You are on page 1of 384

PRINCIPLES

OF SMOKE
MANAGEMENT
This publication was made possible by funds from ASHRAE research.

Principles of Smoke Management by John Klote and James Milke is an exhaustive treatment of smoke man-
agement, including pressurized stairwells, pressurized elevators, zoned smoke control, and smoke manage-
ment in atria and other large spaces. Recent advancements include heat release rate, toxicity of smoke,
natural atrium venting, plugholing, minimum depth of an atrium smoke layer, smoke stratification, smoke
detection, tenability systems, and computer analysis. The book includes numerous example calculations.
Methods of analysis include equations, network flow models, zone fire models, scale modeling, and hazard
analysis. Computational fluid dynamics (CFD) is also addressed. The book includes a CD of computer soft-
ware for ar~alysisof smoke management systems.

This publication was prepared under ASHRAE Research Project 1122.


Cognizant TC: TC 5.6, Fire and Smoke Control.

ABOUT THE AUTHORS


John H. Klote, DSc., P.E., Fellow ASHRAE, is a consulting engineer specializing in the design and
review of smoke management systems, as well as code consulting and teaching private smoke management
courses. He conducted research for 19 years at the National Institute of Standards and Technology (NIST)
and has published over 80 papers and articles on smoke management and other aspects of fire protection.
Dr. Klote headed the Building Fire Physics Group at NIST, which conducted research in smoke niove-
ment in buildings. The tools used for this research included full-scale fire experiments, scale model fire
experiments, network airflow models, zone fire models, and computational fluid dynamics (CFD). Klore
acted as a consultant in the area of smoke movement for the investigations of the MGM Grand fire and the
First Interstate Bank fire. Klote's research was the basis of the 1997 revision to the NFPA Life Safety Code
(section 5-2.13), allowing elevators to be used as a second means of egress from towers.
In 1986, he earned a Doctor of Science degree in mechanical engineerins from George Washington Uni-
versity. He is a member of the National Fire Protection Association (NFPA). a fellow of SFPE, and a fellow
of ASHRAE. He has extensive participation in ASHRAE and NFPA committees, including being a past
chairman of ASHRAETC 5.6, Fire and Smoke Control. Dr. ~ l o t is e a registered professional engineer in the
District of Columbia, North Carolina, California, and Delaware.

James A. Milke, Ph.D., is an associate professor and associate chair of the Department of Fire Protec-
tion Engineering at the University of Maryland. Dr. Mike has been a member of thefaculty and staff of the
department since 1977. He received his Ph.D. in aerospace engineering from the University of Maryland,
with an emphasis in structures. He received an M.S. degree in mechanical engineering and a B.S. degree in
fire protection engineering, both from the University of Maryland. In addition. he has a B.S. degree in phys-
ics from Ursinus College.
Dr. Mike has served as a research fire prevention engineer at the Building and Fire Research Labora-
tory, National Institute of Standards and Technology, as the fire protection engineer for Fairfax County, Vir-
ginia, and as,a consultant to numerous organizations. Dr. Milke is a fellow of the SFPE and is a member of
the National Fire Protection Association. the International Association of Fire Safety Science. and the Amer-
ican Society of Civil Engineers. He is the chairman of the NFPA Technical Committee on Smoke Manage-
ment Systenis and the ASCWSFPE committee on Structural Design for Fire Conditions. He ser\.es on the
Fire Council of Underwriters Laboratories.
PRINCIPLES OF
SMOKE
MANAGEMENT

John H. Klote
0

James A. Milke

American Society of Heating, Refrigerating


and Air-conditioning Engineers, Inc.

Society of Fire Protection Engineers


ISBN 1-883413-99-0

02002 American Society of Heating, Refrigerating


and Air-conditioning Engineers, Inc.
1791 Tullie Circle, N.E.
Atlanta, GA 30329

AI1 rights reserved.

Printed in the United States of America

ASHRAE has compiled this publication with care, but :W-IRAE has not investigated, and ASHRAE expressly disclaims any duty
to investigate, any product, service, process, procedure, design, or the like that may be described herein. The appearance'of any
technical data or editorial material in .this publication does not eonstitute endorsement, warranty, or guaranty by ASHRAE of any
product, service, process, procedure, design, or the like. ASHRAE docs not warrant that the information in the publication is free
of errors, and ASHRAE does not necessarily agree with any statement or opinion in this publication. The entire risk of the use of
any information in this publication is assumed by the user.

No part of this book may be reproduced without permission in writing from ASHRAE, except by a reviewer who may quotc brief
passaees or reproduce illustrations in a revicw with appropriate crcdit; nor may any part of this book be reproduced, stored in a
retrieval system, or transmitted in any way or by any means--electronic. photocopying. recording, or other-without permission
in writing from ASHRAE.

ASHRAE STAFF

Mildred Ceshwiler Barry Kurian


Editor manager

Erin Howard Jayne Jackson


Assistant Editor Pi-od~ictioi~
Assistant

Christina Johnson
Editorial Ass b r n / ~ t PUBLISHER

W. Stephen Cornstock
DEDICATION
This book is dedicated to the memory of George T. Tamura, who conducted pioneering research in smoke control at
the National Research Council of Canada.
TABLE OF CONTENTS
Chapter Page

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X

Chapter I-Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . l

Chapter 2-Fire and Heat Release . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1

Chapter 3-Smoke and Tenability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

Chapter &Evacuation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .49

Chapter 5-Effective Areas and Smoke Movement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Chapter &-Principles of Smoke Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

Chapter 7-Air Moving Equipment and Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

Chapter 8 . 4 omputer Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

Chapter 9-Hazard Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

Chapter 10-Stainvell Pressurization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139

Chapter 1 l-Elevator Smoke Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

Chapter 12-Zoned Smoke Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171

Chapter 13-Fundamental Concepts for Atria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181

Chapter 14-Atrium Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199

Chapter 15-Physical Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

Chapter 16-Computational Fluid Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225

Chapter 17-Commissioning and Routine Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235

Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247

Appendis A-Units of Mcnsurcmcnt and Physical Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259

vii
AppendixB-Bibliography .................................................................... 271

Appendix C-Calculation of Elevator Evacuation Time ............................................. 277


Appendix D-Application of CONTAMW ........................................................ 289

Appendix E-ASMET Documentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321

Appendix F-ASET-C: A Room Fire Program for Personal Computers ................................. 329

Appendix G-Data and Computer Output for Stairwell Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337

Appendix H-Data and Computer Output for Zoned Smoke Control Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349

Appendix I-Inspection Procedures for Smoke Control Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355

Appendix J-Test Procedures for Stairwell Pressurization Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361

Appendix K-Test Procedures for Zoned Smoke Control Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365

Appendix L-Inspection Procedures for Atria Smoke Exhaust systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369

Appendix M-Test Procedures for Atria Smoke Exhaust Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
PREFACE
In 1983, ASHRAE published Design of Smoke Control Systems for Buildings, written by myself and John
Fothergill. This book was the first attempt to consolidate and present practical information about smoke control
design. Judging by the many favorable comments and suggestions about this first book, I feel that it was a success.
The first publication was limited to systems that control smoke by means of the physical mechanisms of pressuriza-
tion and airflow.
In 1992, ASHRAE and SFPE jointly published Design of Smoke Management System written by myself and
James Milke. The term smoke management was used in the title of this publication to indicate that the physical mech-
anisms were expanded from pressurization and airflow to include compartmentation, dilution, and buoyancy. Based
on heightened concerns about supplying combustion air to the fire, a caution was added about the use of airflow for
smoke management.
This new publication addresses the material of the two earlier books plus people movement in fire, hazard analy-
sis, scale modeling, and computational fluid 'dynamics. In addition, the material about tenability and atrium smoke
management has been extensively revised. As with the other books, this new book is primarily intended for designers,
but it is expected that it will be of interest to other professionals (code oficials, researchers, etc.).
This book and its predecessors are different from other design books in a number of respects. This book is writ-
ten in both English units (also called IP, for inch-pound) and S1 units so that it can be used by a wide audience. To the
extent practical, equations are accompanied by derivations and physical descriptions of the mechanisms involved.
The physical descriptions are worked into the text as simple explanations of how particular mechanisms, processes,
or events happen. The goal of the derivations and physical descriptions is to provide information and understanding
so that readers can apply the material of this book in creative and insightful ways.
As with the first two publications, I hope that this book is of value to the engineering community. Further, I
invite readers to mail their suggestions and comments to me at the address below:

John H. Klote, D.Sc., P.E.


I I I I Carper Street
McLean, VA 22 l0 l
ACKNOWLEDGMENTS
This project would not have been possible without the support of the American Society of Heating, Refrigerating
and Air-conditioning Engineers (ASHRAE). Acknowledgment is made to the members of the ASHRAE Smoke
Control Monitoring Committee for their generous support and constructive criticism. The members of this subcom-
mittee are:

Williarn A. Webb, Chairman (Performance Technology Consulting, Ltd., Lake Bluff, Ill.)
John A. Clark (Eagan, Minn.)
Dave Elovitz (Energy Economics, Inc., Natick, Mass.)
Gary Lougheed (National Research Council Canada, Ottawa, Ontario)

The support and advice of the staff of the Building and Fire Research Laboratory (BFRL) at the National Insti-
tute of Standards and Technology (NIST) in Gaithersburg, Md., was invaluable. Particular appreciation is expressed
to Richard Bukowski, Glen Fomey, and Richard Peacock. Special thanks are due to Daniel Madrzykowski for his
advice regarding oxygen consumption calorimetry and heat release rate. The authors are indebted to Kevin McGrat-
tan of BFRL for his valuable advice and constructive criticism regarding computational fluid dynamics.
Richard Gann and Barbara Levin of N E T and Emil Braun of Hughes Associates, Baltimore, Md., provided valu-
able information and insight concerning the evaluation of the effects of toxic exposures. Creg Beyler of Hughes
Associates provided constructive criticism in a number of areas. Special thanks are due to Gary Lougheed for his con-
structi\-e criticism and for tlie body of relevant research conducted by him and his associates at the National Research
Council of Canada.
Students of fire pro~ectionengineering at the University of Maryland have provided insightful comments on
drafts of several chapters of this book In particular, the students Suzelte Hartmann and Julie Naviaser developed the
information about CONTAMW that is included as Appendix D.
The content of this book is heavily dependent upon tlie work of many researchers, design engineers, and other
professionals around the world. So many of these people have provided experimental research results, system con-
cepts, and analytical methods that it is impossible to thank them all individually. Appreciation is expressed to all
those u h o have contributed to the advancement of smoke managemen1 technology directly or indirectly by their con-
tributions to fire science and fire protection engineering.
CHAPTER 1

Introduction

moke is recognized as the major killer in fire situa- The MGM Grand is not unique in this respect, as is

S tions. Smoke often migrates to building locations


remote from the fire space, threatening life and
damaging property. Stairwells and elevator shafts fre-
illustrated by the fires at the Roosevelt Hotel (Juillera~t
1964) and Johnson City Retirement Center (Steckler et
al. 1990). All of these fires were located on the first
quently become smoke-logged, thereby blockin,0 evacu- floor, but the majority of deaths were on upper floors
ation and inhibiting rescue and fire fighting. The MGM (Figure 1.l).'
Grand Hotel fire (Best and Demers 1982) is an example
of the smoke problem. The fire was limited to the first
floor, but smoke spread throughout the building. Some
l , During the intensive activity of fire fighting and
occupants on upper floors were exposed to smoke for rescue, the locations of some of the bodies are not
hours before rescue. The death toll was 85, and the recorded. Thus Figure 1.1 is limited to the deaths for
majority of the deaths were on floors far above the fire. which the locations were known.

Retirement
8 Center Fire
23 L 7
Johnson City. TN
22
21
g Dec 24,1989
E 4
20
1
19
18 2
I
17
0 1 2 3

MGM Grand Hotel Fire Deaths


Las Vegas, NV
Nov 21,1980

Note: Floors Renumbered for

L.,--
I
0 1 2 3 4 5 6 7 8 9101112131415161718 I

Deaths 0 1 2 3 4 5 6 7 8

Deaths
Figure I .I Deaths byjloor for three fires where rhefire was locn~ed017
rile firsrjloot:
Chapter l - Introduction

Figure l .2 Floor plan of the Health Care Test Faciliy at the ArIST Annex.

The general public is unaware of how fast a fire can Such very rapid fire growth and accompanying
grow and of how much smoke can be produced by a fire. smoke production represent a real possibility in .actual
This unawareness extends to many designers and other wardrobe fires and perhaps even closet fires. Many
related professionals. Because such an awareness is nec- other fire scenarios are possible. For example, a latex or
essary to the evaluation of design parameters for smoke a polyurethane filled mattress ignited by an adjacent
management systems, the following example is pro- wastebasket fire would reach about the same stage of
vided. development in six minutes that wardrobe test N-54
reached in two minutes.
This example is fire test N-54, performed at the
Health Care Test Facility at the National Institute of Full-scale fire tests by Bennetts et al. (1997) and
Standards and Technology Annex in Gaithersburg, Md. Lougheed et al. (2000, 2001) have shown that success-
For technical details of this unsprinklered fire test, the fully sprinklered fires can continue to bum and produce
reader is referred to a report by O'Neill et al. ( 1 980). The enormous amounts of dense buoyant smoke after sprin-
floor plan of the test facility is shown in Figure 1.2. kler activation. While it appears this smoke production
is greatest for fires that are shielded from sprinkler
In this test, various fabrics representing common spray, some unshielded fires still produced considerable
clothing materials were hung on wire coat hangers and amounts of buoyant smoke.
arranged loosely in a wooden wardrobe. A cardboard
The concept of smoke management has developed
box containing crumpled newspaper was placed on the
as a solution to the smoke migration problem.2 Smoke
floor of the wardrobe. The test started when the crum-
movement can be managed by use of one or more of the
pled newspaper was ignited by a match. Following igni-
following mechanisms: compartmentation, dilution, air-
tion, the left-hand door of the wardrobe was closed
flow, pressurization, or buoyancy. These mechanisn~s
tightly while the right-hand door was left partially open
are discussed in detail in Chapter 4. The use o f pressur-
resulting in a 3 in. (76 mm) opening along the vertical
ization produced by mechanical fans is referred to as
edge of the door.
snloke control by NFPA 92A (NFPA 2000). By this def-
At one second after ignition, no flame or smoke inition, stairwell pressurization (Chapter 7), elevator
was visible. At 80 seconds, flames were visible flowing pressurization (Chapter 8), and zoned smoke control
from the top of the wardrobe, a layer of smoke was cov- (Chapter 9) are all types of smoke control systems.
ering the ceiling of the burn room, and smoke had The primary emphasis of this book is on systems
flowed into the corridor forming a one-foot-thick layer that cse pressurization produced by mechanical fans.
just below the corridor ceiling. At 110 seconds, flames The use of pressurization to control the flow of undes-
were flowing from the top two-thirds of the wardrobe ired airborne matter has been practiced for at least 50
opening, and the smoke flowing out of the burn room years. For example, it has been used in buildings, such
doorway had increased significantly. At 120 seconds as experimental laboratories, where there is danger of
after ignition, flames were flowing from the entire open-
ing of the wardrobe door, and the layer of smoke in the 2. As discussed later in "Preliminary Design Con-
corridor and lobby had descended to approximately 4 ft siderations," smoke management is only one of many
(1.2 m) below the ceiling. techniques available to h e protection engineers.
Principles of ~ m o k ~ ~ a n a ~ e m e n t

poison gas, flammable gas, or bacteriological material methods are employed to minimize the possibility of
migrating from one area to another; it has been used to doors being propped open.
control the entrance of contaminants where a dust-free While advances in tenability analysis have made
environment is necessary; it has been used wheremdia- engineering analysis of smoke shafts feasible, these sys-
tion migration and contamination could occur; and it has tems are not included in this book. The idea of smoke
been used in hospitals to prevent the migration of bacte- shafts is that smoke flows up the shaft due to bgoyancy
ria to sterile areas. However, the use of airflow and pres- where the smoke flows away from the building, but the
surization to control smoke flow from a building fire is a authors have concerns about the fundamental effective-
fairly recent adaptation. ness of smoke shafts. Further, there seems to be little
interest in smoke shafts.
INTENT The stair systems known as "smokeproof' towers
The primary intent of this book is to provide practi- are misnomers, in that there is nothing about them that
cal state-of-the-art infoimation to engineers who have ensures no smoke migration into stairs. Originally, these
been charged with design of smoke management sys- towers were separate from the building and were con-
tems. The book is also intended to provide information nected to it only by walkways open to the outside. Some
for the review of designs and development of codes and versions of these towers used relatively small openings
standards,. This chapter contains general background
'
in exterior vestibule walls in place of the separate walk-
informati6n; Chapter 2 deals with fire development and ways. In the absence of an engineering analysis of these
the heat :release rate of fires. Chapter 3 discusses the systems, it can only be stated that the benefits of these
nature of:s.moke, including toxicity, heat exposure, and systems are questionable. For these reasons, separated
visibility through smoke. Chapter 4 ciiscusses people stair towers are not included in this book, and it is rec-
movement during- fire evacuation. ommended that the term "smokeproof' towers not be
Chapter 5 is devoted to smoke movement in build- wed.
ings, and the individual driving forces of smoke move-
ment are discussed in detail. Chapter 6 contains a EQUATIONS AND UNITS
fimdamental discussion of topics that are essential for OF MEASUREMENT
the design of systems to manage smoke movement. It Considering that this book is primarily intended for
discusses the mechanisms of compartmentation, dilu- design, it seems most appropriate that units should be
tion, airflow, pressurization, and buoyancy, which are specified for every equation. However, the topic of
used by themselves or in combination to manage smoke smoke management is relatively new, and there is no
conditions in fire situations. test to refer to for the derivation of many of the equa-
Background information is provided about ducts, tions used. Further, it was desired that the text be in both
fans, fire dampers, smoke dampers, and fan-powered Inch-Pound (IP) units and the International System (S[)
ventilation systems in Chapter 7. Chapter 8 is a descrip- units. It would be unacceptably cunlbersome to present
tion of the computer programs that are used for the anal- derivations using both commonly used English units
ysis of smoke management systems. and S1 units. The equations used for derivations are
Chapters 9 through 14 address hazard analysis, dinlensionally homogeneous, and they can be used with
stairwell pressurization, elevator smoke control, zoned the S1 system, the slug pound system, and the pound
smoke control, and atrium smoke management. For mass poundal system (Appendix A). These dimension-
applications for which these conventional methods are ally homogeneous equations are easily identified
inappropriate, the methods of scale modeling and com- because no units are specified for them in the text. How-
putational'fluid dynamics (CFD) can be used (Chapters e\;er, all of the equations t h a ~the reader is IikeIy to use
15 and 16). Chapter 17 addresses the important topic of for design analysis are given in both English and S1
commissioning and routine testing. units. These equations are easily identified because the
It may be noted that pressurized corridors have appropriate units for the equation are specifically indi-
cated in the text.
been omitted. The principIes presented in this book can
~ ~

bz applied to pressurized corridors in a manner similar


to their application to other pressurization systems. The
HISTORY OF SMOKE VENTING
.
-

concern with pressurized corridors is that if a fire room Smoke venting has been used extensively to man-
door is blocked open, the corridor pressurization system age smoke flow during theater fires. The acceptance of
can force smoke into other rooms off the corridor. For such venting resulted from several major theater fires,
this reason, pressurized corridors are not generally rec- including those at the Brooklyn theater, which killed
ommended except for applications where practical 283 in 1877; the Vienna Ring theater. which killed 449
Chapter l -Introduction

A Are Areas 2 and 4 on Floor 10 ServiceTower

Experimental Tower

I
. m . .......
.............
m . m . I
............
iiilial:l
:::::.--p
.....
...........
............
Fire Area 3 on
Floor l 0 3 Smoke Shafl
m m m m ............m
m............ n m m n 4 ElwatorlSiau Lobby Supply

Figure 1.3 Typical floor plan of the office building at Figure 1.4 Typicalfloor plan of 117e NRCC exper-hen-
30 Church Street. talfire tower.

in 188 1; the Theater Royal, which killed 186 in 1887; klered fires. The term "smoke free" is used to mean
and the Iroquois theater, which killed 571 in 1903. All essentially free of smoke, with the possibility of such
of these fires started on the theater stage and resulted in insignificant amounts of combustion products that tena-
major loss of life in the audience. The Palace theater fire bility is maintained.
in Edinburgh in 1911 was an exception. In this fire, Other full-scale fire tests also demonstrated that
smoke venting through the stage roof was credited for pressurization could provide "smoke free" exits during
helping to prevent any loss of life. The buoyancy of the large unsprinklered fires (Koplon 1973a, 1973b;
hot smoke forced the smoke flow through the vent open- Butcher et al. 1976). Cresci (1973) describes visualiza-
ings, and this venting is called natural venting or gravity tion experiments using a model of the stair shaft at the
venting. Church Street building, where stationary vortices \\.ere
Over the past few decades, fan-powered smoke observed at open doonvays. These vortices are the rea-
exhaust has become the standard for almost all atria in son that the flow coefficient through an open stainvell
North America. In other areas, such as Europe, Austra- door is about half of what it \i.ould be otherwise. This
lia, and New Zealand, both natural venting systems and significant effect on airflow is discussed in Chapter 6.
fan-powered exhaust systems have become common for The Research Tower near Ottawa (Figure 1.4) was
atria. Modem atria smoke management designs are used for a joint National Institute of Standards and Tech-
based on engineering analysis developed over the last nology (NIST) and National Research Council Canada
few decades. These analytical methods are primarily (NRCC) study of elevator smoke control. Again, i t was
based on research in smoke plumes andzone fire model- demonstrated that pressurization could control smoke
ing. Information about these analytical methods is pro- from large unsprinklered fires (Tamura and Klote 1987a,
vided in Chapters 13 and 14. 1987b, 1988; Klote and Tamura 1986a, 1986b, 1987).
In the spring of 1989, NIST conducted a series of
HISTORY OF PRESSURIZATION experiments of zoned smoke control at the Plaza Hotel
SMOKE CONTROL in Washington D C , as shown in Figure 1.5 (Klote
1990). A zoned smoke control system is a system that
The idea of smoke protection by pressurization sys- uses pressurization to restrict smoke migration to the
tems is .to restrict the movement of smoke from a build- zone of fire origin. Once again, it was demonstrated that
ing fire. To study the effectiveness of pressurization pressurization could control smoke from large unsprin-
smoke control, the Brooklyn Polytechnic Institute con- klered fires.
ducted a series of fire experiments at a 22-story office An analysis based on first principles of engineering
building at 30 Church street in New York City (DeCicco was made of the pressure differences produced by the
1973). This building was scheduled for demolition. smoke control'system during the fires at the Plaza Hotel.
Materials representative of fuels that would be in an As is done with zone fire modeling, the pressures n.ithin
office were burned on floors 7 and 10, as shown in Fig- rooms were considered hydrostatic. The general trends
ure 1.3. This project demonstrated that pressurization of calculated values were in agreement with the msa-
could provide "smoke free" exits during large unsprin- surements (Figure l h ) , and this indicates a levc.1 of
Principles of Smoke Management

applicability of zone fire modeling for analysis of pres- barriers, such as partitions and closed doors. Atrium
surization smoke control systems. smoke exhausts often are designed to keep smoke from
descending below a specific level. Further, various types
OBJECTIVES O F SMOKE MANAGEMENT of smoke management systems can be designed to
maintain tenable conditions within specific spaces.
Some objectives of a smoke management system
are to reduce deaths and injuries from smoke, reduce
property loss from smoke damage, and to aid firefight- PRELIMINARY DESIGN
ers. Many designers feel that life safety is the primary CONSIDERATIONS
objective of smoke management; however, systems
have been built with the primary objective of protecting Smoke management should be viewed as only one
property--especially high-value equipment. Regardless part of the overall building fire protection systems. Two
of the objective, the methods of design analysis pre- basic approaches to fire protection are to prevent fire
sented in this book are applicable. ignition and to manage fire impact. Figure 1.7 shows a
Theoretically, a smoke management system can be simplified decision tree for fire protection. The building
designed to provide a safe escape route, a safe refuge occupants and managers have the piimary role in pre-
area, or both. However, a pressurization (smoke control) venting fire ignition. The building design team may
system can meet its objectives even if a small amount of incorporate features into the building to assist the occu-
smoke infiltrates protected areas. For this book, pressur- pants and managers in this effort. Because it is impossi-
ization systems are designed on the basis that no smoke ble to prevent fire ignition completely, managing fire
infiltration will occur. Hazard analysis (Chapter 9) can impact has assumed a significant role in fire protection
be used for the design of systems that maintain tenabil- design. Compartmentation, suppression, control of con-
ity even when people come into contact with some struction materials, exit systems, and smoke manage-
smoke. ment are examples. The NFPA Fire Protection
Handbook (NFPA 1997), SFPE Handbook of Fire Pro-
PERFORMANCE-BASED DESIGN tection Engineering (SFPE 2002), and NFPA 550
(NFPA 1995) contain detailed information about fire
In recent years, performance-based codes have safety.
become a topic of considerable attention. Traditional
codes prescribe requirements, while performance-based
codes require a level of performance. A perforrnance-
based design is developed to meet the level of perfor-
mance stipulated in the code.
This book uses a performance-based approach,
where the kind of performance is based on the type of
system. Pressurization smoke control systems are
designed to maintain specific levels of pressurization at

0' S l0 l5 20 25 3;
i7me (minutes)
(a) Pressure Difference Near Ceiling

0 0-
0 5 10 15 20 25 30
lime (minutes)
(b) ~ressureDierence Near Floor

Figure 1.5 Secot7djloor-plnt~oJthe Plnzn Ho~el. Figure 1.6 Co117par-isotio/ tneaszrt-ed and calczrlated
~ I ~ S I I dI f -i t~s n c e s ji-ot~l Plaza Hotel
tests.
Chapter 1-introduction

Objectives

Ignition Impact

TlTl.m/l
Heat-Energy

Sources
Source-Fuel

Interactions
Threat' Exposure'

'Note: Smoke management is one of many fire protection tools that can be
used to help manage the threat of fire and manage the exposure of fire.

~ i g u r e1.7 Sin~plifiedfir.eprotectiondecision tree.

