1995.
Carbon monoxide is the leading killer of people in fire, principally because

of the long-time exposure from smoldering fires (usually > 1 hour) or because
of rapid fire growth resulting in an underventilated fire. An approximate
formula that shows the effect of exposure to CO over time is given in terms
of the resultant COHb:
COHb (%) ≈ 0.33RMV · XCO (%) · t (min) (8.8)
where
t is the exposure time in minutes
XCO is the volumetric CO concentration in %
RMV is the respiration minute volume or inhalation rate in liters/min
Table 8.3 gives typical RMV values. The significance of Equation 8.8 is that
the damage property (COHb) depends on the exposure quantity (XCOt) and
the respiration rate (RMV). Not only does activity effect RMV, but so does the
exposure to CO2, a plentiful combustion product. Table 8.4 shows the effects
of CO2, both on RMV and on narcosis. An increase in CO2 uptake increases
the respiration rate. So as a fire produces more CO and CO2, the hazard to
humans and animals increase.
In a fire smoke environment with a hydrocarbon fuel, the atmosphere can
contain CO2, reduced O2, and CO. All three accumulate to induce narcosis
due to reducing oxygen in the body, hypoxia. If atomic nitrogen (N) is con-
tained in the fuel, HCN will be produced and it too adds to the narcosis. It
is believed that commonly CO is the predominant cause of death due to fire
smoke, but HCN can be a significant contributor. By the time CO2 and O2
become hazardous, the CO and HCN effects have done their damage.
8.5.2 Additive Fractional Incapacitation Doses

The narcotic gases can combine to harm in an additive manner. While CO
alone may not incapacitate or harm, the other gases can contribute to do so.
A model to account for these effects is based on the concept of a fractional
TABLE 8.4
Effects of CO2 on Humans at Normal Atmospheric Pressure
and O2 Concentration
CO2 Concentration
in Inhaled Air (%) Effect
0.04 Negligible, level in normal air
0.5 Safe limit, prolonged exposure
1.8–2.0 30%–50% increase in ventilation rate
2.5–3.0 100% increase in ventilation rate
4.0 300% increase in ventilation rate
5.0 Dizziness, poisoning symptoms, >30 minutes
7.0–9.0 Unconscious, in 15 minutes
10.0–30.0 Unconscious, in <10 minutes, followed by death
Source: Purser, D.A., Toxicity assessment of combustion products,
Sect. 8, Chap. 2, in: DiNenno, P.J., ed., SFPE Handbook of Fire
Protection Engineering, 2nd edn., National Fire Protection
Association, Quincy, MA, June 1995.
incapacitating dose for each. Dose is the accumulation of the species in the
body over time. The fractional dose is the actual dose divided by its cor-
responding incapacitating dose. For example, very approximately from
Figure 8.4, at 10 minutes the incapacitating dose of HCN is about (130 ppm
× 10 minutes) and for CO about (2700 ppm × 10 minutes). So we might take
the approximate incapacitating dose as about 0.13%-min for HCN and 2.7%-
min for CO. Suppose in a fire exposure for an individual, the accumulated
doses were 0.026%-min for HCN and 2.16%-min for CO. The fractional doses
for each would be 0.20 for HCN and 0.80 for CO. Adding to 1 means that the
two combine to incapacitate the individual exposed. Adding to more than 1
definitely means incapacitation and possible death.
The process is more complicated than this example. There are factors to
consider: (1) variations in incapacitating doses for individuals, (2) the accuracy
of the model, and (3) other biological and toxicological complexities. Purser2
ably describes many of these factors.
8.5.3 Irritant Gases

Irritant gases consist of the acid gases HCl, HF, HBr, and other hydrocar-
bon by-products, such as acrolein and formaldehyde which can be derived
from wood. These gases can be painful to the eyes and air passages and can
reduce the walking speed in escape. They are not likely to be fatal during
inhalation. Postexposure fatality is, however, possible. Pulmonary damage
can result in death up to a day following exposure. It is also possible that
these gases tend to reduce the RMV.
8.5.4 Smoke Visibility

Smoke visibility is perhaps the first perceived hazardous effect in a fire,
even before the smoke becomes a serious thermal threat. Light is attenuated
by smoke due to particles, mainly soot, but also due to tarry condensables.
Soot gives smoke a black color. Tars, generated by heating in an ample air
supply, give smoke its whitish color. White smoke likely means unburned
fuel and condensables present in the smoke. The property that measures
the attenuation (absorption and scattering) in smoke is the extinction coef-
ficient, κs. κs is measured in m–1. It can be used to compute the reduction
in light intensity, I, over a light path length, l. The original intensity, Io is
reduced by the smoke to
I = Io exp (–κs l) (8.9)
If the extinction coefficient is 1 m−1 and the path length is 2 m, the intensity
of the light will be reduced about 13.5% of its original value.
Sometimes this equation is presented in the fire smoke literature as based
on the power of 10, and the optical density per unit path length is used
instead of the extinction coefficient.
The ability to see through smoke is called visibility, Lv, and is related to κs.
Figure 8.5 shows the relationship between these variables. Lv is the maxi-
mum distance an object can be seen and recognized through the smoke.
The extinction coefficient is related to a property called the mass optical
density (Dm). The mass optical density is typically measured and reported
in the fire literature. It represents the extinction of visible light due to solid
and liquid particulates in the smoke. This property has been tabulated
in Table 8.1. The extinction coefficient depends on the concentration of the
l
Light, lo l, Detector
κs
.
V, Smoke
Object
.
Fuel m
FIGURE 8.5
Smoke attenuation and visibility.
smoke particulates and can be computed for a given burning rate (m ) and
smoke volumetric flow rate V . It is related through these variables to the
mass optical density for a case of flowing smoke as
m
mD
ks = , (8.10)
V
or for a static case where m is the mass of fuel burned in a closed volume, V
mDm
ks = . (8.11)
V
For each value of κs, there is a corresponding visibility, which is given in
Figure 8.6 as approximately
Cv
Lv =
Do (8.12)
where Do is the optical density per unit path length and is an alternative
parameter to the extinction coefficient. In fact, κ s = 2.3 Do and Cv ranges from
1 to 4 m3/g depending on the illumination of the object. The range of results
10
5
Lv, Visibility (m)
(Range of data)
1.0
0.5
0.2
0.1
0.02 0.05 0.1 0.2 0.5 1 2 5 10
Do, optical density per path length of smoke (m–1)
FIGURE 8.6
Smoke visibility in terms of optical density per path length. (After Quintiere, J.G., Fire Mater.,
6(3, 4), 145, 1982.)
shown in Figure 8.6 was developed from various experiments using people
to sight objects through smoke from different fires. The results vary due to
human interpretation and on the way the sighted object was illuminated.
Example 8.3
Let us consider an extension to our previous example for the PS in the
underventilated room fire. Assume the smoke has cooled to normal air
temperatures so that the density (ρ) of the cool gas mixture is approxi-
mately 1.0 kg/m3.
For the PS fuel burning underventilated in the room
The PS, the volumetric flow rate is
m 500 g/s + 38 g/m 2 -s ´ 1.9 m 2

V = smoke = = 0.572 m 3/s
r 1 kg/m 3 ´ 10 3 g/kg
Table 8.1 only gives Dm for the overventilated case as Dm = 0.34 m2/g.
As the extinction coefficient is related to soot concentration, we take
Dm y soot
=
Dm , over - ventilated y soot , over - ventilated
For PS at Φ = 1.9, approximating the soot yield ratio in Figure 8.2 as 1.5,
and using Table 8.1, we find for the underventilated fire,
Dm = 0.335 m2/g × 1.5 = 0.503 m2/g.
Equation 8.10 gives
38 g/m 2 -s ´ 1.9 m 2 ´ 0.503 m 2/g

ks = = 63.5 m -1
0.572 m 3/s
The smoke visibility is then computed from Equation 8.11 as
Lv = (1 to 4 m 3/g) ´ 2.3/63.5 m 2/g = 0.036 to 0.145 m
or about 4–15 cm: This is poor visibility.

For the wood fuel burning underventilated in the room
Note that the same unventilated computation for the wood fire at
Φ = 1.97 gives
11 g/m 2 -s ´ 15.2 m 2 ´ 1.5 ´ 0.037 m 2/g

ks = = 13.9 m -1
0.667 m 3/s
and
Lv = (1 to 4 m 3 /g) ´ 2.3/13.9 m 2 /g = 0.17 to 0.66 m .

This wood smoke is somewhat clearer than the PS case, but visibility is
still poor. However, these were big fires of 2000 kW.
Suppose the wood fire was at 200 kW, and the air flow into the room is
still at 500 g/s. This is an overventilated case. Then for this case we have
1/10th the area burning,
11 g/m 2 -s ´ 1.52 m 2 ´ 0.037 m 2/g

ks = = 1.2 m -1
0.516 m 3/s
and
Lv = (1 to 4 m 3/g) ´ 2.3/1.2 m 2/g = 1.9 to 7.7 m.
This is much clearer smoke, but it will still impair visibility.
Irritating acid gases can have significant effects on people and their ability
to navigate through such smoke. Jin4 reports that for irritating smoke the
drop off in visibility is nearly precipitously low at an optical density per
path length of about 0. 2 m−1. This is in contrast to smoke (likely without
acid gases) shown in Figure 8.6. Also he reports that walking speed can
decrease from 1 to 0.5 m/s when walking in clear air to smoke. With irri-
tating smoke, this decrease occurs at an optical density per path length
of about 0.17 m−1 but at about 0.35 m−1 for nonirritating smoke. In general,
he reports that people familiar with a building need a visibility of 4 m
(Do ~ 0.8 m−1) for safe egress, and 13 m (Do ~ 0.1 m−1) for those in unfamiliar
surroundings. These criteria were based on emotional and physical fac-
tors for individuals whom he tested in smoke egress experiments. Factors
affecting egress through smoke included irritation to eyes and throat, fear
as smoke density increased, and general anxiety as vision reduced, as well
as the ability to clearly see.
8.5.5 Heat Effects

Temperature can also cause harm by convective and radiative heat
transfer. Two effects are possible: Heat stress is relatively long term; a
burn injury is more immediate. If skin reaches 45°C, humans can feel
pain. At higher temperatures, damage can occur at increasing tissue
depths. Figure 3.10 shows the time response for blister burns due to
radiant heating of bare skin. We need at least 4 kW/m 2 to cause such
burns in a relatively short time (in contrast to long-term exposure to the
sun at ≤1 kW/m 2, which can result in a relatively mild skin burn—first
degree). Purely convective heat fluxes associated with the threshold heat
flux for a blister correspond to a given smoke temperature. This tempera-
ture can be estimated from Equation 3.2—the formula for convective heat
transfer. Selecting a heat transfer coefficient of h = 10 W/m2-°C, which

approximates a stationary person, gives
q ¢¢ = h(Tsmoke - Tskin )
q ¢¢
Tsmoke = + Tskin
h
Solving,
æ 4 kW/m 2 ö
=ç 2 -3 ÷ + 45°C
è 10 W/m -°C ´ 10 kW/W ø
= 445°C.
Similarly, the pain threshold temperature for a stationary person is roughly

200°C. Actually, people have been exposed in tests to such high temperatures.
The ability to perspire increases their tolerance time. But after long heating, the
body core temperature can rise to cause heat stress in spite of perspiration. Heat
stress occurs when the core temperature reaches 41°C; normal core temperature
is 37°C. At 41°C, consciousness can be affected and further increases can lead to
death by hyperthermia. Heat stress occurs when the body can no longer regu-
late its core temperature. This is roughly described by a balance of energies:
[Rate of energy stored in body] = [Metabolic energy release rate]

+ [Radiative and convective heat
rates]
– [Evaporation energy loss rate due
to perspiration]
– [Respiration energy loss rate]
If the left-hand side is not zero, we have an imbalance, and the core tem-
perature can go up (or down), indicating the start of heat stress. Table 8.5
TABLE 8.5
Tolerance Times under Heat Stress Conditions
Exposure Temperature, °C RH, % Tolerance Time
49 10 ~10 days
49 50 ~2 hours
49 100 ~10 minutes
100 0–100 ~10 minutes
Source: Purser, D.A., Toxicity assessment of combustion products, Sect. 8, Chap. 2, in: DiNenno,
P.J., ed., SFPE Handbook of Fire Protection Engineering, 2nd edn., National Fire Protection
Association, Quincy, MA, June 1995.
lists some tolerance limits of people while at rest exposed to high tempera-
tures. Notice that the relative humidity, RH, has an effect on the evaporation
of perspiration, and therefore affects the tolerance time.
8.6 Summary
Smoke is the flow of products and entrained air away from the fire.
Concentrations of combustion products can be computed from their yields.
Yield is the fraction of combustion product produced per mass of fuel sup-
plied. These yields depend on the fuel and on the fire condition: flaming condi-
tions of over- or underventilation, smoldering, or heating. The equivalence
ratio is a measure of fire ventilation and is crucial in controlling the yields of
species and their concentrations in smoke. Concentration is the fraction of
the species product in the smoke. The effect of species concentrations over
time is known as dose. Dose, the product of concentration and time, can be
a direct measure of harm or damage. For narcotic gases, the concept of frac-
tional dose is a model for the combined effects of species. Various criteria are
available to quantify the level of harm. A summary of the primary effects to
humans from combustion products is listed as follows:
Product Effect
Temperature Heat stress, burns
CO2 Increase in respiration
Soot, tars Visibility
O2, CO, HCN Loss of oxygen supply to blood
HCl, HF, HBr Sensory irritant
Review Questions
1. A smoldering fire occurs in a closed room 40 m3 in volume. The smoke
remains in the room and maintains an average density of 0.9 kg/m3. It
is known for this case that the smoldering is steady at 1 g/s. It produces
yield 0.2 g of carbon monoxide per g of fuel smoldered.
a. Compute the mass of smoke gas in the room.
b. Compute the mass of CO produced after 1 hour.
c. Compute the mass (fraction) concentration of CO in the room after
1 hour.
d. For a person resting, compute the COHb (%) using Equation 8.7 and
Table 8.3. Note: The mass fraction (Y) of CO is nearly identical to its
volume fraction (X), as CO has a molecular weight of 28 and air is
about 29.
e. In a 40 m3 room the mass optical density (Dm) of the smoke is 0.33 m2/g
(from Table 8.1). For the 1 g/s smoldering fire, estimate the smoke vis-
ibility in the room at 1 hour, assuming smoke is well mixed through-
out the room. Use Equation 8.11 and Figure 8.6 for your estimation.
2. Will acetylene produce more soot than propane? Explain the basis of
your answer.
3. Compute the equivalence ratio for nylon using Table 8.1 if its burning rate
is 10 g/s and the air supply to the fire is 600 g/s. Is it underventilated?
True or False
1. Yield is the same as concentration.

2. An exposure to 50 ppm (50/106) of carbon monoxide can kill you.
3. Visibility through smoke does not depend on lighting. That is why exit
signs are not lit.
4. Acid gases, such as hydrogen chloride, can cause corrosion of computer
equipment after a fire.
5. Heat stress is usually not a factor to people trapped in a fire but is a seri-
ous problem for firefighters.
6. Yield is directly related to concentration.
7. Visibility in smoke is the first hazard to humans in fire.
8. Lethal species can combine to kill you in fire smoke even though each
one alone could not.
Activities
1. Examine fire incident statistics and draw from your experience to assess
the time span involved in lethal smoldering fires.
2. Why are rats or mice used in smoke toxicity testing to assess the hazard
to humans? Do their results directly apply to humans? You will need to
research this answer.
3. Examine why CO is a recognized fire hazard to the average person, but
HCN is not.
References
1. A. Tewarson, Generation of heat and chemical compounds in fires, Chap. 3-4 in
2. D. A. Purser, Toxicity assessment of combustion products, Sect. 8, Chap. 2 in
3. J. G. Quintiere, Smoke measurements: An assessment of correlations between
laboratory and full-scale experiments, Fire and Materials, 6, 3 and 4 (1982):
145–160.
4. T. Jin, Visibility and human behavior in fire smoke, Chap. 2-4 in SFPE Handbook
of Fire Protection Engineering, 4th edn., edited by P. J. DiNenno (Quincy, MA:
National Fire Protection Association, 2008).
9
Compartment Fires
Learning Objectives
• Explain the processes in the development of fire in a compartment

• Explain flashover, fully developed and ventilation-limited fires in
a compartment
• Explain fire-induced flows, neutral plane, and the ventilation factor
• Compute vent flow rates and compartment smoke temperatures
9.1 Introduction
Fire in a compartment involves the containment of smoke and the potential
for the fire to spread beyond the compartment. Small fires can transport
smoke over great distances. Large fires can be affected by the compartment:
enhanced due to thermal feedback (increased heat transfer) and dimin-
ished due to oxygen vitiation (decreased air flow to the fire). The motion of
smoke and air in these fires is mainly due to buoyant effects as a result of
the increased temperatures. Typically, the forced conditioned air supply in
a building is secondary to this buoyant flow. We examine these effects and
introduce some simplified, but very useful, formulas for assessing smoke
temperatures and flow rates. We will also take a close look at the phenom-
enon of flashover: The transition from an object burning in a room to the
fire suddenly growing to it biggest possible size. Smoldering fires for most
of this chapter are not deeply addressed. They will be briefly discussed at
the end of the chapter.
267
9.2 Stages of Fire Development

Fire in a compartment is initially unaffected by the compartment conditions.
But as the oxygen is reduced in the compartment and the compartment
temperature rises, the compartment conditions play a role in the behavior
of the fire. The stages of fire in a compartment are depicted in Figure 9.1,
in terms of temperature and time. The fire usually develops from a single
item (developing), then grows rapidly and usually involves others (flash-
over), progresses to the biggest fire the compartment and its ventilation can
allow (fully developed), and, once extinguished, the process ends with a cool-
ing stage.
The stages of a fire in a compartment are described in more detail and are
depicted in Figure 9.2. In that figure, the results of computations for that par-
ticular room are shown along with sketches of the fire and smoke behavior.
A range of realistic furniture properties were used to compute the possible
ranges of smoke CO concentrations and visibility. Also, the equivalence ratio is
estimated. The results are intended for illustrative purposes and to give quanti-
tative estimates of conditions at that stage of the fire. (The student might try to
recompute these fires after reading this chapter to independently confirm their
general accuracy.)
Fully
Flashover
developed Cooling
Development
phase
Temperature
Effect on structure
Effect on people
Effect on operations
Time
FIGURE 9.1
Stages of fire in a compartment.
Compartment Fires 269
9.2.1 Developing Fire

Following ignition, the fire grows on an item and may also involve other
items. The oxygen concentration and compartment temperatures are
not much different than normal air. The fire behaves as if it is burning
in the open air for most of this period (see Figure 9.2a). However, for
some fuels, the smoke can have very limited visibility (10 cm) and sig-
nificant CO content (0.3%). The smoke layer in the room would be above
mid-height.
Flames touch ceiling

600 kW
CO ~ 0.01%–0.3%
Visibility ~ 10–250 cm
(a)
Onset of flashover
1000 kW
CO ~ 0.03%–3%
Visibility ~ 3–70 cm
(b)
Fully developed
ventilation, limited
4500 kW
CO ~ 2%–7%
Visibility ~ 0.1–10 cm
(c)
FIGURE 9.2
Typical conditions for furniture items burning in a 10 × 10 × 8 ft high room with a 3.2 × 6.4 ft
high doorway. CO is concentration of carbon monoxide. (a) Developing fire, (b) onset of flash-
over, and (c) fully developed fire.
9.2.2 Flashover
Flashover marks a dramatic increase in fire conditions due to the confine-
ment of a room. Flashover applies to this period of change. It can be caused
by several situations:
1. The rapid ignition, burning rate, and flame spread of materials due
to increasing heat flux arising from the heated room. This is depicted
in Figure 9.2b. This event is commonly called flashover. Note that at
the onset of flashover, the equivalence ratio is well below 1; the fire
is overventilated.
2. The accumulation of fuel-rich gases and their sudden exposure to air
with ignition. This is commonly called backdraft by the fire service
and others but can be regarded as a type of flashover. As a premixed
flame can move rapidly through the gas, this event can release a lot of
energy in a short time. As a result, the pressure in the compartment
can increase, causing a large pulse of flame and possible structural and
glass failures.
In the rest of this chapter, we will regard flashover as that described in 1.

There is a beginning and end to flashover; it is a transition that can occur
over seconds or minutes. It is due to an instability. It is much like an ignition
process. In ignition, first there is fuel and air and then there is a flame. In
flashover, first there is a small fire and then there is a big fire. For flashover,
the instability is caused by the firepower suddenly exceeding the heat loss
and energy flow rates from the compartment. This imbalance results in a
sharp temperature increase in the smoke of the compartment. Flashover will
cause the fire to reach its fully developed state. All of the fuel gases may not
be able to burn within the room because the air supply is limited. Flames
will occur near the origin or seek out available air supply locations. The
flames will not fill the room, but the smoke layer will descend near the floor.
Flashover terminates when all of the oxygen is consumed or when flames
cover all of the available fuel in the room. The former state is termed
ventilation-limited, and latter state is called a fuel-controlled fire. Then flashover
marks a transition in which the fire development was previously dominated
by the fuel materials and subsequently is primarily controlled by the ventila-
tion conditions indicative of the room and building geometry.
9.2.3 Fully Developed

This stage occurs after flashover. If there is sufficient air (fuel con-
trolled), flames fully encompass the room, with the likelihood of flames
extending from windows and doors. A fire in a glass tower office build-
ing, with full fire on a floor and all the floor-to-ceiling windows broken
(e.g., First Interstate Bank fire, Los Angeles, May 1988), is likely to be
fuel controlled. If insufficient air is only available to the room, the fire
is ventilation limited. A ventilation-limited fire can be defined as one in
which the smoke leaving the room has a concentration of 0% oxygen—all
of the oxygen is reacted within the room. When the ventilation-limited
condition occurs, the production of CO, smoke, and energy are at their
highest values. More fuel will be generated by heat in the room than can
be burned inside. Its equivalence ratio is greater than or equal to 1. This
case is depicted in Figure 9.2c. All of the fuel is involved to its maxi-
mum potential. For a ventilation-limited fire, the room’s energy release
rate is established by the available air supply rate. Most compartments
and buildings become ventilation limited at this fully developed stage,
as there is always enough fuel in typical furnished rooms compared to
ventilation.
Note that in Figure 9.2, typical ranges of values for CO, smoke visibility,
and equivalence ratio were computed for various fuels contained in that
particular room. The firepower was selected to match the stage of the fire,
with a 500°C criterion used for the onset of flashover.
The fully developed state marks the point where the structure becomes
vulnerable. Heat flux conditions in the room can reach as high as
270 kW/m2. This corresponds to a smoke temperature of 1200°C that may
occur in large compartment fires. However, Cox1 reports that in a tunnel fire
(Summit tunnel fire, England, December 1984), the incidence of a melted steel
rail and fused brick occurred. This is shown in the photograph in Figure 9.3.
FIGURE 9.3
Melted steel rail in the Summit Tunnel fire. (From Cox, G., Private communications, 2010.)
He indicates that fused brickwork could occur at about 1250°C and steel
could melt at about 1400°C–1500°C. Both of these effects are shown in the
photograph. Such fire conditions would have had extremely high heat fluxes.
There has not been sufficient experimentation in large building fires to indi-
cate such heating conditions as the tunnel, but one cannot rule this out.
9.2.4 Cooling Stage

The fire will eventually extinguish as it runs out of fuel in the fully
developed stage. The compartment surfaces and objects remain hot.
Out-gassing still remains, as fires die to smoldering. The toxic con-
ditions during this period remain hazardous and can be of harm to
firefighters. Most significantly the cooling stage is still a factor in heat-
ing the structure. This is especially true for steel reinforcing rods in
concrete. While the steel rods may not have been heated during the fully
developed stage, the thermal penetration in the concrete continues to
occur. During the cooling stage, the thermal penetration could reach the
vulnerable rods.
9.2.5 Example of Measured Conditions in Room Fire

Let us take a look at the stages of a fire from an experiment reported by
Hartzell.2 This is shown in Figure 9.4. The fire began as smoldering on an
upholstered polyurethane-cushioned chair. The measurements were taken
at a point indicated in the sketch.
Flashover is perceived at 25.5 minutes. The temperature is 500°F (260°C) at
that instant, but reaches about 1000°F (538°C) in less than a minute. At about
27 minutes, the fire reaches the fully developed (and ventilation-limited)
stage as the O2 drops to zero, the CO peaks at 7%, and later, the temperature
ultimately reaches 1600°F (870°C). Note that the HCN concentration from
the polyurethane foam chair does not increase until well after flashover.
Also, the fact that the oxygen drops to zero and remains there clearly shows
the fire has used all of the oxygen in the air that entered the room and is
ventilation limited.
Because fire development in a room is very important, we shall come
back to a more detailed portrayal of it later. Measurements and external
observations give some indication of characteristic behavior, but pictures
give a more definitive and accurate impression. In the early development
of fire research, and this is still relevant today, observations led to concepts
and then generalizations on how to model fire and smoke behavior. The
processes could be broken into discrete pieces for study. First, we will focus
on fire flows and smoke movement.
Room: 12 × 18 × 8 ft high; open doorway
Data at 5.5 ft height
Flaming
ignition Flashover
25
70,000 O2 CO 1800 1800
Temperature (°F)
20 60,000
HCN (ppm)
Oxygen (%)
1400 1400
CO (ppm)
50,000 T
15
40,000 1000 1000
10 30,000
20,000 HCN 600 600
5
10,000 200 200
16 18 20 22 24 26 28 30 32 34 36
Time (min)
FIGURE 9.4
Furnished room fire initiated by a smoldering upholstered chair. (From Hartzell, G.,
Combustion products and their effects on life safety, Sect. 3, Chap. 1 in: Cote, A.E. and Linville,
J.L., eds., Fire Protection Handbook, 17th edn., National Fire Protection Association, Quincy, MA,
1991, pp. 3.3–3.14.)
9.3 Fire-Induced Flows

Flows occur as a result of fire, or in general due to any heat source through
buoyancy. Buoyancy causes a pressure difference that results in a velocity.
These concepts were presented in Chapter 7. The pressure difference Δp, as
related to buoyancy (Δ means a difference in p, i.e., p2 – p1) is
Dp = (ra - r) gH (9.1)
where
ρa is the air (cold) density
ρ is the smoke of fire (hot) density
g is the gravitational acceleration, 9.81 m/s2
H is a vertical height (between positions 1 and 2)
In Chapter 7, we called this pressure difference term relative potential energy

per unit volume. Note the unit equivalence:
Energy J Newton (N)

in 3 = Pressure in
Volume m m2
One newton (N) is the same as 1 kg-m2/s, and 1 Joule (J) is 1 kg-m3/s or
1 N-m; therefore,
N kg-m 2/s kg-m 3/s J

1 = 1 = 1 =1 3 .
m2 m2 m3 m
These equal terms are equivalent units of measure although their names are
different. A newton and a kilogram are related through Newton’s Second Law
of Motion: Force = Mass × Acceleration.
From equating the potential energy to the kinetic energy as shown in
Chapter 7, the resultant velocity due to pressure caused buoyancy is
2(ra - r) gH
V= . (9.2)
r
From Equations 9.1 and 9.2, we can obtain approximate formulas to explain
the flows associated with fire.
9.3.1 Duct Fan Pressures in Fire

Let us roughly explain the relationship of fire-induced flows compared to
fan-generated flows. Such fan flows would occur in air handling systems of
a building. Consider two cases as shown in Figure 9.5. The first is pure fan
flow over height H in a duct of diameter D. The second is purely buoyant
flow where the duct gases are at a higher temperature, T, compared to the air
at Ta. The fan would create a pressure difference to just balance the friction
effects as given by
H raV 2
DpFan = f (9.3)
D 2
where f is a friction factor. (It has no units, dimensionless.)

The buoyant duct would create a pressure difference given by
æ T ö
Dpb = ra ç 1 - a ÷ gH (9.4)
è Tø
(2) (2)
Δ P = P1 – P2
H H
Ta Ta Ta T
D D
(1) (1)
FIGURE 9.5
Ideal fan and buoyant flows in a duct.
This follows from Equation 9.1 where temperature has replaced the density
difference.
A representative fan supply flow is 1 m/s in a duct diameter, D, of 0.3 m for
air at 27°C with a density of 1.2 kg/m3. The corresponding friction factor for
this flow is 0.03. (Some advanced fluid mechanics of pipe flow is needed to
obtain the friction factor.) Then, the pressure needed to drive this duct flow
is computed from Equation 9.3 as
H 1.2 kg/m 3 (1 m/s)2

DpFan = (0.03) = 0.06 H (m) in N/m 2
0.3 m 2
In comparison, the buoyant pressure generated by smoke at 327°C, from

Equation 9.2, is computed as
æ 27 + 273 K ö
Dpb = 1.2 kg/m 2 ç 1 - n N/m 2 .
÷ ´ 9.81 N/kg ´ H (m) = 5.9H (m) in
è 327 + 273 K ø
The smoke pressure greatly exceeds the fan pressure over the same
height, H. The fan pressure would approximately equal the buoyancy
pressure for duct temperature of 28.5°C, only 1.5°C above outside air. It is
clear that, under fire conditions, buoyant flows are likely to dominate
fan-generated duct flows. Coupling the fire induced and duct ventilation
flows is not simple. But as fire grows, it can be expected that smoke can
flow into intake ducts.
9.3.2 Pressure Level Due to Fire

For a moderate value of H, typical of room or building heights, the pressure
differences are very small compared to the absolute atmospheric pressure.
For example, if H = 1 m, Δpb ~ 5 N/m2, compared to normal atmospheric pres-
sure of 105 N/m2. (Americans would be familiar with 14.6 psi as normal atmo-
spheric pressure at sea level.) Even for H = 100 m, the pressure difference due
to buoyancy is small. In general, the pressures due to fire and also due to air
handling systems do not cause any significant change in the overall build-
ing pressure. Very, very small pressure differences drive these flows, and
the overall building pressure does not significantly depart from 105 N/m2.
Incidentally, N/m2 is commonly called Pascal (Pa), as an alternative unit of
measure for pressure.
9.4 Compartment Flow Dynamics

Flow dynamics in a compartment fire were revealed through fire experi-
ments in the early 1970s. For a fire on the floor, a plume was discernable, and
a hot smoke layer stratified above it. The plume acted as “pump” ingesting
air due to entrainment. The smoke stayed on top because it was more buoy-
ant than the cooler air. Buoyant pressures and momentum pushed the flows.
Occasionally, some mixing occurred between the smoke and air, especially as
the smoke entered or exited the room. All of these effects could be observed
in early experiments. Let us examine how smoke flows in a single room, in a
corridor, and in a building.
9.4.1 Flow in a Room

Figure 9.6 displays a sequence of sketches that describe the process of fire
flows in a room with an opening—a vent near the floor.
At the onset of a fire, its turbulent plume would begin to rise toward the
ceiling. As long as its temperature remains higher than that in the room, it
will continue to rise until it hits the ceiling. The speed of this rise is relatively
fast. This can be appreciated as the velocity is related to gH , indicated by
Equation 9.2. For a ceiling at 2 m, this indicates a speed of about 4 m/s at
most. Of course temperature and mixing effects will slow this down, but it
is relatively fast. Similarly, once the plume hits the ceiling, its momentum is
redirected horizontally. Its velocity will slow as mixing occurs and it moves
radially through a larger circumference. But this still is relatively fast. This
ceiling motion is referred to as a ceiling jet. Research has indicated that its
depth varies slightly, but is generally about 1/10th of the distance from the
fire source to the ceiling. This depth is significant because detection devices
(a) (b) (c)
(d) (e) (f )
FIGURE 9.6
Sequence of developing fire flows in a room: (a) plume rise, (b) ceiling jet forms, (c) start of
layer, (d) layer descends, (e) layer reaches vent, and (f) bidirectional flow.
should be placed within the ceiling jet. The flow below it is relatively tranquil.
Detectors should be located within the jet, as they need a reasonable velocity
to respond.
Once the ceiling jet hits the confining walls, the momentum of the flow
wants to drive it back down, but buoyancy keeps it up. For rooms that are
wider than tall, buoyancy always wins. The flow then tries to return to the
plume. As only some of it can be re-entrained, it layers in a circulating fash-
ion. This serves to mix the forming layer beneath the jet. The bottom of the
hot layer interface is stably stratified with the cooler air below. This interface
is fairly sharp and is only blurred by diffusion. The interface oscillates much
like shallow ocean waves. It continues to descend.
At the point when the layer reaches the vent, it will fall below it. The smoke
layer can continue to drop as the temperature increases. (Cooling would
cause the layer to rise.) Up to this point, the pressure of the fire has resulted
in flow completely out of the room. But at some point, the layer settles
down, and flow must enter the compartment through the lower portion of
the vent. By conservation of mass, the outflow cannot continue indefinitely.
The resulting bidirectional flow is most common in established compartment
fires. For ordinary size open windows or doors, this room developing flow
stage occurs relatively quick. Thereafter, the flow is both in and out of the
room. As the fire becomes larger, the resultant fully developed room fire can
have a layer interface several inches above the floor.
Figure 9.7 gives a more detailed description of the pressure difference and
vent flows during the smoke filling process. The energy release by the fire
onset causes a pressure rise, and it remains higher than the surrounding air.
When the bidirectional flow occurs, the pressure at the lower region of the
compartment has now become lower than the surroundings. This sucks the
air into the room. The layer position at the vent in bidirectional flow is higher
than that in the room, as seen in Figure 9.7c. The vent layer interface is called
HL
P
H Ho
P +
Pa
Pa
Pressure
(a)
HL P
Pa
+
Pressure
(b)
+
Pa P
HN
HL P –
Pressure
(c)
FIGURE 9.7
Flow and pressure dynamics in a room with a vent. (a) Layer descending: cold flow leaves the
vent. (b) Hot flow begins to the vent. (c) Layer interface below the vent and cold flow enters
from the surroundings.
Fire plume Temperature distribution
Smoke flow
Neutral plane
Vent air flow
Entrainment
Smoke layer
FIGURE 9.8
Fire-induced flows in a room.
the neutral plane (HN). The pressure difference at the neutral plane is zero.
This height, for a single door opening, can vary from about 1/2 to 1/3 of the
doorway height as the fire gets bigger.
Recirculation within the smoke region in the room drives the tempera-
tures (and concentrations) to become nearly uniform in the vertical direction.
Laterally, heat loss can cause decreases in temperature moving away from the
fire. But for moderate and small size compartments, the concept of a uniform
property smoke layer has emerged as a suitable approximation to the process
for describing a room fire. The process is shown qualitatively in Figure 9.8 and
quantitatively in Figure 9.9. In the latter, the uniformity is shown for the wall
temperatures as well, and the lower layer below the smoke is also fairly uni-
form. In fact, the wall temperatures exceed the gas temperatures in the lower
layer due to radiant heat transfer from the smoke to the lower walls. Notice
too that the gas temperatures in the lower layer are above the incoming air
(at 25°C) in Figure 9.9 due to mixing as the air enters the vent and descends
into the room. This mixing effect can be inferred from Figure 9.7c as air enters
above the height of the layer in the room and must entrain some smoke as
it plunges to the floor. It forms a floor jet that seeks out the fire. This mixing
effect is responsible for bringing vitiated (lower in oxygen) air to the fire. That
vitiation will compete with increasing temperatures to control the fire.
9.4.2 Smoke Filling in a Leaky Compartment