Many factors will affect the design of a slnoke man- FLEXIBILITY A N D RESILIENCY
agement system. Before the actual mechanical design of To help ensure smoke management system perfor-
the system can proceed, the potential constraints on the mance, the approaches of flexibility and resiliency can
system should be determined and the design criteria be employed. The concept of flexibility consists of
established. This section introduces some considerations using design features that allow for easy adjustment of a
peculiar to smoke management system design, some of smoke management system in order to achieve accept-
which are merely listed below, since detailed discussion able performance. A resilient system is one that resists
is beyond the scope of this book. However. published serious adverse effects due to pressure fluctuations.
works on some of these subjects are cited in the bibliog- During the design of a new building, the leakage
raphy in Appendix B. paths throughout the building can only be estimated.
Therefore, the smoke management design calculations
Occupancy type and characteristics constitute only an approximate representation of the
Evacuation plan pressures and airflows that will occur as a result of the
Refuge areas smoke management system in the actual building. The
Distribution of occupant density introduction of flexibility into a smoke management
Human life support requirements system allows for variations in leakage from the origi-
Form of detection and alarm nally estimated values. Because it is difficult to measure
leakage paths in existing buildings, the concept of flexi-
Fire service response-to-alarm cliaracteristics
bility is also useful for retrofit of smoke management in
Fire suppression system characteristics existing buildings. In many systems, flexibility can be
Type of heating, ventilating, and air-conditioning achieved by the use of fans with sheaves3 to allow sev-
(HVAC) system eral flow rates, a variable flow fan for the same purpose,
Energy ~na~iagement system or by dampers that can be manually adjusted to obtain
Building security provisions desired pressure differences.
Controls Pressure fluctuations often occur during a fire when
Status of doors during potential fire condition doors are opened and closed and when windows are
Potential lirc threats opened, closed, or broken. To resist such fluctuations,
Internal compartmentation and arcliitectr~ralcharac- resiliency can, be incorporated in a system by use o f
teristics
3. A sheave is tlic whcel with a groovcd rim, sonie-
Bu~ldmgleakage paths
ti~ncscallcd a bclt whecl. By exchanging a sheave for
Exterior temperatures onc of anothcr dinmetcr. thc rotational spced of the fan
Wind vcloc~ty and its flow ratc are changed.
Principles of Smoke Management

automatic control to reduce the pressure fluctuations. slows down the burning rate, it reduces the smoke prob-
For example, in pressurized stairwells, automatic con- lem. From fires that are suppressed rather than extin-
trol can be used in the supply fan bypass system to guished, smoke is produced. This smoke can move
reduce the effect of opening and closing stairwell doors. through the building due to various driving forces dis-
An alternative is.to keep the exterior stairwell door open cussed in Chapter 5. OII the other hand, well-designed
during pressurization. This eliminates what is probably smoke management systems can maintain tolerable con-
the major source of fluctuations; that is, the opening and ditions along critical egress routes but will have little
closing of the exterior stairwell door. The concepts of effect on the fire.
flexibility and resiliency are discussed further where
In addition to the fact that the systems perform dif-
they apply to specific smoke management applications.
/ ferent functions, it is important that the designer con-
/ W' sider the interaction between smoke management and

a
SAFETY FACT0RS 9,
fire suppression. For example, in the case of a h l l y
Smoke management is still a relatively new field, sprinklered building, the pressure difference needed to
and it should come as no surprise that there is no CO -
control smoke movement is probably less than in an
sensus concerning safety factors, which are commonly
unsprinklered building, due to the likelihood that the
used in many branches of engineering to provide a level
maximum fire size will be significantly smaller than in
of assurance of system performance. Further, the topic
an unsprinklered building.
of safe@ factors has attracted little attention in smoke
control design. A pressurization (smoke control) system can
Safety factors for sizing fans of pressurization sys- adversely affect performance of a gaseous agent (such
tems are very different from those intended to maintain as halon, CO2, or NZ)suppression system when the sys-
a tenable environment in an atrium or other application tems are located in a common space. In the event that
based on a hazard analysis. If a pressurization fan is both systems are activated concurrently, the smoke
undersized, it will not maintain acceptable pressure dif- exhaust system may exhaust the suppressant gas from
ferences. This should be apparent and corrected during the room, replacing it with outside air. Because gas sup-
commissioning. pression systems commonly provide a single application
Ideally, an analysis of a system intended to maintain of the agent, the potential arises for renewed growth of
a tenable environment would be based on detailed and the fire.
accurate capabilities of simulating smoke transport, A general guideline would be that the gaseous agent
physiological effects of fire-related exposures, human suppression system should take precedence over the
response to fire, and evacuation analysis. However, this smoke control system. An extremely desirable feature in
technology is not so advanced, and these calculations such spaces would be the ability to purge the residual
are of necessity based on a number of conservative
smoke and the suppressant gas after the fire is com-
assumptions with conservative design parameters. It can
pletely extinguished and to replace them with fresh air.
be argued that such conservative calculations may result
This ability to replace the atmosphere in these spaces in
in conservative designs even in the absence of any
the post-fire period is very important from a life-safety
safety factors. The specifics of the design and the meth-
06s of analysis would be expected to have a significant viewpoint, since some gas suppressants are asphyxiants
impact on any approach to safety factors. at normal design concentrations.
~ e & u s eof the absence of any accepted approaches
to safety factors, this topic is not included in the meth- ENERGY CONSERVATION
ods of analysis of this book. The smoke management system must be designed
to override the local controls in a variable air volume
FIRE SUPPRESSION SYSTEMS HVAC system so that the air supply necessary to pres-
Automatic suppression systems are an integral part surize nonfire spaces is supplied. Also, if there is an
of many fire protection designs, and the efficacy of such energy management system or a 24-hour clock system,
systems in controlling building fires is well docu- the designer must ensure that the smoke management -
mented. However, it is important to recognize that while system will take precedence over the local control sys-
the functions of fire suppression and smoke manage- tem so that the necessary air is supplied or exhausted
ment are both desirable fire safety features, they should according to the design approach. It is a good general
not be readily substituted for each other. One of the best rule that smoke management should take precedence
ways to deal with the smoke problem is to stop smoke over energy conservation features in both new designs
production. To the extent that a suppression system and retrofits.
Chapter l - Introduction

SYSTEM ACTIVATION get out of the building. On the way down the stairs, he
thinks of his responsibility to the other occupants. He
System activation is probably the major area of dis-
stops on a lower floor long enough to actuate a manual
agreement in the field of smoke control. Primarily, this
station. If that alarm activated the smoke control system,
disagreement is about automatic activation versus man-
the wrong zone would be identified as the fire zone.
ual activation. In the early days of smoke control, there
was general agreement that activation of "pressure sand- Because of the long response time and the mainte-
wich" systems should be automatic upon alarm from nance problem of clogging with airborne particles, it is
smoke detectors. Automatic activation by smoke detec- generally agreed that smoke detectors located in HVAC
tors located in building spaces has the clear advantage ducts should not be the primary means of smoke control
of fast response. system activation. A means of activation of higher rell-
ability and quicker response time is needed. However,
Some building designers and fire service officials
an alarm from a duct-located detector can be used in
began to realize that smoke detectors could go into
addition to such a primary means of activation. A signal
alarm on a floor far away fiom the fire. Thus, automatic
fiom only this secondary means might be unusual, but it
activation by smoke detectors could result in pressuriza-
should be able to activate the smoke control system.
tion of the zone in which the fire occurred. This would
result in the opposite of the desired operation; that is, Most stairwell pressurization systems operate in the
smoke would be forced into other zones. As a result, a same manner regardless of where the fire is located.
Therefore, it generally is agreed that most stairwell pres-
vocal minority of officials feel that smoke control
should only be activated manually by fire fighters after surization systems can be activated by the alarm of any
they are sure of the fire location. However, many fire alarm-initiating device located within the building.
A possible exception to this is large buildings with hori-
involved professionals are concerned that such manual
zontal separations, such that smoke is not expected to
activation could be so late in the fire development that
have an impact on some stairwells remote from the fire.
significant hazard to life and damage to property would
result. Such delayed activation can suddenly transport a It is recommended that zoned smoke control sys-
body of smoke that is highly charged with unbumed tems be equipped with a remote control center from
hydrocarbons, carbon monoxide, and other toxic gases which the smoke control system can be manually over-
and depleted of oxygen to remote locations. This can ridden. This center should be easily identifiable and
result in a wave-like movement of toxic gases or flame accessible to the fire department. Such a remote control
to remote areas. center allows fire fighters to change the mode of smoke
control system operation in addition to system shut-
The most recent view on the subject is that zoned
down. Activation of smoke management systems for
smoke control should be automatically activated by an
atria and other large spaces is addressed in Chapter 10.
alarm from either heat detectors or sprinkler water flow.
This can only be accomplished if the detector or sprin-
RELIABILITY O F S M O K E MANAGEMENT
kler zones are compatible with the smoke control zones.
Using heat detector or sprinkler flow signals for activa- The intent of this section is to provide insight into
tion increases the likelihood of proper identification of the need for acceptance testing and routine testing and
the fire zone. For smoldering fires, this approach would the relative importance of system simplicity: The fol-
result in a significantly longer response time, and smoke lowing should not be thought of as an exhaustive treat-
detectors would probably be better suited for applica- ment of smoke management reliability. Due to the
tions where smoldering fires are of particular concern. difficulty of obtaining data about the reliability of com-
However, for flaming fires, it is believed that the ponents of smoke management systems, the simple cal-
response time with this approach would be short enough culations that follow are only very rough estimates.
so that significant benefit would be realized by the oper- However, it is believed that the insight gained justifies
ation of the smoke control system. It is hoped that this treatment despite these limitations. Further, the
advances in smoke detector technology and application same reliability concerns that apply to smoke manage-
will significantly improve the ability of these detectors ment systems apply to all life safety systems, and the
to positively identify the fire zone. following discussion may be of general interest beycnd
Throughout all of this controversy, there has been smoke management.
complete agreement that zoned smoke control should The discussion is limited to series systems, which
not be activated by alarms from manual stations (pull are systems that operate only if all the components oper-
boxes). The reason can be illustrated by the scenario ofa ate, as is true of many smoke management system
man who, while observing a fire on an upper floor of a designs. Redundancies (such as backup power) are not
building, decides that the first thing he should do is to included in this analysis. The reliability, R, o i a series
Principles of Smoke Management'

Table 1.1:
Estimated System Reliability for New Smoke Management
System That Has Not Been Commissioned

No. of HVAC No. of Other Reliability1 of New System Mean Lifez of Commissioned
System System Fans Components Before Commissioning System (months)
1 3 0 0.97 1 16
2 0 3 0.83 46
3 3 9 0.56 14
4 5 18 0.31 8
5 5 54 0.03 3
1. System reliabilities calculated from Equation (1.1). For purposes of these calculations, the reliabiliti&of fans ofa forced air HVAC system were
taken as 0.99, and other components were taken as 0.94.
2. Mean lives calculated from Equation (1.3). For purposes of these calculations. the failure rates of fans of a forced air HVAC system were taken as
104 per hour, and other components were taken a s I O - ~per hour.

system is the product of the reliabilities, Ri,of the.com- missioned system is if that component is part of an
ponents. .-: HVAC system. In hot or cold weather, building occu-
pants demand that the HVAC system provide comfort
conditions. Thus, for a new building in extreme weather,
it can be considered that the reliability of the HVAC sys-
tem fan will approach unity. Based on field observa-
tions, it is believed that other components will have a
Usually, discussions of reliability progress from this point
lower reliability. The following reliabilities were chosen
with the assumption that all components operate initially
for example calculations for new systems that have not
and that failures occur with time after system installation.
been commissioned:
For this assumption to be appropriate, a program of accep-
tance testing and defect correction is necessary. Such com-
Fans of a forced air HVAC system 0.99
missioning must include an installation check of all
components, tests of system performance during all modes Other components 0.94
of operation, repair of defects, and retesting until all defects
are corrected. Current construction practices are such that These values were arbitrarily selected, but the rela-
system commissioning is not always this exhaustive. For tive values between them are based on the discussion
above. Table I. 1 lists calculated reliabilities of such sys-
this reason, attention is first given to reliability of systems
tems made up of many components. It can be observed
without commissioning followed by a discussion of reli-
from this table that the more components a system has,
ability of systems for which all components operate after
the less likely the system is to operate before it has been
commissioning.
commissioned. The most reliable new system would be
RELIABILITY BEFORE COMMISSIONING one that only uses the HVAC system fans. A large com-
plicated system consisting of many components (Table
For newly installed components, the reliability can 1.1, system 5) has very little chance of operating before
be thought of as the likelihood that the component will commissioning. The trend of lower reliability for com-
both be installed properly and be in good working con- plicated systems agrees with observations of the author
dition when it is delivered to the construction site. There during nunixous field tests of systems of various
are an enormous number of errors that can occur during degrees of complexity. Probably the most important
manufacture, transportation, storage, and installation point to be made from this discussion is the need for
that can cause a component to fail to operate. Problcms commissioning of new systems.
such as motors wired for the wrong voltage, motors not
connected to power, dampers failing to close, fans run- MEAN LIFE OF COMMISSIONED SYSTEMS
ning backward, holes in walls, and automatic doors fail-
ing to close have been observed in newly built smoke For this discussion, all system components are con-
management systems. Based on experience \\lit11 tield sidered to operate-at the end of the commissioning pro-
testing of smoke management systems, it is estimated cess. A commonly used relation for the reliability of
that the reliability of components i n noncommissioned components is the exponential distribution,
systems is 0.90 or highcr. An imporlant consideration
regarding the reliabilily of a component in a noncorn- R; = exp(-),,r) . (1.2)
Chapter l - Introduction

I II "
Circuit Breakers
DistrobutionTransfomen I
Mechanical

I 1 /
Large
I
Electronic Valve Eq
P"U"~""'"S""

Figure 1.8 Typical ranges offailure rates (adaptedfioni Lees [ I 9801).

where ki is the failure rate of the component. The mean selected for example calculations, but their relative val-
life, L, of a system is ues are based on the above discussion:

Fans of a forced air I-[VAC system 1 o - per


~ hr
Other components Io-' per hr

Table 1.1 shows mean lives of systems composed of


various numbers of components. It can be observed that
Some typical ranges of failure rates of some coni- systems composed of a few components have long mean
ponents and systems are shown in Figure 1.8. It can be lives, while those made up of very many components
seen that failure rates vary over large ranges and that have short lives. This tends to support the view that sim-
failure rates vary considerably with equipment type. It ple systems are more reliable, and this view is supported
by obsenations in the field. However, it should be cau-
seems that the failure rate of HVAC system fans would
tioned that systems should not be overly simple; that is,
be lower than those of other components. If these fans
they should have the features needed to achieve desired
fail, building occupalits desiring heating o r cooling tend
performance at likely conditions during a fire. Further,
to put pressure on maintenance personnel to get fans
the above simple analysis did not include the beneficial
repaired quickly. Smoke management systems are only effects of redundancies. However, it is safe to conclude
needed for a short time over the life of a building. Thus, that unnecessary system complexities should be
when an HVAC system fan is called uron for smoke avoided. The mean lives listed in Table 1 .l also indicate
management duty, it seems that it will be more likely to that routine testing and repair of smoke management
operate than other components. To account for this, the systems is needed so that the systems will probably be
effective failure rate of HVAC system fans can be in good working order when they are needed. A similar
thought of a s being much smaller than other compo- statement can be made concerning all life safety sys-
nents. The following failure rates were arbitrarily tems.
CHAPTER 2

Fire and Heat Release


ky
1
robably the most important aspect of a building quences of a fire after ignition but not with the causes of

P fire is the heat release rate (HRR). The tempera-


ture and amount of gases produced by a fire are
directly related to the HRR, and predictive computer
Ignition.
Growth: After ignition, fire growth is determined
by the material burning, with little Or no influence from
models use the HRR as input. When talk the compa*ment. This stage is characterized by an
about the size of a fire or how big a fire is, they almost (
bundance
fire 2startingofin
airafor the fire.
corner of Figure 2.2 showschair
an upholstered an office
and
always are referring to the HRR. Other indicators of fire
growing until it spreads to other objects. As the fire
size are the fire area and fire perimeter, but neither of
grows, the temperature in the room rises. A fire with
these is commonly used to depict how big a fire is in the sufficient combustion air is called a fuel confrolled fire,
predictive models that have gained a high level of and such a fire is also referred to as burning infr-ee air.
acceptance in recent years. For these reasons, the term Flashover: In engineering, most processes of inter-
jr-e size is used in this book to mean HRR. est consist of gradual changes, but flashover is an excep-
The intent of this chapter is to provide basic infor- tion. Flashover is a sudden change from an apparent
i mation about fire size and development that should be steady fire confined to a relatively small space to a fire
l
helpful concerning evaluation and deterniination of that involves a much larger space, such as the entire
I design fires. A design fire is the challenge that a smoke room.
management system is designed to withstand. Because For the office fire of Figure 2.2 (c), materials
the presence of sprinklers often plays a role in the deter- throughout the room are subject to thermal radiation
8
mination of a design fire, sprinklers are also included. from the Flames and the smoke layer under the ceiling.
The design fire can be a steady fire or an unsteady one. When this radiation is sufficiently high, some of these
While the steady fire is not physically realistic, it can materials ignite. This is followed by other materials
result in very conservative designs and it can simplify
design analysis.
I Il I
I II Post Flashover
STAGES O F FIRE DEVELOPMENT I II I
I II I

Fires in rooms or other compartments are often I Gr~wth l1 I Dewy


l f--
described in terms of the stages of fire development,
shown in Figure 2.1. These stages are useful in discuss-
ing fires, but many fires do not go through all of these
stages due to lack of fuel or the action of a suppression
system.
Ignition: Ignition is the period during which the
Time
fire begins. Smoke management deals with the conse-
Figure 2. l The stages offwe developn7etzf.
Chapter 2- Fire and Heat Release

(a) Fire restricted to inside corner of chair and resulting in smoke layer under ceiling
Principles of Smoke Management

Table 2.1:
Approximate Values of CO 7

Yield for Room ~ i r e s * Measure


Temperature,
CO yield** Flow Rate, & Gas
Concentrations.
Flaming fires in "free air" 0.04
Fully involved fire (in a room without cellulosic 0.2
materials on ceiling or upper portion of walls)***
These estimates are based on Pit&(1994). Mulholland (2002),
and Tewarson (2002).
** Keld is in Ib CO produced per Ib of fuel burned (or g o f CO
produced per g of fuel burned).
*** Fully involved fires in rooms with cellulosic materials (wood,
paper, cardboard, etc.) on ceiling or upper ponion o f walls are
expected to have CO yields several times higher (Pi- 1994).
Figure 2.3 Open air calorimeter:
igniting, and then the entire room is involved in fire.
Once a fire gets to the stage depicted in Figure 2.2 (c), it
only takes a few seconds for a room to flashover. measurements would be meaningless. Various schemes .
In a very large room, such as an open office floor to keep the load cell from heating up were devised, but
plan, only a portion of the room may flashover. The they all interfered to some extent with the measure-
smoke layer temperature at which flashover occurs is ments. The situation was even worse when pieces of
generally in the range of 930°F to 1300°F (500°C to burning material would fall from the load cell.
700°C). The criteria for flashover is sometimes taken to To further exacerbate the difficulties with such
be a smoke layer temperature of 1100°F (600°C) or a HRR determinations, many items burned are composites
radiant heat flux of 1.8 ~ t u l f st (20
~ kw/m2) at the floor of several different materials, each with its own heat of
of the fire r o a n (Peacock et al. 1999). combustion. For example, a desk might be made of
Fully Developed Fire: This stage of fire develop- wood, fiberboard, sheet plastic and molded plastic
ment has the highest temperatures. For small and doors, and drawer fronts. Not only do these materials
medium rooms, the HRR of a fully developed fire have different heats of combustion, but they burn at dif-
depends on the amount of air that reaches the fire. Such ferent times during the course of a fire. For these rea-
a fully developed fire is ventilation cotztt-olled. sons, an HRR estimated from measured mass losses is
In a ventilation controlled fire, more volatile gases often unreliable.
are produced by the burning materials than can be
bumed in the room with the oxygen available, and the Oxygen Consumption Calorimetry
fire can be characterized by flames consisting of burn- In the 1980s, fire research laboratories around the
ing volatile gases extending from open doonvays of the world worked to develop a method of calorimetry that
fire room. For very large rooms, as in an open office was not subject to the problems of the old method dis-
floor plan, the fire may not ever become ventilation con- cussed above. The new method is based on the osygen
trolled. Fully developed fires are characterized by ineffi- used up in the fire and is called oxygetz conszrt?zptioncal-
cient combustion and high production of CO (Table oritnetty (and sometimes oxygetz depletion calorinzetry).
2.1). While oxygen consumption calorimeters often have
Decay: As the fuel is consumed, the HRR of the load cells, the measurements from these cells are for
fire and the temperature of the room drop. The fire may sepante information and not for calculation of the HRR.
change from ventilation controlled to fuel controlled. The key to this technology is that the heat released
Strictly speaking, the term post-flashover fire per unit oxygen consumed is almost a constant for most
includes both fully developed and decay stages, but the materials. Huggctt (1980) found that this heat release
term is often used to mean a fully developed fire. constant is 5,630 Btu per Ib of oxygen consunled (13.1
MJ per kg of oxygen consumed). For most materials
MEASUREMENT OF HEAT RELEASE RATE involved in building fires, this constant has an uncer-
In the early days of fire research, determination of tainty of about 6%.
the HRR during a fire was very crude. Typically, materi- Figure 2.3 shows a calorimeter where furniture is
als were burned on a load cell (scale), and the HRR was burned under a hood connected to an exhaust, such that
estimated from the mass loss and the heat of combustion all the smoke is drawn into the exhaust. From measure-
of the material. If the load cell became too hot, tlie mass ments of the mass flow of exhaust and the O2 content of
Chaptei 2 - Fbe and Heat Release

Measure Temperature. Flow Rate,


and Gas Concentrations.

Smoke
Plume

Front View Section View


Figure 2.4 Room calorimeter:

the exhaust, the time rate of O2 consumption can be cal-


culated. From this, the HRR can be calculated. Because
some of the O2 is not completely consumed, gas mea-
surements also include CO and CO2 Parker (1982) pre-
sents equations for calculation of the HRR, for various
applications.
Oxyzen consumption calorimeters are calibrated by
burning a gaseous fuel (methane, propane, etc.) at a
measured flow rate. The uncertainty of the calorimeter
depends on the uncertainties of (1) the operation of the
calorimeter, (2) the calorimeter calibration process, and
(3) the heat release constant. Calorimeter operation is Time (S)
not always as intended. Some of the smoke may not be
Figure 2.5 Three kiosk fires iIIzcstrate iypical repeat-
captured by the hood, or burning materials may fall off
abiIiry of burni~~g materials (data Ji-onl
the fire and away from the calorimeter. With such unin-
MifIer-[I 9961).
tended operation, uncertainties in excess of 20% could
result. For a well-calibrated calorimeter operated as
intended, the uncertainty of measured HRR may be in cant. The cone calorimeter is a "bench scale" laboratory
the neighborhood of 10%. For more information about instrument developed at NIST (Babrauskas 1990).
the uncertainty of ovygen consumption calorimeters, see
Stroqp et al. (2000). HRR OF SOME OBJECTS
Open air calorimeters (Figure 2.3) are sometimes When duplicate objects are burned, there are devia-
called furniture calorimeters because they are often used tions in HRR as illustrated with the three kiosk fires of
for furniture. However, they can be used for any fuel Figure 2.5. These kiosks are for selling T shirts. The
package provided that ( l ) all of the smoke from the fire deviations of HRR are due to a number of factors,
is collected, and (2) the heat released does not damage including (I) minor variations in arrangement of the T-
the calorimeter including the pollution control equip- shirts, (2) variations in composition of T-shirts, (3) vari-
ment. Typically, these calorimeters are located indoors ations in the dimensions of the kiosk, (4) variations in
to protect the fire from the wind. The hoods are usually materials of the kiosk, and (5) variations in the air cur-
l0 to 20 ft (3 to G m) square, but the size is only con- rents near the kiosk. However, the shapes and peak
strained by the practicalities of construction. HRRs of kiosk curves are similar.
Other types of Oz consumption calorimeters are the Figures 2.6 to 2.19 show HRRs of other objects. The
room calorimeter and the cone calorimeter. The room peak HRR of Scotch pine Christmas trees burned by
calorimeter (Figure 2.4) is used when the effects of the Stroup et al. (1 999) were in the range of' 1800 to 5000 Btu1
walls and ceiling on the HRR are thousht to be signifi- S (1900 to 5300 kW), as shown in Figure 2.6. Ahonen et
P r i i p l e s of Smoke Management

al. (1984) burned smal!er spruce trees, and the peak HRRs stages followed by decay. Many other objects b u m 4
were in the range ofabout 40 to 620 Btds (42 to 650 kW). under an open air calorimeter will show the same type
All of these Christmas tree fires had rapid growth stages of rapid growth followed by decay as the material burns
followed by decay as the tree was burned up. UP.
Lawson et al. (1984) burned an assortment of furni-
Data for a burning dresser (Figure 2.7) and bunk lure (Figures 2.9 to 2.16). In general, all these curves are
bed (Figure 2.8) were obtained by Mitler (2000). Like of the s a n e generd shzpe as the proceeding HRR
the Christmas tree fires, the dresser had rapid growth curves, with the exception of one of the chairs. The
upholstered chair of Figure 2.1 1 has two HRR peaks: (1)
950 Btds (1000 kW) at 240 s and (2) 570 Btuls (600
kW) at 400 S. The wardrobe of Figure 2.15 is an even
more pronounced example of multiple peaks: (1) 3500
Btuls (3700 kW) at 120 s and (2) 3100 BWs (3300 kW)
at 360 S.
For objects with two HRR peaks, the second peak is
due to material or materials in the object that bum dif-
ferently from those responsible for the first peak. Also, a
fire consisting of a number of objects would be expected
to have more than one peak, as in Example 2.2.
Time (S) Madrzykowski and Vittori (1992) burned worksta-
Figure 2.6 Scotch pine Chrislrnas tree (adapledfi-orn tions. These workstations are simulated offke work-
S~roupet al. [ l 9991). spaces, including a chair, shelves or a desk, paper,
personal computer, and dividers separating the worksta-

0
O
'V 300 600 960 l ~ O O 1&0 1/00
Time (S)
Figure 2.7 Wooder? dresser- fda/n ji-on? hfitler- Figure 2.9 Innerspring tnat~ressfilled wilh polyure-
[2000]). lhane foam (dala fvom Lawson et al.
[ I 9841).

5000
5000
4000
4C30
..-.
-
-$ 3000
m 2000
3000 z
IL
I
2000 I 5
1000 1OGO

0 0
'0 300 600 900 1200 1500 1800 0 3 6 0 9 0 1 0 1&0 l&
Time (S) Time (S)
Figure 2.8 HEN/1-eleasc rn/cjur- b~rnkbed (dalnfiani Figure 2.10 M C I ~ fi-ame
I chair wilh polyurethane
Miller- [2000]). foani-filled cushions (dalafi-on7 Lawson er
al. [1984]).
Chapter 2-Fire and Heat Release

l ooc l 1. 4 l000

' 360 660 >OO


lime (S)
1;00 1;00
0
18fOO

Figure 2.11 Upholstered chair with polyurelhanefoa~n Figure 2.14 Metal wardrobe w'th cotton andpolyesrer
padding and weighing 25 lb (11.5 kg) garments (data from Lawson cl al.
(datafroni Lawson et al. [1984]). [1984]).

0
0 300 600 900 1200 1500 1800
lime (S) Time (S)

Figure 2.1 2 Upholsfered chair ~ Y t polytrerhn~~e


h foam Figure 2.15 Wardrobe of 0.5 in. (12.7 I ~ Ip!~-~t~ood
)
padding and weighing 62 IB (28.3 kg) wirh cotton nnd po!~wtet-garnienrs (dnro
(dnia.f,.otn Lauson et al. [l 9841). ,/ram Lnwson et al. [l 9841).

8000 .8000
-Unfinished
6000 - Fire Retardant
.A
In 1 Paint:
6000

g 4000 - 'i :.:.: :.


U
1 Coat
2 Coats - 4000 2S
K
K I
2000
.
1
,
S;

8. .

lime (S)

Figure 2.13 Sofa wit/? po~v~tretl~at~e .foam padding Figure 2.16 War-drobeof 0.125 in. (3.2 ~mnjp11.1t~ood
(datafi-0171
Laws017 et al. [19S4]). ~ d t hcotton a ~ i dpol~:este~-
garnze~its(darn
,from Lnwson et al. [1984]).
Principles of Smoke Managemerit

20W
2000

.;
;1500 1500
m 1000
, l000
F
E
I
I
500 500

0
OO 6W 1200 1800 2400 3000 3600
lime (S) Time (min)
Figure 2.17 Two-divider workstation with conven- Figure 2.19 Automobiles (data from Joyeux [1997]).
tional desk and credenza (data from
Madrzykowski and Vettori [1992]).

lime (S)
Figure 2.18 Three-divider wo~kstationwith an open Figure 2.20 Crib made of geometrically arranged
work top and shelf (data from sticks.
Madrzykm~skia17dVettori [l 9921).

tion from other spaces. The two-divider workstation was used for tests of the smoke management system at
(Figure 2.17) has a peak HRR of 1700 Btuis (1800 kW) the Plaza Hotel (Klote 1990). This crib was made of 144
at 140 S. The three-divider workstation (Figure 2.18) has wood sticks, 1 .S in. (38 mm) by 1.5 in. (38 mm) by 2 ft
a peak HRR of 6400 Btuis (6800 kW) at 550 S. A major (0.61 m) long, and it had a peak HRR of aboet 1400
reason for the higher HRR of the three-sided worksta- Btuls ( l 500 kW) when burned in free air. The stack of
tion is probably the increased radiation feedback from nine wood pallets shown in Figure 2.2 1 has a peak HRR
the additional divider and the shelves. For further infor- of about 3,500 Btu% (3,700 kW) when burned in free
mation about the HRRs of workstations, readers are air. Gross (1 962), Block (197 l), and Walton (1988) have
referred to Madrzykowski (1998). burned wood cribs of various sizes and stick spacings.
Babrauskas (2002) provides heat release data of cribs
Figure 2.19 shows HRR data of automobiles mea-
and pallets.
sured by Joyeux (1 997). Joyeux showed that cars made
in the 1990s had a higher HRR than those made earlier,
and this may be due to increased use of polymers and
VENTILATION-CONTROLLED FIRES <h3 d d )
other nonnletallic materials. -Because of these higher As already stated, the HRR of a ventilation-con-
HRRs, a car fire in a parking garage can ignite an adja- trolled fire depends on the amount of air that reaches the
cent car. fire. Further, the HRR can be expressed as a function of
the openings to the fire room as
Cribs and piles of wood pallets are used in research
and testing when reproducible solid fuel fires are needed
(Figures 2.20 and 2.21). Cribs are geometrically
arranged piles of sticks. The crib shown in Figure 2.20 where
Chapter 2-Fire and Heat Release

Q = heat release rate of fire, kW (Btuls); Example 2.1 Ventilation-Controlled Fire


For a room with a single doorway opening that is fully
A, = area of ventilation opening, f? (m2); involved in h, how big will the fire be? The doorway open-
H, = height of ventilation opening, f?(m); ing is 3 ft (0.914 m) wide by 7 ft (2.13 m) high.
H, = 7 ft (213 m);A, = 3(7) = 21 fi2 (1.95 m*)
C
,, = 61.2(1260). Because the tire is ventilation controlled, Equation (2.1) is
Equation (2. l) appiies to rooms of normal construc- applicable.
tion and size with only one rectangular opening. Figure
2.22 shows the HRR of a ventilation-controlled fire as a
I ' 61.2(21)(7)1'2 = 3400 Btuk (3600 kW)
Q = 6 1 . 2 ~ ~ ~ : /=

function of width of the door or other opening. Equation SPRINKLERS


(2.1) provides useful estimates for rooms made with
normal construction materials (drywall, concrete, wood, Figure 2.24 illustrates t-squared fire growth with
etc.), but it is not appropriate for metal rooms, such as the three possible responses to sprinkler spray: (a) sprin-
on a ship with steel decks and bulkheads. For large klers overpowered by fire, (b) constant HRR, and (c)
rooms (over 300 ft2 [30 m2]), the appropriateness of reduction of HRR. Sprinklers can be overpowered by an
Equation (2.1) is questionable. For information about extremely fast growing fire due to burning materials that
the effects of construction materials and room sizes, see exceed the sprinkler design. Sprinklers can also be over-
Walton and Thomas (1995). powered when the smoke reaching the sprinklers has
For a number of rectangular openings with the same cooled due to plume entrainment, as-can happen with
bottom and top elevations, the heights are the same, and fires in spaces with ceilings that are relatively high com-
the effective area is the sum of the individual areas. pared to the arrangement of fuel. For this to happen, the

Door WidUl (m)

where
2
A , = effective area of all the ventilation openings, ft
(m2>;
A,!i = area of ventilation opening from i = 1 to n, ft2
(m').
This is illustrated for two openings in Figure 2.23.
Door Width (ii;;

Figure 2.22 HRR ofafully developedfire it1 a sinall 01-


medium-sizedroomofnot-tnalconstr~rction.

For openings with the same top and


bottom elevations. A, = A,,, + A w 2 .

Figure 2.2 1 Stack o f 17ine/~a//cts. Figure 2.23 Combining vet7tilariotl openings for esri-
mate of the size o f a-firl(pckvelo~~ed~fit~e.
Principles of Smoke Management

'HRR DECAY D U E TO SFRINKLERS


A constant HRR after sprinkler actuation is a con-
servative estimate for many applications. Fire decay
after sprinkler actuation is more realistic. Fire decay can
be expressed as

where
Q = post sprinkler actuation HRR, kW (Btuls);
= HRRat sprinkler actuation, kW (Btuls);
t = time from ignition, s (S);
to,, = time of sprinkler actuation, s (S);

Time r = time constant of fire suppression, s (S).

(a) Sprinklers Overpowered by Fire For a number of fuel packases likely to be found in
offices, Madrzykowski and Vettori (1992) conducted
sprinklered fire experiments with a spray density of 0. I0
I gpm/ft2 (0.07 m d s ) of water. They determined that a
Conservative Estimate
of Constant HRR fire decay curve with a time constant of 435 s had a
higher HRR than most of the sprinklered fires (Figure
2.25). Evans (1993) used these data and data for wood
crib fires with sprinkler spray densities of 0.06 gpm/ft2
(0.041 mmls) and 0.097 gprn'ft2 (0.066 mmls) from
Tamanini (1976) to develop the following correlation:

Time
(b) Conservative Estimate of Constant
HRR After Sprinkler Activation
where
r v = spray density, gpmlf? (mnds);
C, = 6.15 (3.0).
While Equation (2.4) has not been experimentally
verified, it does allow us to adjust the decay time for
sprinkler densities other than those of Madrzykowski
and.Vettori.

Sprinkler Response
While the information in this section is primarily
about sprinklers, it also applies to vents actuated by fus-
Time
ible links and fixed temperature heat detectors.
(c) Fire Decay After Sprinkler Activation The responsiveness of sprinklers is tested by the
plunge test, where a sprinkler is "$mgedWinto a heated
Figure 2.24 Interaction between fire and sprinklers. oven in which heated air is circulated. The nnalysis of
the plunge test is mathematically the sanie as that of a
small piece of hot metal suddenly quenched in a cool
flame height is typically less than the ceiling height, and fluid, as described in heat transfer texts (Kreith 1965:
room air entrainnient cools the gases in the w o k e Incropera and DeWitt 1985). This analysis is based on
plume. Methods of calculating the plume temperature the assumptions that ( I ) the internal resistance of the
are in Chapter 13. If the sprinklers do activate, the spray sprinkler is negligible, (2) the sprinkler is instanta-
could evaporate before the droplets reach the fuel. neously put ill the oven, (3) the convective heat transfer
coefkient is constant, (4) the gas temperature i n the
Chapter 2-Fire and Heat Release

oven is constant, and (5) the only heat transfer is from calculated from Equation (2.6). The RTI of standsrd
the sprinkler to the gas. sprinklers varies from about 140 to 280 fill2 s1I2 (77
The temperature of the sprinkler increases exponen- to 155 m'I2 sln), and the RTI of quick-response sprin-
tially, as shown in Figure 2.26. The time constant, r, of klers (QRS) varies from about 50 to 100 fill2 slR (28 .
the sprinkler is to 55 ,lR ,lR).
The response time index does not account for con-
ductive heat transfer from the sprinkler. To account for
conduction, a virtual RTI can be calculated as
where
Z = time constant, s (S); RTI, = -RTI
--
CRTI
m = mass of the sprinkler, Ib (kg); '+1/2
C = specific heat of the sprinkler, Btuflb"F (Jkg "C);
h, = convective heat transfer coefficient, ~ t d f t s2 "F where

(w/m2 "C); RTI, = virtual RTI, fill2 slR (m1' slR);

A = surface area of the sprinkler, ft2 (m2). CRTI= conductivity factor, f i l n / s " ~(m'l2 IS'").
.
The time constant, r, is the time at which the tem-
.
.

perature of the sprinkler has reached 63% of the way to


the gas temperature. The convective heat transfer coefi-
cient varies with velocity, so that the time constant also
varies with the velocity at which it is measured.
The response time index (RTI) was developed as a
measure of sprinkler responsiveness that is independent
of velocity.

:
I
where u is the velocity, Ws (mds). 'I: is time constant
In the plunge test, the time to actuation and the gas
velocity are measured. Then the time constant can be Time
calculated from the time to actuation, and the RTI is Figure 2.26 Temperatutasfot~n spr-ir~kler-plztr?ge
test

'
Paper Cart Fuel Package
-.--.Secretarial Desk Fuel Package
o Executive Desk Fuel Package
. ---- Office II Fuel Package
- Office I Fuel Package
- - - Sofa Fuel Packdge
- . Work Station I Fuel Package
- - - Work Station I I Fuel Package
X Wood Cribs

0 200 400 600 800


Time, t - t,, (S)

Figure 2.25 Filr decuj' due to spri~ikleraclio/i n.ill7 a spruj, derisi@of 0.10gpn/
f? (0.07 /ii/ii/s)(adupledfiori~Mad-zykowski and kllori [ l 9921).
Principles of Smoke Management

S~rinklerActuation are not intended to be located in the space are referred to


Actuation depends on gas temperature and velocity as tramientfuels.
near the sprinkler. In a fire, a jet of hot gases flows radi- A few examples of transient fuels are Christmas
ally from where the smoke plume intersects the ceiling. decorations, paint and solvents in stairwells during
Computer programs have been developed that use corre- redecorating, unpacked foam cups in cardboard boxes
lations for such a ceiling jet to predict actuation time. after delivery, cut up cardboard boxes awaiting removal,
The program DETACT-QS (Evans and Stroup and closely stacked upholstered furniture after delivery.
1986) assumes that the thermal device is located in a rel- Sometimes, transient fuels remain in place for long peri-
atively large area, that only the ceiling jet heats the ods. Some examples are (1) a number of polyurethane
mattresses delivered to a dormitory and waiting for dis-
device, and that there is no heating from the accumu-
tribution in the next school year, (2) automobiles on dis-
lated hot gases in the room. The required program inputs
play in a shopping mall, (3) boats and campers on
are the height of the ceiling abo:!e the fuel, the distance
display in an arena, and (4) a two-story colonial house
of the thermal device from the axis of the fire, the actua-
built for display inside a shopping mall.
tion temperature of the thermal device, the response
time index (RTI) for the device, and the rate of heat Transient fuels must not be overlooked when select-
release of the fire. The program outputs are the ceiling ing a design fire. One approach to incorporating tran-
gas tkmperature and the device temperature, both as a sient fuels in a design fire is to consider the fire
function of time and the time required for device actua- occurring over 100 ft2 (9.3 m2) of floor space with a
heat release rate density of 20 Btuls ft2 (225 kw/m2).
tion. DETACT-T2 (Evans et al. 1986) is similar to
This amounts to an allowance for transient h e l s of 2000
DETACT-QS, except it is specifically for t-squared
Btuls (2100 kW).
fires. Several zone fire models (such as FAST,
LAVENT, and JET) are capable of calculating ceiling jet
temperatures and predicting actuation (Chapter 8). Steady Fires
It is the nature of fires to be unsteady, but the steady
DESIGN FIRES fire is a very useful idealization. Steady fires have a con-
A design fire curve is the description of the devel- stant heat release rate. In many applications, use of a
opment of a design fire that can be used in a fire sce- steady design fire can lead to straightforward and con-
nario. The curve is for HRR as a function of time. This servative designs.
curve can be as simple as a constant, and it can also be a
simple function of time. The design fire curve can also HRR per Unit Area
be a complicated sequence of lesser cunles for some or Morgan (1979) suggests a typical rate of heat
all of the stages of tire development described at the release per unit floor area for mercantile occupancies of
beginning of this chapter. 44 Btuls ft2 (500 kw/m2). Fang and Breese (1980)
A fire scenario includes more than just the design determined about the same rate of heat release for resi-
fire curve. The word sce17nrio means an outline of dential occupancies. Morgan and Hansell (1987) and
events, as in a play or other theatrical production. A fire Law (1982) suggest a heat release rate per unit floor
scenario can be thought of as the outline of events and area for office buildings of 20 Btds f? (225 kw/m2).
conditions that are critical to detemiining the outcome For smoke management applications, a heat release rate
of alternative designs. In addition to the HRR and fire per floor area of 20 Btuls ft2 (225 kw/m2) is suggested
location, a scenario could include the type of materials for restricted fuel spaces, and 44 Btuls ft2 (500 kw/m2)
burned, airborne toxicants and soot produced, and peo- is suggested for spaces with furniture, wood, or other
ple movement during fire. combustible materials. A firc occurring over 100 ft2 (9.3
m2) of floor space would result in 2000 Btuls (2100 kW)
for restricted fuel space and 4600 kW (4400 Btuls) for a
space with combustibles. The heat release densities of
In many spaces, the fuel loading is severely
Table 2.2 can be useful in determining design fires.
restricted with the intent of restricting fire size. Such
spaces are characterized by interior finishes of metal,
brick, stone, or gypsum board and furnished with Unsteady Fires
--.
objects made of similar materials plus plants. Even for Fires frequently proceed through an incubation
such a /ire1 reswicmf space, there can be an almost period of slow and uneven growth, followed by a period
unlimited number of combustiblc objects that are in the of established growth as illustrated in Figure 2.27 (a).
space for short periods. Such combustible materials that Figure 2.27 (b) shows that established growth- is often
Chapter 2-Fire and Heat Release

Table 2.2:
Heat Release Density of Some Materials
Heat Release Density, q
Material Burned kwlrn2 Btuls f$
I. Wood pallets, stacked 0.46 m (1.5 h) high (6-12% moisture) 1400 125
2. Wood pallets, stacked 1.52 m (5 ft) high (6-12% moisture) 4,000 350
3. Wood pallets, stacked 3.05 m (10 ft) high (6-12% moisture) - 6,800 .600
4. Wood pallets, stacked 4.88 m (16 ft) high (6-12% moisture) 10,000 900
5. Mail bags, filled, stored 1.52 m (5 fi) high 400 35
6. Cartons, compartmented, stacked 4.57 m (15 fi) high 1,700 150
7. PE letter trays, filled, stacked 1.52 m (5 ft) high on cart 8,500 750
8. PE trash barrels in cartons, stacked 4.57 m ( l 5 ft) high 2,000 175
9. PE fibeglass shower stalls in cartons, stacked 4.57 m (15 ft) high 1,400 125
10. PE bottles packed in item 6 6,200 550
11. PE bottles in cartons, stacked 4.57 m (15 ft) high 2,000 175
12. PU insulation board, rigid foam, stacked 4.j7 m (l5 ft) high 1,900 170
13. PS-jars packed in item 6 . . 14,000 1,250
14. PS tubes nested in cartons, stacked 4.27 m ( l 4 ft) high 5,400 475
I . PS toy parts in cartons, stacked 1.57 m (l5 ft) high 2,000 180
16. PS insulation board, rigid foam, stacked 4.27 m (14 ft) high 3,300 290
17. PVC bottles packed in item 6 3,400 300
IS. PP tubes packed in item 6 4,400 390
19. PP & PE film in rolls, stacked 4.27 m (14 ft) high 6,200 550
20. Methanol pool, 0.16 m (0.52 ft) diamcter 2,000 I SO
21. Methanol pool, 1.22 m (4.0 ft) diameter 400 35
22. Methanol pool, 1.74 m (5.7 ft) diameter 400 35
23. Methanol pool, 2.44 m (8.0 ft) diamc~er 420 37
24. Methanol pool. 0.97 tu (3.2 ft) square 745 66
25. Silicone transfornler fluid pool, 1.74 m (5.7 fr) diameter 90 8
26. Silicone transformer fluid pool, 2.44 m (8.0 ft) dianletcr 90 8
27. Hydrocarbon transformer fluid pool. 1.22 nl (4.0 ft) diameter 940 83
28. Hydrocarbon transformer fluid pool, 1.74 m (5.7 ft) diameter 900 80
29. Heptane pool, 1.22 (4 ft) diameter 3.000 270
30. Heptane pool, 1.74 (5.7 ft) diameter 3.200 280
Nn~rc.
I . Abbreviations are: PE = polytl~ylenc.PS = polyslyrsnc. PVC = pulyvinyl cliloride. PP = polypropylene. P U = polyurethane.
2. Items I tlirough I 0 frorn~fl'~ 4 2 0 (2000).
3. ltenis10 tl~rdugh30 rrolii Hcskcs~ad(IYS4).
4. ltcms 25 tlirot~gli28 arc proprietary products

represented by an idealized parabolic equation (Heskes- It is generally recognized that consideration of the
tad 1984). incubation period is not necessary for design of smoke
management systems, and Equation (2.8) can be
expressed as where t is the time after effective ignition,
and fires following this equation are called t-squared
where
fires.
Q = heat rclcasc rate of fire, kW (Btuls);
Ncg-lecting the incubation pcriod, the t-squared fire
a = firc growth coefficient, k w k 2 ( ~ t u l s ~ ) ; can bc written as
1 = time aficr ignition, S;
*<, = cfl'cc[ivc ignition time, S.
Principles of Smoke Management

where t is considered the time from effective ignition. For tains, etc;). The key to selecting the items that make up
I-P units, the following form of Equation (2.9) is often the base h e 1 package is that the radiant flux from bum-
used: ing one of the items will lead to ignition of the other
items in the base h e 1 package but not to ignition for he1
items outside the base h e 1 package.
The point source radiant model (Figure 2.29) con-
I where siders the flame as a small thermal source such that the
intensity of thermal radiation is proportional to the
Q = heat release rate of fire, Btuk; inverse of the square of distance from the source. Ther-
t = time after effective ignition, S; mal radiation also is called radiant heatflux.
tg = growth time, S. The intensity of thermal radiation is
When t = tg, Equation (2.10) gives a value of Q =
1000 Btuls. Table 2.3 lists fire growth values from
NFPA 92B (NFPA 2000) and NFPA 72 (NFPA 1999).
The fire growths corresponding to the NFPA 928 values
are shown on Figure 2.28. Unless otherwise stated in where
this book, the terms slow, mediznn, fast, and zdlra fast = intensity of thermal radiation, ~ t u / f st (kw/m2);
~
4;
fire growth refer to the NFPA 92B values.
Qr = radiant heat release of the fire, B d s (kW);
Fuel Package Approach
R = distance from the center of the fire, ft (m).
The base fuel package is the maximum probable
size of h e 1 package that is likely to be involved in fire Table 2.3:
for a specific application. A fuel package can be made Fire Growth Constants for T-Squared Fires
up of a number of fuel items (sofa, chair, bed, table, cur-
NFPA 9 2 8 I NFPA72
u (8tuls3) cx (l;w/s2) )g(S) Range of (S)
Slow 0.002778 0.002931 600 Ig2400
Medium 0.01 1 1 1 0.01127 300 150 5 tg < 400
Fast 0.04444 0.04689 150 150
Ultra Fast 0.1778 0.1878 75 NIA

Thin Plywood Corrugated Cardboard


Cartons 1.5 ft (4.6 m)
High -Various: Contents

Time, t
(a) Typical HRR curve

0 200 400 600 800


Time, t Time From Ignition (S)
(b) Idealized Parabolic curve Figure 2.28 Relalion of r-sq~iaredjires 10 sotlze fire
Figure 2.27 Fire grorvlh clil.~;cs. 1es1.7(adapled )I-OIU Nelsoti [l 9871).
Chapter 2-Fire and Heat Release

have high radiative fractions. However, for design appli-


cations, values of X, = 0.3 and X, = 0.7 are common.
The idea of separation distance is useful fo; evalua-
Oriented tion of what items should be in the base fuel package.
Fire Normal Using the point source radiant model, the separation dis-
tance is

The point source model is a good


approximation provided that R > 20.
Figure 2.29 Point source radiation model.
where
RSD = separation distance from the center of the fire to
The point source radiant model is appropriate pro-
a target, ft (m);
vided that the distance from the center of the flame is
greater than twice the diameter of the fire (R > 20). The qi. = intensity of thermal radiation needed for nonpi-
radiant heat release of the fire is loted ignition, ~tu/ft2S (kw/m2).
Fuel items less than RSD away from the fire would
be expected to ignite, and fuel items farther than RsD
where away would not be expected to ignite. The radiant flux
needed for nonpiloted ignition varies from about 0.9
Q = heat release rate of the fire, Btu/s (kW); 13tu/ft2S (10 k ~ l m ' ) for thin easy-to-ignite materials to
X,. = radiative fraction. 1.8 ~ t u / f St (20
~ kw/rn2) for thick materials.
For a fire, the heat release rate, Q,..;, that results in
Heat transfer from a flame is by conduction, con- ignition of an object at a distance of R away is
vection, and radiation. For most fires, conductive heat
transfer from the flame is negligible. The radiant frac-
tion can be expressed as
For radiant heat transfer where R is less than twice
the diameter of the fire, a method other than the point
where X , is the convective fraction. source model is needed. Several texts have general
The radiative fraction depends on the material information about radiant heat transfer (Siege1 and
burned and the diameter of the fire, and the radiative Howell 1992; lncropera and DeWitt 1985; Kreith 1965).
fraction varies from about 0.1 to 0.6. Low sooting fuels, For information about radiant heat transfer of fire, read-
such as methanol, have low radiative fractions, and high ers are referred to Quintiere (1998), Drysdale (1985),
sooting materials, such as gasoline and polystyrene, and Mudan and Croce (1995).
Principles of Smoke Management

11 '".-Ruarnnle ---
'r-- 2.2 Race Fuel-
--- Packaoe
-

1) -- a - - "v
m-

The fuel load in a large atrium consists of the polyurethane foam-filled sofas and chairs shown in Figure 2.30. The a i l i n g of the atrium
is sufficiently high so that successful sprinkler suppression is not anticipated. The HRR of the sofas is the same as that of Figure 2.13,
and its peak HRR is 2960 Bhds (3 120 kW). The HRR of the chairs is the same as that o f Figure 2.1 2, and the peak HRR is 20 l 0 Bhds
(21 20 kW). How many sofas and chairs make up the base fuel package, and what is the HRR of the base fuel package?
1 Part I: Initial Estimate of Base Fuel Package

I(11 Use a radiant flux for nonpiloted ignition of qr,

For the sofq Q, = = 0.3(2960) = 888


= 1.8

Bhds (937 kW).


~tu/ft' s (20 kw/rn2).

11 From Equation'(2.14), the separation distance from the burning sofa is

This shows that a fire on sofa I would not be expected to ignite sofa 2, but it would be expected to ignite chair I . Because fires are often
off center, the center of the fire is taken as the "+" on the side near the chair. This is conservative in that ignition of the chair would be
sooner than if the center of the fire were farther away.
For the chair, Q,. = X r =
~ 0.3 (2010) = 603 Bhds (636 kW)

1l From Equation (2.14), the separation distance from the burning sofa is
7

I1 This shows that the fire of chair I would be expected to ignite sofa 2. Because sofas 3 and 4 are at least 18 ft (5.5 m) away from sofas I
and 2, ignition of sofas 3 and 4 would not be expected. For now the base fuel package will be considered to consist ofsofas I and 2 and
chair I.
Part 11: Calculate HRR Base Fuel Package

I On Figure 2.30, the distance from the center of the fire on sofa I is R , = 3.6 ft ( l . l m).
The heat release rate that results in ignition at R , can be calculated from Equation (2.15)

1l This means that when the fire ofsofa I reaches 293 Btds (309 kW), the chair would be expected to ignite. Because R , = R1, ignition of
sofa 2 is expected when the chair I fire also reaches 293 B t ~ d(309

Calculations of the HRR are done graphically on Figure 2.3 1:


s kW).

(a) The HRR of sofa I is taken from Figure 2.13. The ignition time ofehair I is determined at the intersection of the sofa 1 curve and
293 Btds (309 kW).
(b) The HRR of chair I is taken from Figure 2.12.
(c) The ignition time of sofa 2 is determined in a manner similar to step (a), and the HRR curve for sofa 2 also is taken rrom 2.13.
(d) T'le curves for sofas I and 2 and chair I are added to obtain the cunTefor the base fuel package.