The sequence of flow dynamics shown in Figure 9.6 would also apply to a
closed room up to (e), with the layer nearly getting to the floor. If the room
is perfectly sealed the room pressure would increase continuously and
120 kW wall fire in a (2.8)2 × 2.1 m high room

2.25
2.00
Two-layer
Data fit
1.75
Gas temp.
Wall temp.
1.50
Height (m)
1.25
1.00
0.75
0.50
0.25
0.00
0 40 80 120 160 180 200
Temperature (ºC)
FIGURE 9.9
Room temperatures with a wall vent at a steady state. (After Quintiere, J.G. et al., Fire Sci.
Technol., 4(1), 1, 1984.)
significantly as the temperature rises. The pressure would increase by as

much as a factor of about three atmospheres. However, unless it is a nuclear
reactor containment vessel, or a submarine, this pressure rise will not happen.
This is because building structures leak through their membrane, through
cracks and gaps, and even designed vents. Figure 9.10 depicts the process for
leaks at the ceiling and floor. We will later examine the results of modeling
this case in order to estimate the filling time for the layer to descend.
The filling process is the inverse of filling a tub with water. The smoke
interface moves down to fill the compartment. All buildings leak; they are
not made to be perfectly sealed. For normal buildings, the absolute pressure
changes only slightly to cause the appropriate flows within and at the leaks.
Essentially normal atmospheric conditions prevail.
During filling and leaking, the pressure within remains slightly positive
over the surroundings causing flow out of the room. The layer can descend
over the fire, and its decrease in oxygen can affect the fire. If the fire dies due
to lack of oxygen, cooling occurs and the pressure within becomes negative
FIGURE 9.10
Smoke filling a leaky space.
with respect to the surroundings, and consequently draws air in. This air
can revitalize the fire, and the pressure now goes positive again pushing
gases out. The process can be repeated and can appear as if the fire is breath-
ing. Firefighters know this sign and avoid venting such “breathing” enclo-
sures. The room can be filled with unburned fuel, and a sudden rush of air
can lead to backdraft.
The filling process can take many minutes to occur. It can be significant in
the fire development. For a smoldering fire, the smoldering can persist well
after filling, as it requires very little oxygen to maintain the reaction. During
this period, it becomes deadly to occupants. In flaming fires, the flames can
be extinguished and then lead to smoldering. This may later lead back to
flaming if an air supply is found.
In a large space, it may take a long time for filling. I once examined a fire that
occurred in a small storage room open on to a large clothing store with a high
ceiling. As the store was closed for inventory processing, the workers were
unaware of the fire until they smelled it. They smelled it when the smoke layer
exited from the storage room, filled the main store, and the layer descended
to their noses. A small fire may produce nearly invisible smoke, but odor is
surprisingly detected at low levels. I think many have experienced the odor of
wood burning at ground level far from a fire, with no sight of smoke.
9.4.3 Smoke Movement in a Building

The dynamics of smoke movement through a building have not been
sufficiently studied by experiments; however, we can qualitatively describe
the smoke’s behavior. Computer models have sought to do this. The zone
modeling approach views the fire in a room as two uniform layers use the
pressure dynamics to predict the flows. Of course, the process is more com-
plicated than the computation associated with Figure 9.5 for a single ver-
tical duct, but computers can keep track of all the pressure differences at
the vents in a building computer model. Other computer models try to solve
the basic conservation equations, attempting to model all features of the fire
and smoke from first principles. Such a comprehensive computer model is
commonly referred to as a CFD model (e.g., stands for computational fluid
dynamics, but they do much more). The zone model approach is fast, but
the CFD approach has supplanted it as computers have gotten bigger and
faster. Here, we will mainly be qualitative and simple in our view of building
smoke dynamics.
9.4.3.1 Corridors
Figure 9.11 shows steady flow from a room into a corridor that has an exit
vent. A stratified smoke layer whose interface height depends on the size of
the exit opening is formed in the corridor. The smaller the exit opening, the
lower is the interface (see Figure 1.5). Within the corridor, distinct hot and
cold layers form as the smoke stratifies. However, the flow pattern is more
complex as shown in Figure 9.11 (or Figure 1.6). Mixing occurs as the flow
enters the corridor and the room. A ceiling jet occurs as well as a cold floor
jet. In between, the cold and hot layers recirculate. These patterns were deter-
mined by flow trace observations as well as by measurements.
If the corridor has no exit vent, it will fill, much like a single compartment,
as smoke emerges from a room. If the flow is gradual, we have the process
shown in Figure 9.12a, and in the photographs in Figure 9.13a through d.
These photographs taken in a closed corridor for a small fire in an adjoining
room were visualized with artificial smoke. They clearly show the descent of
a smoke layer interface. However, the cold walls introduce a secondary flow.
In Figure 9.13d, the smoke begins to dribble down the cooler walls as the
layer interface approaches the floor. This is buoyancy in reverse; cooler fluid
moves down in this case.
Another possible flow pattern in the corridor is shown in Figure 9.12b.
There, a sudden high velocity release of smoke expands into the corridor
due to either a rapidly increasing fire or the sudden opening of the room
door. The initial momentum, or kinetic energy, of the smoke can cause a
wave front propagation that can turn back on itself. Such a flow would fill
the corridor by a lateral motion, compared to the more tranquil filling by
Room Corridor Exit view

HE = HD
HC
WE
YE
XCS
XCV
YD
XR
Stratified Smoke and air

smoke layer mixing region
FIGURE 9.11
Fire-induced flows in a room and corridor.
(a)
(b)
FIGURE 9.12
Smoke filling a closed corridor (a) slowly and (b) impulsively.
Doorway
Incipient
smoke flow
(a) (b)
FIGURE 9.13
Smoke filling a closed corridor from a room fire on the right. (a) Incipient smoke impinging on
the left wall of the corridor. (b) Smoke spreading along the corridor ceiling. (Continued)
(c) (d)
FIGURE 9.13 (Continued)

Smoke filling a closed corridor from a room fire on the right. (c) Smoke layer at mid-height and
descending. (d) Smoke layer descending near floor.
layering. The velocity of this front can be estimated by Equation 9.2, with H
given as the corridor height. It could be several meters per second.
I once investigated a high-rise fire that had two interconnecting floors of
plan office space occupied by a law firm. The floors were connected by an
open stairway, but the elevator lobby entrances were locked and monitored
by infrared security stations. The offices were unoccupied by people at the
time of the fire. When a fire in an exterior perimeter office broke the windows
and went to flashover, its wood door to the plan office space failed at the
hinges and fell. The momentum of the smoke from the room carried it along
the top upper floor and then down the internal stairway. The infrared motion
detectors at the elevator lobbies alarmed at about 30 seconds apart indicating
this smoke flow occurred.
9.4.3.2 Vertical Shafts

Flow up a tall shaft, such as an elevator conduit, would initially be plume-
like until the walls intercept the plume. This flow is shown in Figure 9.14.
After interception, a complex mixing process occurs that carries the smoke
upward. Once the shaft is filled with hot smoke, it can act like any other
chimney if it is vented at the top and bottom. The subsequent flow in the
Mixing
Plume
FIGURE 9.14
Smoke filling a shaft.
shaft depends on vents and leakage in the shaft. With no venting, the leakage
could bring air in and smoke out.
Figure 9.15 shows the flow processes in a building having been filled with
smoke. The hot shaft carries the smoke to the upper floors, and smoke leaks
out above the neutral plane of the building. The neutral plane is where the
pressure difference between inside and outside of the building is zero. Its
location will depend on the venting from the building. For the simple case
of a closed building of height HB with leakage through the skin, and at a
uniform inside temperature (T) compared to outside (Ta), the neutral plane
(HN) is found by
HN 1
= (9.5)
H B 1 + (T/Ta )1/3
This would be the case of a heated building on a cold day. For summer con-
ditions with a warmer outside temperature, and conditioned cool air inside,
the flow would enter the building through the top skin and exit below the
neutral plane. The location of this neutral plane would follow Equation 9.5
with the temperature ratio reversed, Ta / T. The summer and winter condi-
tions of a tall building present a stack effect that causes flows even with no
fire. The neutral plane would be about half the building’s height for such
temperature conditions.
This stack effect flow pattern due to the air heating and cooling conditions
of a building would mitigate any fire flows. The fire would introduce a more
complex flow. The fire floor would have its own neutral plane when it domi-
nates the building’s stack effect, as shown in Figure 9.15. Also wind can affect
the building smoke dynamics, particularly by altering the pressures around
HB
Building neutral plane

T
Ta
HN
+
– Fire floor neutral plane
FIGURE 9.15
Simplified dynamics of smoke filling a building.
the building. An early small fire in an air-conditioned building in the sum-

mer could experience downward motion of smoke in cool shafts and leakage
of smoke below the fire floor. Fire flows in tall buildings can be complex. It is
surprising that we do not use the available smoke detectors as an intrinsic
diagnostic tool in a high-rise building fire.
9.5 Single Room Fire Analyses

Let us consider how to make some quantitative fire estimates for a single
room. We will consider a room with leakage and one with a vent. We will
show how to estimate smoke filling time and the smoke flow rate through
the vent and its temperature. We will not derive these formulas, as they
can be found in the references. But we will try to show how the variables
that control the results enter the formulas and how to use the formulas
through graphs.
9.5.1 Smoke Filling

Zukoski et al.4 were first to formulate a prediction for the smoke layer descent.
Figure 9.7 gives the notation with the layer height as HL and the height of the
room as H. The fire is considered on the floor in the presentation, and the
effect of a raised fire can be considered too.4 As the Zukoski dimensionless
number Q* was introduced in Chapter 7, it is significant to present the basic
filling formula in these terms. For the simplest case of only a ceiling vent, the
equation gives
-3/2
HL æ 2 ö
= ç 1 + aQ *1/3 t ÷ .
H è 3 ø
The formula is just for presentation purposes. Q* here is based on the height,
not the diameter as in Chapter 7; the term τ is a dimensionless filling time
with α as an empirical constant varying from about 0.19 to 0.21 depending
on entrainment.
In a more practical form for the equation, we write the time for filling as
æ æ H ö2/3 ö S (m 2 )
t fill (s) = 24 ç ç ÷ - 1÷ (9.6)
ç è HL ø ÷ H (m)2/3 Q (kW)1/3
è ø
where
Q is the firepower
S is the surface area of the room floor
(In the development of Equation 9.6, the radiation loss fraction from the fire
was taken as 0.3, and its variation will not introduce any large differences.)
The results are meant to be approximate, and other leakage configurations
at just the bottom or distributed over the height of the room will change
the results. But this formula will give in general a good order of magnitude
estimate with about 50% accuracy at worst. Also the formula is for a fixed
firepower, not a growing fire that is more common. To account for different
fire sizes, Figure 9.16 is introduced to facilitate estimations from Equation 9.6.
The times are computed for two smoke layer interface depth ratios (HL/H)
for a 50 m2 room floor area. The depth ratio of 0.5 is indicative of filling to a
vent such as an open door or window. The depth of 0.2 is getting close to the
floor, and due to the cold wall downward flows, the smoke might be consid-
ered effectively very near the floor. The very small fire powers are indicative
of smoldering fires (less than 1 kW), and fill times to near the floor are more
than 10 minutes. In contrast, a fire power of over 100 kW drops the layer to a
vent location in about a minute or less for the room height of 3 m.
40
35 Floor area, S = 50 m2
Radiation fraction = 0.3
30
Time to fill to depth (min)
25
H=6 m
20 H=3 m
15
Depth, HL/H = 0.2
10 H=6 m
H=3 m
5
Depth, HL/H = 0.5
0
0.1 1 10 100 1000
Firepower (kW)
FIGURE 9.16
Time for smoke filling to various depth in a room heights 3 and 6 m with a floor area of 50 m2.
9.5.2 Smoldering Fire in a Closed Space

Compared to flaming fire, relatively few studies appear for smoldering.
Sometime ago experiments by Birky5 were analyzed. An upholstered chair
smoldered in a small closed room. Photographs and measurements of car-
bon monoxide revealed the nature of the process. Figure 9.17 shows the
smoke dynamics of the smoldering upholstered chair fire in the closed room
8.8 m × 2.4 m high. The smoldering was initiated by a simulated lit cigarette.
The photographs show that the smoke layer interface is not as sharp as would
occur in flaming fires. This is so because the slower process gives more time
for diffusion and secondary wall flows to occur.
Figure 9.18 shows the corresponding CO concentration as the smoke layer
descends. The smoke reaches the floor in 67 minutes. The CO concentration at
that time is about 1%. The accumulation of CO, or dose to a person exposed,
is the area under the CO–time curve. By the time of complete filling, the dose
is likely to be incapacitating. This can be estimated more precisely by refer-
ring to Chapter 8. The theory of filling was used to compute the solid line
curves in Figure 9.18. Oxygen remains high (typically > 16%) and tempera-
ture increases are small (<50°C) in these conditions. Smoldering can occur in
upholstered chairs with cotton or flexible polyurethane cushions with yields
of CO at about 0.1–0.2.
Figure 9.19 shows theoretical results for this same chair fire in 2.4 m high
room for a range of floor areas. The CO concentrations were computed, and the
time to reach an incapacitation level (estimated as CO (%) × time = 4.5%-min)
FIGURE 9.17
Photo sequence of smoldering chair smoke filling a closed room.
1.8
Chair: data
Theory 1.6
1.4
1.2
CO concentration (%)
2.5 1.0
Height of hot layer from floor (m)
2.0 0.8
1.5 0.6
1.0 0.4
Ideal
source
0.5 height 0.2
0 0
0 20 40 60 80 100 120
Time (min)
FIGURE 9.18
Dynamics of smoldering chair fire in a closed room. (After Quintiere, J.G. et al., Fire Mater.,
6(3 and 4), 99, 1982.)
is indicated as t* (and “*” on the curves). The smoldering source on the chair
was placed at 0.4 m above the floor; the CO is sensed at a height of 1.2 m,
representative of a sitting position. The time for the smoke layer to first reach
1.2 m above the floor is indicated by to in Figure 9.19. For residential room size
spaces, incapacitation occurs within 1 hour, but for house size spaces, it may
take up to 2 hours. These times are consistent with actual smoldering fire inci-
dent data. But often the fire will not remain as smoldering, and a transition
to flaming can occur in similar times as those indicated for incapacitation.
This transition to flaming can accelerate the increase in CO content of the
smoke. Such a transition is not currently predictable; it strongly depends on
aeration and flow velocity at the smolder region. Both processes, flaming and
continued smoldering, can be deadly to occupants in these times of 1–2 hours.
Smolder continuously yields low levels of CO over a long period, but a ven-
tilation-limited fire can lead to CO concentrations of 2%–7% in a short time
following flashover. The latter can cause incapacitation in less than 2 minutes.
S = 10 m2 16 m2
25 m2
S = Floor area
0.40 HC
50 m2
H *
L
0.35 y = H/HC
HC = 2.0 m
S (m2) to (min) t* (min)
L = 0.4 m
0.30 * 10 9 48
16 12 53
CO concentration (%)
* 25 17 60
0.25 50 26 72
100 40 86
* 200 61 107
0.20 100 m2
* 200 m2
0.15
*
0.10
0.05 Critical dose: ∫t*YCO dT = 4.5%-min

to
y(to) = 0.4
0 20 40 60 80 100 120
Time (min)
FIGURE 9.19
The increase of carbon monoxide for smoldering chair fire in rooms of different sizes. (After
Quintiere, J.G. et al., Fire Mater., 6(3 and 4), 99, 1982.)
9.5.3 Vent Flows

We have seen in Figure 9.7c that the pressure difference caused by the differ-
ence in smoke and air temperatures drives the flow across the vent. The rate
of smoke flow out is nearly equal to the rate of air flow in. The difference is
the small amount of fuel gases produced in the room. The exact mass flow
rate depends on the temperature difference and the size of the vent opening.
For a vent of height Ho and cross-sectional area of Ao, the maximum air flow
a,max , is given by
rate, m
a ,max (kg/s) = 0.52 Ao (m 2 ) H o (m)

m (9.7)
where
Ao is in square meters
Ho is in meters
This formula was first determined by Kunio Kawagoe, a noted Japanese

fire research pioneer. For large fires or small vents, Equation 9.6 is a
good approximation in estimating the air supply rate. Incidentally, Ao H o
is called the ventilation factor because it is the principal variable controlling
flow rate. Ao accounts for the flow area and the H o accounts for the velocity
as described by Equation 9.2.
Large vents and temperatures well below 800°C have lower flow rates
than this maximum. The relationship with actual flow rate data for crib
fires in a room is illustrated in Figure 9.20.6 The theoretical maximum,
Equation 9.7, places an upper bound on these door flow data. Recently,
a formula was developed for the entire range of flow conditions.7 It has
been shown accurate with data within ±20%. It depends on the location
and type of fire, the temperature, and the nature of the vent—window
.7 Smoke layer
temperatures:
At 794°C
m
.6
mu
axi
lm
.5
tica
At 243°C
ore
The
.4
ma (kg/s)
At 151°C
.
.3
.2
.1
0
1.0 2.0
Ao√Ho (m5/2)
FIGURE 9.20
Airflow rates through door openings for wood and plastic crib fires. (From Quintiere, J.G. and
McCaffrey, B.J., The burning of wood and plastic cribs in an enclosure, Vol. I, NBSIR 80-2054,
National Bureau of Standards, Gaithersburg, MD, September 1980, p. 49.)
Ho
Hw
FIGURE 9.21
Notation for mass flow rate through a window or door, Equation 9.8.
or door. It applies to Figure 9.21 where the notation in the formula is

described and is given as
m kg/s
= 0.52 5/2 e -Cy éë1 + 0.6(1 + (Cy )2 )(1 - e -3 Hw /Ho )2 ùû (9.8)
Ao H o m
where
2/3 1/2
æW ö æ Ta ö
y=ç o ÷ ç T -T ÷
è Ho ø è a ø
C = 0.55 for a centered floor pool fire, 0.85 near a wall, l in a corner or for
a crib.
Let us consider a graphical representation of a special case for this formula.
Take Wo/Ho as 0.5, Hw/Ho as 0 for a door, and 0.5 for a window, and consider
a centered pool fire with C = 0.55. For normal surrounding air temperature,
Ta = 293 K, and representing T – Ta as ▵T, this special case is given as
æ 5.93 ö
m kg/s -ç 1/2 ÷
= 0.52 5/2 e è DT ø
Ao H o m
é æ æ 5.93 ö2 ö ù
´ ê1 + 0.6 ç 1 + ç ÷ ÷ ´ (0 for a door, and 0.604 for the window )ú .
êë ç è DT ø ÷
1/2
è ø ûú
1.5
For a centered pool fire
For a vent with Wo/Ho = 0.5 ∆T = 800°C
For a window sill, Hw/Ho = 0.5
Vent mass flow rate (kg/s)
1 Door ∆T = 250°C
Window
0.5 ∆T = 150°C
0
0 0.5 1 1.5 2 2.5 3
AH1/2
o (m )
5/2
FIGURE 9.22
Flow rates through door and window vents for a centered pool fire.
It is plotted in Figure 9.22. Notice that for the same dimensional of height
and width, the window vent gives a higher flow rate because the smoke
layer is at or below the sill. The pressure difference is higher for the
window than the door. If the constant for a crib fire were used, the flow
would be less than that of a pool fire on the floor due to the differences in
entrainment.
Change the factor C in Equation 9.8 to 1 for a crib and a use doorway and
see how well the data of Figure 9.20 match.
These formulae give an estimate of the mass flow of air into a win-
dow or door vent. Multiple doors and windows, and ceiling or floor vent
will complicate the process. These can only be appropriately computed
by a more involved analysis. For wall vents and a fire on the floor, the
additivity of the flows through each vent is a reasonable assumption. For
more complex vent configurations, a simple formula computation can be
prohibited.
In particular, the case of single roof vent without any additional vent
below or a singular ground opening to an underground basement fire can-
not be predicted by the earlier formulas. The flow is more complicated and
not driven solely by pressure differences. In these “ceiling” vent cases, the
flow can be bidirectional or oscillatory. A single ceiling vent would have a
much less flow rate than a wall vent of the same size. It can be as small as
1/10th the same size wall vent.
These fire-induced flows provide the air supply to the fire. The flow rate
has an upper limit for wall vents by Equation 9.7, and a single ceiling vent
of the same size would be much smaller. While the flow of air and smoke
is governed by pressure differences, the fire supplies fuel gases by heat
transfer, and this process is independent of the air supply. When there is
not enough air, the fire becomes ventilation limited. A ceiling vent is repre-
sentative of the flow that would supply air to a basement or ship fire. Such
limited air supply can make these fires burn longer and more incompletely
and toxic. In such cases of low air supply, the burning rate within the fire
space is controlled by the air flow rate. Essentially, all the oxygen in the air
is used, and excess fuel escapes in the smoke. The volume of flames inside
is limited by air flow, and so then too is the heat transfer to produce the fuel
gases. Therefore, the burning rate within and the fuel gas production rate is
limited by the air flow.
Beyond the flame volume, the temperature of the smoke controls the heat
transfer within the compartment. Heat transfer is significant in driving the
fire growth rate. Smoke temperature will depend on air supply and how
much fuel can burn. Let us examine predictive formulae for the smoke
temperature.
9.5.4 Smoke Temperature

Figures 9.6 through 9.9 depict the smoke layer region in a room fire. This
region is extensive in the room especially compared to the volume associ-
ated with the flame and buoyant plume regions. The smoke layer is analo-
gous to the bath water filling at least half the bathtub with equal in and
out flows of water. So the level stays fixed, but with some wavy motion.
As with the bath water temperature, its recirculation from the inflow would
allow it to achieve a fairly uniform value. Similarly, we would expect the
temperature of the smoke layer, not counting the plume region, to become
fairly uniform. This is substantiated in the vertical direction of Figure 9.9,
and the nature of the lateral mixing demonstrates the tendency toward a
uniform smoke layer temperatures. Indeed early fire research into com-
partment fires showed this to be a reasonable representation of the smoke
layer. Of course, it can vary within the room, but away from the flames, it
can be reasonably uniform. Radiant heat tends to promote uniformity and
convective heat loss tends to drop the temperature as the smoke moves
from the fire to the exit.
Therefore, a first approximation to the temperature in a room fire is to
consider the smoke temperature. The smoke temperature will be affected by
the size of the fire, Q , and the ventilations factor, Ao H o . The larger the fire Q
and the smaller the vent (or more precisely the ventilation factor), the higher
we expect the smoke temperature. In addition, cooling through the enclos-
ing compartment construction area (A) will affect the smoke temperature.
It is not surprising an approximate formula for the smoke temperature will

involve these variables. The temperature rise above the surrounding incom-
ing air is given as ΔT = T – Ta:
1/3
æ Q 2 ö
DT (°C) = CT ç ÷ (9.9)
è ( hA)( Ao H o ) ø
where
CT is 6.85 for centered fires, about 12.4 for a corner fire, and about 16.5 for
room linings
Ao is in square meters
Ho is in meters
A is the interior surface area in square meters
h is a heat loss coefficient taken as the larger of
æ krc kö (kW)
h = Maximum of çç or ÷÷ in 2
è t l ø ( m - °C )
where
kρc is the interior construction thermal inertia (kW/(m 2 -°C))2 -s
k is thermal conductivity
l is the interior construction thickness in meters
t is time in seconds
Table 9.1 gives typical values of k and kρc for common construction
materials.
As seen the temperature rise depends on fire size, ventilation factor, and
heat transfer to the compartment construction. As with plume temperature,
this equation also could have been put in dimensionless terms using a Q*
based on the ventilation factor. It is sometimes given in those terms, but here
TABLE 9.1
Typical Construction Thermal Properties
Thermal Inertia, kρc Thermal Conductivity,
((kW/m 2-K)2-s) k (W/m-K)
Insulation board 0.09 0.041
Wood 0.30 0.15
Gypsum board 0.60 0.5
Concrete 2.0 1
Steel 150 50
it is given in specific dimensions to facilitate its use. The equation gives the
average temperature of the smoke exiting the compartment. It was gener-
ally based on the exit maximum temperature in developing its correlations.
Higher point-wise temperatures will occur in the room. The equation was
developed for room-centered fires by analyzing many different tests. Others
have investigated other fire configurations, such as a fire in a corner, or with
wall and ceiling linings burning.8 The differences among the configurations
are due to differences in the rate of entrainment into the fire configurations.
Also the equation can be used for multiple wall vents by summing the Ao H o
values and can be used for different construction materials by summing the
hA values for the various wall, ceiling, and floor elements.
Example 9.1:
Consider a room 4 × 4 × 3 m high with a simple vent 1 × 2 m high.
The construction is essentially gypsum plaster with properties given in
Table 9.1. The fire is constant at 500 kW. Assume the gypsum is very
thick. Compute the temperature rise of the smoke at 100 s.
Ao = 2 m 2
Ao H o = 2 2 m 5/2
A = 2( 4 ´ 4) + 4( 4 ´ 3) - 2 = 78 m 2
krc 0.60
h= = = 0.0775 kW/m 2 -K
t 100
Therefore, from Equation 9.9,
1/3
é 500 2 ù
DT = 6.85 ê ú = 167°C
ë (0.0775)(78)(2 2 ) û
There is a restriction on the use of Equation 9.9 in that there must be suf-
ficient air for its use. When there is insufficient air the firepower must be
computed by burning all the oxygen in the incoming air flow rate. We will
come back to this shortly.
9.5.5 Flashover
In Section 9.2.2, flashover was defined as a sudden increase in the firepower,
and it has a beginning and end. Many have used temperature in the com-
partment as an indicator of the onset of flashover. Commonly a temperature
rise of 500°C is a selected criterion. Radiation from the smoke layer is the
principal driver of fire growth in a room. The blackbody radiant heat flux
arising from this temperature (500°C + 25°C air temperature) is 23 kW/m 2
(from σT4). At this heating level, we can expect some items to ignite, flame
spread to increase, and burning rate to be enhanced. The onset of this
increase in fire growth (trifecta) could be perceived as the onset of flashover.
Thus, if we use a ΔT = 500°C to indicate flashover potential, Equation 9.9
would give us the conditions required for flashover to start. Substituting
and solving for Q
3/2
é (500) ù 1/2
Q FO = ê é( hA)( Ao H o )ù for a room-centereed fire. (9.10)
ë 6.85 úû ë û
The coefficient is 624 with units as specified in Equation 9.9.
Example 9.2:
For the room of Example 9.1, presuming flashover at near 100 seconds,
we can estimate the firepower needed to cause flashover:
1/2
QFO = 624 éë(0.0775)(78)(2 2 )ùû = 2580 kW.
If the actual fire is constant at 500 kW, it is insufficient to trigger flash-

over, assuming other combustible items are available for fire growth.
Here, we have arbitrarily used a benchmark of temperature rise of 500°C to

indicate the onset of flashover. Flashover is caused by the sudden temperature
rise and its associated heat flux influence on fire growth. This fire growth
can occur due to an increase in burning rate, remote ignitions, and/or by
increasing flame spread. These different growth processes have different
trigger points. It is not uncommon in upward flame spread to launch rapid
spread with a smoke temperature rise of only 300°C. These differences in
temperature at flashover are due to the nature of the growth processes that
cause the instability of flashover. So a temperature criterion for flashover
is only an approximation. Shortly, we will investigate the anatomy of fire
growth through flashover, and some of the issues in this discussion may be
more appreciated.
9.5.6 Ventilation-Limited Fires

The criterion for a ventilation-limited fire is that in the exiting smoke all the
oxygen has been consumed within the room and its concentration is zero.
A measure of this is the equivalence ratio, Ф. When Ф is 1, under complete
combustion all of the fuel and oxygen in the air have been ideally consumed.
As we do not have ideal complete combustion in fire, the equivalence ratio
criterion is an approximate indicator. But as Ф increases past 1, the oxygen
concentration in fires drops to near zero.
Some quantitative estimates can be made to assess when a room fire

becomes ventilation limited. We know how to compute the maximum air
flow rate, which is valid for small vents or large fires. From Example 9.1, we
can compute this upper limit from Equation 9.7,
a,max = 0.52(2 2 ) = 1.47 kg/s.

m
Almost all common fuels produce approximately 3 kJ/g of air (or 13 kJ/g
of oxygen consumed in the air). If all the oxygen is consumed we have a
ventilation-limited fire and the maximum possible energy release rate within
the room is then
Q max = m
a ,max (kg/s) ´ 3000 kJ/kg. (9.11)
Following on the previous example, this is
Q max = 1.47 ´ 3000 = 4, 412 kW or 4.4 MW.
We see that this energy release is greater than the value to initiate flashover.
This difference between flashover and ventilation-limited energy is true in
general for compartment fires. Typically flashover occurs and then the fire
has the potential to become ventilation limited. However, for a small vent,
the fire may never grow to a flashover state. A ventilation-limited fire could
be sustained, or the fire could self-extinguish. This form of fire can lead to a
backdraft if the ventilation is abruptly increased.
Example 9.3
It is useful to consider the combination of Equations 9.9 through 9.11.
Let us take a special case for illustration. Consider a room 7 × 7 m
and 2.4 m in height. Let it be lined with gypsum board, and select, for
illustration, a time of 6 minutes in the fire. Then, hA = (0.60/360)1/2 ×
165.2 = 6.74 kW/K. For this value, Equation 9.9 can give the firepower
for a given ventilation factor. The results are plotted in Figure 9.23.
This graph shows lines of constant temperature rise, with one as the
flashover benchmark. The dashed line gives the maximum possible
firepower due to the fire becoming ventilation limited as computed
from Equation 9.11. A typical room door might have a ventilation fac-
tor of about 2.5 m5/2. This would indicate a flashover condition of about
2500 kW as computed from Equation 9.10. With enough fuel in the room,
the fire would jump to a ventilation-limited state of just under 4000 kW.
The temperature rise at that point could be computed to be about 670°C.
This graph shows the firepower in the room, but the fuel generation rate
after the fire becomes ventilation limited can produce excess fuel that
burns outside the room.
8000
7000 For a centered fire

∆T = 800°C
h = 0.041 kW/m2-K
6000
A = 165 m2
5000
Firepower (kW)
4000 ∆T = 500°C
3000
2000
∆T = 200°C
1000
0
0 1 2 3 4 5 6
AoHo1/2 (m5/2)
FIGURE 9.23
Firepower for fuel-controlled and ventilation-limited conditions, an example.
9.5.7 Fully Developed Fire Size