It should be noted that adding the HRR curves as in step (d) assumes that the objects will bum as they would in frce air under a calorim-
eter and neglects any effect of radiation from other burning objects.
)IPart 111: Check Bare Fuel Package
This part checks to see if the-base fuel package will ignite other materials.
The highest peak of the HRR curve of Figure 2.3 1 (d) is at 3600 Btds (3800 kW).
For the base fuel package, 8,. = %,.Q = 0.3 (3600) = 1080 Btds (l 140 kW).
1 From Equation (2.30). the separation distance from the b a r fuel packaee is
-

The other items in Figure 2.30 are I S It (5.5 m) lion1 the base fuel package, so ignition ol'these items wo~tldnot be expected. So the
base fuel package and its HRR curve can be ~lscddirectly for a design analysis, or a simplified design llRR curvs can be adapted rrom
it. Ifthere were fuel items \\ ithin this separation distance. these items would have to be added to the base rue1 package, and a new HRR
cunre would have to be determined.
Chapter 2 - F i e and Heat Release

time (S)
(a) Draw curve for sofa 1, and locate ignition point of chair 1.

. .
time (S)
(b) Draw wrve for chair 1.
Sofa 4 Sofa 3

Chair 2
7 7 4000

Note:
R, = R ,= 3 . 6 f t ( l . i m)
Figure 2.30 Arrangemen! offurni~urein the aft-iutnof
Example 2.2.

Time (S)
(c) Locate ignition point and draw curve for sofa 2

Time (S)
(d) Get base fuel package by adding the 3 other curves

Figure 2.31 Graphic delem-minafion of [he base file1


package oof Examnple 2.2.
CHAPTER 3

Smoke and Tenability

n this book, the term srnoke is used in accordance ard. Frequently, people become disoriented in fire situa-

I .with the definition of NFPA 92A (2000) and NFPA


92B (2000), which states that smoke consists of the
airborne solid and liquid particulates and gases evolved
tions because they cannot see through heavy smoke. If
they remain in the building too long, they fall victim to
exposure to toxic gases or elevated temperatures. Fur-
when a material undergoes pyrolysis or combustion, ther, in buildings with balconies, smoke obscuration can
together with the quantity of air that is entrained or 0th- result in fatal falls.
erwise mixed into the mass. The products of combustion
Smoke management systems can be designed with
usually include particulates, unburned fuel, water vapor,
the objective of providing a tenable environment in the
carbon dioxide, carbon monoxide, and some other toxic
means of egress or at other locations during evacuation.
and corrosive gases. As smoke moves through a build-
Such a tenability system needs to be designed to meet
ing, air mixes into the smoke mass and the concentration
tenability criteria. Such criteria need to include expo-
of combustion products in the smoke decreases. Includ-
sure to toxic gases, heat, and thermal radiation. Further,
ing air that is entrained or othenvise mixed facilitates
the criteria often include visibility. As discussed at the
discussions about fire smoke management in atriums
end of this chapter, th-e criteria for a tenability design
and other iarge spaces. Generally. smoke is thought of as
depend on the specific application.
being- visible, but the above definition includes "invisi-
ble smoke" produced by burning of materials that pro-
duce little or no particulate matter, such as hydrogen, oBSCURAT1oN
natural gas, and alcohol.
Many different methods of expressing smoke
Information about smoke hazards is useful in evalu-
obscuration are used in fire science and fire protection
ating'the effects of small quantities of smoke migrating
engineering, and this section discusses the common
into "protected spaces," and it is useful in evaluating the
methods. There is a lack of uniformity concerning
consequences of smoke migration without smoke pro-
smoke obscuration, and some engineering publications
tection. This chapter concentrates on smoke hazards due
use different terminology or have different mathemati-
to toxicity, temperature, and smoke obscuration. The
cal definitions for the same terms. These differences
hazards of temperature consist of hear exposwe, which
could result in significant errors, and readers are cau-
can occur when a person comes into bodily contact with
ticned to take care to verify the exact meanings of
hot gases, and thermal racliafiot~e-vposur-e, n.hich can
obscuration terms used in other publications. The termi-
occur when a person receives thermal radiation from
nology that follows was selected with the intent of being
flames or hot smoke that are some distance away from
consistent with most technical publications in this field.
the person.
Exposure to toxic gases, heat. and thermal radiation The fraction of light transmitted through the path-
can be a direct hazard to life, and reduced visibility due length of smoke is called the transniittance and is writ-
to smoke obscuration can be a significant indirect haz- ten as
Chapter 3-Smoke and Tenability

p--
tight Photo
Source

where +- Li9hiBe,=!?- -+- - - -


T = transmittance, dimensionless;
I, = intensity of light at the beginning of the path-
length;
I, = intensity of light remaining after it has passed
through the pathlength.
I Wires From
Power Source
Wlres To Power
Source and Data
Acquisition System
The units for light intensity are arbitrary, and such Figure 3.1 Smoke meter used to measure smoke obscu-
units are unnecessary for discussions of smoke obscura- ration.
tion and even for measurements of smoke obscuration.
Transmittance is measured by monitoring the extinction
of a beam of light passing through a pathlength, X, of
smoke as illustrated in the light meter of Figure 3.1.
Strictly speaking, the discussion jn this section applies
to light composed of only one wavelength, such as a where a is the extinction coefficient per unit distance in
laser beam, but light meters using less exotic light units of ft" (m-'). The extinction coefficient is sometimes
sources (such as incandescent bulbs) have been used called the attenuation coefficient.
extensively for fire tests.
Percentage obscut-ation is occasionally used and is
When the atmosphere is "smoke free," the intensity
defined as
of light remaining after it has passed through' the path-
length is almost exactly the same as the intensity at the
beginning of the pathlength, and the transmittance is
almost exactly one. It follows that the transniittance of a where R is the dimensionless percentage obscuration.
beam passing tliroi~ghL L ~ i ~ ismoke"
b l e is less than one. The specifc optical dens@ is measured in some
Neutral density tiltcrs, wliich allow only a specific frac- laboratory smoke tests and is defined as
tion of the light to pass through, are used to calibrate
light meters. Thus, the voltage (or current) output of the
photo cell can be calibrated to give transmittance
directly. where
O/~fical de17xitj:,6is delined as
6, = specific optical density (dimensionless);
6 = optical density per unit distance, ft-' (m-');
I/, = volume of the smoke test chamber, ft3 (m3);
Substituting Equation (3.1) and rearranging results
in an equation for optical density in terms ol'transmit- A = decomposed surface area of the test sample burned
tance, ft2 (m2).
The specific optical density is a practical measure-
ment of smoke obscuration only when the decomposed
area of the sample is well defined.
where For laboratory tests where the mass loss of the sam-
ple is measured, the mass optical densiry is an appropri-
6 = optical density per unit distance. K' (m-');
ate measure of obscuration. The mass optical density is
T = transmittance. dimensionless; detined as
X = distance of light travel or the pathlength, ft (m).
Thc e.vti17ctiori co@cit'rit pcr unit distance is
defined as
where
4, = mass optical density, ft2/lb (ni2/S);
Substituting E q i ~ a t i o(3.
~ ~l ) and rearranging yiclds d = optical density pcr unit distance, K ' (m-');
Principles of Smoke Management

Table 3.1:
Comparison of Different Methods of Expressing Smoke Obscuration
Pathlength Optical Density Extinction
Transmittance Percentage Obscuration X 6 Coefficient a

V, = volume of the smoke test chamber, ft3 (m3): cern that a disoriented person could fall from a balcony.
Because a person falling 5 ni (16 ft) has about a 50%
AM = mass loss oftest saniple, Ib (g). chance of fatality, falls are a serious concern for build-
The mass concentration of fuel burned in the test ings with balconies.
chamber is Based on the work of Jin (1974, 1975, 1985), the
relation between visibility and smoke obscuration is
nl
AM
- -
/ ' - VC (3.9) K
S = - (3.11)
a
where nyis the mass concentration of fuel burned in units
of lb/ft3 ( g / ~ ~ 3Substituting
). this density into Equation where
(3.8) yields S = visibilitj, fi (m);
a = extinction coefficient ft-l (m-');
K = proportionality constant (Table 3.3).
The visibility is the obscuration threshold, which is
Table 3.1 lists some values of optical density, the distance at which an object c,n just be seen. The
extinction coefficient, and percentage obscuration for proportionality constant is dependent on the color of
different path lengths. Equations for conversion smoke, the illumination of the object. the intensity of
between differefit smoke obscuration terms are listed in background illumination, and visual acuity of the
Table 3.2. observer. Jin conducted tests determining visibility of
light-emitting and -reflecting signs. Signs in a smoke-
VISIBILITY T H R O U G H S M O K E filled chamber were observed from outside through a
When people cannot see because of smoke from a glass window, and the results for illuminated signs are
building fire, they walk slowl!.. \vhich can significantly shown in Figure 3.2. White smoke \\as produced by
lengthen evacuation time, and they can become disori- smoldering fires, and black smoke \\.as produced by
ented and lost, thus prolonging their exposure to toxic flaming tires. Visibility through the \vhite smoke was
gascs. In atrium fire situations. there is the added con- less, probably due to higher light scattering. It is well
Chapter 3- Smoke and Tenability

Table 3.2: '


Conversion Equations for Smoke Obscuration
Convert To From Equation
Optical Density Extinction Coefficient
6 = 0.4343 a
Optical Sensity Percentage Obscuration
6 = - log,, (1 - A1100)

Optical Density Specific Optical Density

Optical Density Mass Optical Density


6 = 6,mf
Extinction Coefficient Optical Density
a = 2.303 S
Extinction Coefficient Percentage Obscuration
a = -log, ( l - A / 100)
X
Extinction Coefficient Specific Optical Density

Extinction Coefficient Mass Optical Density


a = 2.3036,,m,
Percentage Obscuration Optical ensi it^
2. = 100(1- 10-~")
Percentage Obscuration Extinction Coefficient

Percentage Obscuration Specific Optical Dznsity

Percentaze Obscuration Mass Optical Density

Specific Optical Density Optical Density


6, = 6 v , / A
Specific Optical Density Extinction Coefficient

Specific Optical Density Percentage Obscuration


6 =- K Io&~(I-A/100)
Ax
Specific Optical Density Mass Optical Density
6, = 6,,,m lA
Mass Optical Density Optical Density
S,,, = 6 /.m,
Mass Optical Density Extinction Coefficient

Mass Optical Density Percentase Obscuration

Mass Optical Density Specific Optical Density


6 =A
6 A
,,l

m/ V C
I. Norncnclnturc: 6 =oplical densiiy pcr unit distance. rt-l (n1.l): a = extinction coellicient per unit distnncc. fi-' (ni-l); 1 = percentage obscurntion
(diiilcnsio~iless): ?is
= specific optical dcnsity (din~cnsionlcss):& = mass optical density, liZllb (n121g): 1; = volumic oflhe snloke tssr chambcr, lij
(111'): :\.V = nixrs loss ortcst uorplc, Ih (g): A = decomposed arca ot'thc tea mmplc burned. li' (m'): ,U,-= ni,ass concentration ol.l'uel burned. lblli' (g:
m') [ m , - h.tl! 1; I: .Y = distancu of'lipht m x c l or lllc ~;IL.I; li(111).
~ r i n c i ~ lof
e sSmoke Management
. -

Table 3.3: Example 3.2 Visibility of Doors and Walls


Recommended Proportionality In Example 3.1, what is the visibility of walls and doors?
Constants for Visibility Based on
Research of Jin (1974,1975, and 1985) From Table 3.3, K = 3.
Situation K Extinction coefficient = 0.207 m-'.
Illuminated signs 8
From Equation (3.12), S= 31.207 = 14 ft or 4.3 m.
Reflecting signs 3
Building components in reflected light 3

Brightness Kind d 0 Irritating Smoke


of S i n Smhe a NonirritatingSmoke
a 2000 cdlm' Black Smoke
a! SO0 cdlm' Black Smoke
0 2000 c d d Whde Smoke
0 Mo cdlm2 Whmte Smoke
E
>
OWOQ

>

I
0.4 0.5
I
0.7 1
I
1.5
op 4
I I
2
21
0.2
I

0.3
, ,

0.5 0.7
I

1
I I

1.5 2
I l

3
Ednction Coefficient,a (lh) Extinction Coefficient, U (llm)
Figure 3.2 Relatiomhip between the visibility of light- Figure 3.3 Relationship between visibility of liglit-
emitting signs and smoke obscuration emitting signs ar7d smoke obscuratioi7 for-
(adaptedfi-on7Jiu [l 9Sj1). ir-r-itating and 17onir-r-itating smoke
(adaptedfiotn Jin [ l 9851).

known that scattering of background lighting can signif- The above information about visibility does not
icantly reduce visibility of lighted signs, but quantitative take into account the irritating effects of smoke on the
data about the effect of background illumination are eyes. Jin (1985) conducted tests correlating the visibility
needed. Jin found that the proportionality constant and walking speed of subjects exposed to irritating
ranged from 5 to 10 for light-emitting signs. For reflect- smoke with the extinction coefficient. There are short-
ing signs, the constant ranged from 2 to 4 . Jin indicates comings with correlating pl~ysiologicaleffects with an
optical property of smoke since the effects would seem
that the minimum value of visibility for reflecting signs
to be primarily caused by chemical components of
may be applicable for the visibility of other objects,
smoke. However, the effects of eye irritation are so sig-
such as walls, floors, doors, and stairs. Based on Jin's nificant that Jin's work on the topic is discussed below.
research,.the values of K are listed in Table 3.3.
Figure 3.3 shows the relation between visibility and
Example 3.1 Visibility of an illuminated obscuration for irritating and nonirritating smoke for a
light-emitting sign. The irritating smoke was white
smoke produced by burning wood cribs; the less irritat-
ing smoke was produced by burning kerosene. The visi-

1 Fro: Table 3.3, K = S.


I
bility relationships of Equations (2. i l) and (2.12) are
not appropriate when subjects are exposed to irritating
smoke. In thick irritating smoke, subjects could not keep
11 Extinction coelficient is a=2.303 d so a=7.303(.09) = 0.207 11 their eyes open long enough to read the sign. Figure 3.4
shows the relation between smoke obscuration and
From Equation (3.1 1 ), S = 81.207 = 39 li (1 2 m). the distance \valking speed of people walking down a corridor in irri-
tating and nonirritating smoke. Both eye irritation and
smoke density affect walking speed. Walking speed
decreases with cxtinction coefficient for both smokes,
Chapter 3-Smoke and Tenability

but it is much worse for irritating smoke. For an extinc- chapters on compartrnentatioridesign and atrium design.
tion coefficient of 0.4 m-', the walking speed through For laboratory smoke test chambers and simple room
irritating smoke was about 70% of that through nonirri- calculations, the mass concentration of particulate, my
tating smoke. For extinction coefficients greater than 0.5 can be calculated from Equation (3.9).
m-', the walking speed decreased to about 1 ttlsec (0.3
&S)--the speed of a blindfolded .person. The drop in The extinction coefficient can be expressed as
walking speed was because subjects could not keep their
eyes open, and they walked in a zigzag or went step-by-
step as they held the side wall.
where
Jin (1985) developed an empirical relation for visi-
bility in irritating smoke: a = extinction coefficient, fi-' (m-');
K
S = -(Cs- 1.471og,,a)
a, = specific extinction coefficient, f&lb (m2tg);
a (3.12)
rnp = mass concentration of particulate 1b/ft3(g/m3).
{only for a 2 0.076 ft-' (0.25 m-' )
The specific extinction coefficient depends on size
where
distribution and optical properties of :he particulates.
a = extinction coefficient, fi-l (m-'); Seader and Einhorn (1976) obtained values for a,,, of
S- = visibility, ft (m); 2.1 x 104 ft2/lb (4.4 m21g) for smoke from pyrolysis of
K = proportionalityconstant (Table 3.3); wood and plastics and 3.7 X lo4 ft2/lb (7.6 m2/g) for
CS = -0.6255 (0.133).
smoke froc: flaming combustion of these same materi-
als.
Substituting Equation (3.14) into Equation (3.1 1)
results in

the smoke were initating?


From Table 3.3, K = 8.
where
Extinction coefficient = 0.207 A-'.
S = visibility, fi (m);
From Equation (3.12), K = proportionality constant (Table 3.3):
S = -[-S .6255 - 1.47 log(.207)] = 15 ft (1.6 m) = specific extinction coeflicient, ft2/lb ( n ~ ' / ~ ) ;
a,,,
.207

An alternate approach to calculation of visibility n7/, = mass concentration of particulate lb/fi3 (glm3).
from the mass concentration of particulate is obtained Equation (3.15) relates visibility to the mass concen-
from combining Equations (3.10) and (3.1 l ) with the tration of particulate. The comment concerning the utility
conversion from optical density to extinction coefficient of Equation (3.13) also applies to Equation (3.15).
(Table 3.2).

0 Irritating Smoke
Non~mtatingSmoke

where
S = visibility, fi (m);
K = proportionality constant (Table 3.3);
S,,, = inass optical density, ft2/lb(m2/&;
= mass concentration of fuel burned lb/ft3(g/m3)
I

Mass optical densities for some wood and plastics 0 0.2 0.4 0.6 0.8 1.0 1.2

are in Table 3.4. Equation (3.13) can be useful because Extinction Coefficient, a (m")

the mass concentration of fuel burned can be calculated Figure 3.4 I+hlking spcctl it7 irrim1it7got7d not7it-rila1-
from a smoke transport model as discussed later in the i17gs117oke(crthp!ed.fi-orxJir~[I9S5]).
Principles of Smoke Management

Table 3.4:
Mass Optical Densities (adapted from Mulholland2002)

Mass Optical Density, Sample


- 4" Combustion Thickness
Material ft%b m21g Conditions in. cm
Natural Materials:
Plywood Pyrolysis 0.24 0.6
Wood (Douglas fir) Pyrolysis 0.24 0.6
Cotton Flaming1
Cotton Flaming2
Synthetic Materials:
Polymethylrnethacrylate (PMMA; ~ l e x i ~ l a s ~ ~ ) Pyrolysis
Polyvinylchloride Pyrolysis
Polyvinylchloride (with plasticizer) Pyrolysis
Neoprene Pyrolysis
Polypropylene Flaming 1
Polyethylene Flaming1
Paraffin wax Flaming1
Polystyrene Flaming1
Styrene Flaming1
Polyvinylchloride Flaming1
Polyurethane Flaming1
Polyurethane Flaming2
Latex Flaming 1
Latex Flaming2
Neoprene Flaming1
Neoprene Flaming2
Polystyrene Flaming1
Polystyrene Flaming3
Polystyrene foam Flaming1
Polystyrene foam Flaming3
Acrylonitrile-butadiene-styrene (ABS) Flaming1
Acrylonitrile-butadiene-styrene (ABS) Flaming3
1. Samples in horizontal conliguration (0.005 m').
2. The sample is a mattress.
3. The sample is a plastic utility table.
The use of trade names implies neither recommendation nor endorsement ofany product by the authors or publ~sher.
Chapter 3 -Smoke and Tenability

Tiie airborne particulates produced by a fire consist Example 3.4 Visibility Due to a Pillow Fire
primarily of soot, and the production of particulates can
If smoke from the burning of a 0.50 Ib (230 g) polyurethane
be estimated as '
foam pillow were uniformly mixed in a 20 ft (6.1 m)
square, 10 ft (3.05 m) high room, what would be the visibil.
ity of a lightemitting sign?
where
Approach 1: From Table 3.5, the particulate yield of flexiblt
MP = mass of particulates produced, Ib (g); polyurethane foam is 0.1 88. From buation (3.16), the mass 0)
airborne particulate is
My = mass of he1 consumed, Ib (g);
yp = particulates yield (dimensionless).
Values ofyp are listed in Table 3.5 from small-scale From Equation (3.17), the mass concentration of the particu-
experiments of turbulent flaming combustion for a num- lates is
ber of materials. While it is expected that particulate
production will vary with the size of the fire and the ori-
entation of the fuel, the data of Table 3.5 are rccom-
mended h the absence of data from the kind of large
Using a, = 3 . 7 1~o4 fiZAbfor flaming combustion and K = 8
fires for \vhicli smoke management systems are Fable 3.3), visibility is calculated fi.orn Equation (3.15) as
designed.
Considering a \veil rnixed space, the mass concen-
tration of the pa~ticulatesis
visibility of a light-emitting s i y .

4pproach 2: The mass concentration of fuel burned is calcu-


!ated from Equation (3.9):
where
I/, = volume of: h c smoke in the space, li3 (11i3).
Equation (3.17) can be used for a laboratory test
From Tablc 5.4, the mass optical densiy 4,.of polyurethmc
where l/,. is [he volume of the test chamber. This equa-
Foam from a flaming mattress fire is 1600 ft'nb (0.33 m' g).
tion also can be used for a tire in a room or atrium where Visibility is calculated from Equation (3.13):
VC is the volume of tlie smoke layer. In both cases, the
smoke volume is considered to be well mixed so that the
smoke properties are uniform throughout the volume.
For a h e wit11 a constant heat release rate, the mass
kibility of a light-emitting sign.
of fuel consuliied by a fire can be expressed as
WCsee that this is different from !he 9 fr (7.7 m) estimated in
~pproachI, and this is indicative ofthe limitati~nsof this tech-
iology, including availability of a," and d,,, data.

where EXPOSURE T O GASES


M/ = mass of Cue1 consumed, Ib (g); I n the following sections, information about human
responses to exposures to toxic gases applies to an a w r -
= total heat rdcase rats Gtuis (kW); age person. A person's response to an exposure to toxic
gases primarily depends on age, metabolism, health his-
AHch = chemical hcat of cornbustion Btullb (klkg); tory, and respiratory rate.
1 = timc li.om ignition, S (S); Carbon monoxide (CO) poisonin,o accounts for the
majority of total fire fatalities (.Berl and Halpin 19SO;
K/- = 1 (1000). Harland and Woolley 1979). Table 3.6 lists toxicity dsta
Values of' I'or some materials are listed in for several gases, but only a few gases have been incor-
poratcd in predictive toxicity models. The toxic efficts
Tablc 3.5. Iri tires. combustion is never complete. Com-
of CO are probably the most well known, but some
bustion efficicnc is the ratio of tlie chemical heat of o h e r gases included in toxicity models are hydrogcn
combustion LO the ncl Iieat of combustion. Using AH,.,, cyanide (HCN), hydrogen chloridc (HCI), and hydrogsn
eliminates thc nccd to consider conibustion efliciency. bromide (HBr).
Principles of Smoke ~ a n a ~ e m & t

Table 3.5:
Particulate Yield of Heat of Combustion for Well-VentilatedFires of Solid ~ u e l s '
Particulate. Chemical Heat of Combustion, AHch
Yield
Material Yp Btulib kJ/kg
Natural Materials:
Wood (red oak)
Wood (Douglas fir)'
Wood (hemlock)
Fiberboard*
Wool 100%
@ntlretic materials:
Acrylonitrile-butadiene-styrene(ABS)
Polymethylmethacrylate (PMMA; plexiglasTM)
Polypropylene
Polystyrene
Silicone
~o~~este?
Nylon
Silicone mbber
Poly~lrethaneFoam (Flexible))
Polyurethane Foam ( ~ i ~ i d ) )
Polystyrene ~ o a m )
Polyethylene ~ o a m )
Phenolic Foam
Polyethylene (PE)
PE with 25% chlorine
PE with 36% chlorine
PE with 48% chlorine
Polyvinylchloride (PVC)
PVC 1 (L01 = 0.50)
PVC 2 (L01 = 0.50)
PVC (L01 = 0.20)
PVC (L01 = 0.25)
PVC (L01 = 0.30)
PVC (L01 = 0.35)
Ethylenetetrafluoroethylene (ETFE; TcfzelTM)
Perfluoroalkoxy (PFA; TenonTM)
Fluorinated polyethylene-polypropylene(FEP;
TenonTM)
Tetrafluoroethylenc (TFE; ~ e f l o n ~ " )
1. Data from Tewarson (2002) except as othenvise noted.
2. Paniculate yield data from Mulllolland (2002).
3. Values listed are an average o f a nurnhcr ol'd~lTerent nialerials under this general name.
'"The use o f trade n a n m irnplics neithcr reconimendation nor endorsenient o f any product by [lie authors or puhlishcr.
Chapter 3-Smoke and Tenability

Table 3.6: due to building fires tend to change with time. Thus,
Lethal Concentration of Some Gases Haber's rule has limited use for tenability calculations.
In the past few decades, tenability limits have been
Gas LCS0for 30-Minute expressed in terms of time integrated values. Time inte-
Exposure (ppm) grated values account for the effect of exposure to a
Co2 carbon dioxide , 470,000 changing concentration of a particular gas over a period
C2H40 acetaidehyde of time rather than an instantaneous exposure. The E
acetic acid parameter in Haber's rule can be considered a time inte-
C2H402
grated value with a constant gas concentration. If the
NH3 ammonia concentration is variable in time, then an integration
HCI hydrogen chloride must be conducted to obtain the area under the concen-
CO carbon monoxide tration-time curve in order to determine a time inte-
HBr hydrogen bromide grated value.
NO nitric oxide
COS carbonyl sulfide FED from Animal Test Data
hydrogen sulfide While most animal toxicity tests have been con-
H2S
ducted on rats, other animals include mice, guinea pigs,
HF hydrogen fluoride
hamsters, and rabbits. Because of concern for animal
C3H4N acrylonitrile rights, the toxicity research programs used the minimum
COF2 carbonyl fluoride of animals, and most laboratories stopped animal testing
nitrogen dioxide near the end of the 20th century.
NO2
These tests determine the concentration of airborne
C3Hj0 acrolein
combustion products that is lethal to 50% of the test ani-
CH20 fonnaldeliyde mals exposed for a specified time, and this lethal con-
HCN hydrogen cyanide centration is referred to as the LCjo. The specified time
C9H602N2 toluene disocyanate for animal tests is usually 30 minutes, and the number of
fatalities consist of animals that die during the test and
COCl, phosgene
during a post-exposure time, usually 14 days after the
CAFX perfluoroisobutylene test.
Using extrapolated animal test data, the fractional
Hyperventilation due to carbon dioxide (COz) effective dose is
exposure will increase the rate of intake of CO. Oxygen
( 0 2 ) deprivation is a special case, and the reduction in
the amount of O2 available for tissue respiration is
referred to as hypoxia. Because of the interaction of
these gases, exposure effects discussed below consider where
the combined effects of these gases. The effect of expo- FED = fractional effective dose (dimensionless);
sure to toxic gases on a specific individual depends on
C = concentration, 1blft3( & I ~ ) ; -
the physiological characteristics of the individual.
f = exposure time (min);
Exposure and Time LCf 50 = lethal exposure dose from test data, Ib ftJ min
Haber (1924) proposed that the effect of an expo- (g m'3 rnin).
sure to a gas is related to the product of the gas concen- An FED greater than or equal to one indicates fatal-
tration and time duration of the exposure. Haber's rule is ity. The concentration, C, is the density of materials that
expressed as started as fuel that have accumulated at a location at
time I. This concentration has units of mass of the mate-
rial burned per unit volume. The lethal exposure dose,
where LCI ,o, is the product of the LCso and the exposure time.
E = effect of exposure (ppm-min), Table 3.7 lists some values of LCfSOfor a number of
common materials.
C = concentration (ppm), and
The above equation is the time-integrated form of
I = duration of exposure (rnin). the FED equation. For most applications, the time func-
This elementary equation assumes a constant inges- tional relationship of concentration is not known, and
tion rate of the tosin. The effects of some gases do not the following expression can be used for discrete pairs
follow Haber's rulc, and concentrations of toxic gases of concentration and time intervals.
Principles of Smoke ~ a n a ~ e m e ' * t

Table 3.7:
Approximate Lethal Exposure Dose, LCtSO,
for Common Materials (adapted from Purser 1995)
-
Nonflaming Fire ('Fuel-Controlled Fire \ Fully Developed Fire
Material Ib min g m-3 min f lb-ft-3 n i n- g md min)i lb ff3 min g m-3 min
Cellulosics 0.046 730 - 0.19 -3120 d 0.047 750
C, H, 0 plastics 0.03 1 500 0.075 1200 0.033 530
PVC 0.03 1 500 0.0 19 300 0.012 200
WooVNylon (low N2) 0.03 1 500 0.057 920 0.0044 70
Flexible Polyurethane 0.042 680 0.087 1390 0.012 200
Rigid Polyurethane 0.0039 63 0.0062 100 0.0034 54
~odacrvlicl~~~' 0.0 10 160 0.0087 140 0.0028 45
I. PAN is polyacr).lonitrile.