Equation 9.11 gives the energy release rate within the room if the fire is ven-
tilation limited. The maximum fuel burning rate depends on what fuel is in
the room. For typical rooms in a house, fully developed fires are 800°C or
higher. That temperature corresponds to a radiant heat flux (σT4, T = 1073 K)
of about 75 kW/m2. From Chapter 6, we saw that the heat of gasification may
range from about 1 to 10 kJ/g for different fuels. The imposed heat flux and
heat of gasification can allow us to compute the fuel production rate from the
solid or liquid fuels.
Example 9.4
Consider a continuation of Example 9.1 (4 × 4 × 3 m high, and 1 × 2 m
high wall vent). But now the fuel is allowed to burn according to the fire
conditions. Assume an average heat of gasification (L) of 3 kJ/g for the
room contents. Consider a heat flux of 75 kW/m2 caused by the smoke
in the fire’s fully developed state. Consider this fuel distributed over the
floor area A = 16 m2. Then from Equation 6.1 the generated gaseous fuel
rate is computed as
75 kW/m 2
=
m ´ 16 m 2 = 400 g/s.
3 kJ/g
If the average heat of combustion, ΔHc, is taken as 20 kJ/g, then the cor-
responding fire size (Equation 6.3) is
Q = (400 g/s) (20 kJ/g) = 8000 kW or 8.0 MW.
From the ventilation factor of 2.83 m5/2, Equation 9.11 gave a ventilation-
limited fire of about 4400 kW within the room. The remaining energy
release rate, 8.0 − 4.4 = 3.6 MW, must potentially burn outside of the room.
This dramatic increase in the firepower shows the consequences of
flashover leading to a ventilation-limited fully developed fire. If instead
of just considering fuel distributed over the floor we had said the enclo-
sure was completely lined with fuel, for example, combustible interior
finish, then the entire enclosure surface area (minus the vent opening,
2 m2) gives A = 78 m2. Subsequently, we compute the corresponding fuel
generation rate m = 1950 g/s and the fire size Q = 39 MW.
The hazard of combustible interior finish, even thin ones, is clearly demon-
strated. Figure 9.24 shows the possible fire growth behavior of this room fire
example. The fully developed fire state depends on the nature and extent of
the fuel available to burn. Flashover begins at about 2.6 MW, it ends with a
ventilation-limited fire at 4.4 MW, and the fire can produce fuel that continues
to burn outside the room with a total output of 8–39 MW. The latter is for lining
materials that fully expose their surface area to the heat of the fire in the room.
39 MW linings
Burning outside
compartment
8.0 MW floor-based
Firepower, MW
4.4 MW fully developed

ventilation, limited
2.6 MW flashover
Time
FIGURE 9.24
Fire growth of example fire scenario.
These differences in total fire output show the effect of heat transfer; however,
the decrease in oxygen feeding the fire will mitigate these estimates and reduce
them. Here, we have selected a heat flux of 75 kW/m2 indicative of the smoke
layer heating. The flame also causes additional heating, and it is affected by
oxygen. Vitiation reduces the flame temperatures and as a result the burning
rate. The decrease in oxygen feeding the fire is due to the vent mixing into the
lower layer as shown in Figure 9.11. Let us see how to include this effect.
9.5.7.1 Model for Fuel Generation in a Compartment

It is very important to understand the production of fuel gas generated in
a fully developed compartment fire. The rate of its generation controls the
firepower to the surroundings, and it controls the duration of the fire. The
latter is critical to the fire’s effect on the structure. A model that illustrates
the competing effects of oxygen and heat transfer on the fuel generation was
introduced in the work of Utiskul.9 It simply says that the mass loss rate
by direct flame heating depends on the oxygen concentration supplied to
the flame, and the fuel-exposed surfaces generate mass due to radiant heat
transfer from the compartment smoke and ceiling. Some of this heat transfer
cannot fully strike the surface, as it can be absorbed by a thick flame or the
fuel surfaces are not exposed or hidden from radiation. Representative fuel
categories can be a pool fire on the floor whose surface is always exposed
and a crib fire that has hidden surfaces. The pool fire can represent lining
materials or flat surfaces, and the crib can represent structural furniture. An
equation for the model is quantitatively given as
Yox , l q External
m ¢¢F , o AF , b
F =m + (9.12)
Yox , o L
where
Yox,l is the oxygen mass concentration in the lower layer supplied to the
flame
Yox,o is the oxygen mass concentration normal air at 0.233
AF,b is the surface are covered by flame
m ¢¢F , o is the burning flux in normal air
q External is the radiant heat transfer rate from the compartment
The burning flux in normal air can be determined by measurement or esti-

mated from the maximum burning fluxes in large fires. The amount of radi-
ant heat flux that can get through the flame depends on the size of the flame.
In small-scale laboratory experiments, the flame is small, and the smoke
radiant heating effect is large. But for large realistic fires, the smoke radiant
heat has difficulty in penetrating the flame. So the heat feedback effect from
the smoke is less for big fires.
To exercise this formula many things have to be known: The radiant heat,
how much gets through the flame, the oxygen feeding the fire, and the type
of fuel. A complete model of the compartment fire must be used to integrate
all of these effects. This was done in the work of Utiskul,9 and we will shortly
show some of those results to illustrate quantitative effects.
A significant aspect of ventilation-limited fires is that flames do not cover all of
the fuel heating. The area of fuel covered by the flame (AF,b) is less than the total
exposed area of the fuel. It can be estimated by knowing the stoichiometric air-
to-fuel ratio (s = ΔH/(3 kJ/g-air)) and the supply rate of air to the compartment:
air
m æ Y q ¢¢ ö
AF , b = çm ¢¢F , o ox , l + Ext , b ÷ (9.13)
s è Yox , o L ø
with q¢¢Ext , b the compartment radiation heat flux to the flaming surface.
Essentially this equation is derived by equating flaming firepower over the
area, AF,b, (from Equation 9.12) with the maximum possible firepower due to
air supply (Equation 9.11).
9.5.7.2 Fully Developed Compartment Fire Behavior

A collection of data assembled by Utiskul9 taken from Harmathy (Ref. 17
of Utiskul9) for crib fires and from Bullen and Thomas (Ref. 20 of Utiskul9)
for liquid pool fires in compartment experiments is shown in Figure 9.25
along with his data for the same. His data were developed in a rectangular
compartment 0.4 m high (Figure 9.26) with an array of wood cribs or pans of
heptane distributed over the floor. The data in Figure 9.25 scatter due to the
scale of the compartment and the heat transfer losses. But they show a trend
that is very clear. The fuel gasification rate per unit area has a power relation-
ship (about 0.6) with the ventilation factor divided by the total fuel exposed
area, AF. Also notice that the pool fire data (x’s) have higher mass flux levels
than the crib (+’s) data. This is to be expected because of the greater effect of
radiant feedback for the pool fires. The horizontal axis represents a param-
eter related to the equivalence ratio. Recall the equivalence ratio is defined as
F º (m fuel s/m
air ) µ ( AF /Ao H o ) and is therefore related to the reciprocal of the
axis parameter. So as the horizontal axis parameter decreases (Φ increases)
and the fire state becomes more ventilation limited. At high values along the
axis, the fire is fuel controlled. In this region, there is a tendency for the crib
and pool mass fluxes to level off. These tendencies are more clearly indicated
in just the data of Utiskul and his computations shown in Figure 9.27.
Figure 9.27 shows computations for a specific ratio of compartment surface
area (A) to fuel exposed fuel surface (AF), along with representative data for pool
and crib fires. The effect of scale has been computed as well for a compartment
four times bigger (1.6 m high) than the experiments (0.4 m). The accuracy of
the computations can only be judged by comparison to the small-scale data in
1000
Pool, crib
Steady well ventilated
, Steady underventilated
Unsteady underventilated
, From [17,20]
100
Fuel mass loss rate (g/s-m2)
10
0.1
10 100 1,000 10,000
AoHo0.5ρog0.5/AF (g/s-m2)
FIGURE 9.25
Fuel mass flux as a function of ventilation to fuel area ratio for pool and crib fires. (From Utiskul,
Y., Theoretical and experimental study on fully-developed compartment fires, PhD thesis,
Department of Fire Protection Engineering, University of Maryland, College Park, MD, 2008.)
FIGURE 9.26
Compartment fire experiments. (From Utiskul, Y., Theoretical and experimental study on
fully-developed compartment fires, PhD thesis, Department of Fire Protection Engineering,
University of Maryland, College Park, MD, 2008.)
100
Crib, pool Large scale (×4)
, Small-scale experiments 2|1
As/AF = 46
, Prediction Free burn
large pool
2|1
Small scale
As/AF = 46
Fuel mass loss flux (g/s-m2)
Free burn
small pool
3|2
10 Ext | 3 Small scale Free burn

2|1 As/AF = 12 small crib
3|2
3 | 22 | 1
Ext | 3
3|2
Large scale (×4)
Free burn
As/AF = 12
large crib
Ext | 3
Ext | 3
1
10 100 1,000 10,000 100,000
AoHo0.5ρ o g0.5/AF (g/s-m2)
FIGURE 9.27
Effect of fuel type and scale in fully developed fires. (From Utiskul, Y., Theoretical and experi-
mental study on fully-developed compartment fires, PhD thesis, Department of Fire Protection
Engineering, University of Maryland, College Park, MD, 2008.)
Figure 9.27 as well as those in Equation 9.25. The theoretical result has not been
fully validated but its trends have credibility. So it can serve as a good illustra-
tion of variable behavior and scale effects. As ro g Ao H o /AF increases the
fully developed fire state moves from a ventilation-limited to a fuel-controlled
state. As the fuel exposed area becomes very small relative to the enclosure
size, the fire approaches a free burning fire in the open. Here, there is a reverse
in the scale effect between cribs and pools. As the scale increases for a crib,
its thicker sticks burn at a slower rate per unit area compared to small scale
(m ¢¢ µ thickness-1/2.) In contrast, the larger-scale pool fire can achieve a higher
free burn mass flux than its smaller-scale turbulent counterpart.
As ro g Ao H o /AF decreases from its highest value, the fire conditions
moves through several stages in the following order:
1. Steady well-ventilated, fuel-controlled fire

2. Steady underventilated fire
3. Unsteady oscillating, underventilated fire
Ext. Extinction of flames.

The theoretical curves are labeled with these regime boundaries. The bound-
aries depend on scale and are different for the wood cribs and heptane pool
fires, as each requires different stoichiometric values: heptane, 13.7 g air/g
fuel; and for wood about 4 g air/g fuel. Extinction is based on the flame tem-
perature falling below its critical value needed for burning. The oscillating
region occurs for relatively small vents. Oscillation occurs due to vitiated
oxygen that drives the flame to extinction, followed by a subsequent cooling
effect that lowers the compartment pressure causing a new inflow of fresh
air. The new air revitalizes the fire, and the process repeats—sometimes con-
tinuing until all of the fuel is depleted. A photograph showing this sequence
is displayed for a crib fire in Figure 9.28. A more dramatic effect is depicted in
Figure 9.29 for pools distributed over the floor of the compartment in under-
ventilated burning. The flames begin by covering all of the fuel as there is
initially enough oxygen. As the oxygen is reduced, the flaming is reduced
and flames move toward the vent. The flames seek the air. As the flaming
fuel is depleted, the flames would then move back to the remaining fuel.
This condition is commonly responsible for deep damage, in extinguished
Crib2 before ignition

Air inflow
(a)
00:02:47.99 00:02:49.45 00:02:51.38

(b) (c) (d)
00:02:55.72 00:02:56.29 00:02:58.91

(e) (f ) (g)
FIGURE 9.28
Oscillatory burning for a crib fire in a compartment: (a) configuration; (b, d, f) growing; (c, e, g)
dying. (From Utiskul, Y., Theoretical and experimental study on fully-developed compartment
fires, PhD thesis, Department of Fire Protection Engineering, University of Maryland, College
Park, MD, 2008.)
00:00:02.55
(a) (b)
00:00:20.18 00:00:22.06
(c) (d)
00:00:29.05
(e) (f )
FIGURE 9.29
Early fire growth in underventilated burning of heptane pools showing reduced flaming as initial
oxygen is reduced and flames move to the vent. (a) Heptane pan layout. (b) Full area burning.
(c) Burning at ~60% AF. (d) Burning at ~40% AF. (e) Burning at ~20% AF. (f) Ventilation limited burning.
(From Utiskul, Y., Theoretical and experimental study on fully-developed compartment fires, PhD
thesis, Department of Fire Protection Engineering, University of Maryland, College Park, MD, 2008.)
house fires, occurring near the vents as well as at the origin of the fire. Past
investigative principles suggested this to be two fires and arson as the cause!
9.5.7.3 Fuel Load and Burning Duration

An important aspect of a compartment fire is not only its temperature but
its duration. The burning duration can be calculated by dividing the mass
of fuel available by the mass supply rate of gaseous fuel. For a structure to
TABLE 9.2
Fuel Load in Mass Per Unit of Floor Area
Fuel Load (Per Floor Area) Standard
Occupancy Mean Value (kg/m 2) Deviation (kg/m 2)
School classroom 40.5 16.9
Apartment house bed room 44.6 10.7
Apartment house living room 34.7 23.2
Apartment house storage room 55.5 32.6
Office room 39.2 34.7
Office library 146 68
Office meeting room 6.4 1.4
Hotel guest room 10.6 1.6
Hotel banquet room 3.3 1.2
Athletic area 2.5 2.3
Source: Natori, A., Fire Sci. Technol., 27(3), 310, 2008.
be safe, the burning time should be less than the rating time for a struc-
tural element in simplest terms. In other words, the fire should burn out
before the structure is damaged. The available fuel burnable mass has been
associated with the occupancy of a building, and fuel load surveys have
been carried out to characterize occupancies. Each particular occupancy
has variations, and we will express these in terms of a mean value and a
standard deviation. Natori10 presents such information based on fuel load
surveys conducted between 1973 and 1996. These come from Japanese lit-
erature but are indicative of global characteristics. The results are given in
Table 9.2. It is seen that in some cases, the standard deviation is significant
compared to the mean, showing a wide possible range for an occupancy.
These data are crucial in the design of safe structures for fire. Formulas
from fire dynamics can give the temperatures and the burning rate. The
occupancy fuel load then will allow computation of the duration of the fully
developed fire. We have seen that the fully developed fire is usually venti-
lation limited and the fuel mass loss rate is determined by the ventilation
factor. So fuel quantity and ventilation openings are key in determining the
duration of a compartment fire.
9.6 Anatomy of Fire Growth

Observations of fire are crucial to learning. Today inexpensive video cam-
eras that might be sacrificed in the fire are valuable aids. In 2009, the ATF
Fire Laboratory and the International Association of Arson Investigators
conducted fire tests to examine damage after flashover. This was produced
into a training module for CFI-Trainer available at
https://www.cfitrainer.net/Training_Programs/Postflashover_Fires.aspx.
An examination of these data and video records vividly show the behavior
of fire as it moves through its stages. It clearly shows the onset of flashover
and the start of the ventilation-limited fire. It shows the movement of flames
from the origin to eventually the doorway. A review of the data and stills of
the videos will try to illustrate these effects.
A schematic of the bedroom where the fire occurs is shown in Figure 9.30.
The significant instrumentation for this discussion is described as follows:
• Thermocouple tree TC-B: running from the floor (B-0) to the ceiling
(B-8), in 1 ft increments
• Heat flux meters: THF-A (total), RHR-A (radiation only), THF-B
(total), THF-C (total), all at 1.22 m from floor
• Oxygen: GS-A (at 0.19 m above floor), GS-B (as 2.22 m above floor),
GS-C (at 0.15 m above floor), GS-D (at 1.16 m above floor)
The fire is started in a plastic waste basket, filled with paper, that is between
the bed and the chair. The clock recording data is begun before ignition
occurs at 0:19 (minutes:seconds). By 1:00 the sides of the bed and chair are
ignited (Figure 9.31a).
At about 2:30, the temperatures at the room center near the ceiling begin to rise
sharply (see Figure 9.32). This was preceded by the onset of accelerating flame
T T
THF-B THF-C
THF-A GS-D
R
T&R
RHF-A GS-B (high)
GS-A (low)
TC-A
CO-A
TC-B Smoke-A
TC-D Vel-Top
GS-C TC-C Vel-Middle x
Vel-Bottom
y
FIGURE 9.30
Bedroom schematic showing furnishings and instrumentation. (From ATF Fire Laboratory.)
spread over the top of the bed and the back of the chair (Figure 9.31b through
d). At 2:15, a blackened top-left corner of the chair-back indicates the start of
the spread. This flame spread began at about 2:15, accelerated, and ended cov-
ering most of the chair at about 3:02. In Figure 9.32, the beginning and end of
the rapid rise in gas temperatures nearly coincide with the onset of this rapid
spread and its end, occurring between about 2:15 to 2:30 and 3:00, respectively.
This marks the beginning and end of flashover, a 30-second duration.
Notice that before the onset flashover, the heat flux meter on the wall adja-
cent to the fire origin (THF-A) achieves a heat flux of 200 kW/m2 (Figure 9.33).
It sees a thick impinging flame igniting basket flame merging with flames
on the side of the chair, as shown in Figure 9.31b. This heat flux is due to con-
vection and radiation from the flame. In contrast at the same time, the heat
flux away from the origin, halfway up on the rear wall (THF-B) has a much
lower value. This sensor “sees” the upper smoke layer with about a 50% view
angle. This is nearly the same view angle to the upper smoke layer as a heat
flux meter on the floor away from the origin would see. Hence, THF-B on
the wall is approximately indicative of a floor heat flux away from the ori-
gin. At the onset of flashover (2:15–2:30), it primarily reads radiation at about
20 kW/m2. At the end of flashover, it attains over 100 kW/m2 (~3:00). These are
(a) (b)
(c) (d)
FIGURE 9.31
Sequence of events in bedroom fire development. (a) ~ 1:00 Ignition of the sides of the chair and
bed. (b) 2:15 Onset of rapid flame spread on the chair back. (c) 3:00 Flame has spread completely
over top of bed. (d) 3:02 Flame has spread completely over chair back. (Continued )
(f )
(e)
(g) (h)

Sequence of events in bedroom fire development. (e) 2:58 Black layer of smoke drops to near
the floor in the room. (f) 3:44 Flames move away from the origin region. (g) 4:09 Flames begin
to move toward the door. (h) 4:30 Flames emerge from the door.
typical values just before and just after flashover. That is why 20 kW/m2 is a
common benchmark for computing flashover.
It is also instructive to examine the oxygen concentration in the smoke
layer (GS-B, Figure 9.34) at the beginning and end of flashover. At about 2:30,
it is about 0.12 (12%), then begins to drop precipitously to zero (ventilation
limited) at 3:00. At this time, the oxygen levels (Figure 9.34) near the floor
(GS-A and C) are still near pure air levels of 21%.
Also during flashover, the smoke layer interface in the room drops from
about 1 m to about 40 cm and then continues to drop as the compartment
gets hotter. By 2:58 (Figure 9.31e), it is now at 20 cm and remains there. Notice
that the neutral plane in the doorway is higher at about 1 m.
1000 TC-B8 1832
End FO
TC-B7
TC-B6
TC-B5
800 TC-B4 1472
Onset of flashover
TC-B3
Temperature (°C)
Temperature (F)
TC-B2
TC-B1
600 TC-B0 1112
Flames move from center

400 752
Suppression
200 392
0 32
0 1 2 3 4 5 6 7 8 9
Time (min)
FIGURE 9.32
Temperature vertical array at the center of the room. (From ATF Fire Laboratory.)
250
All located on walls THF-A THF-C
at 1.22 m above floor (origin) (rear)
200
THF-B
Heat flux (kW/m2)
(nightstand)
150
Flames moves to door
FO
100
Flames moves from origin
50
Suppression
RHF-A
0
0 1 2 3 4 5 6 7 8
Time (min)
FIGURE 9.33
Heat flux sensors: THF-A (total), RHR-A (radiation only), THF-B (total), THF-C (total) all at
1.22 m from floor. (From ATF Fire Laboratory.)
0.25
B 2.22 m
0.2 (rear high)
A 0.19 m
(rear low
C 0.15 m
Oxygen (mole fraction)
(front)
0.15
)
D 1.16 m
0.1 (rear)
End FO
Start VL
0.05
Suppression
0
0 1 2 3 4 5 6 7
Time (min)
FIGURE 9.34
Oxygen concentrations: GS-A (at 0.19 m above floor), GS-B (as 2.22 m above floor), GS-C
(at 0.15 m above floor), GS-D (at 1.16 m above floor). (From ATF Fire Laboratory.)
Clearly these illustrations and data show that flashover is a process of rapid
fire growth brought on by increases in heat flux from the smoke layer over
the bed and chair back in this case. This sends the smoke layer down toward
the floor, temperatures in the smoke layer sharply rise, and its oxygen drops.
For this fire, the end of flashover is marked by a ventilation-limited state of
zero oxygen in the smoke. If less fuel were in the room, it could have ended in
a fuel-controlled fire with all of the combustible material covered by flames.
During the ventilation-limited fire, the flames do not cover all of the fuel.
Indeed, Figure 9.31f at about 3:44 the flames move away from the origin
toward the right rear wall (as facing from the doorway). A confirmation of
this flame movement is indicated by the drop in the heat flux near the origin
(THF-B) as the flames move away, and an increase in the wall heat flux closer
to the right rear (THF-C) as flames are felt (Figure 9.33). Also the oxygen level
at the origin near the floor (GS-A) starts to drop at about 3:44 and by 4:00 is
about 2%. This indicates that the inflow of air is hardly reaching this left cor-
ner. The flow of air is now directed to the right side of the back wall where
there is flame. (See the increase of oxygen at GS-D in Figure 9.34.) There is
not much to burn in this region, mostly carpeting. Consequently, by 4:35 the
flames move again from the right rear area to the door, and flames emerge
from the door (Figures9.31g and h). At this time (4:35), the flames engage the
thermocouple array TC-B, and as the carpeting is now burning there, their
temperatures over the entire vertical array become nearly uniform.
Some have called the event of the emerging door flame, flashover, as there
is an obvious presentation of flames to suggest all the room is now burning.
This is false according to the concept that flashover is an instability that leads
the fire to a ventilation-limited or fully developed fuel-controlled state. Only a
portion of the room has flames; now, primarily near and in the doorway. This
is corroborated by no flames shown from the view looking at the right rear
floor (Figure 9.31g), only smoke. Also, the center temperatures (TC-B) become
uniform and all start to drop shortly after 4:35 and continue until after 5:00.
Unusual oxygen sensor behavior indicates maxima after the flames move
away from the origin. As flames move to the right rear, the incoming flow is
entrained there and the oxygen (GS-D) at the back wall increases to about
13%, then drop again as the flames move forward to the doorway. This effect
of turbulent flames impinging on the right rear gypsum wall could result in
a clean burn. Such a pattern is caused by the soot being burned off the wall
and remaining in that state. The subsequent oxygen drop in GS-D indicates
that the flames leave this area.
The most unusual behavior is the rise in oxygen for GS-A at the origin
near the floor. After 4:00, it increases to about 18% at 5:00, while in this same
location, the oxygen is still 0% in the smoke layer (GS-B). This unusual effect
might be explained by a sharp increase in airflow rate from much higher
temperatures near the doorway due to the burning carpet. The increase in
airflow could cause a high-velocity floor jet of fresh air with enough momen-
tum to reach the rear, especially by the path under the bed. If oxygen reaches
this origin area and there is still remaining fuel, flames can reappear there. It
appears that they did reappear as exhibited by the heat flux sensors (THF-B
and C) increasing after about 5 minutes. As seen, complex behavior can
occur after flashover and a ventilation-limited state is achieved.
The ventilation-limited state clearly needs more study. It increases the
production of incomplete and toxic combustion products. The duration of
the fire, significant to the safety of structures in fire, depends on the air sup-
ply. The flames within the compartment move to essentially seek oxygen.
The deposition of soot on plaster walls depends on the presence of oxygen-
rich turbulent flames. Eventually the flames emerge from vents with a flour-
ish. This dramatic event is not “flashover.” Hopefully, the portrayal of this
bedroom fire has enlightened an understanding of the processes in room fire
development through flashover to its fully developed state.
9.7 Summary
Fire grows in a room in stages: first, a nearly free-burning stage without any
influence from the room; then flashover can occur; followed by the fully
developed state, which is likely to be ventilation limited for typical furnished
rooms. After extinction of flames, smoldering can persist and this is a cooling
stage. Flashover is an instability between heat released by combustion and
energy losses. It manifests its occurrence by rapid fire growth due to increas-
ing heat flux from the upper smoke layer. A sudden increase in room tem-
peratures marks the onset of flashover. A leveling off of these temperatures
marks the end of flashover and the start of the fully developed fire. Oxygen
depletion in the lower layer causes the fire to diminish and, in cases of poor
ventilation, possibly extinguish. A level of zero oxygen concentration in the
exiting smoke defines the ventilation-controlled fire. Combustion in this state
is controlled by the airflow rate. In this state of insufficient air, the flames in
the room will fill only a portion of the room and will move to the vents where
air is entering. As flames exit the room, some might call this “flashover” but it
is not. If oxygen does not drop to zero in the smoke, the fully developed fire is
fuel controlled and all the fuel in the room can support flames.
Small pressures drive the flows associated with fires in rooms. In multistory
buildings, normal air-conditioning for heating and cooling, as well as fire
effects, leads to a stack effect that controls buoyancy for the entire building.
Without a dominant stack effect, the fire-induced flows can exceed pressures
due to air conditioning ducts and disrupt those duct flows. The flow processes
in a room fire begin with plume rise to the ceiling, a formation of a ceiling jet
about 1/10th of the room height, then a stratified layer forms and descends.
In a closed compartment, it can nearly reach the floor. Otherwise, an open
wall vent will lead to a bidirectional flow with air in and smoke out. The flow
system can approximately be considered as an upper smoke layer and a cooler
lower layer with fairly uniform temperatures and gas concentrations in each.
Formulae exist for estimating smoke filling, smoke layer temperature, and
vent flow rates. These show how to estimate the firepower needed for flash-
over and ventilation-limited conditions.
Review Questions
1. For your classroom estimate the energy release rate needed to initiate
flashover. Assume the door is open. What kind of fuel package might this
represent?
2. What fire size (kW) will cause a 300°C smoke layer temperature in your
room with the door half open? What is the maximum radiant heat flux
associated with smoke of 300°C?
3. If your room reaches flashover and becomes fully developed reaching
800°C, what is the maximum air flow rate through your fully opened
door? If all the windows break, what is the additional air flow rate? What
fire size can be supported by these total air flow rates? Is there sufficient
fuel in the room to do it?
4. A tall building on a winter day has an indoor temperature of 23°C and an

outdoor temperature of 0°C. Its height is estimated as 350 m. What pres-
sure difference would result over a shaft of this height? If the shaft has
an effective diameter of 3 m, what is the average velocity of air moving
in the shaft? Assume it is fully open at the base and top of the building.
(See Equations 9.3 and 9.4.) Let f = 0.02.
5. Assume a smoldering chair fire occurs in your classroom, and the doors
and windows are shut. Using the information in Figure 9.19, estimate the
time the smoldering fire smoke layer reaches the fire location at 1.2 m
above the room floor (to). Also, estimate the time to achieve an incapacita-
tion from CO at that location (t*).
True or False
1. Fires in a room with a door or window open, producing smoke of less

than 200°C, can be considered a developing, almost free-burning fire.
2. Smoldering fires typically in a room produce lethal conditions in less
than 10 minutes.
3. Backdraft is a form of flashover.
4. In the summer, fire in a high-rise air-conditioned building can have
smoke propagate to the floors below the fire.
5. A fire in a vessel eventually has to go out due to oxygen depletion or it
may burst the vessel.
6. Pressures associated with air handling ducts can dominate the flows of a
big building fire.
7. Flashover occurs when flames discharge from a room.
8. Flames do not fill a room in a ventilation-limited fire.
9. Smoke forms a ceiling jet and the inlet air flow can form a floor jet.
10. Oxygen can never get to zero concentration in a room fire.
Activities
In a laboratory with safe operating and extinguishing systems, conduct
some room fire experiments. Build a small compartment between 1 and 2 ft
in height. Allow for various window and door openings. Use small pool or
crib fires. Measure the temperature at the exit. Conduct various experiments
to reveal the stages of fire. Take care in ignition, and never leave liquid fuel
sitting in the compartment for a long time before ignition.
References
1. G. Cox, Private communications, BRE, Watford, UK, 2010.
2. G. Hartzell, Combustion products and their effects on life safety, Sect. 3,
Chap. 1 in Fire Protection Handbook, 17th edn., edited by A. E. Cote and J. L.
Linville (Quincy, MA: National Fire Protection Association, 1991), pp. 3.3–3.14.
3. J. G. Quintiere, K. Steckler, and D. Corley, An assessment of fire induced flows
in compartments, Fire Science and Technology, 4, 1 (1984): 1–14.
4. E. E. Zukoski, T. Kubota, and B. M. Cetegen, Entrainment in fire plumes, Fire
Safety Journal, 3 (1980): 107–121.
5. J. G. Quintiere, M. Birky, F. Macdonald, and G. Smith, An analysis of smolder-
ing fires in closed compartments and their hazard due to carbon monoxide,
Fire and Materials, 6, 3 and 4 (1982): 99–110.
6. J. G. Quintiere and B. J. McCaffrey, The Burning of Wood and Plastic Cribs in
an Enclosure, Vol. I, NBSIR 80-2054 (Gaithersburg, MD: National Bureau of
Standards, September 1980), p. 49.
7. J. G. Quintiere and L. Wang, A general formula for the prediction of vent flows,
Fire Safety Journal, 44, 5 (2009): 789–792.
8. J. G. Quintiere, Fundamentals of Fire Phenomena (John Wiley & Sons Ltd.,
Chichester, UK, May 2006), p. 359.
9. Y. Utiskul, Theoretical and experimental study on fully-developed compart-
ment fires, PhD thesis, Department of Fire Protection Engineering, University
of Maryland, College Park, MD, 2008.
10. A. Natori, Research regarding analysis of the target safety standard for fire
resistance design and its mode of expression—Estimation for the safety stan-
dard targeted by the verification method for fire resistance performance,
Fire Science and Technology, 27, 3 (2008): 310–489.
10
Design, Investigation, and Case Studies
Learning Objectives
• Appreciate analytical applications in fire safety design and in fire

investigation
• Understand performance-based design
• Understand the role of science in investigation
• Explain fire modeling in terms of equations and computer solutions
10.1 Introduction
Having completed the previous chapters, it should be clear that methods
exist for estimating the size of a fire and its growth, Q (t), and for estimat-
ing its damage effects from heat and smoke. Also, it should be clear that
this process is complex, and we have only considered simplified methods
for estimating aspects of fire growth. Nevertheless, it is possible to step
through a fire scenario using our equations with good results and reason-
able accuracy. Most fire phenomena have uncertainties due to circumstances
not completely known (e.g., the extent of a door opening, the proximity of
one item to another). Even if we know all of these factors, our computations
are not perfect, varying in accuracy by as much as ±50%, but usually much
better. Yet these approximate analyses can still be very useful, even with
these limitations. They serve to provide guidance in design and direction
in fire investigation. Both of these applications do not require high preci-
sion. If time is the important factor being considered in a fire phenomenon,
it might be estimated in seconds, minutes, or hours. That level of accuracy
should be enough to make conclusions. This level of accuracy is not exclusive
319
to fire but to all forms of investigation and design. The analysis must give a
means to make informed decisions.
The use of science in fire is increasing. In fire safety design, the perfor-
mance-based design is gaining widespread acceptance around the world.
This approach is an alternative to a plank-by-plank following of detailed
regulations. The performance-based approach considers the safety issues
associated with the specific fire challenge and renders a design to achieve
a quantitative engineering performance. In fire investigation, U.S. Supreme
Court decisions (Daubert and Kumo Tire) have mandated that expert tes-
timony has a scientific basis. This does not mean that science alone must
serve as the investigator, but that opinions must have followed a scientific
method. The facts must be gathered and analyzed, and hypotheses accord-
ingly evaluated for consistency. The analyses must include relevant testing
or computations to support an opinion in court. Usually, the question asked
of an expert is “… to a reasonable degree of scientific certainty what is your
opinion?”
In this chapter, we will consider the science-based aspects of fire safety
design and fire investigation. In fire safety design, a specific fire is not obvious
and many scenarios and their probability must be considered. In fire investi-
gation, we may have some evidence or hypothesis of the fire. An important
aspect of design and investigation is the justification of the relevant fire
scenarios and the probabilities of their occurrence. In investigation, we might
use our calculations to clearly eliminate or accept a certain event and its time
of occurrence. In design, we might rule out a hazard from worst-case fire con-
siderations. We will discuss these two applications more.
To illustrate the applications to design and investigations, some case
studies will be introduced. In addition, a discussion of the use of computer
models will be presented. Such models are part of our new technology and
many people become versed in their use. Some are easy to use, and others
require computer skills. Hopefully this book has provided background to
appreciate the ingredients of computer models. A bit more will be said about
this later, but the user of any model—equation or computer—must under-
stand its relevance and limitations. Misapplied, both can be misleading and
misunderstood.
10.2 Fire Safety Design

Design in fire safety is used principally in establishing compliance with
the local regulations. These regulations, based on codes and standards of
practice, specify the requirements. The architect, builder, or building owner
must ensure compliance. To do more is not usually sought; however,
many situations arise where the regulation is not clearly applicable or the
Design, Investigation, and Case Studies 321
designer wishes an equivalent alternative. The establishment of equivalency

requires at least a technical discussion but more appropriately an analysis.
To truly meet an equivalency to the regulation, it must be interpreted into a
performance-based statement, and the parties have to agree to its correct-
ness. The formulas we have discussed for fire computations can be used for
such equivalency analyses. The SFPE Handbook of Fire Protection Engineering
referenced throughout this book offers more depth in design formulas than
we have had time to discuss.
Performance-based design has taken hold globally in fire safety design.
In my view, its engineering practice began in Japan. Indeed, Japan had
an engineering handbook for fire well before the United States. However,
some attribute the formal thought beginnings of this approach to a study
in Australia funded by the Warren Centre at the University of Sydney in
1987. While that was a sound study, one motivating factor was to give the
steel construction industry a leverage point. Here, that market force helped
to bring about a change in the process of regulations for fire. In the United
Kingdom, the reduction of government regulations in the Thatcher adminis-
tration brought about change there. And as engineers in fire were beginning
to have true engineering tools, the idea of performance codes caught on.
Now in the United States, the International Code Council and the National
Fire Protection Association (NFPA) endorse the SFPE Engineering Guide to
Performance-Based Fire Protection. This modeling-based approach to design
issues requires competency, integrity, and completeness. While consensus
standards have been developed through an empirical process, the misuse of
science can have severe consequences, so performance-based design must be
administered with care.
Many aspects of fire are subject to regulation and therefore open to design
possibilities. Some areas that apply are listed:
1. Detection and alarm

2. Mitigation of growth and suppression
3. Egress
4. Continuity of operations
5. Structural integrity
6. Refuge and rescue
We have not addressed all of these aspects in our discussion of fire behavior.
The SFPE handbook can give a good overview on ways to predict aspects of
these subjects. Egress and human behavior in fire is a subject just as impor-
tant as the growth of fire but is not studied as much. Detection depends on
the technology of the detector, but in investigation nonintended fire detec-
tors may be significant. For example, faults in electric wires can be used to
orient the source of the initial fire. This is known as “arc mapping” in which
beads on copper wires from electrical shorts caused by the fire can show
directionality of fire fire’s attack. Other detectors might be security alarms,

such as motion detectors and CCTV for surveillance. On the subject of struc-
tural integrity, the book seemed to be closed until the devastating WTC fail-
ure and collapse of the twin towers. Engineers are now turning again to such
design challenges for structures.
Perhaps the more significant aspect of fire safety design is the interaction of
fire growth time with egress time. In general, we seek egress time to be less
than the time to cause fire damage to people. For all possible fire scenarios
and their likelihood (probability), we need to determine the egress time of
the building population. Egress time increases as the fire progresses because
smoke slows people. On the other hand, the time to cause damage by the
fire is inversely related to fire size. As the fire grows, the time for hazard to
people decreases. Sprinklers, suppression by fire fighters, and fire resistance
barriers all affect this damage time. A qualitative illustration is shown in
Figure 10.1. Adding sprinklers could cause the fire damage time to become
infinitely long, meaning complete suppression. Adding compartmentation
will increase the time for overall damage, but eventually these barriers will
fail and transmit heat and smoke. If the fire continues, eventually the time to
escape (egress) exceeds the damage time, and people stop moving and suc-
cumb. The graph is qualitative and only serves to show the various param-
eters. But a designer of building safety, in a performance code approach, must
make it quantitative and show there are sufficient factors of safety to present a
“safe” design. No design can be completely “safe.” Note that even sprinklers
can be defeated and the fire damage curve would then turn downward again.
Fire
lers
d
nk
ama
Adding spri
ge ti
Egress time
me
Time
Addi
n g com
partm
entat
Fire ion
damag
e tim
e
0
0
∙
Time after ignition (or firegrowth, Q)
FIGURE 10.1
Fire damage time compared to egress time, for a particular fire scenario.
Sprinkler protection alone is not a sure way to mitigate fire growth. It can fail
from poor maintenance and control, and even by design.
10.2.1 Examples in Design

Here are two examples selected to illustrate design-type analyses. Both are
based on a computer model used often in Japan for big buildings. The com-
puter code is described in Reference1 when Dr. Tanaka was a guest scientist
at National Institute of Standards and Technology (NIST). This modeling
approach is a “zone model” that divides a room into an upper smoke zone
and lower zone, and then computes the conservation laws for the gas zones,
and boundary elements such as the walls. The Tanaka code formed the basis
of the NIST computer code CFAST; however, they are far from the same hav-
ing evolved separately. The Tanaka code has gone on to become a staple
modeling approach by the Japanese construction industry and academia in
Japan. An extensive documentation of it in a 2002-upgraded version BRI-2
is available in Reference.2 That code has widespread use in Japan and has
been extended to urban conflagrations as well. The two examples of fire safety
design analysis have actually been conducted because of fire incidents; they
were not conducted as part of the design process. However, these examples
are introduced to demonstrate that such analysis is applicable in the design
phase. The analysis of an actual fire incident demonstrates design strategies
that could have been applied during development of the building instead of
simply applying the regulations. The two buildings discussed in the examples
undoubtedly were in compliance with the regulations after construction. Such
analyses could have provided cost-effective safety improvements or equiva-
lency to the regulations. Both of these analyses involve the dynamics of smoke
in between many rooms in a big building. Our simple equations for flow in
this text could not do this type of analysis alone, but this code uses similar
physics in a way to account for many flow and fire phenomena at once.
This code is based on a zone modeling approach that solves differential
conservation equations for the upper and lower gas layers in a room.
With today’s computers, this code runs fast, and even a large building can
be computed in minutes. Recently, two of our former University of Maryland
students used the Building Research Institute (BRI) code as a basis to develop
a multiroom fire growth model and its impact on the building structure.
This was done for an insurance-based company to assess fire risk (risk man-
agement solutions); it combined probabilistic and deterministic modeling.
10.2.1.1 Example 1: Effect of Shaft Vents in a Building Fire

The first example was computed by Takeyoshi Tanaka, a scientist of the BRI
of Japan, using his computer code designed to calculate the smoke transport
and its properties through a building.1 The results of the computations are
shown in Figure 10.2. Two shaft vent conditions are considered and lead to
slightly different results for smoke movement and temperature. These com-
putations were initiated following the MGM fire of November 21, 1980, in
Las Vegas, Nevada. At the time, computers could not easily or quickly com-
pute the 65-story MCM building. So these calculations for only five stories
were performed. Their inferences were never used for analyzing that fire or
learning from its consequences. Most people (85) died on the upper floors
of the MGM as smoke traveled up the vertical shafts. The calculations of
Figure 10.2 show that the case of a vent in the center shaft does not affect
smoke temperature very much but does impact the smoke depth on the
Smoke Smoke
308 808
2.28 0.58 0.24
303
0.22 0.22 0.49 0.61 0.43 0.43 0.49 0.49 0.65 0.80 0.73 0.74
0.23 0.29 0.20 0.49
0.18 0.09
0.31 0.06
0.31 0.31 0.31 0.33 0.34 0.34 0.06 0.06 0.06 0.16 0.16 0.17
0.33 0.06
0.34 0.34 0.33 0.34 0.34 0.06 0.07 0.16 0.16 0.17
869 555 369 565

0.38 0.42 0.20
0.02 0.21 0.44 0.16 0.16 0.17
0.01 0.63 0.06 0.33 0.34 0.34 0.25
0.31 0.33 0.06 0.07

0.31 0.31 0.31 0.34 0.33 0.06 0.05 0.06 0.05 0.06
Time: 0.5 min Time: 0.5 min
Smoke Smoke
316 301 317 302
3.68 0.45
1.02
0.24
303 307 308 304 303 307 309 305
0.10 0.60 0.56 1.21
0.80 0.89 1.36 1.27 1.23 1.85 1.88 1.33
0.70 0.72 1.29 0.71
0.08 0.10 0.81 0.04 0.59
882 307 307
0.63 0.08 0.05 0.41 0.29 0.31
0.63 0.63 0.71 0.81 0.76 0.76 0.42 0.51
0.30 0.19
0.72 0.56
0.72 0.72 0.83 0.83 0.83 0.57 0.57 0.87 0.87 0.87
328 394 594 390 399 595

0.25 0.26 0.43 0.26 0.27 0.42
0.49 0.34
0.79 0.81 0.16 0.83 0.83 0.83 0.65 0.67 0.02 0.87 0.87 0.87
0.67 0.72 0.52 0.56

0.67 0.67 0.67 0.72 0.72 0.53 0.53 0.53 0.56 0.56
Smoke Smoke
320 303 323 306
0.11
3.95 189
0.42
303 306 311 314 309 306 306 315 317 312
0.73 0.74 0.83 0.86 1.50 1.37 1.35 1.34 203 204 204
0.08 1.51 1.49 1.33
0.06 0.02 0.02 0.04
0.05 0.03
304 305 301 305 310 313 307
0.85 0.12 0.20 0.58 0.04 0.27 0.37 0.50 1.05 0.98 0.80
0.85 0.85
0.97 0.76 0.55 0.03 0.22 0.23
0.04
1.00 0.92 0.92

1.00 1.00 0.65 0.65 0.61 0.92
320 370 581 323 376 589

0.02 0.30 0.32 0.25 0.61 0.08 0.26 0.23 0.24 0.56
0.52 0.51 0.65 0.65 0.65 0.46 0.40 0.62 0.62 0.62
0.30 0.43 0.07 0.30 0.411 0.11
0.93 1.00 0.85 0.92

0.93 0.93 0.93 1.00 1.00 0.85 0.85 0.85 0.92 0.92
FIGURE 10.2
Smoke movement computations. (From Tanaka, T., A model of multiroom fire spread, NBSIR
83-2718, National Bureau of Standards, Gaithersburg, MD, August 1983.)
third floor. Further calculations could produce more complete informa-

tion on the benefit of shaft vents. By considering different smoke control
options through computations, a designer could determine the best safety
for the occupants. The current BRI-2 computer code now can address forced
ventilation effects that could be used to address pressurization options in
design as well.
10.2.1.2 Example 2: Smoke Movement in the World

Trade Center, New York, New York
Another high-rise building fire that attracted much attention was the explo-
sion and fire in the World Trade Center in New York City on February 26,
1993. These computations were implemented by Yamaguchi et al.3 to analyze
the fire scenario of the 1993 World Trade Center fire. An excerpt is shown
in Figure 10.3. Computers are now faster and bigger, and this type of zone
model can practically compute an entire 110-story building as the WTC tow-
ers. A fire of 20 MW for 25 minutes was selected to simulate the parking
garage fire (shaded black in Figure 10.3). The results show the smoke flow
(black arrows for upper smoke layer flows, white for lower layer flows) at
13 minutes (780 seconds). While this computation analyzed the particular
fire scenario caused by the parking garage explosion, other scenarios can be
considered in a design strategy coupled with an egress model for the build-
ing to assess the impact of smoke and fire on the occupants.
No direct deaths resulted due to the smoke filling the tower building,
although it took many hours for occupants to leave the building through the
stairways. In this particular fire, the smoke affected visibility but apparently
was not toxic. The garage was either well ventilated or the fire never reached
the ventilation-limited state to cause toxic smoke to enter the many floors
and stairwells. But in a design calculation, a bigger fire and its effect could be
considered. I suspect the outcome would be different, and the smoke would
be a serious problem for the occupants.
As we can imagine, such calculations cannot only estimate the smoke con-
ditions under this fire event but could be beneficial in assessing the hazard
of other fire scenarios. Design strategies that might make the building safer
can be evaluated on the computer. So the computer allows the coupling of
various fire phenomena and projects them over time. This we cannot eas-
ily do by the single equation approach alone. Yet computer models can be
checked by the individual formulas to at least see if there is some consis-
tency. This should always be done, or a computer could stray without the
awareness of the user.
As we know now, this explosion in a truck bomb in the basement of the WTC
was a harbinger of what was to come on September 11, 2001 (9/11). The bomb
in 1993 did not dent the massive support column in the basement, but did
send smoke through the buildings to such an extent that it took many hours
for occupants to egress. We will come back later to a discussion of 9/11.
780s
19 110F
12 18
26 28
11
10
20 21 22 23 25
33 78F
8
27
13 14 15 16 17 24
32 44F
4
1
Upper layer flow
Lower layer flow
35
G.L. 31 1F
30
29
34
FIGURE 10.3
Analysis of smoke movement in the World Trade Center fire, 1993. (From Yamaguchi, J. et al., A
study on predicting smoke transport in a high-rise building [in Japanese], in: Proceedings of the Annual
Meeting, Japanese Association for Fire Science and Engineering [JAFSE], Tokyo, Japan, May 1995.)
10.2.2 Performance Codes

Performance fire safety codes as opposed to prescriptive codes now form the
basis of many fire safety designs, especially for big or unusual buildings.
Essentially this means using computational analysis over set specifications
in the regulations. This process is evolving, and now guidelines have been
laid out by SFPE.4 The guide basically encourages the use of methods that are
verified and validated (V&V). The concept of V&V basically addresses that the
ingredients of the method are sound and its limitations clear (verification),
and the output gives a known level of accuracy against quality data or
other valid sources (validation). The Nuclear Regulatory Commission (NRC)
put forth a report on the V&V of several design approaches to fire.5 They
addressed the equation approach similar to the substance of this book, the
use of zone models, and the use of the NIST FDS code. All three methods,
where they were applicable, gave comparable and acceptable results for the
NRC criteria applied.
Because fire computations require certain expertise, the use of meth-
ods, even with a V&V label, does not assure perfection. Indeed, the use of
a sophisticated computer model that gives graphical output that looks like
smoke and fire can be misapplied, especially when the user only knows how
to operate the computer. Tools are becoming user-friendly, and checks and
balances must be considered. For a specified regulatory approach, the check
to compliance could be done by inspection. For a performance-based design
approach the check is not so straightforward. The criteria for design, the
analysis in design, and implementation of the design must all be checked in
some way. Until this expertise is comprehensive and sound throughout the
practice, fire safety design can be problematic. Changing fire safety design
from regulations to computers is technologically feasible but may not be
practical to implement. Such new technology must progress hand in hand
with needed research to fill the knowledgebase and the computer gaps.
My own view is that for special applications performance-based or engi-
neered fire safety designs are desirable. After all this is the basis of engi-
neering. However, in product development, flaws, usually are realized in the
engineering stage. Of course, some flaws get by and they are realized in
use. The correction of such in-use product flaws is usually well discussed in
the public domain. In fire safety, public awareness is usually low. Also fire
events that attract sustained public attention are rare. So how can a perfor-
mance-based design be checked or subjugated to review after a fire event?
On the other hand, the current regulatory approach to fire safety lays out
specific rules. They are mostly devoid of science, and their basis is sometimes
founded in folklore. A historical analysis of specific fire regulations would
show scientific weaknesses. A suggestive alternative to performance-based
codes is to have an engineering basis for current regulatory specifications.
Then the specifications can evolve as science and engineering knowledge
improves.
10.3 Fire Investigation

Litigation has pushed the use of computers and fire analysis into the field of
fire investigation. In some countries when a significant fire accident occurs,
an official board of inquiry is established. Recently, this was done in China
for the Mandarin Hotel fire (February 9, 2009). The hotel was 44 stories and
was nearing completion when fireworks caused the fire. The building was
part of the modern Z-shaped structure of China Central Television (CCTV)
in Beijing. Unfortunately, this official investigation will likely never be made
public. As with legal litigation, after-incident fire reports are rare or devoid
of scientific analysis.
In the United Kingdom, some years ago, the King’s Cross subway escalator
fire (November 18, 1987) was investigated by an official government board
of inquiry. They found through experiments and computer modeling that
a fluid dynamical effect caused the fire to lean down against the wooden
stairs on the escalator (dubbed the trench effect). This close contact with the
stairs was deemed responsible for the rapid fire spread. Computer modeling
became a forensic celebrity after this fire; however, some still believe that
the painted ceiling of the escalator was also critical in causing the rapid fire
spread. It is likely to say that the use of computer modeling in the King’s
Cross fire brought it to the forefront of consideration.
Even before computer models were available, scientists had been drawn
into investigations by litigators many years before. As fire research devel-
oped models, in terms of formulas, lawyers were astute enough to engage
fire scientists. Often, however, litigation never illuminates what happened,
as cases are sealed or are confidential. While the King’s Cross escalator was
of old wood construction, the Mandarin hotel was of modern steel construc-
tion. Apparently, something was not considered in the fire safety design of
the Mandarin to make it so vulnerable to fireworks. Structurally, however,
it survived.
The earliest I recall that twentieth-century fire science tools were used in
a notable investigation was the MGM hotel fire in Las Vegas (November 21,
1980). We discussed some aspect of that in Example 1, but those calculations
were not part of an official investigation; most modeling was done under
litigation. It is likely that sprinklers in hotels and high-rise buildings were
required because of the MGM fire.
As in design applications, there is no simple approach to evaluating
the fire issues in an investigation without a competent knowledge of fire
behavior. Although computer models can be useful, the knowledge gained
in this book can go a long way to create a timeline or proposed scenario in
a fire investigation. Unlike design, this process for investigation is not open
ended. There was a specific fire for investigation, and it did specific dam-
age. These facts must be established through evidence, witness accounts,
and analysis.
10.3.1 Dupont Plaza Fire

My experience in fire investigation was nonexistent until Harold “Bud”
Nelson and I visited the site of the Dupont Plaza Hotel fire in Puerto Rico
(December 31, 1986). Ninety-seven people died, mainly by flame, in the
Casino on the second floor of the hotel. The fire began on a first floor of a
nearby ballroom among newly delivered furniture. Under auspices of the
U.S. Fire Administration and with the permission of the Bureau of Alcohol,
Tobacco and Firearms (ATF) Bud and I examined the scene. The unusual
aspect of this fire was that the ballroom fire was so fuel rich that when it
broke through the glass windows into the Casino, the smoke turned to flame
and a wave of fire engulfed the occupants.
We were able to explain this fire and present a quantitative timeline of
events by equations very similar, if not identical, to those contained in this
book. At that time, Bud Nelson had a computerized rendition of a set of equa-
tions called FPETool. Figure 10.4 shows Bud in a pool lounge chair working
FPETool with the Dupont Plaza in the background. ATF now has some-
thing similar for their fire investigation agents programmed into a scientific
calculator. Bud’s report can be found through NIST.6 It was a landmark for
investigation showing what a scientific analysis could add to an investiga-
tion. The ATF were impressed that we found results consistent with their
FIGURE 10.4
Harold “Bud” Nelson analyzing the Dupont Plaza fire on the scene.
scene and witness data, and we were impressed that we could make this
contribution to a real fire.
I can only relate my emotions to the movie, the Flight of the Phoenix (1965)
in which after a plane crash in the Sahara one survivor says he is an airplane
designer and guides them to build a new plane from the parts. On takeoff,
they learn that he had only made model airplanes. That is the analogy to
Bud and my backgrounds in the laboratory of fire science compared to a
real scene fire investigation. Something resonated between the ATF agents
on the scene and us. It provided a catalyst to interactions with the ATF fire
investigators. Scientists and investigators could learn something from each
other. This spirit is making both stronger. The National Association of Fire
Investigators has sponsored recent scientific meetings, papers on scientific
aspects of fire investigation are appearing in scientific journals, and the
International Association of Arson Investigators has a website for knowl-
edge-based learning: CFItrainer.net.
My next involvement with the ATF came in a course we presented to them
at the University of Maryland in 1992. Bud Nelson was part of that course.
It was a challenge to see if we, as scientists, could present useful information
to investigators. Example 3, presented as follows, was a challenge question to
see if science could support the evidence of a case being considered by ATF
at the time. Later courses with the ATF shaped the first edition of this book.
10.3.2 Example 3: The Case of the Laundry Basket Fire

In May 1993, the murder of a man in Pittsburgh area led to the conviction of
his wife with an additional conviction of arson. The wife apparently hatched
a conspiracy with an associate to torch her residence, while the family was
in Florida. The following gives an account of the events that came to light in
the trial.
The wife was accused of masterminding a botched attempt to burn down
her home and of filing a false theft report so insurance money could be col-
lected. It came to light in the trial that the woman had devised a scheme to
have their house set ablaze, while the family went to Florida. She arranged
to have an accomplice torch the residence, while the family was in Florida.
But the alleged plan didn’t work out. According to a testimony from a county
fire official, investigators believe a fire was deliberately set in the basement,
but a water pipe joint quickly melted and the water extinguished the blaze
in less than 30 minutes. This failed arson event eventually led to the plot
to murder the woman’s husband, and the subsequent murder trial with the
background of a failed arson attempt.
The arson attempt in this was the scenario of fire in a laundry basket that
ruptured a copper water pipe, extinguishing the fire. The evidence showed
that the fire had been intentionally set between the hot water heater and
furnace in the laundry room of the residence. An overhead copper water
pipe opened at an elbow joint as the solder connection melted. Plumbing
Water Water
Furnace heater heater
Furnace
(a) (b)
FIGURE 10.5
The laundry basket fire and copper pipe joint. (a) Clothing fire and (b) added gasoline fire.
solders melt in the range of 250°C–300°C at most. To achieve such a tem-

perature, the flame tip of the fire must reach the overhead copper pipe.
The flame source must be, at least, capable of this temperature at this
height. Moreover, to melt the solder in a water-filled pipe requires the pipe
to be bathed in the flame to maximize heat transfer to the solder connec-
tion. If the flame immerses the pipe over a length, as the flame temperature
is effectively constant, the pipe will have a constant temperature over this
length. No temperature difference along its length means no heat loss along
the pipe, so all the heat from the flame stays in the pipe. Figure 10.5 illus-
trates the scenario.
An ordinary laundry basket fire could be approximated as a large waste-
basket fire from Table 6.4 as 10 g/s at an estimated heat of combustion of
20 kJ/g (assuming plastic and cotton clothing). This gives
Q = (10 g/s)(20 kJ/g ) = 200 kW.
From Equation 7.9, we compute a flame height of
L f = 0.23 Q 2/5 - 1.02D
= 0.23(200 kW)2/5 - 1.02(0.7 m) = 1.2 m
where we have assumed a fire diameter of 0.7 m. This flame height, 1.2 m,
represents probably an upper limit for a laundry basket fire. But we see that
its flames cannot reach the pipe at the ceiling (Figure 10.5a). On the other
hand, if an accelerant—say gasoline—were added to the laundry basket,
then we have (from Chapter 6) a higher burning rate:
Q = (55 g/m 2 -s)(p/4)(0.7 m)2 ( 43.7 kJ/g) for gasoline

or
Q = 925 kW
and
Lf = 0.23(925)2/5 – 1.02(0.7)
or
Lf = 2.82 m
This flame would strike the ceiling and easily bathe the copper pipe joint.
Its temperatures would be sufficient to melt the solder (Figure 10.5b).
It should be clear from this example that some choices had to be made
on the nature and size of these fires. These assumptions may differ among
analysts, but in the end an ordinary laundry basket, even if somehow acci-
dentally ignited, is not very likely to be capable of producing a flame to melt
the solder at the joint. Even without any direct chemical evidence of an accel-
erant, such an analysis would strongly indicate its presence.
The foregoing example arose in the first training course given to fire inves-
tigators in the Bureau of Alcohol, Tobacco, and Firearms (BATF) program.
Agent William J. Petraitis posed the problem after he and Allegheny County
investigator Thomas Hitchings had conducted tests on laundry baskets with
and without accelerants. They were looking for telltale fire signatures that
might occur. The flame height is the key indicator here. The example proved
useful to illustrate the application of the science in this ATF course to a real
case. It showed the agents that a calculation could add more evidence to
their findings done by tests alone. However, once formulas might lead in a
particular direction, if possible, tests should be done for confirmation and
demonstrative purposes.
10.3.3 Example 4: An Analysis of the Waldbaum Fire,

Brooklyn, New York (August 3, 1978)
On August 3, 1978, a fire occurred at the Waldbaum supermarket in
Brooklyn, New York. A sketch of the store is shown in Figure 10.6. At the
time of the fire, an extension was under construction as indicated in the fig-
ure. The store was a typical supermarket with a mezzanine along a portion
of the north wall. The loft was completely of wood construction compris-
ing the floor, roof, and structural support trusses. The trusses were made
of 3 by 12 in. members interlaced together in bundles of 4 or 5. Trusses 2,
4, and 6 were covered on one side with plaster to form firewalls in the loft.
However, to enable passage, these trusses had doorway openings. The roof
#7
#6
Door
#5 Mezzanine
machine
#4 room
6' × 6'8'' door Men

#3 Women
Aven
#2
150΄
ue Y
Truss #1
Fire building Extension

(under
construction)
Ocean avenue
N
FIGURE 10.6
Waldbaum supermarket, Brooklyn, New York, 1978.
had been modified with a rain roof added at the peak. This construction
formed a double layer roof at the peak. Also the new construction required
a splayed roof section to meet the new roof of the extension. This construc-
tion formed another double roof triangular section along the north wall as
shown in Figure 10.7. Due to the ongoing construction and the addition of
new columns along the wall to the extension, construction voids existed
along this wall to the hidden space between the old and new roof.
On the day of the fire, construction work began at 7 a.m. and the store
opened for business at 8 a.m. Flames were first seen at 8:30 a.m. along the
interface between the ceiling and extension wall of the mezzanine men’s
room and the machine room. The fire eventually spread into the loft
between truss sections 4 and 6, and collapsed truss 5 at approximately
9:15 a.m. The collapse caused 12 firefighters working on the roof to fall into
the flames. Six were killed.
A man was tried and convicted in 1978 of setting this fire. His confession
stated that he and two others set the fire near dawn (approximately 6 a.m.) by
making holes in the roof and using newspaper and lighter fluid to initiate the
Roof
New roof extension
Loft
Store extension
4 or 5 3'' × 12'' Truss member under
Mezzanine construction
FIGURE 10.7
Splayed roof extension and structural support trusses.
fire below. This confession was later questioned and discounted in a retrial
that was held in 1994.
A consistent fire scenario matching the timing of these events was never
fully presented. The original fire investigators could not agree on a cause and
later suggested that the cause of the fire may have been of electrical origin.
However, there was no electrical power in the ceiling area where flames were
first seen. Also a workman in the loft, at the onset of the flames to the mezza-
nine, saw no evidence of fire in the loft. The prosecutors sought advice on how
to explain an alleged fire could start at about 6 a.m. but not seen until 8:30 a.m.
I was asked to assist William J. Petraitis, special agent of the BATF in the
investigation of this fire, 16 years after it occurred. Agent Petraitis discovered
the splayed roof extension, shown in Figure 10.7, and reasoned the fire had to
have begun in that space. We then examined the hypothesis of the fire begin-
ning there at approximately 6 a.m. to see if it could be consistent with the
other known events. A fire scenario was developed, and calculations were
examined to support the plausibility of the events and their timing. All of the
analysis is not presented here, but the main results are described. The analy-
ses were performed using the material in this book along with information
from the scientific literature. The fire scenario involved issues of ignition to
flaming, extinction, smoldering, reignition to flaming, and the burning of
wood members to the point of collapse. Here is the proposed scenario sup-
ported by calculations.
10.3.3.1 Early Dawn: Ignition (Approximately 6 a.m.)

At approximately 6 a.m., an intentional fire is considered to have been
set in the roof area adjacent to the construction of the building extension
(Figure 10.8). Based on the original confession, this fire is set by stuffing
Early dawn: Ignition (approximately 6 a.m.)
Holes
Loft
Ignitor fuel
Store extension
under
FIGURE 10.8
Flammable liquid-soaked newspapers inserted into roof holes.
paper through holes in the new roof extension along with a liquid accelerant.
The splayed roof extension built over the existing roof forms a void space
between the two roofs. The new roof is supported by rafters. The fire is set in
channels of the wood rafters that extend between the primary wood trusses
4 and 5. Gaps under the rafters allow air to flow into the fire, but the space
is mainly enclosed with temporary partitions at the wall adjacent to the new
building extension. The accelerant-soaked paper probably caused a fire of
100–500 kW in one or more rafter channels, involving no more than 1 m2.
This fire condition is depicted in Figure 10.9.
Loft
Flaming regions
Store extension
under
FIGURE 10.9
Ignition at approximately 6 a.m.
Under expected heat fluxes of 40–50 kW/m2, the wood members would
ignite in 30 seconds to 1 minute and begin to contribute roughly an addi-
tional 500 kW. The accelerant newspaper fire would burn out in 1–2 minutes.
The wood fire could progress to roughly 1000 kW, but then it would become
limited in further growth by smoke filling this confined space and consum-
ing its oxygen.
10.3.3.2 Smoldering Stage

Oxygen depletion would cause flaming to cease in about 1–2 minutes fol-
lowing the ignition of wood construction. However, sustained smoldering
would be possible, especially in the smaller more confined rafter spaces
where radiant heat transfer would be high (see Figure 10.10). Ohlemiller7
reports that sufficient radiant heat transfer is necessary to sustain smoldering
in wood cavities. He also reports smoldering propagation rates of 1–6 cm/h.
Using the smaller value, smoldering can propagate through the old 3/4 in.
plank roof in roughly 2 hours. A short time before 8:30, a hole would form
between the loft and the roof cavity. With air velocity increased to 25 cm/s,
Ohlemiller7 found smoldering in wood would change to flaming com-
bustion. Such velocities would be realized at the hole as air was naturally
induced to flow from the loft through the hole into the hot cavity space.
Flames would then erupt in the false roof cavity space. Airflow from the
loft through the hole would initially keep flames and smoke out of the loft.
So here we have a plausible scientific explanation of how a 6 a.m. fire initia-
tion finally breaks out into the loft 2½ hours later. Again, the student should
realize that this analysis does not give absolute precision in the timeline,
Unsustained smoldering
Sustained smoldering
Store extension
under
FIGURE 10.10
Smoldering combustion at approximately 6:05 a.m.
but an order of accuracy to within tens of minutes. The estimated timeline

agrees with the actual events to within a reasonable degree of scientific certainty.
10.3.3.3 Onset of Flaming: Shortly before 8:30 a.m.