Example 3.5 Calculation of FED


Would a 20-minute exposure to atmosphere in a room resulting
!?om burning 6 Ib of flexible polyurethane foam in the room be
expected to be fatal? The size ofthe room is 8 R by 12 R by 8 fi
(2.44 m by 3.66 m by 2.44 m).
where
Flexible polyurethane foam would be expected to bum very
Ci = concentration for time interval I, lb/ft3 (eJni3); rapidly compared to the 20-minute exposure time, so the con-
Afi = time interval i, min (rnin); centration in the room can be considered constant.

LCf = lethal exposure dose from test data, Ib rnin C =


mass of fuel burned 6
volume of space - (8)(12)(8)
(g m-' min);
n = number of discrete concentration time pairs.
When the concentration is constant, Equation (3.2 1) From Table 3.7, LCt 50 = 0.087 Ib ft-3 rnin (1390 g m-3 min)
written as for a fuel-controlled fire.
Because the concentration is constant, the FED is calculated as
Cr
FED = - (3.22) Ct
LCt50 FED = - -
LC,,,
- 0.0078(20)
0.087
= l ,

Many references use the term corzcentrafion rime


This indicates that fatality would be expected.
pmdzicf, Cf, to mean the integral term of Equation
(3.22), and this meaning of Cf will be used for the rest o f
this book. Table 3.8:
The question arises, should incapacitation or fatal- Components of Air
ity be used as the design criterion for gas exposure. A constituent' O h by Volume
person who is incapacitated due to exposure to toxic
Nitrogen (Nz) 78.084
gases will continue to be exposed to those gases. Unless
the person is rescued or the gas concentrations improve Oxygen (01) 20.946
dramatically, such exposure will result in fatality. Carbon Dioxide (CO2) 0.033
Incapacitation often is used to mean the condition Argon (Ar) 0.9?4
that self-evacuation is very difficult or impossible. Usu- Trace Gases (He, Kr, Xe, H?,CH,, and N20) 0.003
ally an incapacitaling dose is less than a fatai dose, but
1. alrd Plryics (CRC 1985)
Handhook oj Clrerrrisr~)~
this is not always the case. It is possible that a person
could walk out of the smoke-filled environment only to Components of Air
die some time later.
While a FED of one indicates fatality, Bukowski et Calculations using predictive toxicity gas models
al. (1 989) state that an FED of 0.5 can be considered an involve the components o f air, and these components are
approximation to the incapacitaling dose. I L is possible listed in Table 3 3 . The small concentration of CO2 is
that this approximation is a conservative criterion for essential to control normal breathing, but it does not
smoke nnanage~ncntdesign analysis. have a significant impact on toxicity calculations.
Chapter 3-Smoke and Tenability

For the fire protection purposes of this book, the is seen for any time. In the former case, this would rep-
small quantities of CO2, argon (Ar), and the trace gases resent such physiological effects as breath holding and
are neglected, and air is considered to be composed of the time required for the gas to be transferred to the
20.9% O2 and 79.1% NZ by volume. Some sources use blood and then to the tissues. In the latter case, this rep-
21% 0 2 and 79% N2 by volume, which also yields use- resents an exposure concentration for which the equilib-
ful engineering results. : rium concentration of carboxyhemoglobin (COHb) in
the blood is below the level that causes lethality (Levin
CO and CO2 et al. 1987).
Exposure to CO results in carboxyhemoglobin Following the work with CO, the effect of CO2 on
uptake (COHb) in the blood, which results in decreased the observed CO toxicity was studied. The result of this
oxygen-carrying capacity of the blood. Stewart et al. work was the observation that the "effective toxicity" of
(1973) conducted a series of experiments on humans CO increases with increasing CO2 concentration, dou-
and, based on this research, C O H b uptake can be bling at a level of about 5% (50,000 pprn), as shown in
expressed as Figure 3.6. The physiological effects of the CO2 are to
increase the respiration rate and reduce the blood pH,
producing a metabolic acidosis. The interaction beh~.een

where
CCOHb = concentration of COHb in the blood%;
1
CCOHb,O= concentration of COHb in the blood at time I
I
zero,%; Asymptote
CCo = concentration of CO in air, pprn; - l minute

V = volume of breathed air per minute, Llmin;


Asymptote
Ati = exposure time interval, min. -
1700 ppm 3350 PPm
- - - - - -at-60_min_
*
___
Equation (3.23) does not include the effects of oxy-
gen depletion, increased breathing rate due to CO2 "0 10 20 30 40 50
Time (minutes)
exposure, or exposure to other toxic gases. The volume
of breathed air, V , is called the respiratory minute vol- Figure 3.5 Carbori I I I O I I O S I ~COI~C~I~II-a1io11
~ VS. time to
ume (RMV). The typical RMV of a 150 Ib (70 kg) per- letlinlitj~($SO% of exposed rate (odnyied
son at rest is about 8.5 Llmin. O'Neill et al. (1980) used
a higher RMV of 18 Llmin to account for activity and.
CO2 exposure, but this approach can significantly
underestimate toxic effects, as is discussed later. For cal-
culations, a value of CCONb,O = 0.75% can be used, and
incapacitation and lethality are approximately 25%
COHb and 50% COHb, respectively. However, calcula-
tion of the COHb level from Equation (3.23) is not a .. ..&&-EL.
I ,, Deaths I ..?E%
reliable indication of toxicity Lr incapacitation because
it does not include the effects of other gases commonly
present in smoke (see Example 3.8).
In the development of predictive toxicity gas mod-
els for fire applications, the first pure gas to be studied
was CO. Rats were exposed to varying concentrations of
pure CO for various times, and the concentrations nec- 0 l , . . 1 , . '.'..,.."-.m

essary to produce deaths of 50% of the exposed animals 0 1000 2000 3000 4M)O 5003 E
Carbon Monodde (ppm)
(the LC50)for each exposure time was determined. The
plot of these data (Figure 3.5) shows that the curve has Figure 3.6 Dearlls,fi.o~ne.vposi~reto CO alone and CO
two asymptotes-an exposure time (about I minute) p1zr.s COz (udoptcd ,%.on7 Lcvin er ul.
below wliich no cl'fect is seen for any concentration and
( 1 9s 71).
a concentration (about 1700 ppm) below which no effect
. . .,... .
.;,:A;:

>.

. . Principles of Smoke ~ a k g e m e * t

CO and CO2 is apparent from the formulations of the N- The model incorporates the ~ncreasedbreathing rate
gas and FINmodels that follow. due to CO2 exposure. It is apparent that there is a unique
interaction between HCN and NO2. For many of the
N-Gas Model gases, the contribution to lethality is expressed as the
The N-gas model was developed at the National ratio of the gas exposure to the LCS0. This is how O2 is
Institute of Standards and Technology (NIST) and treated, except that it is in terms of oxygen depletion.
-

relates fatality with animal test data of exposures to pure


gases and mixtures of gases (Levin 1996; Levin et al. The toxicity of CO2 is not included in the N-gas
1995; Babrauskas et al. 1991). For mixtures of gases, model because fire-generated atmospheres do not con-
including NO2, the N-gas model can be stated as tain toxic concentrations of CO,. The LCso of CO, is
47% and the maximum concentration of CO2 in a fire
atmosphere is 20.9% if all of the oxygen in the air is
converted to CO2.
For animal tests, it was found that when the NG,,
value was approximately I, some of the animals died.
For values below 0.8, there would be no fatalities, and
for values above 1.3, all of the animals would be
and for mixtures not including NO2, the N-gas model can expected to die.
be stared as The time-integrated average exposure to CO is
,,,[CO] 20.9 - [ 0 2 1 1 '
-1 = l..
+

NGas = [ C O 2 ]- b 20.9 - L C p ( 0 2 ) , [CO] = Ccodf


(3.25) fe,=o
[HCN] [HCf] [HBI-]
+ LC,,(HCN) +

LC,,(HCf)
+

LC,,(HBr) where I, is the exposure time. The other time-integrated


averages can be expressed in a similar manner. For discrete
where
concentratio11data, the time-integrated average can be writ-
N ~ a ~= N-Gas model indicator (dimensionless);
ten as follows:
n7 = -18 for CO2 S 5% and 23 for CO2 > 5%;
B = 122,000 for COz < 5% and -38,600 for
CO, > 5%; [CO] = -
f
1
e
2 Cco, ;At;
i: l
LCSO(OZ) = lethal concentration of 0 2 % ;
LCjo(HCN)= lethal concentration of HCN, ppm;
LCS0(NOZ)= lethal concentration of NO2, ppm;
LCSO(HCI)= lethal concentration of HC1, ppm;
LCSO(HBr)= lethal concentration of HBr, ppm;
[CO] = time-integrated average exposure to CO,
PPm;
[CO?] = time-integrated average exposure to COz,
PPm;
L021 = time-integrated average exposure to O-,,
W);
[HCW = time-integrated average exposure to HCN,
PPm;
[NOz] = time-integnted average exposure to NOZ,
PP'T
I
[HCI] = time-integrated average exposure to HCI,
ppm;
[HBr] = -
l'> 2 C,,,., ;Ati
i= l
[I-IBr] = time-intcgrated average exposure to H Br,
ppm.
&apte; 3 -Smoke and Tenability

Cco,i = concentration of CO, pprn; Equation (3.27) can be used where the time inter-
vals are either uniform or nonuniform. For uniform
CCm,i = concentration of CO2, ppm;
intervals, the time-integrated average terins of these
C02,i = concentration of 02,%; equations become mean averages. When the concentra-
tion of any of the gases other than 0 2 is zero, the contri-
CHCNi = concentration of HCN, ppm;
bution of that gas to the NW value is also zero. This is
Cm,, = concentration of NO2, ppm; to be expected, but it is not so for the fractional incapac-
CHCISi= concentration of HC1, ppm; itating dose method discussed later.
Equations (3.24) and (3.25) apply when the expo-
CHB,+ = concentration of HBr, ppm; sure time is the same as the duration of the LCS0data.
fe = exposure time, min; Example 3.6 demonstrates the use of the N-Gas model
for four gases, but Table 3.9 has LCSo values for all of
At = time interval i, min;
the gases in this model for many exposure times. For
n = number of concentration values for each gas exposure times between those listed in this table, LC50
and time interval. values can be interpolated.

Example 3.6 Using the N-Gas Model . . .


lalculate /VGrrsfor a 20-minute exposure to the mixture of gases listed below.
Time C0.i CCOZ.; GO.
i c ~ ci ~ ~ .
I (m in) Yo . PPm PPm PPm
0 0 20.90 0 0 0
I 2 20.72 5SO 40 2
2 4 20.30 1900 60 3
3 6 19.80 3200 120 6
4 8 19.70 3600 120 6
5 10 19.60 3800 I60 8
6 I2 19.60 3800 500 25
7 14 19.60 3800 600 30
S I6 19.60 3800 600 30
9 IS 19.60 3800 600 30
10 20 19.60 3800 600 30
The time-integrated avcragc exposures can be calculated from Equation (3.27). Bccausc the intervals are unifomi, the time-
integrated average ternis are mean averages of the concentrations as listed below.
[02]= 19.8 1 [CO] = 340
[CO,] = 3208 [HCN J = 17
Bccausc rhcl-e is no exposure to HCI and H&, Equation (3.25) becomes

Bccausc COz is less than 5% (50.000 ppm), t11 = -1 8 and b = 122,000.


For a 70-minute exposure. lethal concentrations from Table 3.9 are LCS0(02)= 5.2% and LCjO(HCN)= 170 pprn.

This exposure ii'ould not be expected to cause fatalitv.


Principles.of Smoke Management

Table 3.9: ' FIO,i = hction of an incapacitating dose of ~ O W -


Lethal Concentration, LCSO,of Various Gases oxygen hypoxia per unit time (min-l);
Exposure FImSi = hction of an incapacitating dose of CO2 per
unit time (min-l);
Time HCN Oz HCI HBr NOz
Ati = exposure time interval i (min);
min PP"' PP"' ' PP"' PP"'
1 3000 - - - n = number of concentration values for each gas
2 1600 - - 1450 and time intervals.
5 570 4.0 15900 12600 830 The following terms are calculated as
10 290 4.8 8400 6600 510
15 230 5.0 6900 5400 380
20 170 5.2 6400 5100 320
25 160 5.3 5900 4700 290
30 150 5.4 3800 3000 200
45 120 5.6 3300 2600 150
60 90 5.8 2800 2200 100
I. Note: LC50 values based on data from Levin et al. (1988 and
1989). Levin (1996). Levin (2000). and Hanzell et al. (1990)
except for HBr. Because o f the chemical similarities o f HCl
and HBr, they are expected to have similar toxicological effects,
and most o f the above LC50 values for HBr were extrapolated
from those o f HCI.
where
Fractional Incapacitating Dose Cco,; = concentration of CO (ppm);
Purser (2002) developed a model to calculate a CHC.\r; = concentration of HCN (ppm);
fractional incapacitating dose for exposures to CO, CC02,; = concentration of CO2 (percent);
HCN, CO2, and reduced Oz. The notation in this section Co., = concentration of Oz (percent).
A value of FlIr of I or more indicates incapacita-
has been modified from that of Purser to facilitate com-
tion. and the incapacitation time based on can be
puter programming. taken as the time it takes for FINto become I.
Equation (3.29) represents incapacitation due to the
toxic effects of COz, and this equation was included for
completeness. As previously stated, fire-generated
atmospheres do not contain toxic concentrations of COz
Equation (3.29) may be useful for fire scenarios that
include sources of CO? other than the fire. For applica-
tions where there are no nonfire sources of CO1, Equa-
tion (3.28) should be used for the calculation of F/,,,.
As previously stated, the FINmethod is based on air
composed of 20.9% 0 2 . Any combustion calculations or
whichever is greater, where test measurements that are used for input to calculations
of F/,, should be consistent with this O2 concentration.
FIN = fractional incapacitatingdose of all narcotic Examination of Equation (3.30) sho\vs that
gases (dinlensionless); 1. for zero CO, Flco,i has a value of zero;
FIco,; = fraction of an incapacitating dose of CO per
2. for zero HCN, FICjV,,
has a value of about 0.0045 nii11-l;
unit tinie (min-l); and
= fraction ofan incapacitatingdose of HCN per 3. for zero 20.9% O1, FIO,,has a value of about 0.002 1
unit time (min-l);
For item I , it would be expected that a zero concen-
Vcoz,, = factor for CO1-induced hyperventilation; tration of CO would result in a zero contribution to the
Chapter 3-Smoke and Tenability

FIN. However, items 2 and 3 were unexpected. A z e r o - . about 3.3 hours can be calculated for exposure to an
concentration of HCN results in a positive contribution atmosphere of normal 0 2 and zero concentrations of
to the FIN,and no oxygen depletion ( 0 2 = 20.9%) also CO2, CO, and HCN. This exposure can be thought o f a s
results in a positive contribution. For the short exposure breathing normal air, and no such exposurewould result
times characteristic of most fire protection applications,
in incapacitation. This indicated that the FIN approach is
these positive contributions are small and should not be
of concern. inappropriate for long exposures. However, the FED
some are measured in hours as was and the N-gas model are based predominantly on test
the case for the World Trade Center explosion. From data with 30-minute exposure times, and applying these
Equation (3.28) and (3.30), an incapacitation time of models for long exposure times is also questionable.

Example 3.7 Using the F[,,, Model


For the gases of Example 3.6, calculate the FIN

Use Equations (3.28) and (3.30) to calculate the table below. Remember for FIN, CO2has units of percent.
i Time (min) FICO.i F ~ ~ ~ . i '~02,; Floei FIN
0 0 NIA NIA NIA NIA 0
l 2 0.00 13 0.00475 1.053 0.000325 0.013
2 4 0.00 19 0.00486 1.080 0.000407 0.029
3 6 0.0039 0.0052 1 1.107 0.000534 0.050
4 8 0.0033 0.0052 1 1.115 0.000563 0.072
5 10 0.0053 0.00545 1.119 0.000594 0.097
6 12 0.0 1 73 0.00806 1.119 0.000594 0.155
7 14 0.0209 0.00904 1.119 0.000594 0.223
S 16 0.0209 0.00904 1.119 0.000594 0.291
9 18 0.0209 0.00904 1.119 0.000594 0.359
10 20 0.0209 0.00904 1 119 0.000594 0.427

At 20 minutes of exposure, the FIj,,is about 0.43. This indicates that this exposure is not expected to cause incapacitation.

Example 3.8 Comparison of To~icitvhlodels


For the gas concentrations listed below, calculate NG,,, FIAi,and COHb.
Time c~?,i Cc02.i Cc0.i C ~ i ~ ~ . i
i (min) % PP"' PP111 PP"'
0 0 20.90 0 0 0
1 2 20.18 2320 320 8
2 4 18.50 7600 480 12
3 6 16.50 12800 960 24
4 8 16.10 14400 960 24
5 10 15.70 15200 1280 32
6 12 15.70 15200 4000 100
7 14 15.70 15200 4800 120
S. 16 15.70 15200 4800 120
0 IS 15.70 1 5700 4800, 120
10 20 15.70 15200 4800 120

Part I: In thc snnic manner as Example 3.6, Nh = 1.1 is calculated. This means fatality \\xiuld he expected
from rh~scsposurc.
Principles of Smoke Management

Example 3.8 (Continued) Comparison of Toxicity Models

Part 11: Calculations of FIN are similar to those of Example 3.7.


i Time (min) F l ~i ~ , F ~i ~ ~ vN
~ ~ 2 . i Flo.i
0 0 N/A N/A N/A N/A 0.000
I 2 0.0109 0.00545 1.OS8 0.00043 0.036
2 4 0.0166 0.00598 1.203 0.00 108 0.093
3 6 0.0340 0.00788 1.328 6.00317 0.210
4 8 0.0340 0.00788 1.369 0.00393 0.333
5 10 0.0458 0.00947 1.390 0.00488 0.496

I At F,,,.,,= 1, incapacitation is expected. From the above table, incapacitation is expected at about 12 minutes.

Part 11: COHb in the blood is calculated from Equation (3.23);where


11

- 1.036 . .
A C ~ = 3.3 I 7 X I O-'CCO, lfAt and CCoHh= CCOHb,o + ACCOHb.l
i= l

c' = 18 Urnin; At = 2 rnin: CCOH&,= 0.75%.

Time CCO.: ACco~b.i c,,,,


(min) * .;, % O/a
0 0.0Oil0.00000 0.7500
2 0.032 0.00003 0.7500
4 0.048 0.00005 0.750 1
G 0.096 0.000 1 1 0.7502
S 0.096 0.000 1 1 0.7503
10 0.12s 0.000 14 0.7504
12 0.400 0.00046 0.7509
14 0.450 0.00056 0.75 15
16 0.4SC1 0.00056 0.7520
18 O.IS(:I 0.00056 0.7526
20 0.480 0.00056 0.753 1
This lcvcl of COlib is below hat \vhich would resuit in either incapacitation or fatality, and these calcula-
I/ rams sl~ow[hat C@!{!, c~lculakdl h ~ Equation
n (3.23) is not a reliable indication of incapacitation or fatality.

EXPOSURE T O HEAT burns can be expected to be the dominant effect for d q


Exposure to elevated temprl.ratul-e atmospheres can air temperatures greater than 250°F (1 21°C).
lead to skin burns and hypenhermia (heat stroke). A The effect o.f esnosure to elevated tem~eratures
temperature limit of 250°F (1 2 1 'C) for d ~ air
y is used as depends on the of the a i r and the type and
a rule of thumb to dctermins \\.hicl1 of thcse two possi-
ble efl'cc~swill dominate. Generally, extent of clorhing worn. Physiologically, exposure to an
. csposure
. to hcatcd
dry a i r a, a ,empcratllrc less approximately 2jO"F elevated tcrnperature environment Can cause an increase
(121°C) lcads only to Iiypmhc.rmia. Pain from skin in body or blood temperatuce. Also affecting the thermal
Chapter 3-Smoke and Tenability

tenability limits is the presence of clothing. Perspiration


is a key mechanism used by humans to resist the effects
of exposure to a high-temperature environment. How-
ever, clothing may inhibit the efficiency of this natural
cooling process. Conversely, clothing provides insula-
tion from high-temperature environments to protect the
skin from becoming burned. Thus, at temperatures in
excess of 250°F (121°C), where pain from skin burns is
the dominant effect, the presence of clothing can be con-
sidered to be beneficial. However, at the lower tempera-
tures, where hyperthermia is the dominant effect, Exposure Time (minutes) .
clothing is detrimental. Figure 3.7 Ther-~nal~olerance for humans at rest,
nnked, with low air movement (adapted
As in the case of exposure to toxic gases, consider-
fi-on7 Blockley [1973]).
ation of the time duration of exposure is necessary to
accurately assess thethreat. A limit of approximately
300°F (150°C) is often stated for exposure durations of
five minutes. The thermal tolerance of humans at rest, The cumulative dose is the sum of the doses for
naked, with low air movement is shown in Figure 3.7. each of the intervals:
Purser recommends the following relationhip for time
to incapacitation based on averaging the time to inca-
pacitation for exposures to huniid air arid dry air:

where
where F1,ll = total cumulative dose (dimensionless);
1~1, = time to incapacitation due to thernial exposure,
F,,,,, = incapacitating dose for time interval i (dinien-
min;
sionless).
C, = 5.670 (5.185);
Incapacitation would be expected for FI,l, greater
C2 = 0.0 152 (0.0273);
than or equal to one. Substituting Equations (3.31) and
T = temperature of air, "F ("C). (3.32) into Equation (3.33) yields
Equation (3.3 l ) applies when the teniperature is not
changing witli time. To deal witli changing tempera-
tures, the same concept of a fractional incapacitating
dose that was used for gases can be applied to heat
exposure (Purser 2002). During any one time step, the whcrc
incapacitating dose is given as
Fl,/, = total cumulative dose (dini~nsionless);
G A/;
Flrh = - (3.32) Ati = cxposure time interval i, n:in;
[/A. i
= temperature of air in interval i, "F ("C);
where
C, = 5.670 (5.185):
F = incapacitating dose for the time interval (di~nen- C? = 0.0 157 (0.0373).
sionless);
Equation (3.34) is in a forin uscful for calculation
Ati = exposure time intcrval i, min; with lcmpc'raturcs ~ I - O ~ L I by
C Ca~smoke transport model

tlh,, = timc to incapacitation for tcmpcralureof'intenral or tcmpcrarurcs f.1-on1 lire tcsls 1,ccorded with a data
i, min. acq~risitiorl?.stem.
Principles of Smoke Managemelit

1
- --
--
Example 3.9 Cumulative Exposure to Heat
Determine if incapacitation would be expected for a petson exposed to a smoke layer where the average smoke layer tem-
perature during the first minute is 125°F(52°C). During each ofthe next four minutes, the smoke layer temperature increases
25°F (14°C).

I in the following table, t f i , ,Fldl ,.,and Fit,, were calculated from Equations (3.31), (3.32), and (3.33).
I C 1111,i F I Ii ~ Ft~h

min "F min


1 125 43 0.02 0.02
2 l50 30 0.03 0.06
3 175 20 0.05 0.11
4 200 14 0.07 0.18
5 225 9 0.11 0.28
Since the total F,,,, is well below 1.0, incapacitation would nor be expected.

EXPOSURE TO THERMAL RADIATION


Thermal radiation can cause pain, blistering. and
burning of exposed skin. Exposure to thermal radiation observers be able to approach the fire?
is often not addressed in discussions of tenability for From Chapter 2, the separation distance for nonpiloted
smoke control applications because of the limited ignition due to thermal radiation can be adapted for the
smoke temperatures for such designs. Gas temperatures separation distance to prevent skin pain as
that are tenable for contact with skin are also tenable
with respect to thermal radiation.
Stoll and Chianta (1969) show that the exposure
time to pain and blistering can be represented by
where
Q, = radiant heat release of the fire, Btuls (kW);
RsD = separation disrance from the center of rhe fire to a
III
person, fi (m);
q,, = limit of radiant flux to prevent pain, ~ t d f t 's (kW1

Ill2).
Calculare Q, = 1000 (0.3) = 300 Btds (320 kW), and use
where
I,:[,
= exposure time to pain, s (S);
1'0"
;,,h = exposure time to blister, s (S); RsD = -= 10 ft (3 m) separation distance.
d4d.22) I
qr = intensity of thernial radiation, ~ t u &s ( k ~ l m ' ) :
TENABILITY AND PERFECT DILUTION
C,:, = 3.20 (85);
It is cornnion to encounter situations where the dilu-
C,, = 8.39 (223). tion necessary to meet some visibility criterion results in
sucn ION gas concentrations that toxicity. is not an issue.
The above relationships are shown in Figure 3.S. .A
Generally, such dilution also results in smoke tempera-
value of q): = 0.22 ~ t u l f t 's (2.5 kwlrn2) is often used as tures so low that heat exposure and thermal radiation
the value that can be tolerated for a few minutes \vithout exposure are not issues. However, this is not so for fuels
unbearable pain. that produce low amounts of soot.
Chapter 3-Smoke and Tenability

Radiant Flux (kW/m2) In Equatiorl(3.22), the concentration C is the same as the


mass concentration of fuel burned, mj So that equationcan
be written as

where
FED = fractional effective dose (dimensionless);
m/ = mass concentration of fuel burned, l b l g (g/
m3);
f - exposure time (min); and

LCt 5, = lethal exposure dose from test data, Ib ftJ min


(g m-3 min).
Without heat transfer, the smoke temperature will
be
Radiant Flux (Btuk fi2)
Figure 3.8 Tolerai7ce of humat1 skin to thermal radiant
J11u (adapted fj-otn Stoll a17d Chianta
[l 9691). where
Tg = smoke temperature, "F ("C);
To = ambient temperature, "F ("C);
Klote (1 999a) developed equations based on perfect
dilution that allow relative comparison of visibility, tox- Q = heat release rate of fire, Btu (Id);
icity, and temperature for a particular fuel. This section M, = mass of smoke, Ib (kg);
presents a similar but more straightPonvard approach. Cp = specific heat ofsmoke, 0.24 BtuAb "F (I .O I d k g
The analysis considers that the products of conibus- "C);
tion (particulates, gases, and heat) are diluted by air.
The follo\ving equations are needed in order to get
This analysis neglects smoke panicle aging (agglomera-
the desired expression for the smoke temperature:
tion and deposition), reduction of specific gases, and
heat transfer. These are all conservative assumptions in
that they result in higher predicted levels of dilution to
meet tenability criterion. Further, almost all smoke
transport calculations neglect smoke particle aging and
reduction of specific gases.
This analysis consists of putting visibility, the effect
where
of toxic esposure, and smoke temperature in terms of a
common variable so comparisons can be made. The Mj = mass of fuel bumed, Ib (g);
variable selected is the mass concentration of fuel AHc11= chemical heat of combustion, Btu4b (kJ/kg);
burned, 11.;: Equation (3.13) already has visibility in such
terms: PS = density of smoke, lb/fi3 (kg/m3);

i"2 = ambient pressure, lb/$ (Pa);


R = gas constant of smoke, 53.3 ft Ibfllbm O R (287 JI
kg K);
C7. = 460 (273);
visibility. ft (m);
proportionality constant (Table 3.3);
v, = volume of smoke, h3(m3);

mass optical density, li2/lb (m2/%);


9 = 1 ( 1000).
Substitu~ingEquations (3.39) into Equation (3.38)
mass concentration of furl burned 1b/ft3(g/rn3). and rearranging yields
Principles of Smoke Management

m
Rmch
The design criterion for visibility can be put into
T,=-
+
- CT where a = L . (3.40)
l-a K/ CpPa Equation (3.41) to get the maximum value of the mass
concentration of fuel burned to meet the visibility ciite-
Equations (3.36), (3.37), and (3.40) are in terms of rion, and Equations (3.37)-and (3.40) can be used to cal-
the mass concentration of fuel burned, m/:Equation (3.36) culate the upper limits of the FED and T, resulting from
can be solved for mass concentration of fuel burned: this mass concentration of-fuel burned. This approach is
used in Example 3.1 1.

Example 3.11 Evaluation of Toxicitv and Heat Exposure from Visibilitv Criterion
For a visibility criterion of being able to see an illuminated exit si.p 30 fi (9.1 m) away, are toxicity and heat exposure
)/ calculation needed in addition tovisibility calculations? The fuel ispolyurethane. .

I Part I: Calculate m/

From Table 3.4, the mass optical density, ,a


, is 1600 f&lb (0.33 m'@.
From Table 3.3, K = 8 for an illuminated sign. Visibility, S, is 30 fi (9.1 m).
From Equation (3.41),

11 This is the mass concentration of fuel burnedthat satisfies the visibility criterion.
1 Part 11: Calculate FED

Use an exposure time of 20 minutes.


From Table 3.7, the lethal exposure dose from test data. LCI jo, is 0.087 Ib fYz tiiin (1 390 g m-' min).
From Equation (3.37),

I This is an upper limit on the FED in that it is at the highest value of mass concentrarion of fuel burned.

I Part 111: Calculate T,


From Table 3.5, the chemical heat of combustion, AHd,. is 7570 Btdlb ( I 7,600 kJkg).
P,= 14.7 (144) = 2120 lb/ft2. To = 75 "F (24 "C).
From Equation (3.40),

This temperature is the upper limit for the smoke based on dilution, aud it is not a s o n c a n wirh regard to heat exposure. This
example shows that calculations for esposure to toxic gases and heat exposure arc not necessary. provided that the systcm \\as
designed to meet the visibility criterion. Because heat exposure is not an issue. exposure to thernial radiation is also not an issue.
Chapter 3 -Smoke &d ~ e i a b i l i t ~

TENABILITY CRITERIA remain in the toxic environment until fatality or rescue,


In the most general sense, the criterion for all tena- it seems that the criterion should at a minimum be based
bility systems could be stated as: tenable conditions are on incapacitation. Exposures to some gases (for exam-
to be maintained in spaces where people are expected to ple HC1 and HBr) can result in post-exposure fatality,
be for the expected duration of their time in those such that a person might not be incapacitated while
spaces. However, such a criterion is too general to be being exposed but die some time after exposure. Con-
useful for design applications, and more specific criteria sidering both the dominance of CO among toxic fire
are needed. More detailed criteria deal with one or more gases and that CO does not result in post-exposure fatal-
of the following: exposure to toxic gases, exposure to ities, incapacitation could be a sufficient criterion for
heat, exposure to thermal radiation, and visibility most applications.
through smoke. It is the nature of such detailed criteria
that it depends on the specific application.
The time for exposures can be mandated in codes,
and Chapter 4 provides information about people move-
ment that can be used to calculate this time. For the con-
ditions of Example 3.1 1, the exposures to toxic gases,
The visibility distance for exit
heat, and thermal radiation are insignificant provided
signs depends on the distance between the exits in a spe-
that the system was designed to meet the visibility crite-
cific building. The visibility distence for seeing balcony
rion. For such insignificant exposures, detailed tenabil-
walls and railings might be taken as two or three times
ity criteria have no real purpose. Whenever possible,
the width of the balcony. In many applications, the crite-
this approach can sgnificantly simplify design analysis.
rion for seeing the exit signs wodd be expected to be
For applications where exposure to toxic gases is
the more stringent of the two.
significant, it might seem that the tenability criterion
should be based on prevention of both incapacitation For additional material about survival of exposure
and fatality. Because a person who is incapacitated will to fire produced environments, see Gann (2001).
CHAPTER 4

Evacuation Analysis

his chapter presents information about evacuation Quarantelli (1979a) provides the following state-
analysis for application to smoke management ment concerning behavior in fire incidents:
systems. In hazard analyses, evacuation behavior Overall my point has been that in both absolute
needs to be assessed to estimate the time duration and relative terms, human behavior in disasters in
in which an individual is exposed to a particular envi- modem, industrial societies is fairly good by any
reasonable criteria one could use. There is little
ronment. The evacuation time is composed of at least evidence beyond anecdotal stories, and none of a
the following three periods of time: systematic, comparative and quantitative nature
that suggests that behavior under stress is any
more illogical, irrational or dysfunctional than
Becoming aware of the tire everyday behavior.
Preparing for movement Part of the problem is that sometimes the behavior
Movement to an exit under stress is compared not with everyday
behavior, but with an idealized conception of
Generally, an evacuation analysis considering only behavior. Of course along that line it does not
come out well. But this is a match of real disaster
these three steps assumes that the individual's only behavior with the ideal, when the honest compari-
action is to evacuate. In addition to evacuating. an indi- son should be between real disaster behavior and
vidual [nay investigate, attempt extinguishment, assist actual everyday behavior. If the last kind of match
is made, there is not that much difference between
others, call the fire department, etc. An evacuation anal- the two.
ysis could account for many of these other actions in the While panic is perceived by nontechnical individu-
"preparing for movement" step. als to occur quite frequently in fires, it actually occurs
During building fires, elevators are almost always very infrequently. As noted by Quarantelli (1979b) and
taken out of service and vertical evacuation is by stairs. confirmed by Bryan (2002) and Keating (1982), most
In a few situations, elevators are used for e\.acuation. commonly people respond adaptively to the fire incident
For information about calculating evacuation time by and are often altruistic in their behavior. In Wood's
elevators, see Appendix C. (1971) study of human behavior in fires. he noted that
peoplz acted to increase their level of risk in only 5% of
THE MYTH OF PANIC all fire incidents. According to a panel convened to
Often, movies, television. and the press present the address panic, the characteristics of panic behavior
unrealistic image that panic bchavior in fire situations is include the following:
common. However, extensi\.t. research supports the con-
clusion that panic behavior in fire situations is \.cry rare. Acute fear
Even in large building fires resul~ingin multiple deaths, Perception ofxrisis
people experiencing fear still usually act in pi~rposeful Fear of separation (exceeds that of self-preserva-
ways. tion)
Confusion
Chapter 4-Evacuation Analysis

Table 4.1:
Types of Fire Alarm Signals Used in Drills in London Subway Station

Type Description
Bell only Alarm bell rung, no staff or PA
Staff Alarm bell rung, two staff members gave PA announcement to "evacuate station"and
then directed evacuation
Public Address Each 30 seconds, PA announcement said twice, "please evacuate the station immediately"
Staff + Public Address PA announcement instructing people to leave via trains or exits, with staff directing
people following the directions of the announcement
Directions + Public Address Same as stafffPA,except occupants were also told about the type (fire) and location of the
incident

Table 4.2:
Comparison of Response to Various Fire Alarm Signals
Time (min, S) to Start Time (min, S) to Start
to Move From to Move to Bottom of
Evacuation Alarm Concourse Escalator Comments
Bell Only 8:15 9:OO Delayed or no evacuation
Staff investigates, makes PA announcement. 2:15 3:OO Occupants directed to con-
directs evacuation course
Plain "recorded" PA announcement. repeated . 1:15 7:40 Occupants stood at bottom of
every 30 seconds escalator
PA directive + staff directing evacuation 1:15 1 :30 Occupants evacuated
P;\ directive plus status 130 1:00 Occupants evacuated

Extreme frustration difficult to estimate reliably, being a function of the fire


Chaotic/antisocial behavior scenario, building characteristics (compartmented ver-
Entrapment sus open-plan), and thc proximity, alertness, and mental
Flight abilities of the occupants.
- Contagion
PRE-MOVEMENT
The panel indicated that all nine characteristics may Interpretation o f t h e alarm signal as an indication of
not be evident for every individual who does engage in a threatening fire by building occupants is dependent on
panic type behavior (Quarentelli 1979b). However, they the type of signal provided by the alann system (Ram-
also caution against quickly labeling any particular achandran 1991; Proulx and Sime 1991; Prouls and
action as panic behavior that has only a few of these Fahy 1997). Bells and horns arc often ignored, being
characteristics. considered to indicatc a drill, test, or false alarm. In a
laboratory exercise, Ranlachandran found that only 13%
B E C O M I N G A W A R E O F T H E FIRE
of 96 individuals considered bells to signal an actual
Bryan (2002) discusses several ways that occupants alarm. Similarly, Pauls' survcv of occupants of office
become aware of'a fire. In most cases, the initial cues of buildings indicatcd that only 17% of occupants responded
a fire are ambiguous, involving a different odor, a slight to traditional fire alann signals in high-rise office build-
haze, or strange noises. In some cases occupants may 1r:gs.
observe the flames. In still others, occupants may be The response of people to various types of fire
alerted by an alartii system. alarm signals was observed by Prouls and Sime in drills
Evaluating the rime to become aware o f the tire via at mid-afternoon in a London subway station. Cameras
an audible or visual fire alarm signal actuated by a fire recorded the responses of the individuals, with inter-
detector or sprinkler waterflow switch may involve an views conducted to supplement the video recording. The
analysis of the response time of automatic detection five types of alarms used in thc study are described in
equipment or sprinklers. Several computzr models dis- Table 4.1. Alarms were initiated tivc seconds after a
cussed in Chapter 8 are capable of calculating sprinkler train arrived at the station. It can be seen from Table 4.2
dctection. In contrast. manual detection is ~iiuchmore that pre-movcment ~ i r n ewas as much as nine minutes
Principles of Smoke Management

for an alarm bell only, but the pre-movement time was Evacuation Analysis Using
much less with verbal. announcements. For guidance on Hydraulic Analogy
the use of verbal announcements, see Keating and Lof-
Evacuation analysis using the hydraulic analbgy
tus (1977).
assumes that people follow a directed route of travel to
Given the predominance of ambiguous cues during
their destination, which is typically outside or an area of
the early stages of a fire, building occupants often inves-
refuge. As such, the occupants are assumed to travel
tigate these cues or ignore the initial cues completely,
along a route where the distance to the destination is
thereby delaying initiation of evacuation. Pre-movement
continuously decreasing, neglecting the possibility of
time may also be dependent on the time of day. Proulx
traveling in circles, proceediag in the "wrong" direction,
and Fahy measured the pre-movement time to be up to
and retracing steps, etc. Consequently, an "efficiency"
10 minutes long in a mid-rise apartment fire drill during
factor may be applied to evacuation times estimated
the day. During an early morning high-rise apartment
using this approach to account for the possibility of an
fire, the pre-movement time was 15 minutes for numer-
evacuation process where the occupants may divert
ous occupants and up to five hours for others.
from a directed route.
EVACUATION TIME ANALYSIS Evacuation modeling following the hydraulic anal-
There are three principal approaches for estimating ogy requires information on the people movement char-
the evacuation time for a building: acteristics of velocity, flow rate, and specific flow.
I . ~&&ical correlation of total evacuation time for rate of travel along a corridor, ramp, stak4
Veloci~:
building.
2. Model movement applying hydraulic analogy, simu-
- Flow rate: number of persons passing a particular
segment of the egress system per unit time (for
lating people as fluid particles.
example, persons per unit time passing through a
3. Model movement applying hydraulic analogy, with doorway or over an imaginary line drawn across a
consideration of the behavioral aspects of the people. corridor).
Specificflow: flow rate per unit width of the egress
Empirical Correlations
component (for example, per unit time per unit
The first method consists of correlations that were width through a doorway).
developed from a regression analysis of evacuation data
from 50 fire drills in high-rise office buildings ranging The movement of people has been examined for
from 8 to 15 stories in height. The two correlations travel on stairs (mostly downward travel), in corridors,
developed by Pauls (1980) (one from a linear regression and through doonvays. Virtually all of the information
analysis and the other from a nonlinear regression anal- on people movement has been collected from observa-
ysis) are tions of fire drills or normal movement.

Population per Effective WidM (plm)

and

where . -
T = evacuation time (win);
Equations:
C, = constant, 0.193 (0.08 1 ); - - - (4.1)
- (4.2)
Cz = constant, 0.0394 (0.01 2);
P = population using the stair (p);
W = effective width of stair, ft (m) (see discussion on "0 50 190 150 200 250

efective width later in this section). Population per Effective Width (plft)

The unit of population above is persons, and the Figure 4.1 Estinlared evncmtion time jor- high-rise
symbol used in this chapter for persons is p. The predic- buildings (Pauls 1980).
tions of Equations (4.1) and (4.2) are very close to each
other, as shown in Figure 4.1. Becausc Equation (4.2) is 4. Thc v c l o c i r on stairs rcSers to the rate o f ~ r a v calong
l
the simpler form, i~ is IUOI-ccommonly uscd. ;I di;lgonalp;trh obtaincd by con~wctingthe tips ofthc stairs.
Chapter 4-Evacuation Analysis

Considering that people tend to move faster in v = 0.85k


emergencies than they do in fire drills (Figure 4.2), it
might seem that evacuation time estimates based on fire where
drill data would be conservative. However, this does not v = velocity, @m (mk);
account for the possibility of exit routes being blocked a = constant, 2.86 (0.266);
by smoke or fire. An "efficiency" factor also may be k = velocity factor, fpm (mls); and
applied to account for blocked exits routes.
-D = density of occupant flow, @/m2).
Velocity Equations (4.3) and (4.4) apply to flow on horizon-
The velocity has been shown to be a function of the tal surfaces and on stairs. For horizontal surfaces and the
density of the occupant flow, type of egress component, stair tread and riser types listed in Table 4.3, the velocity
and mobility capabilities of the individual (Gwynne et factors are listed in Table 4.4. On stairs, the distance of
al. 1999; Nelson and MacLennan 2002; Predtechenskii travel is the diagonal of the stair (Figure 4.3), which is
and Milinskii 1978). Nelson and MacLennan propose
correlations of velocity for mobile individuals consider-
ing the available data collected by numerous research- where
ers. LD = diagonal distance of the stairs, ft (m);
For a density greater than 0.05 1 p/ft2 (0.55 p/m2),
Lv = vertical distance of travel, ft (m);
B = angle of the stairs.
For densities less than 0.05 1 p/ft2 (0.55 p/m2), other The dependence of the velocity on density, as pre-
occupants do not interfere with the walking speed of an dicted by Equations (4.3) and (4.4): is presented in Fig-
individual. The maximum walking velocity for level ure 4.2.
walkways and stairways is The velocity correlations prejznted in Equations
(4.3) and (4.4) principally relate to adult, mobile indi-
viduals. Prouls (1995) indicates that the mean velocity
for children and the elderly is on the order of 90 fpm
(0.45 d s ) . The velocity for an "encumbered" adult is in
the range of 45 to 155 fpm (0.22 to 0.79 rnls), which is

Table 4.4:
Velocity Factor, k

-
Egress Component k (fpm) k (mls)
Corridor, aisle, ramp, doorway 275 1.40
Area Density. 6 Riser and Tread Type
7.5110 196 1 .OO
Figure 4.2 Cornpar-ison o f nor-nzal velociq and veloc-
711 1 212 1 .08
ity during emergencies (P,-edtechenskii
6.5112 229 1.16
and Milirukii 1978).
6.511 3 242 1.23

Table 4.3:
Dimensions of Stair Risers and Treads
Riser and Riser, LR Tread, LT Stair
Tread Type in. mm in. mm Angle, 8 Sin, Q
7.5110 7.5 190 10 254 36.9" 0.600
~ r i n c i ~ l e s 'Smoke
of Management

tions because emergencies can happen during unusually


crowded conditions. The number of people expected to
occupy a particular space is dependent on the use oftthe
space. The number of people expected to occupy a space
can be estimated for design purposes based on occupant
load factors, which are included in the U.S. building
codes (ICC 2000; ICBO 1997; BOCA 1999; SBBCI
1999) and the NFPA Life Safety Code (2000). The occu-
pant load factors included in each of the referenced
codes are similar and these occupant load factors repre-
Figure 4.3 Stair geometry. sent average maximum density of occupants. Occupant
load factors from the NFPA Life Safety Code are listed
in Table 4.6.
also appreciably less than the maximum velocity noted
in Equation (4.4).5 Table 4.5 lists mean velocities for Predtechenskii and Milinskii use a definition of
impaired individuals. density based only on areas. rea density &e
ratio of the floor area occupied by each individ~ualper-
Density,., son in the group divided by the tdal floor area occupied
by the gouk(including the area between individuals).
Density is the ratio of the number of people in a
This can be expressed as
group in an egress component divided by the total floor
area occupied by the group (including the area between
individuals). This can be expressed as

where
S = area density (dimensionless);
where
7 7
P = population, p (p); A,, = average area occupied by an individual, ft- (m-).
7 7
A = total floor area occupied by the group, ti- (m-). The average area occupied by an individual
Typical densities of people nlovenlent range from includes the floor area directly under the individual and
0.1 to 0.2 p/ft2 (1.0 to 2.0 p/n~2)(Predtechenskii and the floor space around the individual.
Milinskii 1978; Frantzich 1996; Pauls 2002; Fruin The relationship between these two density tenns is
1987).
The. normal occupant loading may not be consid-
ered an appropriate population for evacuation calcula-
For the areas that people occupy. see Tables 4.7 to
5. An encumbered adult is an individual c a v i n g pack-
ages, luggage, o r a child. 4.9.
Table 4.5:
Mean Velocity for Impaired Individuals (Shields et al. 1996)

- - .. lnipairment Level Walkway Stairs down Stairs up


fpni nits f ~ m nils fpni nils
Electric wheelchair 260 0.89
Manual wheelchair 200 0.69
Crutches 280 0.94 43 0.22 43 0.22
Walking stick 160 0.8 1 63 0.32 67 0.3
Walking fialne 100 0.5 1
Rollator I10 0.6 1
No aid I S0 0.93 65 0.33 SI 0.4 l
No disability 2-10 1.24 140 0.70 I40 0.70
Chapter 4-Evacuation Analysis

Table 4.6:
Occupant Load ~actors'
Occupant
Load ~ a c t o ?
Space Use perslf? pers/m2
Assembly
Less concentrated use without fixed seating 15 net 1.4 net
Concentrated use without fixed seating 7 net 0.65 net
Waiting space 3 net 0.28 net
Library-stack areas l00 gross 9.3 gross
Library-reading areas 50 net 4.6 net
Mercantile
Street floor and sales basement 30 gross 2.8 gross -
Multiple street floors 40 gross 3.7 gross
Other floors 60 gross 5.6 gross
Storage, shipping 300 gross 27.9 gross
Educational
Classroom area 20 net 1.9 net
Shops 50 net 4.6 net
Daycare centers 35 net 3.3 net
Business (offices), industrial l00 gross 9.3 gross
Hotel and apartment 200 gross 18.6 gross
Health care
Sleeping departments 120 gross I l . l gross
In-patient treatment departments 240 gross 22.3 gross
Detention and correctional 120 gross 11.1 gross
l. Data from Table A-S-3.1.l of NFPA l01(2000).
2. The populalion of a space is the product o f [he load factor and the net area or gross area oftha! space as indi-
cated above.

Table 4.7:
Area Occupied b y people1

. Age
10 to 15 15 to30 Crcater than 30
2
ft m= ft' m2 ft2 .,l
Walking Female 1.36 0.126 1.63 0.151
Male 1.3 1 0.122 1.78 0.165
All 1.33 0.124 1.72 0.160 2.08 0.192
Standing All 1.57 0.146 1.87 0.174
,411' 2.00 0.186
I . Data are from Kendik (IYSj).
2. Wih coats
Principles of Smoke Management

Table 4.8:
Area Occupied by People in IP units1
Person Type Horizontal projection2 Shoulder Breadth Body Depth

Adult 1,I-1.4 1.5-1.6 0.92-1.1


Youth
Child
Encumbered ~ d u l t ~ 2.5-8.9 1.6-3.6 1.3-2.6
I . Data are from Predtechenskti and Milinskii (1978).
2. The horizontal projection is dctcrmined by representing the body shape by an ellipse.
3. An encumbered adult is an individual c a v i n g a child, l u a a g e . or packages.

Table 4.9:
Area Occupied by People in SI units1
Person Type Horizontal projection2 Shoulder Breadth Body Depth
m2 m m
Adult 0.10-0.13 0.46-0.50 0.28-0.32
Youth
Child
Encumbered ~ d u l t ~
I. Data are from Predtcclienskii and klilinskii (1978).
2. The horizontal projection is dctenninsd by representing the body shape by an ellipse
3. An encumbsrsd adult is an individual canyin: a child, luggage. or packages.

Densily ( p h i ) Density (plrn')

3 0 0 ° ,
1
I I
2
I I
3
, , :d 1.50 30
0 1 2 3 4
16

- - 25 Stair Riserand Tread Type:


Stair Riser and Tread Type: =
F
- 1.00 ='5 C
$
- 20
12

- 0.75 .$
- a a
Z
15 0.8

- 0.50 3
-
0
-
U
3 10
0
c
0
W
a
m 0.4
50 - - 0.25 5

-0 0 0
0.4 0 0.1 0.2 0.3 0.4
Density ( p l f f ) Density ( P I U )

Figure 4.4 Velocir!.ns a/imction o f densip. Figure 4.5 Spec~$cflowas afilnction o f d e n s i ~ .

Specific Flow F,. = DV = ( 1 -aD)kD (4.10)


The specific tlo\v, F,, is analogous to the mass flux \\.here
in hyd;dillic systems. As such. the specific flow is F, = specific flow, plmin-ft (pls-m).
defined as the product of the velocity and density of the
flow, For densities less than 0.05 1 p/ft2 (0.55 p/m2),

Expr-essions Ibr thc specific flow as a function of The specific flow predicted by Equations (4.10) and
density call o n l y be obtained by for [he (4.1 1) is presented in Figure 4.5. The width referenced
velocity 1.1-0111Equations (4.3) and (4.4). in the units for the specific flow equations relates to the
FOI-a dcnsity yeatcr than 0.05 i p/li2 (0.55 p!m2), ..effective width" as defined by Pauls (2002). The con-
Chapter 4- Evacuation ~ n a l ~ s ~ i s

cept of effective width is based on the observation that Flow


people do not generally occupy the entire width of an Nelson and MacLennan (2002) present a method to
egress component, staying a small distance away from obtain a first order approximation of the egress time in
the walls or edge of the component. Nelson and buildings. The method involves determining the maxi-
MacLennan refer to this small distance as a "boundary mum flow rate for each of the egress components in the
layer," in keeping with the hydraulic analogy for people egress system.
movement. The width of the boundary layer for the vari- For a density greater than 0.051 p/ff2 (0.55 p/m2),
ety of egress components is presented in Table 4.10. The the flow rate for a particular egress component is given
as
boundary layer and effective width are illustrated in Fig-
ure 4.6.