At approximately 8:30 a.m., flames are observed in the mezzanine area at the
wall and ceiling near truss 4. It is believed that this was due to the expansion
of the flames as flashover caused a rapid rise in energy release rate and a
pressurization of this confined space (Figure 10.11). The associated pressure
increase forced flames through any voids from the cavity space. As the hole
to the loft became larger, the increased airflow rate into this hole would also
promote a larger flaming fire in the rafter spaces. The flow of the flaming
combustion products would follow the rafter channel to the end points at
trusses 4 and 5. The first flames in the loft were seen at an opening near truss
4. As the hole to the loft became even larger, flames would lap upward under
the sloping wood ceiling of the loft. The person who had been in the loft just
before 8:30 would not have been aware of the fire until smoke and flames
emerged from the larger hole. In fact, he only saw flames when he descended
to the machine room after being warned of the fire.
Evidence to support this stage of the scenario is shown in the photograph
of Figure 10.12. The small white vapor plume rising from the roof indicates
a hole in the roof. This hole is unlike the larger roof holes, attributed to the
fire fighters in ventilating or from burn through of the roof. These large holes
emanate dark gray smoke. Flames are also seen to have emerged from the
roof on the other side of truss 4. This flaming region is likely just above the
first and highest hole made by the arsonists into the confined rafter space as
shown in Figure 10.8. Smoldering would have been sustained in this more
Flame extension
into loft
Store extension
under
construction
First sign of flames
Mezzanine
FIGURE 10.11
Flaming erupts in the roof cavity shortly before 8:30 a.m.
FIGURE 10.12
Fire on the roof of Waldbaum store before collapse. (From Hughes, B.J., Private communications
of statements and records, Office of the District Attorney, Kings County, Brooklyn, NY, 1994.)
confined space. The hole indicated by the white vapor plume is in the area of
the larger void space where fire was likely initiated, but smoldering was not
sustained there due to less severe heat transfer.
10.3.3.4 Fire Growth in the Cockloft

Flame spread would rapidly move from the hole, up and under the loft
wood ceiling. It is estimated that flames would move from the hole region
up and along the peak of the loft to truss 4 and through truss 5 in about
2–3 minutes. This time was estimated from a flamespread model9 using
an ignition time of 30 seconds and an energy release rate per unit area of
150 kW/m2. Such calculations are beyond the information presented in
Chapter 5, as a solution to a differential equation is needed. However, the
spread model is based on the theories explained there. This rapid spread
(Figure 10.13) seemed incredulous. With the low confidence in flamespread
calculations, an experiment was performed in a similar but smaller loft.
Flashover occurred in the loft in approximately 1 minute. In the store loft, flash-
over was estimated to occur when the wood fire contribution was 6–12 MW in
the section of the loft between trusses 4 and 6. The flame spread calculation
indicated that this would occur in approximately 2½ minutes. So the esti-
mates from the theory are not so different from the smaller loft experiment.
Full involvement was estimated to take up to 5 minutes more. From the
onset of flaming at the hole, full involvement of the loft section would take
approximately 5–10 minutes. Because of the false roof sections and the truss
plasterboard firewall sections, the firefighters on the roof would be unaware
of the raging fire in the loft section below them.
Truss 6
Firewall
Doorway
Truss 5
Truss 4
Firewall
Doorway
FIGURE 10.13
Fire growth in the loft at approximately 8:35 a.m.
10.3.3.5 Collapse of the Roof due to Truss Member Failure ~ 9:15 a.m.
If one assumes that the roof collapse is due to the failure of a truss member
as a result of fire degradation, the burn-through time of a member can be
estimated. A fully developed fire in the loft space occurs at about 8:40 to
8:45 a.m. This fire exposure would give the wood its maximum burning rate.
From Table 6.2, a mass burning rate per unit area of 11 g/m 2-s is selected
as typical for wood under these conditions. For a 3 in. thick truss member,
for an estimated density of about 500 kg/m3, a burn through (1½ in.) would
take about 29 minutes. Collapse occurred at about 9:15 a.m.
10.3.3.6 Concluding Remarks

This fire scenario was a complex series of phenomena that are not usually
appreciated in the study of fire. Experienced fire investigators at the time of
this event in 1978 were not able to deduce this process. Yet relatively simple
scientific analysis could produce a plausible and consistent series of events
within the periods of the recorded observations of this fire. Some may ques-
tion the details, but the scenario fits the timeline of what is known and
alleged. Only accuracy to the order of minutes is scientifically supported in
this analysis, yet the estimated times are comparable to the real events as
they are known here.
The photograph in Figure 10.12, which shows the small white smoke plume
in the lower right below the bottom edge of the flames, is a telltale indicator.
It would appear to be one of the original ignition holes where smoldering
was not sustained in the splayed roof structure. But for a hole above, the fire
went from flaming to smoldering and back to flaming. The low rate of down-
ward flame spread kept the fire in the vicinity of the upper hole, but the more
rapid upward spread drove the fire into the loft with tremendous speed.
This scenario cannot be described or defended without a comprehen-
sive knowledge of fire behavior and available data to support the calcula-
tions. Smoldering research and data are rare but were critically needed in
this analysis. The background of this fire was provided by the Office of the
District Attorney, Kings County, Brooklyn, New York.8
Even if the student cannot perform or confirm all the calculations in this
analysis, the qualitative aspects should be generally supported by the subject
matter of this book and hopefully the student can accept them. First, the set
fires in the roof cavity go out without a sufficient supply of air, but in wood,
they can resort to smoldering. Smoldering in wood needs good heat transfer
to be sustained and that was realized in the narrow section of the new roof
cavity space. Then as smoldering burned through the wood roof into the loft,
new air brought the fire to flaming again. As these flames entered the under-
side of the loft roof, the fire spread quickly into the loft section. Flashover in
that section led to a powerful fire with high heat flux that attacked the beam
structure causing it to burn through and collapse. This is the basic scientific
description of this fire and explains a timeline that extended over several
hours.
10.3.4 Example 5: The Branch Davidian Fire near

Waco, Texas (April 19, 1993)
This fire involving the Branch Davidian compound near Waco Texas in 1993
was a notable tragic event. The siege preceding the fire brought the attention
of the world to this standoff between the U.S. government and a religious
group. The siege was prompted by an abortive raid by the ATF for guns
acquired by the Branch Davidians. Deaths on both sides occurred as a result
of the raid. The siege followed and lasted for many days until the fire.
The opportunity to investigate this fire was afforded by ATF support that
allowed Dr. Fred Mowrer and me to join an assembled team of fire scene
investigators at the site shortly after the fire. A forward-looking infrared
(FLIR) video recording by an aircraft was key to the early investigation in
identifying the locations of the initiated fires. Our involvement led to a report
to ATF that was later sent to the Department of Justice (DOJ) by the ATF.
The fire investigation played out over an extended period of years as various
investigations and trials took place. Initially, there was a criminal trial on
the deaths of four ATF agents. This trial was followed by a Congressional
hearing on the role of the government in the fire and the use of possibly
pyrotechnic tear gas. A civil action by the surviving Branch Davidians
then led to a trial in 2000. A Special Counsel report by John Danforth to the
DOJ summarizes many of these events and states “…what is remarkable is
the overwhelming evidence exonerating the government…”.10 What is also

remarkable is that criminal action by the local government having jurisdic-
tion did not investigate the fire for arson nor attempt to hold any individuals
accountable. My involvement is summarized chronologically as follows:
• April 22, 23, 1993 Visit scene of the fire that occurred on April 19
• April 25, 1993 Visit FBI headquarters to obtain video and pho-
tographic data
• September 8, 1993 Report to AFT and DOJ
• February 1994 Testified in the criminal trial
• July 1995 Testified to Congressional panel
• July 2000 Testified in the civil trial
The scene after this fire was a burned out flattened pile of debris, ash, and
horribly, bodies too. Many Branch Davidians lost their lives, not in the fire
but by gunshot wounds intentionally inflicted. The nature of the scene is
depicted in Figure 10.14. It is clear from that scene that “fire pattern” analysis
suggested by fire experts for the Branch Davidians on the fire would not be
applicable. The key evidence of this fire was visually documented by both
the government and the news media as the compound burned. That evidence
and later testimony by surviving Davidians confirmed that the fire started at
several locations within the compound. The analysis of that evidence could
be amplified by scientific analysis. Some fire events could even be used as
FIGURE 10.14
The fire scene at the Branch Davidian compound after the fire. (From Waco Tribune-Herald,
Waco, TX, August 24, 1993.)
data to cross-check with predictions for scientific correlations. I will relate

the story of this fire in scientific terms and illustrate a few analyses.
It is rare that data are recorded during a fire where investigators have a
continuous visual documentation of what took place. Because of the circum-
stances of the siege at the Branch Davidian compound near Waco, Texas, in
1993, much video and photographic recording equipment was being used.
The news media were recording video approximately 1–2 miles from the
compound, and the government authorities had overhead surveillance
aircraft shooting still photographs and a precise time-coded near-infrared
video (FLIR) camera. Because of the thermal sensitivity of the FLIR, it easily
recorded the fire inception. From these video data, the information in this
text, and the behavior of fire, it was possible to reconstruct the characteris-
tics of this fire.
10.3.4.1 Congressional Committee Statement on the Mount

Carmel Branch Davidian Fire (April 19, 1995)11
A summary of the results was contained in testimony before the
Congressional Subcommittee on Crime of the Committee on the Judiciary,
and the Subcommittee on National Security, International Affairs, and
Criminal Justice of the Committee on Government Reform and Oversight.
That testimony is reproduced here, as it lays out my role, analysis of the fire,
and my conclusions. The figures used to annotate that testimony have been
labeled to agree with the book text. The civil trial later brought some new
information to light that confirmed and added to support to these conclu-
sions. For orientation it is useful to recognize the nature of the compound
through a scale model of the building, shown in Figure 10.15; and a sche-
matic of the state of the building just before the fire with areas demolished
by government activity indicated.
10.3.4.1.1 Background
My name is James Quintiere. I am a professor of Fire Protection
Engineering at the University of Maryland, College Park, MD. Before
coming to the university, I was a division chief in charge of fire research
at the Center for Fire Research of the National Institute for Standards
and Technology (NIST). I have 25 years of experience in fire research,
education, and in the science of fire growth. I am currently the chairman
of the International Association for Fire Safety Science, a world organiza-
tion of scientists and engineers for the promotion of fire research and its
beneficial applications.
Shortly after the fire of the Branch Davidian compound at Waco, Texas,
on April 19, 1993, I was asked to contribute to the fire investigation. In doing
so, I enlisted the support of Dr. Fred Mowrer, also of the Department of Fire
Protection Engineering, University of Maryland. We visited the Waco fire
site during April 22, 1993. At that time, we joined the team under Paul Gray
FIGURE 10.15
Scale model depicting the state of the building before the siege.
(Houston), which also consisted of Thomas Hitchings (Pittsburgh), William

Cass (Los Angeles), and John Ricketts (San Francisco). The group under Paul
Gray would focus on the cause and origin of the fire. We would analyze the
development of the fire and draw interpretations and conclusions from that
analysis.
10.3.4.1.2 Visual Data

The fire had completely leveled the compound so that no significant struc-
tural remains were available to establish the development of this fire.
However, this fire was probably one of the most extensively recorded fires in
history. Not only were commercial television stations continuously record-
ing this event, but surveillance government planes were taking still photo-
graphs and using a FLIR video. These visual records became the principal
source of data for our analysis.
The video and photographic data were made available to us by the FBI. Video
copies of data that we requested were given to us at the FBI Headquarters in
Washington, DC, on April 25, 1993. Subsequently, the FBI video and photo
laboratories supplied additional materials and support as requested dur-
ing our investigation. The data included television coverage of the fire by
the Canadian Broadcast Corporation, Channel 10 of Waco, the FLIR video
recording, and aerial photographs. These covered the time period of the fire,
approximately 12:00–12:30 p.m. CDT.
The principal source of data to establish the inception of the fires and
their locations is the FLIR video. Based on the calibrated clock of the FLIR,
the other video and photographic records could be correlated, and a com-
prehensive visual record of this fire could be established. From this visual
data, I was able to determine the point of origin of the fires, the growth
rates, and estimates of the fire energy output rates at critical transition
points in their development. I also drew conclusions of the nature of the
ignition sources, the role of the tear gas, the effect of the wind, and the sur-
vivability time of the occupants. I will summarize these conclusions and
how they were determined. In addition to this statement, I would like our
official report and a video I made for the criminal trial to be submitted for
the record of this hearing. If you wish, I can review the video as well.
10.3.4.1.3 Ignitions
At least three separate fires began in the compound on April 19, 1993.
Fire 1: The first began at 12:07:42 p.m. CDT in the front of the second floor
right tower. This is believed to have been a bedroom. We can expect the
furnishings to be indicative of a crowded bedroom. I counted about seven
mattress box springs remains in the fire debris at this general location,
presumably from this and adjoining rooms.
The precise time of the onset of this fire can be determined because of the
characteristics of the FLIR camera. The FLIR camera records the intensity of
light and heat radiation in the wavelength range of 8–12 μm. This is in contrast
to what our eye sees, which is in the range of 0.4–0.7 μm. As a result, the FLIR
operated on auto-ranging, which would set the center of its black and white
shading, or gray scale to the ground temperature (say roughly, 81°F). Then it
set its range 40° above and below this mid-temperature. As an object in the
field of view emitted more radiation due to a temperature increase, the object
would appear more white on the IR video. For an 81°F midrange temperature,
this would mean that a change from gray to white color would indicate a tem-
perature increase to 120°F or higher. Reflected sunlight could also cause white
images, and the FLIR could penetrate smoke; but as the smoke became hotter
and thicker, it would see it as white smoke. The FLIR sensor would become
saturated at 194°F, above which the image would not be distinguishable.
Consequently, the FLIR could detect, by a color change to white, temperatures
as low as approximately 160°F (±30° due to auto-ranging). And, it could see
through much of the early light smoke of the fire that would obscure the build-
ing by normal viewing. The FLIR is the definitive key to the detection of these
fires about to start in the building configuration as shown in Figure 10.16.
The image of the temperature rise of the first fire is seen in Figure 10.17,
in the second floor south corner bedroom. The first sign of this temperature
rise was seen at 12:07:42 in the front side of this room. The image in this
photograph occurred 9 seconds later and is due to the transport of hot gases
within the room.
FIGURE 10.16
Schematic of building following demolition action before the fire.
Reflection—
not a fire
9 seconds after
start of
first fire.
Shown in
second window
FIGURE 10.17
FLIR image 9 seconds after fire is seen in the second floor south bedroom. A white heat image
is seen at the corner of the window.
In a similar manner, the other fire starts were determined. It appears that
they all began on the perimeter of the building.
Fire 2: The second fire began in the dining room on the first floor level,
approximately 1 minute after the first fire, at 12:08:48 p.m. CDT. This is seen
on the FLIR video by a hot plume rising from the rear of the dining room. On
surveying the fire debris, I counted 20 burned stacked chairs in this general
location within the dining room.
Fire 3: Nearly one minute after the dining room fire began, the third
fire is seen in the chapel window on the right side of the building at
12:09:45 p.m. CDT. This is shown in Figure 10.18. The dining room fire
is also visible, and the bedroom fire has now affected adjoining rooms,
adjacent, and above.
Less than a minute later, a related or separate fire is seen to occur in the
debris area behind the chapel at 12:10:23 p.m. CDT, as shown in Figure 10.19.
This fire could have been connected to the previous chapel fire. The time
difference between the two fire observations is comparable to the time asso-
ciated with flame spread on a liquid fuel poured between the two points.
Figure 10.20 shows an aerial view at about the time of the possible fourth
fire start. By comparing this figure to the previous figure, it can be seen
3rd fire,
chapel
FIGURE 10.18
FLIR image showing hot (white) region at the first floor chapel window indicating the start of
the third fire.
FIGURE 10.19
FLIR image showing possible fourth fire start in debris area behind the chapel.
FIGURE 10.20
Normal photograph showing smoke obscuration of the compound.
that the visible smoke is much more evident than in the FLIR image of
Figure 10.19. This figure shows the advantage of the FLIR in being able to see
through this smoke.
10.3.4.1.4 Flashover
Following ignition of these fires, the next significant event is flashover,
which marks the transition point of a discrete fire in a room to a fully
developed fire in which flames now fill the room and emerge from the
windows. It is rapid and can take place in seconds. It occurs after the
room is sufficiently heated. It marks the difference between survivable
and nonsurvivable conditions in that room. These events can be seen,
directly and indirectly, from the video records. The first is seen directly
for Fire 1 as shown as window flames appear in the split screen images of
Figure 10.21.
Flashover occurs at 12:09:42, 2 minutes after the start of that fire.
Calculations show that this fire growth rate for the initial burning item
would be rated as “fast” according to NFPA standard 72E. Its energy release
rate would be about 2 MW at flashover, compared to an estimated 50 kW that
was necessary for detection by the FLIR. The detectable fire is like a 1 ft2
spill of gasoline, compared to a 10 ft2 gasoline fire at flashover.
Fires 2 and 3, in larger rooms, grow much more rapidly than the bedroom
fire. Flashover occurs in about 2.5 minutes for the dining room (12:11:07)
and in 4 minutes for the chapel (12:13:49). Figure 10.22 shows the effect of
flashover for the chapel by black smoke that suddenly emerges from the
front opening in the building. This smoke, pouring into the 25 mph wind,
is due to the overpressure caused by the sudden increase of energy associ-
ated with flashover in the chapel.
FIGURE 10.21
Split screen showing a comparison between a FLIR image and a normal video image at 12:09:42.
FIGURE 10.22
Black smoke is pushed out of the front (right) opening as a result of flashover in the chapel.
10.3.4.1.5 Fire Cause

It is concluded that these three fires, occurring nearly at 1 minute inter-
vals, were intentionally set from within the compound. Even if the tank
battering had caused the spillage of fuel from lamps, a match would be
needed to initiate the fire. An electrical spark is ruled out because the elec-
tric power was shut off in the compound. It is obvious that these three fires
needed an ignition source deliberately placed in each of the three loca-
tions. Also, none of these three fires could have caused any of the others
because their growth rates would not provide sufficient heating to cause
such remote ignitions. Any external heat source that might have been used
to start the fires would have clearly been visible on the infrared video.
This heat source was not seen. Although normal furnishings and interior
construction characteristics would provide a means for fire propagation,
the more than usual rapid spread of these fires, especially in the dining
room and chapel areas, indicates that some form of accelerant was very
likely used.
10.3.4.1.6 Tear Gas

Methylene chloride, used as a dispersal agent for CS tear gas (2-chloro-
benzalmalononitrile), is flammable as a vapor at a concentration of 12%
in air; however, it is not easily ignited as a liquid. In fact, it will put out a
match on attempting to ignite the liquid. Although fire spread was rela-
tively rapid in the compound, these rates are not indicative of the much
more rapid propagation that would be associated with a flammable mix-

ture in air. Those rates would be in excess of 2 ft/s and would be seen as a
fireball moving through the atmosphere of the interior of the compound.
No such characteristics were observed in this early fire growth.
Recently, I conducted additional experiments to access the role of
methylene chloride as a wetting agent to available fuel types in the com-
pound, such as wood and paper. Since methylene chloride is a liquid at
normal temperatures, it could have been absorbed into the furnishings of
the compound. From my experiments, I can conclude that the methylene
chloride had no enhancement effect on the fire spread over room furnish-
ings. Also, I can conclude, from the flashpoint data (197°C or 387°F) of
CS itself, that its deposition on furnishings should not have had a signifi-
cant effect on fire propagation either. The tear gas had no bearing on the
propagation of this fire.
10.3.4.1.7 Wind
Wind effects did have a profound effect on the external fire spread over the
compound. An approximate 25 mph wind from the south caused the fire
plume to be bent approximately 65° from the vertical when the fire fully
involved the compound. It is estimated that the fire was expending 3600 MW
at this time with an observed length of approximately 240 ft.
Wind effects did not appear to have had a significant effect on the fire
growth within the compound, as seen in Fire 1 where flames and smoke
emerge periodically from the right tower windows into the wind. This
effect could have been as a result of closed doors or windows on the down-
wind side of the compound. The tank-made openings on the front of the
compound could have had some effect on fire growth over the first floor,
but more significantly could have provided air to areas of refuge for some
of the occupants.
10.3.4.1.8 Survivability
It is estimated that the occupants would have had sufficient warning of the
fire to enable them to escape, for at least up to 5 minutes from its incep-
tion, and up to nearly 20 minutes in some more protected locations. This is
dramatically indicated by one occupant who jumps from the second floor
12 minutes after the start of the fire. Although smoke would have impaired
visibility, exits were within 30 ft of most occupants, with additional openings
made by the battering tanks.
Carbon monoxide in the smoke would have been the primary threat to
the occupants. However, preliminary autopsy reports made available to me
indicated that only 5 of 31 victims with recorded CO levels had lethal levels
of carbon monoxide (CO). The remaining 26 victims, with recorded CO data,
stopped breathing before lethal CO levels were attained. If these data are
correct, at least 26 victims did not die due to the fire. The autopsy report
goes on to indicate that, in at least 27 of the victims, the cause of death could
be attributed to gunshot wounds.
10.3.4.1.9 Concluding Remarks

During the weeks preceding the fire at the Branch Davidian compound,
we were all bystanders to the drama of the standoff and wondered how it
would end. The eventual outcome was a horrible event. In the 2 years since,
many theories about the fire have been proposed; some quite bizarre. I hope
this presentation, our report, and the video I would like to submit will help
explain the events of this fire.
10.3.4.2 Follow-Up to Waco

Following that testimony, I thought that was the end of my involvement.
But in August 1999 I was retained by DOJ to serve as a fire expert in the
civil trial of July 2000. New information came to light. I will not go into
it all. But one event struck me: the rescue of a Branch Davidian woman
who jumped out of the second floor window 12 minutes into the fire. She
then crawled back into the flaming first floor, only to be reluctantly rescued
by the heroic action of a FBI agent. That section of the building collapsed
in flames several moments later. While the rescuing FBI agent was inter-
nally recognized, I do not think that action received any outside public-
ity. Indeed, an immediate criminal investigation of the fire in 1993 never
emerged. It was as if the entire event should be forgotten. Of course, for
some, it was only an inspiration that led to the bombing of the Alfred P.
Murrah Federal Building in downtown Oklahoma City exactly 2 years later
on April 19, 1995. That bombing was the most destructive act of terrorism
on American soil until the 9/11 attacks.
10.3.4.3 Scientific Analyses of Some Aspects of the Waco Fire

Let me conclude the story of this fire with some excerpts of analyses
that illustrate the use of science. In some cases the illustration may not
have been critical to the conclusions of the fire. Consider the following
phenomena.
10.3.4.3.1 Fireball
At about 19 minutes from the onset of the first fire, while the compound was
nearly completely covered in flames, a large fireball was seen, as shown in
Figure 10.23.
FIGURE 10.23
Fireball occurring in the midst of Branch Davidian fire.
From Chapter 7, the equations describing the dynamics of a fireball are

given as (here the units have been changed from cm to m)
Height of fireball: Lb (m) = 12.7[Vf (m3)]1/3 (7.23)
Maximum diameter: Db (m) = 7.7 [Vf (m3)]1/3 (7.24)
Burn time: tb (s) = 2.8 [Vf (m3)]1/6. (7.25)
From the video records and from length scale estimates, the corresponding
results were estimated:
Maximum diameter about 22 m

Height about 38 m
Duration about 3 seconds
The consistency and accuracy of the formulas can be examined. By combin-

ing Equations 7.23 and 7.24, a diameter is computed to be 23.2 m when using
38 m as the height. The observed maximum diameter is 22 m as estimated
from the visuals. Using now 22 m for the diameter, the duration is computed
to be 4.7 seconds compared to 3 seconds from the visual records. Moreover,

accepting 22 m as the diameter, the volume of fuel released can be calculated
as 23 m3. As it was believed that the fireball was the result of a breached
propane tank into the fire, taking the density of propane in normal air to be
1.82 kg/m3, the mass of propane released is 42 kg or about 93 lbs. This is
consistent with a 100 lb capacity tank. There was such a tank outside the
building where the base of the fireball appears to originate. This is quite
a remarkable confirmation of equations developed in the laboratory using
much smaller scale data.
10.3.4.3.2. Wind Effects on the Flames

A curious phenomenon in this fire was that flames exited the window
directly into the wind at the corner room of Fire 1. The wind was nearly
continuous at about 25 mph into that corner of the building as depicted in
Figure 10.20. The only way that could happen is that the door to the fire
room on the second floor was closed, or the second floor corridor had limited
ventilation. This means that the room is being pressurized by the wind, and
the flow acts like normal buoyancy-driven flow with a neutral plane at the
window. A flow analysis of the effect of wind and demolition penetrations to
the building was done and concluded this bidirectional flow of smoke and
air on the wind side of the Fire 1 room was supported as long as it was closed
from the second floor corridor.
For Fire 2 on the first floor in the dining room, the flow analysis predicted
that wind would blow through the large demolition hole at the front of
the building and then out through voids in its rear wall. Indeed, flames
were seen penetrating this rear wall. In addition, this flow analysis indi-
cated that the wind would push the fire in the chapel area (Fires 3 and 4)
toward the demolished gym in the rear. While these effects explained
aspects of the fire and supported their behavior and timing, a calculation
of the overall wind effect on the building’s flames was not so critical to
the investigation.
However, another calculation showed the accuracy of a scientific formula.
Equations by Thomas for wind-blown flames developed for wood cribs of
a finite width give the length of the flame and its angle as a function of the
wind speed and firepower.12 These equations were applied to the building
when flames completely engulfed it. The flame angle from the visual records
indicated it as about 65° from the vertical. For that angle and wind speed, the
Thomas equations indicated a fire of about 4000 MW and a flame length of
64 m. The visual record showed a flame of about 80 m long. This is quite a
consistent result, with again small laboratory-based formulas giving credible
results in real fires.
10.3.4.3.3 FLIR Result

The FLIR was key to this fire in identifying the precise timeline and the
location of the fires. In addition, together with the media coverage, the
indication of full fire involvement could be discerned for the compart-

ment in which each fire originated. Hence, an indication of fire growth
rate could be determined. Analysis showed that the characteristics of
this FLIR would allow a visibility through the early gray smoke of the
fire at 17 times what normal human sight could discern. In addition, the
FLIR would indicate a gray spectrum that moved from a black image to
white for temperatures of about 8°C–56°C, respectively. This meant that
in viewing the first fire from the aircraft height, the floor through the open
window would be in view and when it indicated a white image the floor
would be about 56°C. This level of temperature would result from a fire
very early in its growth and would be small. Following this event, flames
were seen to emerge from another window of the room some distance
away in about 2 minutes later. In this way, the growth rates of all the fires
could be determined.
An interesting estimation of the growth rates of these fires (1–3) was
determined in terms of the t-squared growth time to reach 1000 kW (1 MW),
as explained in Chapter 6. The FLIR and other visuals could allow the time
at inception and the time of flashover by indications of full room involve-
ment. From the room size and its ventilation openings, the firepower at
flashover could be estimated from Equation 9.10. Assuming that these
fires grew as t-squared fires, then this information could be used to calcu-
late the growth rates in that format. Figure 10.24 gives the 1 MW growth
times for the three fires and compares them to other fire growth rates, as
given in Chapter 6. It is clearly seen that these separate distinct fires occur-
ring in the Branch Davidian compound were relatively very fast, and that
speed of growth makes them suspicious and not accidental. The addition
of ample liquid fuels in the compound and evidence given by surviving
Branch Davidians indicating the fires were purposely set makes the scien-
tific results compelling.
10.3.5 Patterns
Patterns in fire investigations have long been an indicator for drawing
conclusions. This pattern analysis for fire investigation begs a scientific
basis. Yet relatively little research has been exercised. Many have come to
realize that such indicators as the spalling of concrete or the deep char-
ring near an open door do not necessarily mean arson and an accelerant.
The student should recall that many liquid fuels, in particular gasoline,
have low boiling points. At those temperatures, a wood floor would be
protected, while the liquid is still burning. Of course, some type of pat-
tern may emerge, but this needs some research to make any discrimi-
nations. The damage of spalling in concrete is due to water trapped in
the concrete and stresses caused by boiling and also by thermal stresses
alone. These stresses break the concrete sometimes in a violent fash-
ion. They are caused by the high heat flux in rooms that go to flashover.
Growth time in seconds to reach 1000 kW

Paper products,
densely packed,
20 ft high
mattress
Mail bags, filled

5 ft high
Large sofa
20 stacked chairs
Wood pallets, 10 ft high
Fire 1
Fire 3
PS jars in cartons
15 ft high
Fire 2
Ultrafast Fast Medium Slow
Gasoline
0 100 200 300 400 500 600

t1, Growth time (s)
FIGURE 10.24
Fire growth rates in terms of time to reach 1 MW by a t-squared rate.
Speaking of flashover, there seems to be confusion among investigators on

what it is. Many claim no flashover occurred in rooms that are damaged
and charred throughout. Perhaps they are working from a different defi-
nition of the term. We should recall flashover is a transition of the fire from
an object or discrete region burning to a fully developed fire controlled
either by ventilation or by the fuel in the room. That accelerated transition
is flashover. Its end result produces much damage and of course patterns.
We now know that ventilation-limited conditions can cause the fire to
seek out the sources of air. Hence, deep burn damage near ventilation
openings occurs.
In the end, fire leaves patterns and they may have forensic value, but
this process must go beyond the reading of tea leaves. Let me introduce
some ideas on patterns from cases or fire experiments that I have experi-
enced. In some instances, they are hypotheses that need research.
10.3.5.1 Soot Patterns

We know that soot is produced on the fuel side of a diffusion flame, and in
turbulent flames, it will break out of the flame and move away with the smoke.
As it is born by a given fuel, it may carry a chemical signature of that fuel.

I know study on tracing the soot chemistry to the fuel has been done in China.
Maybe this will be a fruitful path for forensics someday. Let us look to the
physical nature of soot.
The deposition of soot on surfaces commonly occurs in fire. It can occur
due to turbulent impaction and settling due to gravity. However, the main
mechanism for its deposition is thermophoresis. This process is stronger
than turbulent diffusion alone. It is a process where soot particles move
from a high to low temperature. The mechanism is much like heat moving
from high to low temperature, but in this case the soot particles move too.
In the early stage of a fire, when the surfaces are cooler than the smoke,
the soot deposits by thermophoresis. Later, as the surfaces heat up, or the
fire dies, little more soot is deposited. In the area of the early fire before
flashover, the flames may strike a wall. But we know from Chapter 7 that
about 10 times more air than needed to burn the fuel is entrained into the
flame. This means the hot turbulent flame is rich in oxygen. As the flame
strikes, the wall where soot is deposited, the high temperature and avail-
ability of oxygen can now burn the soot. Hence, a clean burn mark can
occur on gypsum walls as the soot is burned off. Now as the fire moves to
flashover, and ventilation effects take the flames away from this area and
its oxygen drops, the clean burn pattern can survive the fire, especially if
suppression intervenes.
10.3.5.2 Clean Burn Pattern

Figure 10.25 shows the onset of a bedroom fire experiment at a clothes-
basket against a gypsum board wall. The aftermath of that fire involv-
ing the full bedroom and into the rest of the house leaves a clean burn
where the fire began. This is common in fire experiments as such a clean
burn can be definitively traced to the origin area. The oxidation of steel to
rust might be another indicator of early fire growth. These indicators need
FIGURE 10.25
Bedroom fire leaving a clean mark on the wall at fire origin.
study to confirm and identify their meaning to fire growth. As we know

after flashover the flame can move about the room in a ventilation-limited
fire. This can produce additional clean burns as the soot is burned off. If
this hypothesis is correct on the formation of clean burns, then the clean
burn farthest from the room vent is likely to be at the origin of the fire.
10.3.5.3 The Hands of Time in a Fire

An unusual pattern occurred in a scenario of a sofa fire in a small apart-
ment. Because no doors or windows to the outside were open or broken
during the fire, the apartment filled with smoke and consumed the oxy-
gen to a level where the fire went out. The fire was set to cover up a
murder in the apartment. A man was later tried and convicted. A curious
aspect of this fire was the soot on a clock in which, for lack of a glass
cover, indicted a clean image of the clock hands. The clean image on the
clock indicated a time for which the convicted man could not have set
the fire.
To investigate this further, a fire experiment that mimicked this apart-
ment scenario was conducted. It was prompted by the way thermopho-
resis causes soot deposition. In the experimental fire, that had nearly the
same volume as the apartment fire, the fire began on a sofa at 11:20 and
extinguished from lack of oxygen at 11:25. Its smoke barely reached 300°C,
and the soot was strongest just before the fire died. Several open-faced
clocks were placed on the walls of the room. They all gave similar pat-
terns exemplified by the photograph in Figure 10.26. The hour hand’s clean
image clearly indicates a time of 11:20–11:25. It points to the initiating time
of the experimental fire.
Initially, as the fire on the sofa began, the clock was cold and the soot
deposited. The fire grew and consumed the oxygen, and then died. When
it died, the clock was now at a higher or equal temperature to the smoke.
No more soot gets deposited. As the temperatures never got high enough to
destroy the clock workings, the clock continued to run.
In the murder case, a similar clock image was left on the functioning clock.
Its cause was never traced back to the nature of thermophoresis.
10.3.5.4 Gasoline versus Fire Damage

In another case involving alleged arson and murder, a man was convicted
and recently died in prison. Guilt or innocence is not the question here, but
the nature of the evidence and its interpretation by experts in the trial are
important to examine.
One expert reported that burn patterns in the living room, dining room,
and kitchen were indicative of the use of a liquid accelerant. Core samples
of the concrete floor in the living room revealed the presence of gaso-
line. The investigator determined that the fire had been intentionally set.
FIGURE 10.26
Clock hand image due to soot deposition in a fire began at 11:20 and died at 11:25.
This investigator also eliminated the possibility of a flashover as having

occurred in the apartment.
An expert for the defense claimed that the fire started in the kitchen
on the stove and then spread to the living room as a result of a flashover.
He testified that the burn patterns on the living room floor were the result
of a normal fire and that the gasoline could have been in the concrete long
before the fire.
I came into this fire by a civil case and was drawn to the background
of the murder conviction. In my opinion, the fire began in the kitchen
and flashover surely occurred. Indeed, this fire spread into the hall and
other apartments in the building. The apartment of origin had a fully
developed ventilation-limited fire that had to pass through flashover.
The kitchen had the most damage, but that alone is not sufficient to tag
it as the origin. The telling bit of evidence is that the kitchen had no win-
dows. For the fire to have originated in the living room from gasoline,
with an open door to the hall and sliding glass doors broken by flames to
the patio, means that the flames would move first out of these openings.
Then later it would move back into the closed confines of the kitchen as
the fire burned out in the surrounding areas or as fire suppression would
intervene. The kitchen would have less damage than the living room and
the intervening dining room. But the kitchen had the most damage, and
that could only occur by a fire originating there. Its only source of ven-
tilation was the rest of the apartment, and as the fire became ventilation
limited, the flames would move from the kitchen through the apartment
to the openings.
The extreme disparity in opinions by these investigators is troubling.
One relied on a gasoline signature and claimed flashover never occurred.
The other was unable to articulate that the kitchen damage was due to
ventilation effects on fire behavior. An underlying question is whether a
chemical signature of a liquid accelerant always indicates arson.
10.3.6 World Trade Center Terrorism and Fire (9/11)

We all remember the day of September 11, 2001. We all know the cause of
the fires that brought down the twin towers. Two commercial Boeing 767
airplanes traveling over 500 mph struck each tower. The North (WTC 1) hit
after the South (WTC 2) fell in 102 minutes and the South in 56 minutes.
Later fires in WTC 7 fires by falling flaming debris caused it to fall in about
7 hours. The question of interest to designers and investigators should be
why such notable buildings would not have the inherent fire protection to
sustain these fires and not cause them to catastrophically fail. Let us take a
look at this question.
The WTC twin towers were constructed in the 1960s with a unique
design to open the floor. They might be characterized as a steel “tube”
construction. Its central core columns took most of the gravity load with
sharing by 59 perimeter columns that were mainly to handle the wind
loads. A “hat” truss at the roof allowed for load sharing among the
perimeter columns. The floor system between the perimeter frame and
the central core consisted primarily of a concrete metal deck supported
by a double truss (bar-joist) network, 900 mm deep, with web rods of
27.7 mm (1.09 in.) diameter that spanned to 18.3 m. The main trusses
were cross-braced by similar secondary joists. The entire floor truss sys-
tem was aligned by bolts and welded to seat connections at the perimeter
and core columns. Figure 10.27 shows a plan view of a typical floor with
overall dimensions of the open office and core spaces along with its struc-
tural design aspects.
In an early design analysis, a calculation considered the effect of a Boeing
707 traveling at 600 mph hitting a tower. It concluded that it would only cause
local damage and most certainly not total collapse. This was confirmed by
the slightly bigger 767 planes that hit the two towers; they did not fall under
impact. Of course there was structural damage, but it was the long burning
fires that eventually weakened the steel structure. Why? A lot has been said
on these issues since; perhaps more study is needed.
63.5 m 13 14 15
0
16
10 11
17
63.5 m
24 m
12
11
42 m
FIGURE 10.27
Plan view of typical floor and structural design (number labels: 10—office, 11, 12—lifts, 13—
perimeter frame, 14—bar joists, 15—secondary joists, 17—core columns). (From World Trade
Center building performance study, Federal Emergency Management Agency, FEMA 403,
2002; Quintiere, J.G. et al., Fire Safety J., 37, 2002.)
10.3.6.1 Investigative Efforts

A terrorist attack on the United States fell under the jurisdiction of the FBI
on 9/11. While ATF had a mandate to investigate federal crimes of arson, and
National Transportation Safety Board investigates air crashes, they were not
enlisted to conduct any official investigations. The American Society of Civil
Engineers (ASCE) sent a voluntary team to the WTC site. Later they were
funded by Federal Emergency Management Administration (FEMA) and
produced a good first attempt at the cause.13 Nearly a year after the attack
on the buildings the federal government funded two investigative actions:
(1) the 9/11 Commission14 and (2) NIST.15 The latter organization was to focus
on the fires and building collapse, while the 9/11 Commission focused on all
other aspects. The Commission was eventually funded at $12 million and
completed its work by a mandated constraint in 2 years. It held many formal
open hearings with witnesses under oath. The NIST investigation finished
its report on WTC 1 and 2 four years after the event in 2005 with initial
funding of $16 million. They produced a 10,000-page report. The result took
longer than the U.S. involvement in World War II and cost about $1600 per
page or more. With added funding, NIST completed a report on WTC 7 in
2008. In contrast to the Commission, NIST held open workshops to display
progress and allowed visitors 5 minutes for formal comments and allowed
no questions.
Two books by NY Times reporters are very illuminating on the WTC

event. The first by Glanz and Lipton16 addresses the development of the WTC
buildings up to their collapse. The second by Dwyer and Flynn17 portrays,
in riveting fashion, the actions of those in and around the buildings on 9/11.
I got involved in the investigation in a very nonofficial way. I was not
enlisted in any part of its official investigation or in litigation. I did get an
offer from the ATF agents at Federal Law Enforcement Training Centers
(FLETC) in Georgia to drive through the night with them to the Pentagon on
9/11. I was about to teach a course on fire dynamics. Instead I opted to fly
back the next day. Alas, all airports were shutdown, and they opted to stop
and sleep in Motel 6 that night. I spent about a week waylaid in Tampa, FL.
The ATF only remained at the Pentagon site for several days and were
not requested to go to NYC. Meanwhile, I settled into business, and
learned I had been volunteered for the backup ASCE team. I never went to
the site, but I soon learned that the steel was being carted off to the dump
with no attention to forensics. I was appalled. As I spoke out, I attracted the
attention of the Skyscraper Safety Campaign founded by Sally Regenhard
and Monica Gabrielle, who were members of the 9/11 families, losing a son
and husband, respectively. They were bent on determining the cause of the
collapse. And they were instrumental in their crusade in securing the NIST
funding from Congress.
First, drawn by curiosity to see if the formulas of fire science could shed
light on the collapse, we published a paper.18 I enlisted the assistance of
Dr. Rachael Becker of the Technion. She had studied the fire resistance
of steel structural members and the role of its insulation. In 2000, she
had been a guest worker at NIST and in our department at University of
Maryland. We could only do a sketchy job, having no full information.
Later I went deeper as the information from the FEMA report became
available,19 and finally, I published a document challenging aspects of the
NIST investigation.20 I will not go into all the details, but I will take you to
places where formulas and information in this text can lead you to make
your own analyses.
Recently, NIST published the results of the WTC investigation in a peer-
reviewed journal. This opened their analysis and conclusions to public
comments. After a Letter to the Editor of the journal was redacted beyond
recognition, it was submitted instead to another fire journal. The authors
of the Letter were myself and Professor F. A. Williams, a noted combustion
scientist who had served on the NIST WTC advisory committee.
In short, our conclusion is that the bar-joist floor trusses failed due to
lack of sufficient insulation as specified in the design; NIST concludes that
the trusses had sufficient insulation and that the aircraft dislodged criti-
cal insulation on the core columns and allowed them to weaken by the fire.
The reader can examine the Comments to NIST and their Response in the
Forum section of the Journal of Fire Sciences.21
Now, let us examine some aspects of this event in terms of formulas dis-
cussed herein, and factual information supported from References,19,20 and in
general from the FEMA13 and NIST reports.15
10.3.6.2 Role of the Jet Fuel

Many laypeople today believe the jet fuel played the dominant role in the
fires. It was the igniter, but not the fuel for the fire of long duration. On impact,
fireballs erupted. There were about three or four emanating from the sides of
a tower. Anyone can find a dramatic photograph of this event and then scale
the size of the fireballs. We have in Chapter 7 a formula that relates maxi-
mum diameter to initial fuel volume. Making an estimate of the bulk den-
sity for the fuel in the dispersed spray at the impact, the fuel burned in the
fireballs can be determined. From Chapter 7: Db (m) = 7.7[Vf (m3)]1/3 and with
an aerosol density of 2 kg/m3 gives about 1000 kg burned in a single fireball.
Considering four fireballs, that is, 4000 kg burned outside. Each aircraft
carried about 10,000 gal of jet fuel or about 28,500 kg. The remainder of the
fuel, about 24,500 kg, burned in the building. The maximum burning rate
per unit area for JP-4 is taken as 60 g/m2-s from Table 6.2. Over a single floor
of office area of about 3000 m2, this results in about 181 kg/s. This means
that the jet fuel could burnout if just spilled over one floor in 136 seconds
(24,500 kg/181 kg/s). It likely takes longer due to ventilation and transient
burning effects, but the result clearly indicates that the jet fuel is short lived.
If a spill over several floors is considered, less fuel per floor means even a
shorter burn time. By all accounts, and confirmed in the NIST analyses, the
jet fuel is the starter only. Indeed, it acted as the igniter and an extensive
enough igniter to eliminate the normal fire-developing phase of traditional
building fires. Hence, it can be concluded that large fully developed fires
immediately occurred on the impacted floors of the towers.
10.3.6.3 Fuel Load for the Fire

Now, the question turns to what did burn. In this case, it is the contents of
those floors impacted and set on fire by the jet fuel. NIST concentrated on
reconstructing the fuel load of the Marsh McLennan offices struck in WTC 1.
They used the same fuel load for the trading floors struck in WTC 2. From
Table 9.2, the average fuel load for an office building is 39.2 kg/m 2, but with
large a standard deviation of 34.7 kg/m 2. NIST obtained an estimate for
the Marsh McLennan space in the WTC from architectural drawings. I was
able to obtain the same from a 9/11 Family source. Kate Stewart, a student
at the time at Maryland, who was very skilled in building plan review,
examined the plans and computed a load. She contacted the furniture sup-
pliers for their weight information, estimated transient materials in a typi-
cal office, and discovered NIST did not include a ring of common metal
file cabinets. She computed the furniture load without the common files
at 61,054 kg compared the same furniture computed by NIST at 60,643 kg.

This gave us confidence in our accounting methodology. There were 170
4-drawer files left out by NIST. We computed with only 50% paper capacity
in the files that they constituted another 75,000 kg per floor. So the NIST
analysis used a fuel load of 19.5 kg/m 2 of flow area, while we think there
was at least 44 kg/m 2. NIST apparently only used the architectural draw-
ings as their source information. I talked to people who had been on the
Marsh McLennan floors before the fire. A manager told me they were paper
hogs, and a security guard said that stacks of paper covered the window-
sills. Some at NIST told me paper does not burn in the file cabinets. So I
did a test with a file in a furnace at typical fire temperatures; after 2 hours
about 60% of the paper was consumed with only ash and char remaining.
The importance of the fuel load in an investigation of fire involving struc-
tural damage is that the duration of the fire and its temperature control the
response of the structure. The longer the fire, the more vulnerable is the
structure. Let us make a rough estimate here. In Table 6.2, the maximum
burning rate of wood cribs per floor area (they sit on) is listed at 11 g/m2-s
for burning in the open. Assuming the furniture and paper burned as a
crib fire, the NIST fuel load would give a fire burning time of 0.48 hours
(19.5 kg/m2/0.011 kg/m2-s). Our larger estimate gives 1.08 hours. Ventilation
effects would make these estimated times longer, and we know the fires
lasted on a floor for up to the collapse time of WTC 1 at 102 minutes. The
importance of the fuel load should be clear, and relatively simple estimates
can indicate the duration and temperatures in the fire.
10.3.6.4 Fire Effect on the Structure

In the case of a steel structure, it must not exceed 500°C–600°C, as its
strength diminishes dramatically at those levels. Indeed, this temperature
level is usually a criterion of failure in standard furnace tests. These furnace
tests subject a particular structural element with its applied insulation to
a severely increasing temperature and then rates its failure point by time.
When the World Trade Center was built in the 1960s, the required rating by
such testing was relaxed from 3 to 2 hours. Most tall buildings built around
the world today require 2 hours. Many argue that the standard furnace tests
do not represent a real fire, but the test temperatures typically achieve 600°C
quickly, 800°C in less than ½ hour, and 1000°C in less than 1 hour. The WTC
fires are surely represented by these temperatures. So the rating by standard
tests has some meaning in reality.
Good engineering practice would be to first determine the fuel load of
occupancy, estimate its burning time, and specify a rating in hours for the
structure that exceeds the burning time. This can be done by formulas,
computer modeling, or by test data. This would be the performance design
approach. Usually, regulations require a rating, and for the WTC, it was
2 hours for the floor trusses.
Something unusual happened in the life of the WTC towers. The required
insulation on the steel truss rods of the trusses of about 1 in. in diameter
(27.7 mm) was changed. How this happened never came out in the NIST
investigation, but it is known that the required thickness was changed in
the mid-1990s, nearly 30 years after the towers were built. This change was
implemented as floors became available due to changing tenants. The change
made was to increase the original specification of ½–1½ in. of thickness for
the spray-on insulation. Apparently this was changed on examining recently
tested structures representative of the WTC trusses. Why such testing on the
actual structural elements was not done in the design of such an unusual
building is unknown.
The NYNJ Port Authority (PA), responsible for the WTC, did inspections of
the sprayed on insulation. Their reporting to NIST indicated that the original
½ in. specified was about ¾ in. applied, and the 1½ in. was more like 2¼ in.
applied. NIST used the applied levels in their calculations. I used ¾ in. in
mine but stuck with 1½ on the change, as I found it difficult to believe that
a spray-on company would add so much more. The PA specified amount
would result in a 4 in. diameter, while the applied PA amount would have a
diameter of 5½ in. No photographic evidence of the newly applied insulation
was ever presented in the NIST report.
Here is the punch line to all of this. The Marsh McLennan floors in WTC 1
had the newly applied insulation to the trusses. The fire area of the South
tower (WTC 2) had the original insulation. It is simple to wonder if WTC 2 had
¾ in. and WTC 1 had 1½ in., then WTC 1 might fall in twice the time. Well,
WTC 2 fell in 56 minutes and WTC 1 fell at 102 minutes. Indeed, I will leave
you with the results computed in Reference19 using formulas for to predict the
time the steel rod of the truss would achieve a failure point of 600°C with the
different amounts of insulation. The formula accounts for conduction through
the insulation with all of that heat to uniformly raise the temperature of the
steel. The fire temperature is nearly the temperature of the outer insulation,
as the heat resistance of the insulation dominates the heat transfer from the
fire to steel. The calculation specified the fire temperature at 800°C, likely low
for this fire. A heavy core column was also included, and the effect of losing
the insulation was examined. Table 10.1 gives a summary of the results for the
different truss rod insulation thicknesses and also a core column.
It can be seen from Table 10.1 that the computed times for the trusses at
the fire floors of WTC 1 and 2 give times consistent with the collapse of the
buildings. The calculations support the fire as the cause of the collapse. The
failure of one rod is manifested as the sagging of the trusses over the floor.
The connections at their ends cannot tolerate this new loading configuration
with the steel in a fire-weakened state. As the connections fail, the floors
begin to fall in total or in part. In the Glanz and Lipton book,16 they cite
observations of “sudden smoke puffs” out of the fire floors up to 10 minutes
before total collapse of the buildings. I interpret this as local floor failures.
As one or two floors fell on to the next, the connections at the truss ends were
TABLE 10.1
Heating Times for Structural Elements
Insulation
Thickness Time to Reach Time to Reach
600oC with 600oC with No
Element mm in. Insulation (min) Insulation (min)
27.7 mm steel rod 12.7 ½ 55 8
27.7 mm steel rod 19.1 ¾ 73 8
27.7 mm steel rod 38.1 1½ 111 8
55.8 cm box column with 28.6 11 8 1640 75
7.6 cm thick steel
Source: Quintiere, J.Q., A predicted timeline of failure for the WTC towers, in: Proceedings
of the 10th International Conference on Interflam 2004, Interscience Communication,
London, U.K., 2004, pp. 1009–1022.
not strong enough to carry the weight even for unaffected fire connections.
A domino effect resulted.
10.3.6.5 Afterthoughts on WTC

In any investigation, one cannot know if truth prevails in the conclusions.
It is interesting to observe that the NASA Columbia mission fell to Earth
in February 2003, and its accident report was concluded in August 2003.
This was a fire too; the insulation protecting the shuttle on reentry was dam-
aged on launch. Scattered debris, flight evidence, recording evidence, and
definitive tests concluded the cause was the damage at launch. In the TWA
800 crash (July 17, 1996), a fuel tank fire was found to be its cause. It took a
reconstruction of the debris raised from the Atlantic Ocean to piece back
the evidence for supporting its cause. In both cases, actions were taken to
prevent accidents in the future. NASA revised its launch procedures with
attention to the effects of foam insulation on the cryogenic fuel tanks; the
aircraft industry developed fuel tank inerting systems. In the case of WTC,
the NIST conclusion was that the impacting aircraft dislodged key insulation
on the core steel columns, and the insulation specified on the trusses was
sufficient. So while NIST made many recommendations for improvement,
the issue of the floor truss insulation was not a consideration. Was the insuf-
ficient insulation the cause? Why did such a significant building not have
more attention paid to its structural fire protection? Why the changes in the
insulation? Why the discarding of steel from the site into new products from
abroad and memorial artifacts? Why not a reconstruction with the steel from
the fire floors (as the primary structural elements were marked to location)?
If steel is examined by metallurgical testing after a fire, its maximum tem-
perature reached in the fire could be determined with good accuracy. NIST
did this late in their investigation, but having only anecdotal limited steel,
the tests were insignificant.
My intent here has been twofold: first to cast doubt on the NIST findings,
and second to demonstrate that formulas developed from fire research can
produce information even in a complex event as the WTC fires on 9/11.
10.4 Computer Fire Models

Computer models are mathematical solutions implemented and displayed
on computers. They attempt to alleviate the user from the direct develop-
ment of the mathematical solution, and usually make it convenient and easy
for the user. In addition, if they are very easy to use, the user may not need to
know anything about fire behavior to get a result. The same issue can apply
to use of equations for a particular fire phenomenon as given in this book.
The user need only know how to compute the algebra. In the end, compe-
tency requires understanding of the application as well as skill in imple-
menting the model.
In all models for fire, the models are approximate. There are no exact solu-
tions in the complex area of fire. The mathematics offers an approximate
solution to a physical problem, either through an equation or by a computer.
In this book, the equations have been developed from theory of the phenom-
enon, data from or confirmed by experiments, and then the formulation of a
generalized formula. In a computer model, these equations may be used as
part of the model to synthesize many phenomena together. Alternatively, a
computer model may be based on the fundamental equations representing
conservation of mass, momentum, energy, and subsidiary equations govern-
ing properties and other effects.
10.4.1 Zone Models

The first type of computer model used in fire was a system model, often
called a zone model because it treats fire in a room as upper and lower zones
exchanging mass by the fire plume and at vents (see Figure 9.6). The model
by Tanaka et al.,1–3 and illustrated in Figures 10.2 and 10.3, is representative
of a zone model. This type of model is based on fire flows being stratified in
a room and on known physics of fire phenomena. For example, a pressure-
difference model governs flow at vents, an entrainment model governs flow
into the fire plume, heat transfer to surfaces is based on specific equations
for convection and radiation, and so on. As more phenomena are known or
required, they can be folded into the system model. Sometimes the folding-in
involves approximations. For example, the fire is assumed to have no volume
in the two zones, but separately its volume can be computed. One may think
this is not so good, but a similar technique can be used for calculating drag
over an airfoil. The pressure distribution is computed with no regard for
friction in the boundary near the surface of the foil. Then that pressure
distribution is used to compute the friction at the boundary.
Such system models are used in many fields, and they give an overview
based on the component phenomena that can be mathematically expressed.
They form commonly ordinary differential equations in time. Quantities in
each zone or boundary are computed as a function of time. The computing
process requires the speed and storage capacity of a computer. As its compu-
tations only deal with time, it can be a fairly fast process. In addition, such a
system model can be extended beyond rooms in a building. One application,
illustrated in Figure 5.10, shows how the fire plumes and emitting firebrands
from burning wooden houses can be synthesized to model fire spread in a
conflagration. In Japan, such an approach has been used to assess the risk of
fire to people and buildings in fire following earthquake. A dramatic appli-
cation has been to model the fire and its effect on people seeking refuge
following the Great Kanto earthquake and fire of 1923.22 A conception of the
model is shown in Figure 10.28. The model considers the growth rate of the
fire over the houses, the effect of radiation, and convection heat transfer to
people moving to areas of lower hazard potential and refuge sites. Statistical
factors are also considered in the model.
The accuracy of the model simulation is indicated by its prediction for
the number of people reaching the various refuge sites compared to those
estimated by a survey taken in 1923. These results are shown in Table 10.2.
It should be noted that most of those reaching the Hifukusyo Ato site were
killed by the large fire whirl (see Chapter 1).
Zone models are only as good as the system of equations coupled together.
Again, the user must look under the hood to understand such a model.
The model led by Tanaka2 is popular in Japan but does not have a wide distri-
bution. A CD disc comes with the original reference that explains the model and
Fire plume axis
U∞ Temperature rise due to fire plume

ΔT0
ΔT
Fire
Fire plume
qR̋ exposure
Thermal radiation
Evacuation site
FIGURE 10.28
Model concept for the effect of fire, following earthquake, on people in evacuation to refuge
sites. (From Nishino, T. et al., A study on the estimation of the evacuation behaviors of Tokyo
city residents in the Kanto earthquake fire, in: Proceedings of the Ninth International Symposium
on Fire Safety Science, 2008, pp. 453–464)
368
TABLE 10.2
Number of People Reaching the Refuge Sites in the Kanto Fire
Number of Fukagawa Imperial Yasukuni
Evacuated Person Ueno Park Asakusa Park Landfill Hifukusyo Ato Seicho Park Palace Park Shiba Park Park
Simulation 174,100 70,880 52,020 40,290 20,010 215,430 56,020 26,660
Survey3,4,9,10 500,000 70,000 50,000 40,000 20,000 300,000 50,000 50,000
Source: Nishino, T. et al., A study on the estimation of the evacuation behaviors of Tokyo city residents in the Kanto earthquake fire, in: Proceedings of the
Ninth International Symposium on Fire Safety Science, 2008, pp. 453–464.
Principles of Fire Behavior
allows for its use. A more readily available zone model developed by Colleen A.
Wade, called BRANZFIRE, can be downloaded at http://www.branz.co.nz/
cms_display.php?sn=74&st=1&pg=9456. It is a well-supported code.
10.4.2 Field Models

The second type of computer fire model is termed a field model. It solves for
every point in the domain of the problem and advances the results over time.
These field models are generally called computational fluid dynamics mod-
els. Usually, they use the fundamental equations of physics and chemistry
in an exact mathematical form. The form is partial differential equations in
time and three spatial coordinates. The partial differential equations must be
solved. They cannot be solved into formulas; they can only be approximately
solved by repeated numerical computations. This requires large comput-
ers. The first step in this approximate solution is to divide the domain space
into finite cells. Then the model predicts temperature and other variables for
each cell and steps forward in time over specified increments. The mathe-
matical cell approximates results for ideally each point. The cell size may be
larger than phenomena that the model purports to predict. When this is the
case, the model must adjust with special provisions. Unfortunately, issues
of turbulence and combustion occurring at millimeter scales are not fully
resolved; to get them correct requires special models that are not perfect and
the subject of continuing scientific development.
A popular fire field model is the fire dynamics simulator (FDS). It is described
in literature and can be downloaded from NIST along with the computer code
in a user-friendly rendition. It was released in 2000 and is currently being
used by several thousand people around the world for a wide range of appli-
cations from design to investigation. A review paper on the model describes
its limitations and its ability.23 It cites three issues with such a model of fire: (1)
fire has many scenarios, models cannot do all; (2) the computational power
of computers is limited, but always improving; and (3) models are needed to
predict how real objects burn, and this is not possible today. A big limitation
is the ratio of largest to smallest scale that is needed to model all fire phe-
nomena. Fire in a building may need 10’s of meters to indicate geometry, and
predicting combustion from fundamentals requires a cell size smaller than a
millimeter. This requires a ratio of scales, largest to smallest, of about 100,000.
To capture all of this range is impossible with current computers. So in the
rendition of large building spaces the cell size is practically selected to be of
the order of a meter or slightly less to get the code to complete the calcula-
tion in less than a day or two. This means that unless phenomena such as
combustion that occur at much smaller scales are not dealt with in special
ways to insure their accurate prediction, the results pertaining to combustion
and other small-scale phenomena can be poor. Despite these limitations such
codes can be very powerful in giving a detailed output for the fire and smoke.
An example is its ability to compute the turbulent flame shape over intervals
FIGURE 10.29
FDS prediction of a 1-meter-diameter methane turbulent flame. (From McGrattan, K. et al.,
Int. J. Comp. Fluid Dyn., July 2011.)
of time as shown in Figure 10.29. The model captures the dynamics of the
turbulent flame and its oscillatory nature very well. On average, its predicted
flame height agrees well with the formula in Chapter 7, and that formula cor-
relates a wide range of experimental data.
Another field model under development is FIREFOAM by FMGlobal.
Its goal is to predict commodity fires in warehouse settings and their sup-
pression by sprinklers. Its progress to date has been very impressive.
Modeling of fire by computer is essential to get all of the combined effects
of fire. Smoke flow through a building can be modeled, fire growth aspects
can be considered, and toxic and thermal effects on people evacuating
over time are possible. But all of this must be done with understanding
and validation. The user cannot just be skilled at using the computer. This
book has attempted to give some understanding of fire and its effects to a
student with limited mathematical and scientific training. Today, people
with those limitations are likely to be more endowed with computer skills.
To them I say, learn and know what you are using when considering
fire models.
10.5 Summary
A consideration of fire modeling in applications to fire safety design and fire
investigation has been discussed. Generally, fire safety is dealt with by reg-
ulations, and equivalency principles allow an alternative with engineer-
ing analysis. Performance-based design is being used as an alternative to
regulatory specified rules. In a performance-based method, formulas and
computer models seek to achieve a safe design based on a quantitative
benchmark. Many of the principles and processes quantitatively described

in the previous chapters can be used in various design or investigative prob-
lems. In addition to the formulas considered in this book, computer models
offer an approach to fire modeling that may require no understanding of
fire. It should be apparent to achieve a competent level of fire modeling, and
knowledge of the subject needs to be mastered. Modeling by computer or
formula needs competency. Several examples of fire investigation were pre-
sented including the Branch Davidian fire near Waco, Texas, of April 19, 1993.
The examples try to illustrate the approach to investigative analysis and to
show that formulas can give credible predictions. In both design and inves-
tigation, perfection in prediction is not essential; ballpark estimates are usu-
ally sufficient. Predictions in terms of seconds, minutes, or hours, and mm,
m, and km are usually good enough to exclude a fire scenario or set limits
in design. The example of the Waldbaum fire is a good illustration to show
how order-of-magnitude estimates can explain a timeline that extended over
hours. The example of the WTC 9/11 events speaks both to issues in design
and investigation. Computer models have been discussed in terms of the
system-type zone model represented by the Japanese BRI-2 and the field
modeling approach represented by FDS.
Review Questions
1. What is performance-based design?
2. Do models need to be verified and validated? What does this mean?
3. What is a zone model? What is a field model?
4. How did the example of the fire in the laundry basket indicate arson?
5. How was the fire from the bombing of the WTC in 1993 different from
the fires in 2001?
6. Do fires leave patterns that have meaning?
Activities
1. Consider one or two examples from your fire experience to analyze. Have
the class discuss these and work up calculations to quantify features of
the fire scenario.
2. Identify computer models sold or made available to the fire community
such as FDS and BRI-2. Investigate their contents. Discuss.
3. Identify regulations in fire safety that could be enhanced by analyses.

Discuss.
4. What additional aspects of fire safety would you like to see computed?
Are these formulas available?
5. Consider each of the examples presented in the chapter. Raise questions,
comments, and other calculations to challenge or defend the conclusions.
6. Discuss the investigative approach to the 9/11 WTC collapses and the
design of fire protection in those buildings.
References
1. T. Tanaka, A model of multiroom fire spread, NBSIR 83-2718 (Gaithersburg,
MD: National Bureau of Standards, August 1983).
2. T. Tanaka and S. Yamada, BRI2002: Two layer zone smoke transport model,
Fire Science and Technology, 23, 1 (2004): 1–44.
3. J. Yamaguchi, T. Fujita, T. Tanaka, and T. Waka-matsu, A study on predicting
smoke transport in a high-rise building (in Japanese), in Proceedings of the Annual
Meeting (Japanese Association for Fire Science and Engineering [JAFSE], Tokyo,
Japan, May 1995).
4. SFPE Engineering Guide to Performance-Based Fire Protection, 2nd edn. (Society of
Fire Protection Engineers, Bethesda, MD, 2007).
5. Verification and Validation of Selected Fire Models for Nuclear Power Plant
Applications (NUREG-1824), Nuclear Regulatory Commission, Washington,
DC (2007).
6. H. E. Nelson, Engineering analysis of the early stages of fire development—The
fire at the Dupont Plaza Hotel and Casino—December 31, 1986, NBSIR 87-3560
(National Institute of Standards and Technology, May 1987).
7. T. J. Ohlemiller, Smoldering propagation on solid wood, in Proceedings of
the Third International Symposium on Fire Safety Science, edited by G. Cox and
B. Langford (London, U.K.: Elsevier Applied Science, 1991), pp. 565–574.
8. B. J. Hughes, Private communications of statements and records, Office of the
District Attorney, Kings County, Brooklyn, NY, 1994.
9. J. G. Quintiere, Estimating fire growth on compartment interior finish materials,
Honors Lecture, SFPE Engineering Seminars, Society of Fire Protection Engineers,
Spring Meeting, San Francisco, CA, May 1994.
10. J. C. Danforth, Interim Report to the Deputy Attorney General Concerning
The 1993 Confrontation at the Mt. Carmel Complex, Waco, TX, July 21, 2000,
Pursuant To Order No. 2256-99 of The Attorney General Special Counsel.
11. J. G. Quintiere, Statement on Matter of the Branch Davidians near Waco, Texas
before the Subcommittee on Crime of the Committee on the Judiciary and
the Subcommittee on National Security, International Affairs and Criminal
Justice of the Committee on Governmental Reform and Oversight, House of
Representatives, First Session, 104th Congress, July 26, 1995.
12. P. H. Thomas, R. W. Pickard, and H. G. Wraight, On the size and orientation of

buoyant diffusion flames and the effect of wind. Fire Research Notes 516, Fire
Research Station, Boreham Wood, U.K., 1963.
13. World Trade Center building performance study, Federal Emergency
Management Agency, FEMA 403, FEMA Region II, New York, NY (2002).
14. The 9–11 Commission Report, Final Report of the National Commission
on Terrorist Attacks Upon the United States, Official Government Edition
(August 2004).
15. Final Report of the Federal Building and Fire Safety Investigation of the World
Trade Center Disaster, Drafts for Public comment, NIST NSTAR 1, NIST, DoC
(September 2005).
16. J. Glanz and E. Lipton, City in the Sky: The Rise and Fall of the World Trade Center
(Times Books, Henry Holt and Co., LLC, New York, NY, 2003).
17. J. Dwyer and K. Flynn, 102 Minutes The Untold Story of the Fight to Survive Inside
the Twin Towers (Times Books, Henry Holt and Co., LLC, New York, NY, 2005).
18. J. G. Quintiere, M. di Marzo, R. Becker, A suggested cause of the fire-induced
collapse of the World Trade Towers, Fire Safety Journal, 37 (2002).
19. J. Q. Quintiere, A predicted timeline of failure for the WTC towers, in Proceedings
of the 10th International Conference on Interflam 2004 (London, U.K.: Interscience
Communication, 2004), pp. 1009–1022.
20. J. G. Quintiere, Questions on the WTC investigation, Urban Structures Resilience
under Multi-Hazard Threats: Lessons of 9/11 and Research Issues for Urban Future
Work (the Netherlands: Springer, 2005).
21. J. G. Quintiere and F. A. Williams, Comments on the National institute of
standards and technology investigation of the 2001 World Trade Center fires,
Journal of Fire Sciences, 32 (May 2014): 281–291.
22. T. Nishino, S.-I. Tsuburaya, K. Himoto, and T. Tanaka, A study on the estimation
of the evacuation behaviors of Tokyo city residents in the Kanto earthquake fire,
in Proceedings of the Ninth International Symposium on Fire Safety Science (2008),
pp. 453–464.
23. K. McGrattan, R. McDermott, J. Floyd, S. Hostikka, G. Forney, and H. Baum,
CFD modeling of fire, International Journal of Computational Fluid Dynamics,
26 (July 2011): 349–361.
Appendix: Mathematics of Science
Learning Objectives
Upon completion of this section, you should be able to
• Appreciate why mathematics is needed for science

• Be able to perform algebraic operations
• Be able to read a graph
• Know the importance of units and how to convert to SI units
A.1 Introduction
This appendix has been included to provide a brief review and introduction
to algebra. It is to help the student with algebraic symbols and operations.
In addition, it will illustrate how algebraic equations come about in science.
The ability to manipulate equations requires a basis in algebraic principles
and the ability to use consistent units for physical quantities. Hopefully,
this addition will give the student a better learning pathway to the material
in the text.
Science is the organization of knowledge into general laws and princi-
ples that apply over a broad range of conditions. For example, on the Earth,
rules may apply, but when traveling in space, they may be slightly different.
This deals with your frame of reference, geometrically and in time. Albert
Einstein showed time can be distorted, while Isaac Newton treated time as
invariant. In other words, we might shorten time (and aging) under high-
speed interplanetary travel, but on Earth we all age with the same time
standard. As we are addressing time mainly on Earth for fire considerations,
we do not have to concern ourselves with Einstein’s time stretches. However,
conditions on the Moon or Mars will subject fire to other gravities, and in an
orbiting vehicle no gravity is experienced. Scientists are now studying how
to expand the material in this book to those non-Earth-bound gravitational
conditions.
375
376 Appendix: Mathematics of Science
A.2 Mass and Energy Concepts

Now let us return to science and only consider the rules that apply on the
Earth. First, let us address mass and energy. While these are common terms,
they are not easily defined because they are concepts. Yet all matter contains
mass and energy. Suffice it to say that mass pertains to a measure of the quan-
tity of matter. Mass can be considered as that amount of matter expressed as
weight; weight being the force necessary to suspend an object of matter in
the Earth’s gravitational field. Energy might be related to the amount of work
it takes to raise an object of matter to a higher elevation on Earth. Work itself
is a form of energy defined as the weight of the object times the distance it
is raised. Work is a very tangible quantity involving a force (weight) and a
distance. Energy in general may not be so tangible. It has many forms, and
its identification with work can be simply a way to recognize that it is present
in some particular form. For example, we might have energy due to motion
(kinetic energy) or that due to the flow of electrons in a wire. The modern
physicist makes no distinction between mass and energy. Indeed, 1 Joule (J)
of energy is equivalent to 1.11 × 10−17 kilograms (kg) of mass. This follows
according to Einstein who related them through his groundbreaking equa-
tion E = mc².
Most have heard of this equation and here we have our first scientific law
given as an equation. The letter E stands for energy and m stands for mass—this
seems logical. And c is the speed of light in a vacuum, 299,792,458 m/s (m stands
for meter, and s stands for seconds) to be exact. We might also approximate this
large number as 300 million m/s. That is the same as 3 with 8 zeroes following
it, or 3 times 100,000,000, or 3 × 108. Einstein’s equation justifies why 1 Joule (J)
of energy is equivalent to 1.11 × 10−17 kilogram (kg) of mass as
m = E/c2 = 1 J/c2 ~ 1 J/(3 × 108 m/s)2 = 1/(9 × 1016) = 1/9 × 10−16 = 1.11 × 10−17 kg
If this is not understandable, hopefully by the time you finish this chapter
it will be clearer. First we need an appreciation of mathematics in science;
then we need to learn how to manipulate equations into solutions.
In the representation of the speed of light, meters per second were used as
its measure. Others might prefer different units as 670,616,629 mph (mph is an
abbreviation for miles per hour). Ordinary people in the United States would
better relate to the mph units, but most scientists use the metric system, or
more precisely the Standard International (SI) system of units. Many coun-
tries, with the notable exception of the United States, use the SI units in com-
mon practice. Now here we see that scientific terms have units, and the units
are somewhat arbitrary and vary with tradition and country. But of course,
without equivalence between the various units of measure for the same
quantity, technological exchange would be impossible. So commerce must
rely on precise measurements of mass and energy, and laboratories around
Appendix: Mathematics of Science 377
the world ensure that we have standard scales of reference and exact conver-
sions between the various different units of measure.
A.3 Conservation Laws

Back to energy and mass. In E = mc², we see E and m are related through c²
(c-squared, or c × c). As c is a universal constant (never changing), it means
E and m can be exchanged. Of course, they only can significantly exchange
under atomic reactions, as in the atomic bomb where the fission of an atom
(a form of mass) leads to a small loss of mass that produces a big amount of
energy. This is seen from the formula, as c is a very large number.
A.3.1 Conservation of Mass

Fire deals with chemical reactions only, so E = mc² will never apply.
Chemical reactions rescramble the atoms of the molecules of the reaction,
and the original mass of the reacting molecules is the same as the molecules
formed. The atoms in the molecules at the start of the chemical reaction and
the newly produced molecules at the end remain the same. For example,
carbon might react with oxygen and burn to carbon dioxide. If a perfect
mixture of carbon and oxygen is introduced (such a mixture is called “stoi-
chiometric”), none of the carbon and oxygen will remain after the reaction.
In terms of mass, we might consider this as one atom of carbon combin-
ing with an oxygen molecule that has two atoms of the element oxygen.
The carbon dioxide that forms has exactly these same atoms in its molecule.
These original atoms have not been destroyed in the chemical reaction.
The matter in terms of the atoms remains the same. As mass is the general
measure of the quantity of matter we say: mass has been conserved. This is a
general law of nature called the “conservation of mass.” By deduction, as
all matter has energy as well as mass, we might conclude that there is also
a “conservation of energy law.”
A.3.2 Conservation of Energy

Energy, however, can take many forms and each has to be recognized and
accounted for in applying conservation to energy. For example, an object
that is dropped in a vacuum, offering no resistance, will have all of its
potential energy converted into kinetic energy. Suppose the object is per-
fectly insulated and is moving without the aid of gravity in a vacuum and
now strikes an immovable plate. The kinetic energy must go somewhere.
The temperature of the object will rise, as this kinetic energy is now put
back into the matter of the object. The temperature is a manifestation of
the internal energy of the object. Thus, the internal energy of the object has
increased while its potential energy has dropped an equal amount.
In fire, the chemical reaction releases energy in the form of heat. What
actually happens in this chemical reaction is internal energy within the orig-
inal molecules is converted to a new form of energy (heat) as the chemical
reaction proceeds. The final internal energy of the molecules that form is less
by that amount of heat that was lost to the surroundings. The total energy
stays preserved and is redistributed as
Energy of reactant molecules = Heat transferred + Energy of product

molecules
The difference between the energy of the reactants and products is the
energy lost to the surroundings as heat. Thus, the sum of energies at the end
is the same as the energy at the beginning. This is an example of the conser-
vation of energy law. As no atoms were destroyed or converted into energy
in the chemical reaction, the mass is also conserved. Only by rescrambling
the atoms of the reactants into new product molecules has energy been
redistributed.
The conservation of energy is commonly called the “first law of thermody-
namics.” These laws for mass and energy cannot be proven; they are based
on scientific observations of which no exception has been found. Also energy
and mass are invented terms to express these concepts from observation and
measurement. Thus, we need to appreciate that science is the generalization
of observations into laws that govern the behavior of the universe. We can
relate mass to the quantity of matter (stuff), and all mass has energy that can
be recognized in various forms. For example, kinetic energy is defined as
mass times one-half of its velocity squared. A weight falling a certain dis-
tance is said to do work and the rate of doing work is commonly called power.
Water flowing over a dam can be converted from work to kinetic energy of
a water wheel that is now spun. But friction robs us of some of this water-
power generated by the wheel. By measuring these different forms of energy,
the conservation of energy has been validated. In this way, scientists have
built up the identification of many forms of energy and have shown that
energy is conserved when it is transformed to a new form.
A.3.3 Conservation of Momentum

There is one final universal law, and it is known as “Newton’s second law”
(after Isaac Newton) where
F = ma (A.1)
Here, a is the acceleration of the object mass, m, subjected to a force, F.