Maximum Specific Flow where


F, = flow rate p/min, @/S);
Considering that Equation (4.10) is a quadratic
function, a maximum specific flow is achieved at a den- w = effective width, ft (m).
sity of For a density less than 0.051 p/ft2 (0.55 p/m2), the
flow rate for a particular egress component is given as

Because a is indenendent of the type of egress com- The maximum flow rate occurs when the specific
ponent, according to this correlation. the specific flow is flow is maximized (i.e., where D,,,, occurs (see Equa-
maximized at the same density for all types of egress
components. Predtechenskii and Milinskii provide data 4

that indicate differences in the density where the spe- L Effective -;


I
l width I
cific flow is inaxinlized for different types of egress / Boundary
components. Layer
Generally, evacuation of a building requires that
building occupants traverse several egress components.
For example, for an individual located in a room on an
upper floor, evacuation involves travel

along aisles or through an open space in the room,


through the room doorway into a corridor,
along a corridor to the stair doon\.ay,
down the stairs, and
through the exterior door to the outside.
/ Boundary
Table 4.10:
Layer
Boundary Layer Width
Component Boundary Lager
in. mn!
Theater chairs, stadium benches 0 0 Water
Railings, handrails' 3.5 S9 'cooler
Obstacles -1 100
Stainvays, doors, archvays 6 l50
Corridor and ramp walls S 200 Effective
Width?
I. \Vllcrc Ilandr;~itsarc present. Nelson and I I x L c ~ l n msug:cst
that the boundxy laycr a-id111i'or Iwndr:~ilsshould he used i i d i c
Iwundary laycr \r id111lirr 11;lndrailsis Ics; [ h ~ n 1 1 1 Ibr
~ 111c
egress C(III~(IOIISIIIwlicrc I ~ I C Iiandrail i s prcrtxl.
Principles of Smoke Management

Table 4.11:
Maximum Specific Flow
Maximum Specific Flow,
Fs,m,
Egress Component plmin-ft pls-m
Corridor, aisle, ramp, doorway 24.0 1.32
Riser and Tread Type
7.5110 17.1 0.94
711 1 18.5 1.01
6.5112 20.0 1.09

For a particular egress system composed of several


components, the maximum flow rate, F,, of each com-
ponent can be determined as FS,,w. The flow time
associated with each component is P/FS,,,,w, where P
is the population passing through the component. The
component with the greatest value of P/Fs,,,,,w is
defined as the controlling element where the constraint
Figure 4.7 Comti-ained flow ir7 evacuotion of a five- is expected.
story building. In many situations, the point of constraint can be
identified easily. For example, consider a stainvell dis-
charging directly to the outside that has doors of only
tion (4.12)). Maximum specific flow, F,,,,,,w, for a vari- 0r.e width (see Figure 4.7). For staiiwell and door
ety of egress components is provided in Table 4.1 1. Tlie widths designed to comply with the Life Safety Code or
controlling egress coniponent is tlie component with tlie model building codes in the US., the minirnuni flow
smallest maximum flow rate, relating to where a queue will be associated with the doorways. Tlie entire popula-
is expected to form if D,,,, occurs ill an upstream com- tion using this stainvell would have to pass through this
ponent. exterior door. Because the interior stainvell doors on the
upper floors would only serve a fraction of this popula-
EVACUATION TIME tion, they would be less congested. The exception is the
stair that is used for evacuation of only one floor, and
Constrained Flow Approach this stair would have constraints at both stairwell doors,
The constrained flow approach is based on the provided that both doors are of the same width. In such a
assumption that there is a point along the egress system case, the evacuation analysis could be conducted witli
where a queue forms. Tlie evacuation flow envisioned the constraint at either location.
when applying this type ofevacuation model is depicted
When the exterior stairwell daor is the constraint in
in Figure 4.7 where the egress system is funneled into a
particular point, such as an exterior doorway, before the tlie egress system, the modeled evacuation time
evacuees depart from the building or affected area. becomes
Assuming that all occupants start their evacuation
simultaneously at time zero, the niodeled evacuation
time using the constrained flow approach can be esti-
mated as Example 4.1 illustrates tlie constrained flow
approach. This example is appropriate for situations
where a queue is expected to form at the exterior stair-
well door. Generally this happens when an appreciable
where number of people occupy tlie area of the building being
I,,, = nodel led evacuation time for an egress route. niodeled. Conversely, in buildings with low occupant
f, = time for first person to arrive at constraint, loads, a queue is unlikely. In cases with low occupant
loads, a more complex analysis is needed to examine the
I, = time for population to pass through constraint.
occupant flow on a component-by-co~iiponent basis.
I, = time for first person to travel fi-om constraint and These analyses also may be applied to provide a more
proceed to outside (or area of'reli~ge). accurate assessment in cases whcre queuing is likely.
Chapter 4- Evacuation Analysis

Example 4.1 Evacuation Time


Determine the evacuation time for a five-story building with the following characteristics (see Figure 4.8):

There are 200 people on each floor. Each floor is served by two 44 in. (1.12 m) wide stairways. The doors leading into and from
the stairway are 32 in. wide (0.81 m). The stair riser and tread type was 7/11. The floor-to-floor distance is 12 ft (3.7 m) and the
landing behveen floors is 4 X 8 ft (1.22 X 2.44 m). Handrails are provided on both sides of the stairways.

Solution:
Component Effective Width Specific Flow Flow Rate
ft (m) plft-rnin (plm-S) ptmin @/S)
Door into stairway 1.67 (0.51) 24.0 (1.32) 40 (0.67)
Stairway 3.08 (0.94) 18.5 (1.01) 57 (0.95)
Landing 2.67 (0.82) 24.0 (1.32) 65 (1.08)
Door from stainvay 1.67 (0.5 1) 24.0 (1.32) 40 (0.67)
Time for population to move out of exterior stair door:
The controlling component is selected as the door leading from the stairway The time required for the half of the buildmg occu-
pants on the upper floors (400 persons) to pass through this doorway is estimated to be 400140 = 10 minutes.
Time to travel down one flight of stairs:
The time required for the first person traveling at a velocity associated with the maximum density is given by the time ro travel
do\vn one flight of stairs and two landings.
The vertlcal distance of the stairs is 12 ft (3.7 m). From Table 4.3, sin 0 is 0.537 for 711 1 stairs.
From Equation (4 S), the diagonal distance along the stairs is LD = Lr,/sin0 = 12/0.537 = 22.3 ft (6.8 m).
2
The density on the stairs is taken at D,,,,. From Equation (4.12), D,, = 0.175 p!ft (1.88 plm2).
From Table 4.4, k is 21 2 fpm ( l .08 rnls).
.
From Equation (4.3), v = X-- akD = 212 - 2.86(212)(0.175) = 106 fpm (0.539 1~1s)
The length of travel along each of two landings is 8 ft (2.4 m) (assuming an average length oftravel on the middle of the land-
ing). Because thc velocity on a stairway is less than that for a horizontal component, such as a landing, the velocity on the land-
ing is limited to that achieved on the stainvay. As such, the length of travel on the landing can be added to that for the stairway,
giving a total length of travel of 38.3 ft (11.7 m). The time required to traverse this distance at the velocity achieved on the stair-
ways is 38.31106 = 0.36 min (22 s). This is roundsd up to 0.4 min (24 s).
Total evacuation time:
The total evacuation rime is 10 + 0.4 = 10.4 min (624 S).
Principles of Smoke Management

(a) Elevation View

Figure 4.9 Merging egress,flows.

I I
If the conibined flow rate of egress components
' '(b) Plan Mew leading to the intersection is greater than the capacity of
the f l o n rate for the egress component leading from the
Figure 423 Diagram o f building for- Esati7ple 4.1. intersection, a queue is expected to form. If a queue
forms, the analysis can continue, considering that the
flow rate in component #3 is equal to the maximunl
Component-by-Component Analysis capacity of the component.
Questions are often asked concerning the composi-
The component-by-component analysis involves a
tion of the queue relative to the incoming flows (i.e.,
determination of the time for the population to traverse does any one group have a "right-of-way" while most or
each egress component. In this case, the density of the all of the other group stops). The total evacuation time
flow along each egress component must be determined of the building is not dependent on which group has the
so that the velocity and floiv rate can be determined. right-of-way. Alternatively, if the intent of the analysis
The starting point of such an analysis is to assume is to determine the time required to clear a particular
an initial density of the population. If such a calculation floor level and the merger is nith people from another
is to be done using algebraic equations (instead of one floor level, then the right-of-nay decision will impact
of the computer nlodels described in the last section of the results. Unfortunately, there is no technical support
this chapter), a reasonable assumption is to consider all for establishing any rules co~lcerningthe right-of-way or
building occupants on a particular floor to be uniformly proportion of the tlorvs from the entering streams that
distributed in the corridors. As the population starts to Gccurs at a merger. Ho\ve\,er. given the observation
move, the density of the people may change as a result from human behavior studies that people tend to react
of t'lree types of transitions: altruistically, it is reasonable to expect that people trav-
eling from other floor levels nould yield to people leav-
mergers o f flows at corridor intersections or where ing the fire floor.
people entering a stair merge with people traveling Where the \vidth o f the egress component changes,
in the stairs from other floors, as indicated in Figure 4. IOa and 4.10b, the density of the
changes in the widtl; of the egress component, flow also changes. The flow rate of people entering the
egress component equals that leaving it:
changes in specific flan,, resulting in a transition
from one type of egress component to another, e.g.,
a corridor to a stair.
For converging flow. as illustrated in Figure 4.IOb,
The new density after a transition may be deter- -
a queue might be espected to form at the transition.
mined by applying one of the following principles. When there is a queue, the flow downstream from the
The combined flow rat< of people entering an inter- transition is equal 1.0 the ~llasi~nurn
capacity of the com-
section equals the flo!!, rate of people from the intel-sec- ponent.
tion (see Figure 4.9). When a queue forms \\.it11converging flow of Figure
4. lob, the density ofa tlow ofoccupants proceeding away
fiom a transition isdetermined by solvingeither Equation
Chapter 4-Evacuation Analysis

(4.13) or (4.14). Where Equation (4.13) applies, solution assumption needs to be made of the distribution of occu-
of the quadratic equation results in two possible solutions pants among the available paths. Often; an equal propor-
for the density. The lesser value for density should be tion of the group is assumed in each of the available
selected as the correct value. The lower density is correct paths. Alternatively, the distribution may be determined
for reasons indicated in the following example. If an in propxtion to the respective capacities or other char-
occupant flow at the maximum density was approaching acteristics of the available paths (Predtechenskii and
a widening comdor (Figure 4.10a), the solution of Equa- Milinskii 1978; Murosaki et ai. 1986).
tion (4.13) would yield one density greater than the max-
imum and one less. However, in the case of a widening The following model can be applied if the order of
corridor, it's unreasonable to expect the density to evacuation is arbitrarily determined to proceed from
increase (and velocity to decrease) when proceeding from highest floor to lowest floor. At time zero, all people
the narrow to the wide corridor. move to the stairs on all floors and travel to the next
In either of these types of analyses where multiple floor level. If the stairwell capacity is exceeded as a
egress paths are available to a group of occupants, some result of the merger, then the maximum flow proceeds in
the stairwell with the right-of-way given to the occu-
pants on upper levels. (The total evacuation time is inde-
pendent of whether people from upper floors have or
surrender the right-of-way.) Consequently, the merged
flow in the stairwell is composed predominantly of peo-
ple from the upper level, supplemented by additional
(a) Diverging Flow people from the next floor to provide the maximum per-
mitted flow rate for the stairwell. Occupants on all other
floor levels stop their movement into the stair as a result
of the stairwell having achieved maximum capacity
Once the last occupant from the upper floor reaches the
Icvel below the upper floor, the flow from this next floor
is increased to its maximum value.
(b) Converging Flow
The component-bycomponent approach is illus-
Figure 4.10 T,-ansiliorzin egress componcrit. trated in Example 4.2.
Principles of Smoke Management

Example 4.2 Evacuation Time


Determine the evacuation time for the same five-story building as in Example 4.1 (see Figure 4.8):

Solution:
Assume that all occupants initiate movement simultaneouslyand half of the building occupants are located in the corridor at a distance
of at least 100 R (15.2 m) fiom the stair door. Other occupants are in the spaces adjacent to the corridor and are assumed to join the peo-
ple in the corridor promptly upon notificaiion. Assume an equal number of occupants use the two stairs.

The density of the people in the corridor is 0.125 (1.35 p/m2).


Given this density, the specific flow of the people in the corridor is 22 p/ft-min (1.20 p/m-S) < F,.
The velocity in the corridor is 177 @m(0.90 &S).
The flow rate in the corridor is 58.7 p/min (0.98 p/s).
Time to reach stainvay is 100/177= 0.56 rnin (339 S).
The maximum specific flow of the door leading into the stainvay is 40 p/min (see example 4.1) (0.67 PIS).

As such, a queue forms at the doorwa~,given that the flou. in the corridor toward the door is 58.7 p/min (0.98 p/s). The queue builds at
a rate of 18.7 p/min (0.3 1 p/s).

Given flow of 40 p/min (0.67 p/s) in stairway, density is 0.099 (1.07 p/m2).
The vel&ity in the stair for lea\-ingthe fifth floor approaching the fourth floor is 152 @m (0.77 m/s).
Time to cave1 38.4 ft (11.7 m) to reach fourth floor is 0.25 min ( l 5 S).
-
-,

At this point, flows from the fourth and fifth floors merge at the landing of the fourth floor, as well as every other floor level.
The total time required for the last person from the fifth floor ro enter the stair at that floor level is 2.79 rnin (167 S).
The time required for the last person from the fifth floor to reach the 4th floor is 3.04 rnin (182 S).

With a flow proceeding down the stain From the fifth floor of 40 p/min (0.67 p/s) and 40 p/min (0.67 p/s) entering the stairway ffom
the 4th floor, the outflow from the point of merger would be 80 p'niin (1.33 p/s) if no queue occurs. However, since the flow capacity
in the stainvay is 57 p/min (0.95 p/s), the flo\v in the stain\-ay \\-ill be limited to 57 p/min (0.95 p/s). Priority of flow in the stairway is
given to occupants from the top floor Ie\.el.

Thus, prior to the queue fonning in the stainvay (i.e., 031 rnin [19 S]),32 people exited from the second, third, and fourth floors.
Because the flow capacity in the stain\a!. is limited to 40 plmin (0.67 p/s), the flow ffom all lower floors is stopped. Once the last per-
son from the fifthtloor reaches the founh floor. the flow of the GS remaining people from the fourth floor recommences.

The time required for the last person from the fourth floor to enter the stair at that floor level is 4.74 rnin (284 S).
The time required for the last person from the fourth floor to reach the third floor is 4.99 rnin (299 S).

I Similarly:

I The time required for the last person frotn the third floor to enter the stair at thzt floor lcvel is 6.69 min (40! S).
The time required for the last person from the third floor to reach the second floor is 6.94 rnin (4 16 S).
The time required for the last person from the second floor to enter the stair at that floor level is 8.64 rnin (5 18 S).
The time required for the last person from the second floor to reach the first floor is 8.89 rnin (533 S).

COMPUTER-BASED paths (where choices are available). The flow distribu-


EVACUATION MODELS tion between multiple paths may be determined by
The lbllowing three types of e\ acuation models are -
occupant behavior considerations. Optimization models
available: minimize the evacuation time by considering an optimal
distribution of occupants among multiple flow paths.
Sh~lation
The current optimization models neglect behavioral
Optimization
considerations. The risk assessment models quantify the
Risk assessment
risk posed to building occupants by conducting a fire
Si~nulatio~,
modcls predict 1n0\sment and bella\.ior Ilazard analysis, combined with an elementary evacua-
of occupants by assessing the t l o ~disrribution among tion analysis. The risk assessment models need to be
Chapter 4- Evacuation Analysis

applied numerous times to address the probability of


various scenarios and their outcomes.
The characteristics of existing evacuation models
are described in a review by Gwynne and Galea (1999).
A summary of the chakcteristics of the evacuation
models is indicated in Figure 4.11.
Building spaces may be represented as coarse or
fine networks. A coarse network usually uses a single
node to represent each space. Additional nodes are used
only for large rooms or rooms that have connections to
several other rooms. In the coarse network approach,
rooms (or nodes) are connected by arcs. Coarse net- Figure 4.11 Evacuation models (adapted from
works assume unifonn conditions on each node and a Gwynne and Galea 1999).
constant traversal time along arcs. Alternatively, fine
networks divide each room into several small sections.
"patience" and motivation factors) be entered. Justifica-
In some cases, a small grid is created over the entire
building space where the size of a particular area may be tion of such input is subject to much debate. Most of the
as small as the area occupied by an individual. models assume that occupants only become engaged in
Theevacuation models assess movement of the evacuation behavior. Neglecting the variety of nonevac-
building occupants by two perspectives. A global per- uation behavior that occurs will result in a smaller evac-
spective tracks the occupants anonyn~ously. In this uation time, perhaps substantially, than if such behavior
approach, the iiiodel does not distinguish which individ- is considered. None of the models currently considers
ual leaves the room or building at a particular time. The the possibility of two-way flow in a corridor, either as
global perspective models assume uniform characteris- the result of emergency personnel or some building
tics for the entire building population. In contrast, mod- occupants moving opposite to the evacuating occupants.
els with the individual perspective track each person, As a prerequisite to any evacuation analysis, the
identifying where any particular person is during the
number of people in the building must be established.
evacuation period. The models with the individual per-
The location of the occupants also needs to be specified,
spective consider individual traits (e.g., mental and
though at varying levels of detail, depending on the
physical capabilities, tolerance to smoke, and group
model. Location of individual occupants can be
interactions).
"placed" at a specific point for applications involving
Behavioral characteristics included in the models
may be done by several methods (e.g., deterministic fine network niodels. For the coarse network models,
equations [functional analogy], pre-established behav- people only need to be located in a room or floor of a
ioral patterns, and iflthen rules, \vhich may or may not building. When using a first-order approach with hand
be subject to change by the user). calculations, the calculations become very tedious when
One principal area of concern with the evacuation placing people in individual rooms. As such, for first-
models relates to the reliability of input parameters. order estimates, people may be placed in a queue at the
People movement characteristics need to be provided. esit door from the floor or large section of the floor to
Where a constant velocity is required, the results of the simplify the calculations. The loss of accuracy with this
analysis will be dependent on whether the mean or max- assun~ptionrelates to the time for people to travel from
imum velocity is included. Some of the niodels require their respective starting points to fonn a queue at the
personal characteristics of building occupants (e.g., as door. In many buildings, this time is relatively short.
CHAPTER 5

Effective Areas and Smoke Movement

n building fires, smoke often migrates to locations The total flow, vT,from the space is the sum of the
I remat; from the fire space. Stairwells and elevator
shafts can become smoke-logged, thereby blocking
evacuation and inhibiting fire fighting. In this chapter,
flows through the leakage paths:

several of the driving forces of smoke movement are


discussed, methods of determining the neutral plane Ere
The effective area, A,, for this situation is that
provided, and some general
- comments are made con-
-
ceming smoke movement. The information in this chap-
ter is also applicable to the migration of other airborne
which results in the total flow, vT.Therefore, the total
m m
flow can be expressed as
matter, such as hazardous gases, bacteriolog~calmatter,
or radloactlve matter in laboratories, hospitals, or indug-
trial facilities. However, the discussion in this chapter is
pr~marilyaimed at smoke movement. The concept of_
ettectlve flow areas is quite usehl for analysis of smoke
movement and of smoke control systems, and this topic
1s addressed next.

EFFECTIVE FLOW AREAS


The paths in a system can be in parallel with one
another, in series, or in a combination of parallel and
series paths. The effective area of a system of flow areas
is the area' that results in t h e s a m d o w as the system
when it is shbjected to the same pressure difference over
the total sistem of flow paths. This is analogous to the
flow of electric current through a system of electrical
resistances. The following analysis is for the same flow
coefficients for each flow path and for constant air tem-
perature. Variations in flow coefficients and temperature
7

are addressed later.

Parallel Paths
Three parallel leakage areas from a pressurized
space are illustrated in Figure 5.1. The pressure differ-
ence, Ap, is the same across each of the leaka,oe areas. Figure 5.1 Flowpafhs in parallel.
Chapter 5-Effective Areas and Smoke Movement

where

vT = volumetric flow rate through the path, c h (m3/s);

m = mass flow rate through the path, Ibis (kgk);


C = dimensionless flow coefficient;
A, = effective flow area (or leakage area), ft2 (m2);
Ap = pressure difference across path, in. H 2 0 (Pa);

p = density gas in path, lb/@ (kg/m3);


K, = 776. (1.00).
The flow PI through area A , can be expressed as

Figure 5.2 Flow paths is series.


The flows V* and v3 can be expressed in a similar
manner. Substituting the expressions for PI, V * , and The effective area for flow paths in series is the
V3 into Equation (5.1) and collecting like terms yields
flow area that results in the flow V for a total pressure
difference of Apr. Therefore, the flow V can be
expressed as

Compari~~g
this with Equation (5.2) yields

Solving Equation (5.8) for ApT yields


The above logic can be extended to any number of
flow paths, and it can be stated that the effective area of
17 individual leakage paths in parallel is the sum of the

individual flow areas.


The pressure difference across A , can be expressed
as

In Figure S. l. if A I is 1.08 R* (0.10 ni') and A? and A3 are both The pressure differences Ap2 and Ap3 can also be
0.54 ft' (0.05 m*),what is the effective flow area ofthe system?
expressed in a similar manner. Substituting Equation
I Fro111Equation (SS), A , = 2.16 R' (0.20 m'). (5.9) and the expressions for ApI, Apz,and into Equa-
tion (5.7) yields anexpression for the effective flow area.
Series Paths
Three leakage paths in series from a pressurized (5.11)
AJ.
space are illustrated in Figure 5.2. The flow rate. l', is
the same through each of the leakage areas. -

The total pressure difference, Ap7, from the pressur- This same reasoning can be extended to any num-
ized space to the outside is the sum of the pressure dif- ber of leakage areas in series to yield
ferences Ap ,. Ap?, and Apj across each of the respective
flow areas. .-l ,,A?, and Aj:
Principles pf Smoke Management

where n is the number of leakage areas, Ai,in series. In These two effective flow areas are in series with A , .
smoke control analysis, there are tiequently only two paths Therefore, the effective area of the system is given by
in series, and the effective flow area for this case is

Example 5.2 Two Equal Series Paths following flow areas: AI =A2 = A3 = 0.22 9 (0.02 m2) and A4
1 Calculate the effective leakage area of two paths of 0.22
(0.02 m2) in series.
ftL = A ~ = A ~ = o 9(0.01
.II m2).

From the equations above, A23, = 0.44 ft2 (0.04 m2), A4& =
For two equal flow areas (A = AI = A?), Equation (5.13) 0.33 9 (0.03 m2), and A, = 0.17 9 ( 0 316 m2).
becomes A, = 0.707, A = 0.707 (0.22) = 0.156 9(0.0145 m').
Effects of Temperatures and
Example 5.3 Two Unequal Series Paths Flow Coefficients
Calculate the effective flow area of two paths in series, where For most calculations involved in smoke control,
11 the flow areas are
II the a&umptions o f constant temperature and unifomi
flow coefficient are appropriate, but it may be desired in
I AI = 0.100 ft2 (0.00929 m') and A2 = l .OO f;(0.0929 m2).
..- I/ some cases to consider the effects of these parameters.
From Equation (5.13), A, = 0.0995 (0.00924 m2).

This example illustrates that, when two areas are in series and
I For parallel and series flow paths, the equations for
effective flow area are

one is much larger than the other, the effective area is approsi-
mately equal to the smaller area.

P- P

Example 5.4 Effective Flow Area of Four Series P:tths for parallel patlis and
Calculate the effective flow area of the folln\\ Ing areas that are

From Equation (5.13), A, = 0.0704 liZ(0.00651 m2). for series patlis where

Combination of Paths in Parallel and Series

The method of developing an effective area for a


system of both parallel and series paths is to combine,
systematkally, groups of parallel paths and series paths.
The system illustrated in Figure 5.3 i s analyzed as an
..
l example: '

This figure shows that A2 and A3 arc parallel; there-


fore, their effective area is

Areas A+ As, and A6 are also in parallel, so their


effective area is
Figure 5.3 Cornbina/ion qf/low p / h s in parallel a d
series.
Chapter 5-Effective Areas and Smoke Movement

A, = effective flow area o f system, fl? (m2); W: Arrrrws indicate direc(rmof air movement

T, = absolute temperature in effective flow path, "R Q;

C, = flow coefficient for effective path, dimensionless;


= absolute temperature in path i, "R (K);

Ai = flow area of path i, fl? (m2);

Ci = flow coefficient of path i, dimensionless.


For the case of two areas in series with the same
- . ,. ,.; ,..' . ,.:,;.< . 7 ,. .< ,?,; ,#, :,,;c y,f..?.,9.,>;$...'.;,9
flow coefficients, the effective area is . ,. ,,

Normal S@& Effed Reverse Stack Effect

Figure 5.4 Air movement due to no/-nlaland reverse


slack effect.

Stack Effect
1. What'is the effective area of nvo paths in series, both of 0.22
ft? (0.02 m
) area with one at 70°F (21°C) and the other at
' Frequently, when it is cold outside, there is an
100°F (3S°C)? Use c of 70°F (21°C). upward movement of air within building shafts, such as
stainvells, elevator shafts, du~nbwaitersshafts, rnechani-
cal shafts, and mail chutes. Air in the building has a
buoyant force because it is warmer and therefore less
dense than outside air. The buoyant force causes air to

I From Equation (5.19),A, = 0.153 9 (0.0 142 m').

With both temperatures the same, the effective area of this sys-
rise within building shafts. This phenomenon is callzd
by various names, such as stack effect, stack action. and
chimney effect. These names come from the colnparlson
tem is 0.156 ft2 (0.0145 m"). as calculated in Example 5.2. with the upward flow of gases in a smoke stack or chim-
Considering the degree of uncertainty associated wit11 flow ney. However. a downward flow of air can occur in air-
areas; adjustment of the effectiie flow area is unnecessary. conditioned buildings when it is hot outside. For this
manual, the upward flow will be called normal stack
2. What is the effectix area above if the elevated temperature effect and the downward flow will be called re\srse
is 1000°F (538"C)?
stack effect as illustrated in Figure 5.4.
Most building shafts have relatively large cross-
sectional areas and, for most flows typical of those
induced by stack effect, the friction losses are negligible

I From Equation (5.19). A , = 0.1 I1 @ (0.0105 m').


in comparison with pressure differences due to buoy-
ancy. Accordingly, this analysis is for negligible shaft
friction. but shafi friction is specifically addressed later.
DRIVING FORCES OF Pressure within a shaft is due to fluid static forces and
SMOKE MOVEMENT can be espressed as

The driving forces of smoke movement include nat-


urally occurring stack effect, buoyancy of combustion
where
gases, expansion of combustion gases, the wind effect,
= air pressure inside the shall,
fan-powered ventilation systems, and elevator piston
effect. This section discusses these driving fcrces and, in g = acceleration of gravity,
particular, addresses smoke movement due to the stack z = elevation.
effect process, either naturally occurring or that of com- = gas density inside the shafi.
bustion gases. Generally, each driving force is discussed For the ele\.ations relltvant to buildings, the accslsr-
here as acting alone in order to facilitate discussion and ation of gravity can bc considered constant. For constant
lead to an understanding of smokc transport. density. Equation (5.70)can hc integrated to yield
..
Principles of Smoke Management

Appendix A. Substituting Equation (5.24) into Equation


(5.23) and rearranging results in the following equation.
where p, is the pressure at z = 0. To simplifL the analysis,
the vertical coordinate system was selected such that p, =
p, at z = 0. In the absence of wind effects, the outside pres-
sure,po, is
where
To = absolute temperature of outside air,

where p, is the density outside the shaft Pressures inside T, = absolute temperature of air inside the shaft.
the shaft and outside the building are graphically illustrated Equation (5.25) was developed for a shaft con-
in Figure 5.5 for normal stack effect. This figure also nected to the outside. The neutral plane is a horizontal
shows the pressure of the building spaces, and methods of plane located at z = 0, where the pressure inside equals
calculating this are presented later in this section. The pres- that outside as stated above. If the location of the neutral
sure difference: 4,:from the inside to the outside is plane is known, this equation can be used to determine
expressed as the pressure difference from the inside to the outside
regardless of variations in building leakage or the pres-
ence of other shafts. Methods of determining the loca-
Because i~ariationsin pressure within a building are tion of the neutral plane are discussed later. Tables 5.1
very small compared to atmospheric pressure, atmo- and 5.2 are comparisons of pressure differences due to
spheric pressure, p,,,,,, can be used in calculating gas various driving forces. For standard atmospheric pres-
density from the ideal gas law. sure of air, Equation (5.25) becomes

where where
Aps0 = pressure difference from shaft to outside, in. H 2 0
p = air density.
p,,,,, = absolute atmospheric pressure, (Pal;
To = absolute temperature of outside air, "R (K);
R = gas constant of air,
T, = absolute temperature of air inside shaft, "R (K);
T = absolute temperature of air.
Values for the gas constant and of standard atmo- h = distance above neutral plane, ft (m);
spheric pressure for several systems of units are given in K, = 7.64 (3460).