A mass receiving a fixed force will accelerate according to this formula (law).
Equation A.1 is an expression of the “conservation of momentum law” as it is

called. “Momentum” is defined as the product of mass and velocity. The gen-
eral law is expressed as the sum of the forces on an object is equal to the
rate of change of its momentum. Equation A.1 also provides a way todefine
mass as that quantity in which the force on an object enables its acceleration,
accordingly.
Equation A.1 is given in symbols. These symbols are like acronyms of
science. The symbols are used as simplified and shortened nicknames.
Often the nicknames have been chosen in an obvious way: F for force,
a as acceleration, and m for mass. Many scientific texts will use the same
symbols for the same quantity. But soon one runs out of the alphabet, as
single letters are generally used, and other alphabets, mainly Greek, are
employed. The beginning student in science needs to be aware of such
Greek terminology to converse smoothly with equations of science.
Table A.1 lists the upper and lower case letters of the Greek alphabet and
their English names. For comfort when dealing with scientific expressions
it pays to know some Greek.
Let us investigate Newton’s law and consider the meaning of the
terms. Acceleration is the rate of change of velocity. For example, if the
car speeds up from 0 to 60 mph in 6 seconds, its acceleration is 60 mph/6
seconds or 10 miles per hour per second. It is proper to express this in
consistent units, not hours and seconds together. Also if seconds are to
be used, generally the foot is a better choice of length over mile. Now we
have an exercise in converting from one set of units for length, miles and
feet, and one set of units for time, hours and seconds. This can be done
in a number of ways. We need to look up how units of one measure are
converted to alternative units in that measure. Hours can be changed to
seconds (3600 s = 1 h) and miles can be changed to feet (1 mi = 5280 ft),
TABLE A.1
Greek Alphabet Names, Uppercase/Lowercase
Α/α Alpha N/ν Nu
Β/β Beta Ξ/ξ Xi
Γ/γ Gamma Ο/ο Omicron
Δ/δ Delta Π/π Ρi
Ε/ε Epsilon Ρ/ρ Rho
Ζ/ζ Zeta Σ/σ Sigma
Η/η Eta Τ/τ Tau
Θ/θ Τheta Υ/υ Upsilon
Ι/τ Iota Φ/φ Phi
K/ĸ Kappa Χ/χ Chi
Λ/λ Lambda Ψ/ψ Psi
Μ/μ Mu Ω/ω Omega
or we might use 1 ft/s is 0.682 mph. Let us use these conversion factors to
change from 10 miles per hour per second to ft/s/s or ft/s2:
1 ft/s
60 mph ´
0.682 mph ft/s ft
= 14.67 or 2
6s s s
Note how (mph) as a unit measure cancels in the numerator and denominator,
and we are left with ft/s²—the units of acceleration.
Instead of numbers we could express acceleration in terms of the ratio of
velocity to time, or more precisely
Change of velocity
Acceleration =
Change of time
Symbolically, we could represent a for acceleration, v for velocity, and t for

time. Further, we could use the symbol capital delta, Δ, in the Greek alpha-
bet, synonymous with D in the English alphabet, to mean the operation:
change of. Then the previous equation in words can be expressed completely
in symbols as
Dv (A.2)
a=
Dt
More precisely, Newton realized that the interval in time needed in the for-
mula should be very small if he was to compute the precise acceleration of
an object as it moves. As a consequence, he invented the calculus in which
Equation A.2 becomes the time derivative of v as Δt becomes very small.
The acceleration can be expressed as the instantaneous rate of change of
velocity over time. This book will stop short of the calculus, but one should
appreciate that there are higher forms of mathematics that enable more
complex and precise computations.
A.3.4 Laws of Nature

In general, the three conservation laws can be expressed in terms of rate
equations. These rates are time derivatives. For example, the rate of energy (E)
is expressed as ΔE/Δt as Δt gets small is written as dE/dt for its time derivative.
The three laws of nature, conservation of mass, energy, and momentum, can
be expressed for an isolated system as follows:
dm
1. = 0, since the mass does not change over time.
dt
dE
2. = 0, since energy does not change over time.
dt
dv
3. F = m , as now acceleration is expressed as the rate of change of
dt
velocity.
These laws, using the calculus, allow scientists to mathematically express

the laws of nature. They apply to all processes and must always be satis-
fied. Consequently, if we wish to explain or predict an aspect of any pro-
cess, we must apply these equations to that process and render a solution.
In this way mathematics is the tool of the scientist to describe aspects of
nature in terms of concrete numbers, each with appropriate units for a
specific quantity. For most applications, this set of three equations cannot
be solved exactly to produce a simple formula. Thus, the laws of nature
can be expressed in mathematical terms and in conceptual and measurable
quantities but do not easily render specific solutions. Indeed, the process
of expressing these laws in complete mathematical terms has evolved over
time as more phenomena could be represented. Computer models solve
equations representing these laws in terms of the derivatives. It is not a
trivial process.
Scientists began with simple studies to represent a law in mathematical
terms. For example, Newton (~1700) used solid objects to first elucidate the
conservation of momentum principle. About 60 years later, Daniel Bernoulli
extended it to flowing water, but he ignored the frictional effects of water
flowing through a duct. About another 60 years later Navier and Stokes
included this friction. Their description became much more complicated, but
the law of nature was the same. The expression of Newton’s law for a flowing
fluid, such as water or air, must consider all of the forces present and must
represent the rate of momentum at every point in the flow and at every time.
The Navier–Stokes equation does this in very complex mathematical terms.
In general, the basic three laws of nature that govern all motion, mass, and
energy have been expressed in deeper and more complex mathematics over
the last 300 years.
Don’t get nervous. You will not be asked to solve them. In fact, the biggest
computers today can only solve them in very limited ways. We do not yet
have the magic universal calculator. Computers solve them approximately.
The approximations are not just in the numbers, but also in how to express
the unsteady phenomena of fluid turbulence, the rate of the combustion
reaction, and the complexities of heat transfer. In addition, the properties of
materials are not always precisely known.
A.4 Solution by Algebraic Formula

The alternative to precise mathematical solutions is to make observations
and measurements, use the laws of nature to the extent possible, and develop
formulas for specific, rather than general, problems. These formulas are the
results of a solution for limited aspects of a specific process. They have been
developed using implications of the conservation laws and careful experi-
mental data so that an algebraic equation gives a result that holds over a
wide range of conditions. These are the formulas you will meet in this text.
An algebraic equation will likely have some unusual algebraic operations,
at least to the user that has forgotten or never learned algebra. We will review
the principles of elementary algebra here.
A.4.1 Algebraic Operations and Formulas

The formulas you will encounter in this text will be in terms of elementary
algebra. This algebra involves operations with numbers and symbols con-
sisting of normal arithmetic, special functions, powers, and roots. We have
implicitly been using algebra to describe the laws of nature so far. But let us
backup and become more elementary.
As an example, consider the problem of estimating the time to drive from
Washington, DC to New York City by car averaging 50 mph. The distance
is about 230 miles. We can formulate the problem into an equation. Let us
choose
v as the velocity of the care

x as the distance
t as the time
Then,
x
t=
v
or computing
230 miles
t= = 4.6 hour
50 miles/hour
Notice we have the units
miles
(miles/hour )
If we multiply this complex fraction in terms of units by hour/hour (any-

thing divided by itself is unity or one), we get
miles hour 1
× = × hour = hour (h)
miles hour 1
hour
Thus, the answer comes out in the units of hour.
A.4.1.1 Power Operations

Let us consider another example. What is the area of a track of land 10 meters
(m) × 10 m? Let x be the length of one side of the square track and A stands
for area. Our formula is
A = x ´ x = 100 m 2
The symbol × means multiply, but here we are multiplying x by itself. Algebra
introduces a shorthand for this operation with x × x as x² which means
x squared or to the power 2. Here, the power 2 means multiply x by itself.
Let us carry this further. Extending this example to the calculation of
volume (V) of a cube have each side of length x
V = x³
is the formula. Now we have x , or this is called “x-cubed” or x to the power 3,
3
meaning it is multiplied by itself three times. In general, we might write xn.

But what if n is not an integer? That could prove troublesome. Let us inves-
tigate this issue.
A.4.1.2 Roots and Fractional Powers

What if we know that the volume of the cube is 1000 cubic meter (m3)? How
can we find the length of a side? We need to find the cube root of V. This is the
inverse of the cube operation or power-3 operation. What number, when mul-
tiplied by itself three times, will give 1000? Of course, we easily see that num-
ber as 10, but the cube root of 965 is not something you can do in your head. We
could do it by trial and error, and estimate the answer. However, today we are
blessed with electronic calculators that can perform all algebraic operations
at the stroke of a button. Let us proceed with the language and symbolism of
algebra. The operation of taking a cube root is denoted as follows:
3
V = 3 x3 = x
The cube root has annihilated the cube operation on the x, and we return to
x alone.
Alternative, we can write this root operation as a fractional power 3 V = V 1/3.
The 1/3 power is equivalent to the root operation.
Note: x3 = x × x × x, so raising it to the 1/3 means undoing the cube operation,
( x 3 )1/3 = x1 or V 1/3 = x
Thus, we have preformed algebraic manipulation of the formula, by reversing

the process of finding V from x, to finding x from V.
A.4.1.3 Scientific Calculators

The operation of having a fractional power implies a root operation. But
what if the power is something like 5/2 or 2.5? Consider the number 4 to the
power of 2.5. We can dissect its operation as follows:
4 2.5 = 4 ´ 4 ´ 40.5 = 16 ´ 2 = 32
Perhaps that was relatively easy, as we could readily recognize the square
root of 4 as 2. If the power was 2.4 instead the answer is lower, and can be
found as 27.8576…. Our only recourse for obtaining this number is to use an
electronic calculator. These devices can now be purchased at low prices and
can even be downloaded free from the web. They are referred to as “scientific
calculators.” Figure A.1 is an example of such a calculator. It allows for many
algebraic operations.
The process of computing 42.4 is done by using the yx -key. In this case 4
is typed for y and 2.4 is inputted for x. Pressing the equal button gives the
answer. In this class you will need to perform operations like this, but in
sequence, such as 42.4/10.2. So getting familiar with a calculator is a must.
While the precise answer to 42.4/10.2 is 2.731…, it could well have been
approximated by using 42.5/10 giving 32/10 or 3.2. We know it should be
lower, and this approximation shows we are on the right track in using the
electronic calculator. It is always prudent to do such approximations when
performing a series of operations on the calculator. The student should be
aware that scientific calculators differ in their mechanics, so there is learning
process for each.
A.4.1.4 Negative and Power Zero

Before leaving power operations we should consider what happens if a
power is zero or negative. It should be obvious that x/x = 1. Writing this as
10x ex 2x
log ln lg2 HEX INFO
sin–1 cos–1 tan–1 e
INV sin cos tan π
x2
yx n! √ 1/x x++y
( ) Min MR M+
7 8 9 C AC
4 5 6 × ÷
1 2 3 + –
0 · EXP +/– =
FIGURE A.1
Typical scientific calculator.
powers, we consider the term in the denominator as a negative power. So the

equivalent of x/x = 1 can be written as
x1 ´ x -1 = x1-1 = x 0 = 1
Thus, we see that the zero power for any quantity gives 1, and multiply-
ing like terms with powers can be simplified by adding the powers. For
example,
x 3/2 xy x 3/2 x1/2 y 1/2

= = x 3/2 +1/2 y 1/2 - 2/5 = x 2 y 1/10
y 2/5 y 2/5
The beginning student of algebra should master these power operations to

be better versed at manipulating fire formulas.
A.4.1.5 Constants in Science

In nature, some constants often show up and they are given a special symbol.
For example, the area of a circle of radius, r, is
A = pr 2
where π (pi, the Greek letter) is one such constant.

Rounded off to five places,
π = 3.14159.
In Figure A.1, it is seen that the calculator has an individual button for the
exact value of π.
Another constant is the quantity called e given to five places as
e = 2.71828.
For the calculator shown in Figure A.1, e shares the same button for π and can
be accessed by pressing a shift key. These are special numbers you are likely
to see in the formulas found in fire science.
While π shows up in formulas related to circular effects, it is not obvious
why the e shows up. In science, there are many natural processes that, by the
conservation laws and the calculus, can be shown to occur with e to some
power involving length or time. The expression ex is called the “exponential
function of x,” meaning the number e is taken to the x-power. This function
occurs in a process where a quantity depends on its own rate of change.
For example, if the power produced in a fire depends on the rate of change of
that power, then with P standing for the firepower, it will behave as
P = Pe
i
Ct
(A.3)
where
Pi is the initial value of P
t is time
C is a constant
If C is negative, the power decreases in time. Of course this happens for fires
that die out. On the other hand, if C is positive, the power will continue to
increase indefinitely over time. Mathematically, this can only reach a very
large number at the end of time called “infinity.” Here, a mathematical solu-
tion gives an answer that is not physically meaningful. With C positive, the
fire will grow, but the formula does not take into account the availability of
air that will eventually limit this fire growth. We say P grows (or decays)
exponentially in time.
A.4.1.6 Logarithms
Consider the exponential decay process. We may wish to compute the time
it will take for the power to decay to one-half of its original value. This is
commonly called the “half-life.” We need an operation that can reverse the
power process on the number e. This inverse operation is called the “natural
logarithm” given by the symbol ln. The process is described as follows:
ln éë e Ct ùû = Ct (A.4)
Logarithms of numbers have been computed and can be found in mathemat-

ical tables, but today we can easily access them on the scientific calculator.
In Figure A.1 it is seen that there is a button for ln, meaning taking the loga-
rithm of x as ln(x). Also notice on the same button is the inverse operation of ex.
With this ability, we can compute the half-life:
æPö æ1ö
ln ç ÷ = ln ç ÷ = ln(e Ct ) = Ct
P
è iø è2ø
It can be determined from a table or the calculator that ln(1/2) = −0.693…

roughly. Consequently, the half-life is about −1.44/C. For this to give a posi-
tive number for time, the C must be negative for decay and must have units
of reciprocal time, for example, 1/seconds or s−1. For example, if C = −0.01 s−1,
then the half-life time is 144 seconds.
This ln operation implies taking the logarithm to the base e. There is
another logarithm taken to the base 10 and its operation is denoted by log. Its
inverse operation is log(10x) = x. It too is listed on the calculator. These special
numbers e and π, along with the logarithmic operations are likely to show up
in scientific formulas, and therefore a scientific calculator is essential.
A.5 Graphs
Graphs are often employed as an alternative to interpreting formulas
through algebra. In fact, many formulas for fire calculations have resulted
from plots of data, and then a formula was developed directly from the
graphical fit to these data. So reading a graph is also one way of addressing
an equation. For example, a plot of P = Pie Ct , where Pi is taken at 100 kW and
C is taken as (−0.01 s−1), is given in Figure A.2.
In Figure A.2 a linear scale is used on each of the axes in which the incre-
ments on the axes are all equal. On the horizontal axis, the smallest increment
is 20 seconds (s). On the vertical axes, the smallest increment is 5 kW. The graph
also shows the half-life solution when the power has decreased to 50 kW. But
as the graph approaches zero below a power of about 5 kW, it is difficult to read
with any precision.
An alternative graphical presentation that allows more accuracy in read-
ing these low power results is to employ a graph with a logarithmic scale
100
80
P = 100e–0.01t
60
P (kW)
40
20
0
0 100 200 300 400 500
t (s)
FIGURE A.2
Linear-scale plot.
on both axes. This graph is shown in Figure A.3, and it greatly expands the
low-scale precision. In a logarithmic scale the major increments are broken
into decades, and within each decade there are increments inclusive of 1–10.
These increments within a decade are not equal in length. Therefore, extrap-
olation of results must be done with care.
A.6 Units and Conversions

A formula and a graph will contain various symbols for specific quanti-
ties and they will have units. These units must be consistently used or
the results will be wrong. If the answer on the left-hand side is to be in
meters, the formula must be algebraically computed from the terms on the
right-hand side to also yield meters. A simple example can illustrate this.
Suppose you know the area of a room floor and the length of one of the
sides, and you wish to find the length of the other side. You are given that
the area is 18.3 m2 and a side is given as 12.5 ft. These units are inconsistent.
Our formula for computing the side, call it s2, with the given side as s1, from
the area, A is
A
s2 =
s1
100
10
P = 100e–0.01t
P (kW)
0.1
0.1 1 10 100 1000
t (s)
FIGURE A.3
Logarithmic-scale plot.
Let us maintain meter units, as you will find this more common in science.
We need to convert ft to m. Conversion tables are available to making these
changes and must be utilized. Table A.2 is a listing of conversion factors that
you might find most pertinent to the subject of this course. It emphasizes the
conversion of units to the Standard International (SI) system. That system is
most often used in science and the formulas of fire are mostly represented
relative to SI units. The SI unit for length is m. The other key units are mass
in kilograms (kg) and time in seconds (s). All of these units have shorthand
symbols for them as indicated. From Table A.2, 1 ft = 0.305 m as equivalent
lengths. This equivalence can be used to formulate an algebra for converting,
that is,
0.305 m
12.5 ft ´ = 3.81 m
1 ft
In this process, the conversion factor 0.305 was rounded off to three places
and the answer has been rounded off to three as well. This too is consistency.
Now the problem can be completed in consistent units of m. Substituting
into the equation gives
A 18.3 m 2
s2 = = = 4.80 m
s1 12.5 ft ´ 0.305 m
1 ft
TABLE A.2
Approximate Conversions from English to SI Units
Length
in. Inch 25.4 Millimeters mm
ft Feet 0.305 Meters m
yd Yards 0.914 Meters m
mi Miles 1.61 Kilometers km
Area
in.² Square inches 645.2 Square millimeters mm²
ft² Square feet 0.093 Square meters m²
yd² Square yard 0.836 Square meters m²
Volume
ft³ Cubic feet 0.028 Cubic meters m³
yd³ Cubic yards 0.765 Cubic meters m³
Mass
oz Ounces 28.35 Grams g
lb Pounds 0.454 Kilograms kg
Force and pressure
N Newton kg-m/s2 Alias
Pa Pascal N/m2 Alias
lbf Pound-force 4.45 newtons N
lbf/in² Pound-force per sq. inch 6.89 kilopascals kPa
Energy
J Joule N-m Alias
Btu British thermal units 1.055 kilojoules kJ
cal Calorie 4.187 joules J
Power
W Watt J/s Alias
hp Horsepower 0.746 kilowatt kW
Temperature
°F Fahrenheit (°F − 32)(5/9) °C
Celsius
K Kelvin °C + 273.15 Alias
Notice in the algebra of dealing with units that ft cancel, and the m2 cancels
with m to give the length in units of m.
A.6.1 SI Units: Aliases or Derived Units

The conservation laws provide a basis for giving the equivalence of units
and minimizing the number of independent units actually needed. However,
as in all practices, names spring up to simplify certain collection of units into

aliases. In the SI system we need only deal with kg, m, and s as the primary
units. Here we have covered mass, length, and time. But what units should
we give to force, a distinct quantity? From Newton’s law or the conservation
of momentum, we know force is always equal to mass times acceleration.
This means that in terms of the primary SI units we might give force the
units of kg times m/s2. However, this name, kg-m/s2, is cumbersome, so it
is given an alias name of Newton (N), as 1 N = 1 kg-m/s2. This is a so-called
derived unit.
How about energy? Well, energy is measured by work, and work is the
product of force and length. Therefore, we can now call energy units in the
SI system as N-m. This already has the alias N in it, yet we wish to be sim-
pler, and the alias joule (J) is introduced: 1 J = 1 N-m, another derived unit.
Finally, power is the rate of energy and is a commodity that is purchased in
kilowatts—another alias, the watt (W). 1 W = 1 J/s. In this way many terms
in science can be represented in terms of the basic units of mass, length, and
time: kg, m, and s. These, and other aliases, with their conversions to SI units
are listed in Table A.2.
A.6.2 Equations and Dimensions

Let us consider one final example to show how consistency can be main-
tained in an equation. We take an example of a frequently used formula for
the ratio of flame length (L) to diameter (D) for a pool fire given as
L
= 3.87Q*2/5 - 1.02 (A.5)
D
As L/D is the ratio of two length dimensions, flame length to diameter, they
must both be in the same units. Also these same units will cancel, leaving no
units for L/D. The quantity L/D is called a dimensionless quantity or variable.
Of course the L and D each have units, and as they must be consistent, they
can both be in ft or both in m.
In Equation A.5, the left-hand side is dimensionless, and that means
the right-hand side must also be dimensionless. The numbers here have
no units, and the variable Q* (Q-star, some have given it the name, the
Zukoski number) must also be dimensionless. However, Q* is a quantity
composed of many factors, each having units. Often scientists compose
equations that are dimensionless with the intent that they are valid over
a wide range of conditions. Also the dimensionless nature of the equation
makes it more compact and reduces the number of variables. Here Q* is
composed of the several properties and the variable for the firepower,
Q in this fashion:
Q
Q* = (A.6)
rcpT gD5/2
where here we have the properties of air given as

Density, ρ = 1.20 kg/m3
Specific heat, cp = 1.00 kJ/kg-K
Temperature, T = 293 K
The SI unit for temperature is in kelvin (K) and can be considered as an

alias for Celsius (°C). In fact, K is a temperature scale that has its zero point
associated with no more energy in the material. Celsius is basically a scale of
reference for temperature that uses the freezing and boiling points of water
as 0 and 100, respectively. When the quantity of temperature stands alone in
an equation, the absolute temperature scale of K must be used over °C. If it
appears as a temperature difference, either scale can be used.
One more term is contained in the denominator of Equation A.6. It is g,
standing for the gravitational force on a unit mass. For the Earth, this is
about 9.81 N/kg. Note, as a newton is an alias, this can be also equivalently
written as 9.81 m/s2. This now has units of acceleration, and g is also called
the acceleration due to gravity.
Let us compute these terms in the denominator of Equation A.6 with the
air properties specified. We note that the units combine as follows when
treated with algebraic operation:
1/2
kg kJ æ mö kJ kW
1.20 3
´ 1.00 ´ 293 K ´ ç 9.81 2 ÷ = 1101.2 5/2
or 5/2
m kg-K è s ø s-m m
Here we have recognized the alias: kW for kJ/s. Now if we express the
firepower in kW and the diameter in m, Q* will remain dimensionless and
consistent with L/D.
Often the same equation as (A.5) will be given in specific units. This follows
from substituting the air property and gravity term as 1101.2 kW/m5/2 into
Equation A.5 to obtain
2/5
L (m ) é Q (kW) ù
= éë0.235 m/kW 2/5 ùû ê 5/2 ú
- 1.02
D (m ) ë [D (m)] û
or
2/5
L (m) = ëé0.235 kW -2/5 ùû éëQ (kW)ùû - 1.02D (m) (A.7)
Equation A.7 is the dimensional form of Equation A.5 in the units of m and kW.
Here the units are all indicated, and even the number 0.235 has units, in
contrast to the numbers in the dimensionless equation (A.5).
When an equation is given as a formula, the units are very important.
If it is dimensionless to begin with, consistent units must be used for each
term. If it is dimensional, specific units must be used for the variables in the
equation and nothing other. In Equation A.7, the units of kW must be used
for the firepower and the units of m for diameter; then the flame length is
computed in m also. I was told of the use of Equation A.7 in a court case
where two experts got very different answers. The opposing expert used ft
for diameter and got a negative flame length. It may have scored points in
that expert’s testimony initially, but a lot of credibility was lost on rebuttal.
Units are very important. They must be used consistently, or the results of
the equation will be wrong. Take care.
A.7 Summary
Mathematics is required to solve the laws of science. The key laws of science
embrace
1. Conservation of mass
2. Conservation of energy
3. Conservation of momentum
These laws must hold for every natural process and can be expressed in
mathematical terms. Yet, even modern computers are not capable of solv-
ing all aspects and can only give approximate answers. Alternatively, sci-
ence has found specific solutions in terms of single algebraic equations. The
solution of these equations into precise numbers requires an understanding
of algebra and the appropriate manipulation of units. SI units are mostly
used in science, and conversion to these units is essential in applications.
The basic SI units, for purposes of fire science, involve mass in kg, length
in m, time in s, and K for temperature. Combinations of these units are suf-
ficient for describing all scientific variables and properties of fire. However,
it is practice to give shorthand alias names to the various combinations. This
occurs for force (N), energy (J), and power (W). In the computation of any
algebraic equation, it is essential that the units are consistent so that the units
on the right and the left side of the equation reduce to the same quantity. A
brief review of algebra was presented with emphasis on the use of powers
and roots. Exponential and logarithmic functions were discussed. A scien-
tific calculator is essential for performing these operations. Alternatively, it
was shown how graphical displays of the solution of an equation could be

presented in linear and logarithmic scales.
Review Questions
1. What equation shows the equivalence of mass and energy?
2. What equation relates force and mass?
3. How can energy be recognized?
4. What are the basic SI units?
5. Can energy or mass of the universe ever be destroyed?
x 5/2 y 1/2e 2 x
6. Simplify .
y 2.3 x
7. Compute πe0.6.
8. Compute the height of a flame in inches from Equation A.7 for firepower
of 18,450 W and a diameter of 60 cm.
9. What force, in N, is required to hold 100 kg in Earth’s gravitational accel-
eration field of 9.81 m/s2?
10. Name several derived units in the SI system.
True or False
1. In the chemical reaction of fire, mass is destroyed.

2. Atoms are a form of mass.
3. Heat produced in a fire is due to a rearrangement of the atoms into new
molecules.
4. SI units can include feet for length.
5. Newton’s law is the conservation of momentum.
6. Units in an algebraic scientific formula must be consistent.
7. Energy is basically just heat alone.
8. The boiling point of water at normal atmospheric pressure is 80°C.
9. The units for acceleration are ft/s.
10. Logarithmic scales on graphs are arranged in decades.
Activities
1. Buy or download an inexpensive scientific calculator and learn how to
perform simple algebraic operations.
2. Explore tables and conversion techniques found in books or on the
internet.
3. Explore the conservation laws and their history.
4. Review algebra.
5. Practice manipulating units in scientific equations.
Glossary
Absorption coefficient: That property that pertains to the amount of radia-

tion absorbed per unit length.
Activation energy: Energy per mole of fuel needed to initiate ignition.
Adiabatic flame temperature: The maximum possible temperature in the
chemical reacting zone with no heat lost.
Arc mapping: A method of locating beads due to electric wire circuit faults
due to fire in order to try to determine the location of the early fire.
ASTM E 84: A standard providing a test (Steiner tunnel test) developed
by Underwriters Laboratories to examine wind-aided flame spread
under ceiling mounted materials.
ASTM E-162: A standard providing a radiant fuel test to measure the
flammability of materials under downward spread.
ASTM E-648: A standard providing a flame test for floor-covering materials.
ASTM E-1321: A standard providing a test to determine data for ignition
and flame spread.
Autoignition: Initiation of fire by chemical process inherent in the mate-
rial without a pilot; specific fuel concentration and temperature are
usually needed.
Autoignition temperature: The lowest temperature at which a mixture of
fuel and oxidizer can propagate a flame without the aid of an initi-
ating energy source (pilot).
Backdraft: The sudden eruption of fire in a compartment due to the accu-
mulation of fuel gases and the introduction of fresh air; results in a
pressure increase.
Black body: A material emitting the maximum possible radiant energy.
Blow-off: Extinguishment of a flame due to wind or high-speed flow.
Boiling point: The maximum temperature at which a liquid can evaporate
under normal atmospheric conditions; equilibrium temperature for
a liquid and its vapor to coexist at 1 atmosphere of pressure.
Boundary layer: Fluid region near a surface in which viscous or heat transfer
effects dominate.
Buoyancy: An effective force on fluid due to density or temperature differ-
ences in a gravitational field.
Burning rate: The mass of fuel consumed in the fire per unit time.
Burnout: Point at which flames cease.
Carboxyhemoglobin: Carbon monoxide hemoglobin compound found in
the blood due to inhalation of carbon monoxide.
Chain branching: Propagating sequence of chemical reactions steps that
produce more unstable species than it consumes; terminates in
ignition.
397
398 Glossary
Characteristic combustion length: A length scale representative of the fire

size; proportional to the power 2/5 of the firepower.
Charring: The production of a solid carbonaceous residue on heating or
burning a solid.
Chemical equation: An equation showing the proportions (by mass or
moles) of the fuel, oxygen, and products in a chemical reaction in
which the atoms are conserved.
Chemical kinetics: Refers to the study of chemical reaction rates.
Chemical reaction: Chemical interactions between molecules to form new
molecules while conserving atoms.
Clean-burn: A less sooty region on a noncombustible wall after a fully
developed room fire; appears lighter in shade compared to sur-
rounded wall.
Combustion: A chemical reaction, involving oxygen, that produces a per-
ceptible amount of energy—fire or controlled fire.
Completeness: Pertaining to a combustion process going to its most stable
state, an ideal reaction; water and carbon dioxide would be the
complete products of combustion for hydrocarbon fires.
Concentration: The percentage of material per unit mass (or volume) of its
mixture.
Conduction: Heat transfer due to molecular energy transfer following
Fourier’s Law.
Configuration factor: Fraction of radiation received by a target compared to
the total emitted by the source—view factor.
Conflagration or mass fire: A fire over a large tract of land; the flames are
much shorter than the horizontal extent of the fire.
Convection: Conduction heat transfer from a moving fluid (gas or liquid) to
a solid surface.
Convective heat transfer coefficient: A quantity that represents the abil-
ity of heat to be transformed from a moving fluid to a solid surface
expressed in terms of heat flux per unit temperature difference.
Cracking: Pyrolysis; breaking gaseous molecules into other molecules,
usually by heat.
Critical heat flux: A threshold level of heating below which ignition (or in
another context, flame spread) is not possible.
Crown fire: Fire spreading in the tops of trees in a forest fire.
Deflagration: A propagating premixed flame at speeds below the speed of
sound.
Detonation: A high-speed premixed flame preceded by a shock wave.
Developing fire: The early stage of growth (in a compartment fire) before
flashover and full involvement; almost like burning in the open.
Diffusion: Process of species transport in a mixture from its high to low
concentration.
Diffusion flame: A flame in which the fuel and oxygen are transported
(diffused) from opposite sides of the reaction zone (flame).
Glossary 399
Dimensionless: Having no units of measure (terms combine to produce

no units).
Dose: The accumulation of product concentration inhaled over time; the
integral of concentration over time.
Eddies: Rotating regions of a fluid.
Emissivity: That property (having values 0–1) that gives the fraction of
being a perfect radiation emitter due to temperature.
Energy: A state of matter representative of its ability to do work or transfer
heat.
Energy release rate: The energy produced by the fire per unit time; fire
power; HRR.
Entrainment: The process of air or gases being drawn into a fire, plume, or jet.
Equation of state: An equation to relate thermodynamic properties—
commonly pressure, temperature, and density for a gas.
Equivalence ratio: The ratio of available fuel to air times the stoichiometric
air-to-fuel ratio or (fuel/air)available divided by (fuel/air)stoichiometric.
Equivalency: The statement in regulations allowing for alternatives by design.
Evaporation: The process of gas molecules escaping from the surface of
a liquid.
Field model: A type of computer fire model that attempts to predict condi-
tions at every point; also known as a computational fluid dynamics
model.
Fire: An uncontrolled chemical reaction producing light and sufficient
energy.
Fireball: An increasing, short-lived, rising ball-shaped flame due a sudden
release and ignition of fuel gases.
Fire plume: The flame and gases emanating from a burning object.
Fire point: The minimum surface temperature needed by the liquid to
sustain a diffusion flame.
Firepower: Chemical release rate of energy in a fire.
Fire spread: The process of an advancing fire front: smoldering or flaming.
Firestorm: Large conflagration that creates large winds and other weather
effects.
Fire tetrahedron: A three-dimensional representation of the fire triangle
with chain branching.
Fire triangle: A concept describing fire as consisting of three ingredients:
fuel, oxygen, and energy.
Fire whirl: A flaming tornado.
Flame height: The vertical measure of the combustion region, usually
luminous.
Flame spread: The process of advancing the fire front in air, along surfaces,
or through porous solids.
Flammable: Indicates that burning can occur; originates from inflammable.
Flashover: A dramatic event in a room fire that is due to energy produc-
tion exceeding energy losses; manifested by a sharp rate of increase
400 Glossary
in temperature of the smoke; ends with a fully developed fire that

involves all the contents in flames if sufficient air or terminates
with a ventilation-limited fire without sufficient air to burn all of
the gaseous fuel produced in the room.
Flashpoint: The temperature of a liquid fuel, theoretically corresponding to
the lower flammable limit of its evaporated vapor and the point of
piloted ignition.
Fluid mechanics: The study of fluid motion.
Flux: Pertains to mass or heat flow rates per unit area.
Forced flow: Refers to air flow produced by wind or a fan.
Free burning: Burning in open air.
Frequency: Cycles per unit time (measured in hertz, cycles per second).
Fuel lean: Description of fuel burning in an excess supply of air.
Fuel limited: State of a compartment fire where the air supply is sufficient
to maintain combustion.
Fully developed: State of a room fire at which the maximum possible energy
release is attained; either all the fuel burns or all of the incoming air
is consumed.
Heat: Energy transfer due to temperature difference.
Heat flux: Heat flow rate per unit area of flow path.
Heat of combustion: The energy released by the fire per unit mass of
fuel burned.
Heat of gasification: Energy required to produce a unit mass of fuel vapor
from a solid or liquid.
Heat release parameter, HRP: Ratio of heat of combustion to heat of
gasification.
Heat release rate: Same as firepower.
Heat transfer: The transport of energy from a high-to-low temperature.
Hemoglobin: Compound in blood that transports oxygen or carbon
monoxide.
Hifukusho-ato district of Tokyo: Area where many died in a fire whirl in 1923.
Hot surface ignition: Ignition of a flammable mixture due to a hot surface;
usually higher than the AIT.
Humidity: The property of the water–air mixture that measures the amount
of water present relative to the equilibrium concentration.
Hyperthermia: Heat stress.
Ignition temperature: The surface temperature needed to cause ignition
in solids.
Incomplete combustion: A combustion process that does not go to the most
stable species such as H2O and CO2.
International System of Units or Standard International (SI) Units: The
system of units for measurement generally adopted by the science
community.
Irritant gases: Acid gases and other hydrocarbon by-products that can
cause pain on contact or with inhalation.
Glossary 401
Jet flame: Flame due to a high-velocity fuel supply; exceeds buoyancy effects.
Kelvin (K): Absolute Celsius temperature scale, ~273 + °C.
Laminar: Refers to orderly, unfluctuating fluid motion.
Line fires: Elongated relatively thin region fires on a horizontal fuel surface;
indicative of wildland fires.
Mass burning flux: Burning rate per unit area.
Mass concentration: Ratio of species mass to mixture mass; also referred to
as mass fraction.
Mass loss rate: The mass of fuel per unit time vaporized under fire heating
but not necessarily burned.
Mass optical density: Optical property related to the yield of particulates
in smoke.
Narcosis: Effect of inducing sleep.
Natural flow: Refers to air flow induced by buoyancy.
Neutral plane: The height above which smoke can flow out of a compart-
ment or building wall; the height of zero pressure difference across
a partition.
Newtons Second Law of Motion: Relates force on a body to its mass and
resulting acceleration.
Opposed flow: Refers to air flow in the direction opposite to the fire spread.
Overventilated: More than stoichiometric air is available.
Oxygen bomb: A device for measuring the maximum energy released in
combustion for a fuel.
Oxygen consumption calorimeter: Device to measure energy release rate
in fire by utilizing the approximate constancy of the heat of combus-
tion per unit mass of oxygen, ~13.1 kJ/g.
Oxyhemoglobin: Oxygen-bearing hemoglobin in blood.
Parts per million, ppm: Concentration based on 106 parts of the mixture.
Pattern analysis: The process of attempting to interpret patterns in the
debris of a fire.
Performance codes: Regulations providing for engineering analysis.
Performance-based design: Design process to meet specific engineering
objectives.
Piloted ignition: Ignition of a flammable fuel–air mixture by a hot spot,
electric arc, or small flame (pilot).
Pool fire: Fire involving a horizontal fuel surface, usually symmetrical.
Premixed flame: A flame in which fuel and air are mixed first before
combustion.
Prescribed burning: Intentional burning of forest terrain to improve
conditions.
Products: Chemical compounds produced by fire.
Pyrolysis: The process of heating fuel to cause decomposition to a gas.
Radiation: Heat transfer due to electromagnetic energy transfer, such as light.
Radicals: Short-lived unstable molecules, such as OH.
Rankine (°R): Absolute Fahrenheit temperature scale, ~460 + °F.
402 Glossary
Reradiation: The radiation reemitted from a heated surface.

Respiration minute volume, RMV: Inhalation rate.
Shock wave: Abrupt change in temperature and pressure due to a flow
instability caused by speeds in excess of the speed of sound.
Smoke: Gases, no longer chemically reacting, that emanate from the fire.
Smoke visibility: Distance over which a human can perceive objects
through smoke.
Smoldering: A slow combustion process between oxygen and a solid fuel.
Soot: Carbonaceous particles produced in flames.
Species: Another name for chemical compounds, usually gases.
Specific heat: Property that measures the ability of matter to store energy.
Spontaneous combustion: A process by which combustion occurs after a
very low heating self-incubation period between a fuel and oxidizer,
or decomposing fuel; leads to smoldering or flaming.
Stack effect: Motion of air and smoke due to buoyancy, usually in a tall
building.
Stoichiometric: Refers to the amount of air needed to burn the fuel (and to
combustion products formed).
Stoichiometric air-to-fuel mass ratio: The ratio of air to fuel by mass needed
to burn all the fuel to completion.
Surface tension: A force within the surface of a liquid.
Synthesis: Recombination of molecules.
Thermal conductivity: The property of matter that represents the ability to
transfer heat by conduction.
Thermal energy: Energy directly related to the temperature of an object;
internal energy.
Thermal inertia: A thermal property responsible for the rate of temperature
rise, kρc.
Thermally thin: Refers to a material that has a nearly uniform temperature
during heating.
Thermal response parameter: A convenient property for predicting ignition
involving thermal inertia and ignition temperature.
Thermal runway: An accelerating chemical reaction due to an imbalance
between heat loss and energy production.
Thermocouple: Device made of two dissimilar metal wires to measure
temperature.
Thermodynamics: The study of energy and states of matter.
Thermophoresis: The process by which particles (soot) are transported due
to a temperature difference.
Thermoplastic: A synthetic polymer that usually softens and melts on
heating.
Thermosetting: Refers to a synthetic polymer that contains cross-link bonds
and usually chars on heating.
Trifecta of fire growth: The combination of ignition, flame spread, and
burning.
Glossary 403
Turbulent: Refers to randomly fluctuating fluid motion around a mean flow.

Urban–wildland fire: A forest fire involving homes built in or around the
forest.
Underventilated: Less than stoichiometric air is available.
Upper and lower flammability limits: Highest and lowest concentrations
of fuel in air in which a premixed flame can propagate.
Vaporization temperature: The temperature of a vaporizing fuel while
burning or needed to cause vaporization.
Ventilation factor: The parameter controlling air and smoke flow rate
through a door or window, Ao√Ho.
Ventilation limited or ventilation controlled: State of a compartment fire
where the air supply is limited, below stoichiometric; exiting smoke
will have zero oxygen concentration; underventilated.
Viscous: Refers to fluid friction.
Vitiation: Refers to the reduction of oxygen concentration in air.
Volume fraction: Species concentration based on volume; or equivalently
moles.
Vortex: A ring of eddies or swirling motion.
Wavelength: The distance traveled in one cycle or the speed of light divided
by frequency.
Wind aided: Refers to air flow in the same direction as the fire spread.
Work: The movement of mass over a distance; mass times distance.
Yield: The mass of product produced per unit mass of fuel supplied.
Zone model: A type of computer fire model that approximates the fire
conditions in a room as two uniform gas layers with a fire plume
energy source.
Index
A energy release rate

Christmas trees, 192
Absolute zero, 27–28, 75–76, 85, 99
estimating, 184–187
Absorption coefficient, 86
heat release parameter, 182–184
Activation energy, 37–38
materials properties, 182–184
Adiabatic flame temperature (AFT), 60,
peak burning rate and firepower
234–235
values, 190
AFT, see Adiabatic flame temperature
rubbish, 190
AIT, see Auto-ignition temperature
sofa burning, 189
Algebraic operations
television sets, 191
constants, 385–386
upholstered chair, 189
logarithms, 386–387
wood, 181–182
negative and zero power, 384–385
extinction, 207–209
power operations, 383
fire growth rate
scientific calculators, 384–385
HRR parameters, 203–206
Arc mapping, 321
mattress, 197
ASTM E-84, 157–158
1 MW for t2 fires, 201
ASTM E-162, 157–158
NFPA design categories, 201–203
ASTM E-648, 158
sofa, 196
ASTM E-1321, 127, 142, 144, 158
upholstered chair, 193–195
Autoignition, 104, 108–115, 131–132
warehouse commodities, 200
Auto-ignition temperature (AIT), 58–59,
wood pallet stack, 198–199
108–109, 111, 113, 115–116
heat of gasification, 168–169, 173
mass burning flux
B burning walls, 175–176
computing, 170
Backdraft, 270, 281, 299 definition, 167
Blackbody, 85, 97, 170, 297 maximum burning rates, 178–179
Blow-off, flame, 145 nonuniform burning, 168
Boiling point, 75, 107–109, 166, 168–169, 173 pool fires, 176–177
Boundary layer, 82, 155 mass loss rate of fuel, 166
BRANZFIRE, 369 materials properties, 174–175
Buoyancy, 214–218 measuring, 166–167
Buoyant plumes; see also Fire plumes steady burning, 169
eddy turbulent effects, 219–221 trifecta, fire growth, 165–166
entrainment, 219, 222 unsteady burning, 171–172
flame height, 220 vehicle fire behavior, 206–207
liquid fuel fire, 221 Burnout, 136–137, 187
vortex shedding, 220
Burning mass flux, 167–168, 170, 172–173,
C
175–178, 180, 224
Burning rate Carboxyhemoglobin (COHb), 254
definition, 165 Chain branching reaction, 37, 50
difficulties in computing, 172–173 Characteristic combustion length, 227
405
406 Index
Charring, 62, 113, 169, 171, 173, 178, 354 stages, 268
Chemical equation, 244 cooling stage, 272
Chemical kinetics, 34–35, 37–38 developing fire, 269
Chemical reaction, 3–4, 34–38, 48, 50, 53, flashover, 270
56, 59, 64–66, 245, 248 fully developed, 270–272
CHF, see Critical heat flux polyurethane-cushioned chair,
Christmas Lectures, 43 272–273
Clean burn, 314, 356–357 Completeness, 244, 321
Combustion products; see also Natural Conduction
fire combustion steady, 78–80
concentrations, 251–252 thermal penetration time, 80–81
equivalence ratio, 246 Configuration factor, 88, 96
fuel lean/overventilated, 244–250, Configuration/view factor, 86–88, 95
253, 260–261, 270 Conflagration or mass fire, 10
fuel rich/underventilated, 244–252, Conservation laws
256, 260, 305–307 conservation of energy, 74, 77,
fuels, 245 377–378
hazards conservation of mass, 74, 77, 377
additive fractional incapacitation conservation of momentum, 378–380
doses, 256–257 laws of nature, 380–381
heat effects, 261–263 Convection, 81–84, 106, 115, 118, 120, 125,
irritant gases, 257 145, 148, 150, 155, 168, 208, 214,
narcotic gases, 254–256 310, 366–367
smoke visibility, 258–260 Convective heat transfer coefficient,
processes, 245 82–83, 96–97, 176, 208–209, 262
stoichiometric air-to-fuel mass Cooling stage, 268, 272, 315
ratio, 244 Cracking, 51, 54, 112
yields, 246–249 Critical heat flux (CHF), 114, 118–121,
Compartment fires 125–128, 131
fire growth, 301, 308–314 Crown fire, 148
fire-induced flows
duct fan pressures, 274–275
D
pressure level, 276
resultant velocity, 274 Deflagration, 61
single room analyses Detonation, 61
burning duration, 307–308 Developing fire, 269, 277
flashover, 297–298 Diffusion flames, 38–39; see also
fuel generation, 302–303 Premixed flames; Smoldering
fuel load, 307–308 anatomy, 48–54
smoke filling, 287 candle flame, 39–47
smoke temperature, 295–297 cigarette smoke plume, 41–42
smoldering, closed room, 288–291 gravity, 40–42
vent flows, 291–295 process, 39–40
ventilation-limited fire, 298–299 pyrolysis, 39
smoke flow shapes, 43
corridors, 282–284 turbulence, 40–42, 54–56
leaky compartment, 279–281 Dimensionless number, 218, 227, 274,
room, 276–279 287, 296
vertical shafts, 284–286 Dose, 257
Index 407
E Dupont plaza fire, 329–330

laundry basket fire, 330–332
Earthquake, 5–7, 149–150, 152–153,
Mount Carmel Branch Davidian Fire,
237, 367
340–342
Eddies, 219–221
background, 342–343
Eddy turbulent effects, 219–221
clean burn pattern, 356–357
Egress time, 322
fire cause, 349
Emissivity, 85–86, 97, 177
flashover, 348
Energy release rate
follow-up, 351
Christmas trees, 192
gasoline versus fire damage,
estimating, 184–187
357–359
heat release parameter, 182–184
hands of time in fire, 357
materials properties, 182–184
ignitions, 344–348
peak burning rate and firepower
scientific analyses, 351–354
values, 190
soot patterns, 355–356
rubbish, 190
survivability, 350–351
sofa burning, 189
tear gas, 349–350
television sets, 191
visual data, 343–344
upholstered chair, 189
wind, 350
wood, 181–182
Waldbaum fire, 332–334
Entrainment, 219, 222–227, 230–231, 276,
cockloft, fire growth, 338–339
287, 294, 297, 366
ignition, 334–336
Equation of state, 217
roof, 337–338
Equivalence ratio, 246, 248, 253, 268,
roof collapse, 339
270–271, 298, 303
smoldering stage, 336–337
Equivalency, 321
World Trade Center, 359
Evaporation, 104–106
afterthoughts, 365–366
Extinction, burning, 207–209
fuel load, 362–363
investigative efforts, 360–362
jet fuel, 362
F
steel structure, 363–365
Federal Trade Commission (FTC), 15 Fire phenomena
Field models, 369–370 ceiling flame plume interactions, 27
Fireball, 238–239 room fire growth, 23–24
FireFOAM, 21 smoke streaks, 26
Fire growth rate Fire plumes; see also Buoyant plumes
HRR parameters, 203–206 ball, 238–239
mattress, 197 buoyant plumes
1 MW for t2 fires, 201 eddy turbulent effects,
NFPA design categories, 201–203 219–221
sofa, 196 entrainment, 219, 222
upholstered chair, 193–195 flame height, 220
warehouse commodities, 200 liquid fuel fire, 221
wood pallet stack, 198–199 vortex shedding, 220
Fire investigation; see also Fire safety definition, 213
design flame height
computer models jet flames, 222–224
field models, 369–370 line fires, 222
zone models, 366–369 pool fire flames, 224–229
408 Index
flame lengths, 236–237 in United States

fluid dynamics costs, 17–19
Bernoulli’s equation, 214–216 fire prevention infrastructure, 13–14
buoyancy, 216, 218–219 flammability tests, 15–17
chimney, 214–216 motivation for improvement, 15
white streamlines, 215 statistics, 10–13
jets, 219 and world, 13
temperatures units of measure, 25–28
flame height and temperature and war, 9–10
calculations, 232–234 Firestorm, 9
predicting, 230–232 Fire tetrahedron, 37
turbulent flame temperature Fire triangle, 36–37
nature, 234–236 Fire whirl, 237–238
turbulent, 218–219 First Law of Thermodynamics, 74
whirl, 237–238 Flame blow-off, 145
Fire point, 107–108 Flame height
Firepower, 180, 187–192 jet flames, 222–224
Fire Prevention and Control Act of line fires, 222
1974, 15 pool fire flames, 224–229
Fire safety design; see also Fire Flame lengths, 236–237
investigation Flame spread
egress time, 322 definition, 136
performance fire safety dwelling, 155–156
codes, 327 fire spread rates, 157
SFPE Engineering Guide to flame and burnout fronts,
Performance-Based Fire 136–137
Protection, 321 liquids, 151–155
shaft vents, 323–325 model, 138
World Trade Center, New York, natural flow, 136–137
325–326 opposed-flow, 136–137
Fire science porous solid arrays
fire, definition, 3–4 bulk densities, 149
natural causes crown fire, 148
earthquake, 5–7 prescribed burning, 148
lightning, 4 Rothermel model, 148
meteors, 7–8 urban wildland fire, 148
underground fires, 8–9 wildland fires, 148
volcanoes, 8 wind, 149, 151–152
phenomena wood crib, 147
ceiling flame plume pyrolysis, 136
interactions, 27 school gymnasium, 158–160
room fire growth, 23–24 on solid surfaces
smoke streaks, 26 downward/lateral wall, 142–145
research heat flow rate, 140
computer simulations, 20–21 thickness effect, 141–142
disciplines underlying, 20 upward/wind-aided, 145–146
fire science history, 21–22 standard test methods, 157–158
scientific notation, 28–29 velocity, 136, 138–139, 143, 145, 192
symbols, 28 wind-aided, 136–137
Index 409
Flammable, 17, 63, 107, 109, 111 Heat transfer; see also Heat flux
Flashover, 270 conduction
compartment fires, 297–298 steady, 78–80
Mount Carmel Branch Davidian thermal penetration time,
Fire, 348 80–81
stages, 270 convection, 81–84
Flashpoint, 59, 107–108, 111, 139, 152, heat flow/chemical energy rate, 76
154, 166 law of heat conduction, 77
Fluid mechanics, 20, 82 radiation, 77, 84–92
Forced flow, 65, 137, 140 absorption coefficient, 86
FPETool, 329 blackbody, 85
Free burning, 249, 305 configuration/view factor,
FTC, see Federal Trade Commission 86–88
Fuel chemistry, 36 emissivity, 85
Fuel-controlled fire, 270, 313 frequency, 84
Fuel lean, 224, 230, 244 opaque surface, 85
Fuel lean/overventilated, 244–250, 253, pool size, 90–91
260–261, 270 soot particles, 85–86
Fuel rich, 244, 270, 329 to target from hot surface, 86–87
Fuel rich/underventilated, 244–252, 256, transparent surface, 85
260, 305–307 thermal energy, 75–76
Fully developed fire, 268–272, 277, Hemoglobin, 254
300–308, 314–315, 339, 348, 355, Hifukusho-ato district of Tokyo, 5
358, 362 Hot surface ignition, 109–110
Furniture calorimeter, 188, 193–197 HRP, see Heat release parameter
Humidity, 106, 263
Hyperthermia, 262
G
Glowing ignition, 112–114 I
Graphs, 387–389
IAFSS, see International Association for
Fire Safety Science
Ignition; see also Piloted ignition
H
evaporation in liquids, 104–106
Heat effects, 261–263 liquid fuels
Heat flux; see also Heat transfer autoignition, 104, 108–110
conduction, 80 piloted ignition, 104, 107–108
convection, 81–84 pure, 106–107
damage, 93–94 solid fuels
definition, 76 CHF, 112
from flames, 98 ignition temperature, 112–114
radiation, 85–86, 88–90, 92 physical thickness, 116
smoke polyethylene, 111
ceiling, 96–97 polymers, 111
fire source to cold object, 97 thick materials, 119–123
floor, 95 thin objects, 117–118
Heat of gasification, 168–169, 173 wood, 111–112
Heat release parameter (HRP), temperature, 112–114
182–184 time for flaming, 114–116
410 Index
Internal energy, see Thermal energy logarithms, 386–387

International Association for Fire Safety negative and zero power, 384–385
Science (IAFSS), 22 power operations, 383
International System of Units, 25–27, 74, scientific calculators, 384–385
376, 389–393 conservation laws
Irritant gases, 257 conservation of energy, 377–378
conservation of mass, 377
conservation of momentum,
J
378–380
Jet flames, 222–224 laws of nature, 380–381
graphs, 387–389
mass and energy, 376–377
K
units and conversions
Kanto earthquake, 5, 237, 367 equations and dimensions,
Kelvin (K), 27–28, 75, 85 391–393
SI units, 388–391
Meteors, 7–8
L
Mount Carmel Branch Davidian Fire,
Lateral/downward flame spread, 142–145 340–342
Law of heat conduction, 20, 77 background, 342–343
LFL, see Lower flammable limit clean burn pattern, 356–357
Lightning, 4 fire cause, 349
Line fires, 222 flashover, 348
Liquid fuels; see also Solid fuels follow-up, 351
autoignition, 104, 108–110 gasoline versus fire damage, 357–359
piloted ignition, 104, 107–108 hands of time in fire, 357
pure, 106–107 ignitions, 344–348
Lower flammable limit (LFL), 57–60, 104, scientific analyses, 351–354
107–108, 111–112 soot patterns, 355–356
survivability, 350–351
tear gas, 349–350
M
visual data, 343–344
Mass and energy, mathematics wind, 350
concepts, 376–377
Mass burning flux
N
burning rate
burning walls, 175–176 Narcotic gases, 254–256
computing, 170 National Bureau of Standards (NBS), 13
definition, 167 National Fire Protection Association
maximum burning rates, 178–179 (NFPA), 10–11, 14
nonuniform burning, 168 National Institute of Standards and
pool fires, 176–177 Technology (NIST), 20
Mass concentration, 252–253, 302 Natori firepower diagram, 205
Mass loss rate, 166, 169, 171–172, 187, Natural fire combustion; see also
302, 308 Combustion products
Mass optical density, 258–259 definition, 34
Mathematics of science diffusion flames, 38–39
algebraic operations anatomy, 48–54
constants, 385–386 candle flame, 39–47
Index 411
cigarette smoke plume, 41–42 solid fuels, 111–113, 116, 119–122

gravity, 40–42 solid properties
process, 39–40 asphalt shingle, 125
pyrolysis, 39 gypsum board, 126
shapes, 43 plywood, 128
turbulence, 40–42, 54–56 polyurethane, 127
energy levels, 34–35 TRP, 126–131
fire triangle, 36–37 time for flaming, 115
fuel chemistry Pool fire flames, 224–2249
complete/ideal reactions, 36 PRC, see Product Research Committee
incomplete reactions, 36 Premixed flames, 38–39; see also
products, 34–35 Diffusion flames; Smoldering
premixed flames, 38–39 examples, 57
examples, 57 flame temperatures, 59–61
flame temperatures, 59–61 laminar flame propagation, 57–59
laminar flame propagation, 57–59 turbulent propagation to
turbulent propagation to detonation, 61
detonation, 61 Prescribed burning, 148
smoldering, 38–39, 62–64 Product Research Committee (PRC), 15
spontaneous, 64–69 Pyrolysis, 39, 111–112, 131, 135–136
temperature levels, 34–35
time and extent, 37–38
R
Natural flow, 136–137
NBS, see National Bureau of Standards Radiation, 77, 84–92
Neutral plane, 279, 285–286, 311, 353 absorption coefficient, 86
Newton’s Second Law of Motion, 274 blackbody, 85
NFPA, see National Fire Protection configuration/view factor, 86–88
Association emissivity, 85
NIST, see National Institute of Standards frequency, 84
and Technology opaque surface, 85
pool size, 90–91
soot particles, 85–86
O
to target from hot surface, 86–87
Opposed flow, 136–137, 142, 145–146, 149, transparent surface, 85
155, 157 Radicals, 37, 50
Oxygen bomb, 181 Radio waves, 84–85
Oxygen consumption calorimeter, 187 Rankine (°R) scale, 27, 75
Oxyhemoglobin, 254 Relative humidity, 106
Reradiation, 170, 172–175
Respiration minute volume (RMV), 256
P
RMV, see Respiration minute volume
Parts per million (ppm), 253 Rothermel model, 148
Pattern analysis, 341, 354
Performance-based design, 320–321, 327
S
Performance codes, 321, 327
Performance fire safety codes, 327 SFPE, see Society of Fire Protection
Piloted ignition; see also Ignition Engineers
definition, 104 SFPE Engineering Guide to Performance-
liquid fuels, 107–108 Based Fire Protection, 321
412 Index
Shock wave, 8, 56, 61 Thermodynamics, 20, 74

Smoke visibility, 258–260 Thermophoresis, 356–357
Smoldering, 33–36, 38–39, 62–64, 66–69, Trifecta of fire growth, 166, 298
112–114, 245, 272–273, 288–291, TRP, see Thermal response parameter
336–340
Society of Fire Protection Engineers
U
(SFPE), 21
Solid fuels; see also Liquid fuels UFL, see Upper flammable limit
ignition Underground fires, 8–9
CHF, 112 Units and conversions
ignition temperature, 112–114 equations and dimensions, 391–393
physical thickness, 116 SI units, 388–391
polyethylene, 111 Unsteady burning, 171–172, 208
polymers, 111 Upper flammable limit (UFL),
thick materials, 119–123 57–60, 108
thin objects, 117–118 Upward/wind-aided flame spread,
wood, 111–112 145–146
Soot, 50–54 Urban wildland fire, 148
Species, 37–39, 50, 52–55, 63
Specific heat, 27, 29, 79–80, 117, 132, 139,
V
156–157, 226–227, 230
Spontaneous combustion, 34, 64–69, 109 Vaporization temperature, 173, 175
Stack effect, 285, 315 Vapor pressure, 105–108, 112
Standard International (SI) Units, see Vehicle fire behavior, 206–207
International System of Units Ventilation factor, 292, 295–296, 299, 301,
Stefan–Boltzmann constant, 85 303, 308
Steiner tunnel test (ASTM E-84), 157–158 Ventilation limited, in compartment,
Stoichiometric air-to-fuel mass ratio, 249, 270–272, 295, 298–299
224, 244 Viscous, 151
Stoichiometric mixture, 38 Vitiation, 254, 267, 279, 302
Surface tension, 151–152 Volcanoes, 8
Synthesis, 22, 54 Volume fraction, 52–53, 252–253
Vortex, 220–221
Vortex shedding, 220
T
Temperature scales, 75
W
Thermal conductivity, 27, 29, 78–79,
119, 296 Waldbaum fire, 332–334
Thermal diffusivity, 29, 79–81, 141 cockloft, fire growth, 338–339
Thermal energy, 75–76, 226, 230 ignition, 334–336
Thermal inertia, 119, 296 roof, 337–338
Thermally thin, 116, 181 roof collapse, 339
Thermal penetration time, 80–81 smoldering stage, 336–337
Thermal resistance, 79 Water density, see Water flux density
Thermal response parameter (TRP), Water flux density, 168
126–131 Water vapor, air, 105
Thermal runaway, 35 Wavelength, 84–86, 344
Thermocouple, 48, 121–122, 232, 235, Wind aided spread, 136–138, 145–147,
309, 313 149, 152, 157, 159–160
Index 413
Work, 74–75 Y
World Trade Center fire, 359
Yields, 246–249
afterthoughts, 365–366
fuel load, 362–363
Z
investigative efforts, 360–362
jet fuel, 362 Zone model, 366–369
steel structure, 363–365 Zukoski number, 227, 239

James G. Quintiere - Principles of Fire Behavior, Second Edition-CRC Press (2016)

Uploaded by

Copyright:

Available Formats

James G. Quintiere - Principles of Fire Behavior, Second Edition-CRC Press (2016)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

James G. Quintiere - Principles of Fire Behavior, Second Edition-CRC Press (2016)

Uploaded by

Copyright:

Available Formats

Principles of

Boca Raton London New York

CRC Press is an imprint of the

© 2017 by Taylor & Francis Group, LLC

No claim to original U.S. Government works

Printed on acid-free paper

International Standard Book Number-13: 978-1-4987-3562-9 (Hardback)

Library of Congress Cataloging‑in‑Publication Data

Names: Quintiere, James G., author.

Visit the Taylor & Francis Web site at

and the CRC Press Web site at

and the father of fire science. He developed the science for

many aspects of fire before others. He set the path.

1. Evolution of Fire Science ...............................................................................1

2. Combustion in Natural Fires...................................................................... 33

3. Heat Transfer ................................................................................................. 73

4. Ignition ......................................................................................................... 103

5. Flame Spread ............................................................................................... 135

6. Burning Rate................................................................................................ 165

7. Fire Plumes .................................................................................................. 213

7.6.2 Flame Height and Temperature Calculations.................. 232

8. Combustion Products ................................................................................ 243

9. Compartment Fires ..................................................................................... 267

9.4.3 Smoke Movement in a Building......................................... 281

10. Design, Investigation, and Case Studies ............................................... 319

10.3.4 Example 5: The Branch Davidian Fire near Waco,

AIT Autoignition temperature

a Constant associated with t-squared fires, Q = at 2

m a,max Maximum air flow rate

• Describe what is fire

詹姆士 G. 昆棣瑞著 = Fundamentals of fire phenomena/James G.

1.2 What Is Fire?

and O; polyvinyl chloride (a plastic) contains H, C, and O plus Cl atoms; and

1.3 Natural Causes of Fire

FIGURE 1.1 (Continued )

Fukushima nuclear facility caused explosions in three reactors. These were

1.3.5 Underground Fires

1.4 Fire and War

1943 at Dugway Proving Grounds in Utah on a simulated city known as

1.5 Fire in the United States and Abroad

1.5.1 U.S. Statistics

308 ´ 106 people

Playing with heat source

Exposure to other fire

Electrical and lighting

1.5.2 United States and the World

1.5.3 U.S. Fire Prevention Infrastructure

b From Guo and Fu.11

to investigate the consequences of fire on building construction materials.

1.5.4 Motivation for Improvement

1.5.5 Flammability Tests

The disparity among flammability tests for construction materials is

Flammable, an oddity, chiefly useful in saving lives. The common word

1.5.6 Cost of Fire

1.6 Fire Research

1.6.1 Disciplines That Underlie Fire

1.6.2 Computer Simulations and Physics

for combustion and turbulence. FMGlobal has embarked on an alternative

1.6.3 Brief History of Fire Science

In 1949, England developed one of the most advanced scientific labo-

a Constant associated with t-squared fires, Q = at 2

m a,max Maximum air flow rate

q (0.034 W/m-K )( 40 - 20)° C