Building Pressure.p,

Pressure

Figure 5.5 Pirsswcs nt~dpresszoadiJZwlices dut-ing normal stack efecf


Chapter 5-Effective Areas and Smoke Movement

Table 5.1: 7
Comparison of Pressure Differences Due to Various Driving Forces (IP Units)

Driving Force Location of'Ap CoriciiCions (in. H20)


Stack effect, Shaft to outside For all stack effect examples, T,= 70 "F and To= 0 "F:
Equation (5.26)
h=30ft 0.07
h=30Oft 0.7
Buoyancy of combustion Fire room to adjacent . For Tf= 1600 "F and To= 70 "F:
gases, room at ceiling
h = 5 ft 0.05
Equation (5.3 l)
h = 10 ft 0.1 1
Wind effect, Across building
(windward to leeward
For all wind examples, p = 0.75 lblf?,
-
c,, = 0.8, and
Equation (5.34)
wall) C,,z = 4 . 3 :
U H = 15 mph 0.12
UH= 30 mph 0.48
Ventilation systems Across barrier of Note: Values based on experience. 0.05 to 0.35
smoke control system .
Elevator piston effect, Elevator lobby to For all the examples of the upper limit of pressure differ-
building ence due to elevator car motion, p = 0.75 lb/ft3, A,, = 1.60
Equations (5.41) to (5.43)
ft', A, = 0.42 ft', A,; = 0.54 ft':

For a single-car shaft with C, = 0.83, A, = 60.4 ft', and A,


= 19.4 ft2:

U = 700 fprn
For a double-car shaft with C, = 0.91, A, = 120.8 ft2, and

U = 700 fpm 0.05


Principles of Smoke Management

Table 5.2:
Comparison of Pressure Differences Due to Various Driving Forces (S1 Units)
Driving Force Location of Ap Conditions &(Pal
Stack effect, Shaft to outside For all stack effect examples, T,= 21 "C and T, = -18 "C:
Equation (5.26)

Buoyancy of combus- Fire room to adjacent For Tf=870 OC and To = 21 OC:


tion gases, room at ceiling
h = 1.5m
Equation (5.3 1)
h=3m
Wind effect, the
(windward to leeward
For all wind examples, p = 1.20 kg/m3,
-
c,, = 0.8 and
Equation (5.34) Clp2= - 0.3:
wall)

U H =14 m/s 130


Ventilation systems Across barrier of Note: Values based on experience. 12 to90
smoke control system
Elevator piston effect, Elevator lobby to For all the examples of the upper limit of pressure difference
Equations (5.41) to (5.43)
building ' due to elevator car motion, p = 1.20 kg/m3, A, = 0.149 m2,

For a single-car shaft with Cc = 0.83, A, = 5.61 m2, and A, =


2
1.80 m :
U = 2.03 m/s
U = 3.56 m/s

For a double-car shaft with Cc= 0.94, A, = 11.22 m2, and A, =


2
7.41 m :
Chapter 5-Effective Areas and Smoke ~ o v e m & t

Example 5.7 - Stack Effect in a Tall Building


The neutral plane is located at mid-height of a 600 ft (185 m)
tall building with inside and outside-temperatures of 70°F
(21°C) and 0°F (-18°C). What is the pressure difference at the
top of the building?

Because of the neutral plane location, h = 300 ft (91.4 m).


Using Equation (5.26), the pressure difference due.10 stack
effect is 0.66 in. H20 (164 Pa) from the top of the shaft to the
outside.

Note: Figure 5.6 can also be used for this calculation. In using
this figure, the term Apso l h is positive for normal stack effect
and it is negative for reverse stack effect.
Outside Temperature. To?F)
For the building illustrated in Figure 5.5, all of the -ro -30 - 20 -10 o 10 10 so 4) 50
vertical airflow is in the shaft. Of course, the floors of Outside Temperature. To('C)
-real buildings have some leakage and there is some air-
flow through these floors. The discussion of stack effect Figure 5.6 Graph of pressure difference due to stack
to this point has been general and it applies to buildings effect.
with o r without leakage through floors. To analyze the
pressure differences on building floors, an idealized stituting Equation (5.27) into this relation and rearrang-
building model is used that has no leakage between
ing, the effective area is eliminated.
floors. For nonnal buildings, airflow through floors is
much smaller than that through shafts. The following
analysis develops some useful equations based on this
zeroji'oor- leakage idealizatiot~.
For the system of flow paths illustrated in Figure
5.5, the effective flow area per floor is In general, the ratio A,;/A,, varies from about 1.7 to
7. The pressure differences from a shaft to the building
space are much less than those from the shaft to the out-
side, as can be seen from the examples listed in Tables
5.1 and 5.2. In the event that many windows on the fire
where
floor break due to the fire, the value of A,, becomes very
A, = effective leakage area between the shaft and the out- larse on the fire floor. When this happens, the ratio A;;/
side, fi2 (ni2); Aio becomes very small, and Q,; approaches
A,; = per floor leakage area between the shaft and the Thus, when a large number of windows break on the fire
floor, the pressure from the shaft to the building is
building, ft2(m2);
almost the same as that from the shaft to the outside.
A;, = per floor leakage area between the building and the
The development of Equation (5.28) considered the
outside, ft' (m').
pressure difference uniform with height at each floor,
The mass flow rate, ril , for a floor can be expressed
which introduces an error-the maximum value of
by the orifice equation as C A , ( ~ ~ A , L I ~ , ) where ' / ' , C is a
which can be calculated by Equation (5.26) for a value
dimensionless flow coeficient that is generally in the
range of 0.6 to 0.7. For paths in series, the pressure dif- of /7 equal to the distance between floors. In the exam-
ference across one path equals the pressure difference ples of Tables 5.1 and 5.2, if the floors were l0 ft (3. I
across the system times the square of the ratio of the m) apart, the maximum error of Equation (5.28) is about
effective area of the system to the flow area of the path 0.01 in. H 2 0 (2.5 Pa). In general, this error is not signif-
in question. Thus, the pressure difference from the shaft icant. Equation (5.2s) can be rcwritkn for the pressure,
to the building space is A / J , =~ ~A p , , , ( A , / . - l , i ) - . By sub- p,. at the building space.
Principles of Smoke Management,

In unusually tight buildings with exterior stairwells,


reverse stack effect has been observed even with low
outside air temperatures (Klote 1980). In this situation,
the exterior stairwell temperature was considerably
The series flow approach to determining building lower than the building temperature. The stairwell was
pressures described above can be used for buildings the cold column of air and the other shafts within the
with multiple shafts if all the shafts are at the same pres- building were the warm columns of air.
sures and if all the shafts have the same starting and Smoke movement from a building fire can be domi-
ending elevations. nated by stack effect. During normal stack effect (Figure
Pressure measurements on several buildings 5.4), smoke from a fire below the neutral plane moves
(Tamura and Wilson 1966, 1967a, 1967b) verify the with the building airflow into shafts and up the shafts.
stack effect theory presented above for conditions This upward smoke flow is enhanced by anv buovancy
encountered in the field. Further, these studies show that - above
forces on the smoke due to its temperature. Once
the neutral plane, the smoke flows out of the shafbinto
the zero floor leakage idealization is generally appropri-
theupper floors of the building, as illustrated in Figure
ate for determining pressure differences on building
floors due to stack effect. Additionally, Igmura and -5.9b. As discussed in Chapter 1, this kind of smoke flow
can have fatal conseauences. as in the fires at the MGM
Klote (1988) have conducted full-scale stack e;Fect
Grand and other buildings. Leakage between floors
U U -..--
experiments at the Canadian ten-story Fire Research results in smoke flow to the floor above the fire floo~.If
Tower near Ottawa, which verified the stack effect the- lkakage between floors is negligible, the floors below
ory for a'iange of temperatures and of leakage condi- -the neutral plane+xcept for the fire floor-will be
tions they considered r-presentative of most buildings. &sentially smoke-free. For significant leakase b e t w e e ~
Figure 5.7 shows comparisons of measured and calcu- flls, s x e
lated pressure differences due to stack effect for outside floor will be much greater than to other floors below the
temperatures of 12°F (-I 1"C), 27°F (-3°C). and 45°F kutral plane, as is shown in Figure 5.9h
(7°C). Figure 5.8 shows comparisons of measured and For a fire above the neutral ,lane. the buildinn air-
> 2
'

calculated pressure differences for ratios A,, /A,, of 1.7. tlo\vs due to normal stack effect tend to restrict the
2.4, and 7. Further, this stack effect theory provides a extent of smoke flow. Airflow from the shafts to the fire
useful approximation for buildings in which all of the floor can prevent smoke infLLtdQnafihasahafts ( ~ i g -
shafts do not have the same starting and ending ele\a- ure 5.9c), but leakage between floors can result in some
tions. smoke movement. If the buoyancy forces of the hot

Pressure Difference (in H,o)


-.l2 -.cd 0 .cd .08 .l2
28

24

20

16 g
s
.-m
12 q
8
Note: Solid lines are
calculated valuas.

0
-30 -20 -10 0 10 20 30
Pressure Difference (Pa)
Chapter 5 --Effective Areas and Smoke Movement

Pressure Dierence (in Y O )

Inside Temperature
72 OF(22 DC)
Outside Temperature
27 O F (-3%)
7

Neutml Plane

, Note: Solid lines are


calculated values.

I I I 1
-15 -10 -5 0 5 10 1s
Pressure Difference(Pa)

Figure 5.8 Presszwe differences across outside wall of the


Canadiarl Reseal-c11 Tower for different bltildjtig
leakages ladapiedflmn Tamura and Klote [l 9881).

Figure 5.9 Air- atid smoke movement it? a high-rise brrildiug h e to slack e#ec/: (U) oit:j'Io~i.due /a s/ack effect. (0)
jir-Ebelow the ne~rtralpla~ie, (c)fire above /he neutt-a1plane. a d (d).fir-e above 111e~ieirlt-c11
plam 11i1h
smoke entering a sliafl due to b u o p q ~ .
Principles of Smoke Management

Figure 5.10 Pressure during afully involved compartment fire.

smoke overcome the stack effect forces at the shafts on perature. For standard atmospheric pressure, the above
b e fire floor, smoke can infiltrate the shafts and flow to relation becomes
upper floors (Figure 5.9d).
The air currents of reverse stack effect (Figure
. - 5.4)
tend to affect the movement of relatively cool smoke in
the reverse of normal stack effect. In the case of hot where
smoke, buoyancy forces can be so great that smoke can = pressure difference from fire compartment to sur-
flow upward even during reverse stack effect. Further
roundings, in. H 2 0 (Pa);
information about smoke flow due to stack effect and
other driving forces is presented by Klote (1989). To = absolute temperature of outside air, OR (K);

Buoyancy of Combustion Gases


Tf = absolute temperature of gas inside fire compart-
ment, OR (K);
High-temperature smoke from a fire has a buoy-- h = distance above neutral plane, ft (m);
ancy force due to its reduced density. The pressures IS = 7.64(3460).
occurring during a fully involved compartment fire are
Fang (1980) has studied pressure differences
illustrated in Figure 5.10, and these pressures can be
caused by the stack effect of a room fire during a series
analyzed in the same manner as pressures due to stack of full-scale fire tests. During these tests, the maximum
effect. In the same manner as Equation (5.26) was pressure difference reached was 0.064 in. H 7 0 (163a)
developed for stack effect, the foIlowing equation for across the bum room wall at the ceiling.
the pressure difference Apfi from the fire compartment Observation of Tables ..5.1. and 5.2 can pro_~~.@e
to its surroundings can be developed: - -

insight on conditions for which buoyancy, as opposed to


s%ck effect,iFTikeTy. to be the dciffiiitlriving force.
\ ~ ~ f j j o k ~f,ao-G~s;-th~e-
~ - w i bbU.oyan.c wi lr-di,-iriinxe
-.
f@rlarge values of As; /Aio at almost any location from
the neutral olane. For low values of-A~/d;, at locations
where a. S"

fa_rfrom the neutral plane,>&ck effect can dominats


To = absolute temperature of gases surrounding the fire even when windows are unbroken. When windows are
,

compartment; broken, stack~effectis even more likely to dominate.


Tj = absolute temperature gas \vithin the fire compart- Stack effect p
can.only be the dominant driving-- force- dur--
ment; ing times of significant inside-toloutside temperaxe
di tlerence.
h = distance above the neutral plane. -------
Much larger pressure differences are possible for
The neutral plane is a horizontal plane where the tall fire compartments where the distance, h, from the
pressure inside the fire compartment equals that outside. neutral plane can be larger, as illustrated by the follow-.
Equation (5.30) is for a constant fire-compartment tem- ing example.
Chapter 5- Effective Areas and Smoke Movement

II Example 5.8 Buoyancy Pressure in a 11


I -Fire ~o&ar&ent
For a firecompartment temperature of 1470°F (800°C), what

l is the buoyant; pressure difference at 6 It (1.83-m) above the


neutral plane?
l
11 Using Equation (5.3 l), the buoyancy pressure difference is 0.06 1)
1 in. H20(15 Pa). Figure 5. I I can also be used forthis
.-.-p

1 Example 5.9 Buoyancy Pressure Difference for Very I


Tall Fire Compartment
Fire CunpmenlTfmw&m (TJ
if the fire compartment temperature is 1290°F (700°C),what is
1) the pressure difference at 3 5 R (10.7 m) above the neutral I( IOC 2w 300 U10 SM WO 700 800 900

11 plane?.
Fire C-Temperahrre(T)

Figure 5.11 Graph of pressures dueio buoyancy.

pressure difference across these openings due to expan-


sion is negligible because of the large flow areas

l
)I
Using Equation (5.3 1) or Figure 5.11, bbis 0.35 in. H 2 0 (88
Pa). This represents an extremely large fire that is probably
unrealistic for most applications, but it was included to illus-
rrarr the exr-nt to which Equation (5.31) can be applied.
.
I
I(
involved. However, for a fire space without open doors
or windows, the pressure differences due to expansion
may be important, provided there is suficient oxygen to
support combustion for a significant time. Gas expan-
sion in such a closed space subject to the exhaust o f
Expansion of Combustion Gases zoned smoke control, is addressed in Chapter 12.
In addition to buoyancy, the energy released by a
fire can cause smoke movement due to expansion. In a Wind Effect
fire compartment with only one opening to the building,
air will flow into the fire compartment and hot smoke Wind can have a pronounced effect on smoke
will flow out of the compartment. Neglecting the added movenient. The pressure, p,,, that wind exerts on a wall
mass of the fuel, which is small compared to the airflow, of a building can be expressed as
and considering the thermal properties of smoke to be
the same as those ofair, the ratio of the volumetric flows
can be simply expressed as a ratio of absolute ternpera-
tures.
where
v,,,,, TO,/, p,,. = wind pressure, in. H20(Pa);
v;,, - Tit, C,,. = dimcnsionless pressure coefficient;
where p, = outside air density, lb/$ (kgh3);
ire,,, = volumetric flow rate of smoke out of the fire U / , = wind velocity at the upwind wall of height H, rnph
compartment, cftn (rn3/s); (mfs);
, = volumetric flow rate ofair into the fire compart- K,,. = 0.0 129 (1 .OO).
ment, chm (m3/s); I t is thc nature of wind to be variable with peak val-
To,, = absolute temperature of smoke leaving the fire ues &at can be two or three times that of the average.
conipartnient, "R (K); The peak values arc important for structural loads, but
- --___-
T,, = absolute temperature of air entering the fire com- theX5aTaverage -.
wind-velocity
. - .
-- .
is more appropriate for
-

-
--W---

partment, "R (K). the calculation OS smoke transport and evaluation-- of


For a smoke temperature of I 1 10°F (600°C), the smOEnlanagenlent s y ~ e r n In ~ . this discussion of wind
. .---
gas will expand to about three tinies irs original volume. effects, the tern1 \*cloci~yi s ~ u s e dto indicate the nreo,~
For a tire conlpartnlcnt with open doors or windows, the
Principles of Smoke Management

The pressure coefficient depends on building geom-


etry and local wind obstructions. For a low-rise building
without local wind obstructions, a typical distribution of
the pressure coefficient is shown in Figure 5.12.
Because the wind is blowing directly at one of the walls,
the distribution of the pressure coefficients is symmetri-
cal, and the pressure coefficients only need to be shown
for half of the building. It can be seen that the pressure
coefficients are positive for the windward wall and neg-
ative for the other walls.
For a tall building without local wind obstructions,
typical distribution of the pressure coefficient is shown

Pressure coefficients is also symmetrical. Values of


__
-
in Figure 5.13. As with Figure 5.12, distribution of these

oressure coefficient C.... averaged over the wall area. are


L .v U

li$ed in Table 5.3for rectangular buildings, which are


free of local obstructions.
L

An approximation of the overall pressure difference


from one side of a building to another due to wind effect Figure 5.12 Typical distribution ofpressure coefficient
can be obtained from over a low-rise building free of local
obstructions.

W2d
where -
L' -0.6
C,", = average pressure coefficient for windward wall;

Clu2 = average pressure coefficient for leeward wall.


Above the surface of the earth, the wind velocity
increases until it reaches the gradient winds. This layer I
of increasing wind speed is referred to as the wind
boundary layer. In the absence of local obstructions to
the wind, the relationship between velocity and height in
the boundary layer is often approximated by the power
law equation,

where
U = wind velocity, @m ( d s ) ;
U. = velocity at reference elevation, @m ( d s ) ;
/
z = elevation of velocity, U, ft (m); \\\\\\\
Side Front
zo = reference elevation, ft (m);
Figure 5.13 q ~ i c adistribution
l ofpressure coefficient -
a = wind exponent, dimensionless.
over a tall buildingj.ee of local obstt-uc-
Some general values of the wind exponent, a , are lions.,

the city center and category 2 (Figure 5.14) for winds


edge of a large city center could be considered terrain Trom the opposftedirection. I nere has been a la%of
category 1 (Figure 5.14) for winds from the direction of consistency among authors regarding recommended val-
Chapter 5 -Effective Areas and Smoke Movement

Table 5.3:
Average Pressure Coefficients for Walls of Rectangular Buildings Free of Local Obstructions
(adapted from MacDonald [1975])

Wind
Building Height Building Plan Angle c,,,for Surface
Ratio Ratio Elevation Plan a A B C D

Note: h = height to eaves or parapet P = length (greater horizontal dimension of a building); W = width
(lesser horizontal dimension of a building).
Principles of Smoke Management,

-
Terrain Category 1: Terrain Category 2:. .
Large City Center Urban, Suburban, Wooded
50% of Buildings Higher Than Areas & Other Areas With
70 ft (21 m); Over at Least Closely Spaced Obstructions
6600 ft (2000 m) Upwind Compared tc or Laps: Than
n Single Family Homes; Over at
Least 6600 fi (2000 m) Upwind

~ ~ ~p ~ ~~

Terrain Category 3:
Open Terrain with Scattered
Obstacles Generally Less Than
33 ft (10 m) High

Terrain Category 4:
Flat, Unobstructed Areas
Exposed to Wind Flowing Over
a Large Body of Water;
m No More Than 1600 fi (500 m)
Inland
-7
-
-
-7
rc

a = 0.10
Wind Velocity
6 = 7OOft (210m)
Profile
Note: a is the the wind exponent, and 6 is the wind boundary layer
thickness.
Figure 5.14 bl4ricl frlwiri cofegor-ies.
Chapter 5-Effective Areas and Smoke Movement

ues of wind exponent, and the values of Figure 5.14 For building and wind measurement sites that are
were chosen to be consistent with those o f the 1997 near each other, the velocities of the gradient winds are
ASHRAE Handbook-Fundamentals, Chapter 6, "Air- equal. Equating Equations (5.36) and (5.37) and rear-
fl;w Around Buildings." ranging results in
Using Equation (5.35) with z = H (where H is t h e
upwind height of the wall of a building), the average Smet H a

velocity of the gradient wind can be expressed as 'H = 'met


met
(:) (5.38)

Substituting this into Equation (5.33) yields

where
UgeH= velocity of the gradient wind above the building, where
fpm (&S);
UH = wind velocity 2t the top of the wall, fpm (&S);
H = upwind height of the wall, fi (m);
6 = boundary layer height in the vicinity of the build- It can be seen that Equation (5.39) has the advan-
ing, ft (m); tage in that it can be used to calculate wind pressures
a = wind exponent in the vicinity of the building, based on measured design wind data.
dimensionless. The above discussion is for buildings without large
General values o f boundary layer height, 6, are local obstructions. For buildings with such obstructions,
listed inFigure 5.14 for the terrain categories. and these specialized wind tunnel tests are needed to determine
were also chosen to be consistent with those of the pressure coefficients due to the wind. Such tests are
ASHRAE Fundamentals. The weather service measures routinely conducted for structural analysis of large
wind data at airports and other locatio~is,typically at 33 buildings. For both structural and smoke management
ft (10 m) above the ground. The average velocity of the purposes, the wind flow around buildings is fully devel-
gradient wind can also be expressed as oped turbulent flow, and the flow coefficients are inde-
pendent of the Reynolds number. Thus, the flow
coefticients obtained from wind tunnel tests for struc-
tural analysis are applicable for smoke management
analysis. While the tern~inologyof a wind tunnel test
where report may ditkr from that o f this section, the results are
applicable to smoke management analysis.
Ug,,,,,, = velocity of the gradient wind above the wind
For tnforniation about wind and smoke manage-
anemometer, fpm ( d s ) ; ment, readers are referred to Kandola (1986a, 1986b)
U,, = measured wind velocity, fprn ( d s ) ; and Klote (1995). For additional information about wind
H,,,,, = height of wind measurement, ft (m); pressures on buildings see Aynsley (19S9), Shaw and
Tamura (1977). and Kandola ( 1 9 8 6 ~ ) .Several civil
6,,,,, = boundary layer height in the vicinity of the
engineering tests provide useful information about wind
wind anemometer, ft (m); engineering-for example, Dyrbye and Hansen (1997);
a;,1et = wind exponent in the vicinity of the wind ane- Liu (1 99 I): MacDonald (1 975); and Simiu and Scanlan
mometer, dinlensionless. ( I 996).

Example 5.10 Wind Pressure in a Suburban Area


A building is located in the center of a large suburban area. and the design velocity from measurements at a nearby airport
is 22 mph (&S). Tlie i~eightof the windward wall is 120 fr. the wind coetticient is 0.8, and tlie outside air density is 0.075
lb/m3 (1.2 kg;lm3). Calculate the wind pressure.

From Figure 5.14, the city center is terrain category 2 with a = 0 . 2 and 6 = 1200 FI (370 IN),a ~ l dthe airport is temain catcgory 3 with
,,,,= 0.14 and 6,, = 900 fi (270 ni). The height of the wind anemometer is H,,,,,= 33 ti (l0 m).
a,

Note: Da~nliom R u i n d t w ~ n e ltest would be more accurate than rhcse calculations. and such wind tunnel daL1 should bc used wlm
available.
Principles of Smoke Management

III Example 5.11 Wind Pressure in an Urban Area


For the conditions of Example 5.10, what is the wind pressure if the building were located in a large city?
From Figure 5.14, the urban area is terrain category 1 with o = 0.33 and 6 = 1500 ft (460 m). 11
From Equation (5.40), C - (-,
h - Hme
(9 (-1
time, 2amct
6
2a
=
33
(-1
g00 2(0.14) 120 2(0.33)
l500
= 0.476.

11 From Equation (5.39),pWis 0.09 in. H 2 0 (22 Pa). II


Note: As with Example 5.10, data from a wind tunnel test would be more accurate than these calculations, and such wind tunnel data
should be used when available.

Forced Ventilation Systems


Machinery
Heating, ventilating, and air-conditioning (HVAC) Room Lobby
systems frequently transport smoke during building Building +-
fires. When a fire starts in an unoccupied portion of a Space
building, the HVAC system can transport smoke to a
space where people can smell the smoke and be alerted 4--

to the fire. 'Upon detection of fire or smoke, the HVAC


system shbuld be designed so that either the fans are +-
shut down'or the system goes into a special smoke con-
trol mode of operation. The advantages and disadvan-
tages of these approaches are complex, and no simple
consensus has been reached regarding a preferred
method for various building types. However, if normal
HVAC operation continues, the HVAC system will
transport smoke to every area the system serves. As the
fire progresses, smoke in these spaces will endanger
life, damage property, and inhibit fire fighting. Although ' " // /v h' h' I
/< L' /'/
Note: Arrows indicate the
// L
shutting down the HVAC system prevents it from sup-
plying oxygen to the fire, system shutdown does not /,y// ,b direction of flow.
prevent smoke movement through the supply and return
Figure 5.15 Airflow due to the downward moveinenf of
ducts, air shafts, and other building openings due to
an elevator cal:
stack effect, buoyancy, or wind. Installation of smoke
dampers can help inhibit this smoke movement. A third
alternative fire mode for HVAC systems consists of con-
tinued HVAC operation, while dumping return air to the
outside in an attempt to minimize smoke transport
throughout in the building by the HVAC system. While
this third approach has not been experimentally or theo- where
retically verified, it seems that it may have the potential
P = air density within the shaft, lb/ft3 (kg/m3);
to minimize smoke transport through the HVAC system.
Computer 'simulation of smoke movement through A, = cross-sectional area of shaft, ft2 (m2);
HVAC systems is discussed by Klote (1987).
U = velocity of elevator car, Fpm (mls);
N, = number of floors above the car, dimensionless;
Elevator Piston Effect
Nb = number of floors below the car, dimensionless;
When an elevator car moves in a shaft, transient
pressures are -produced. A downward-moving elevator C = flow coefficient for building leakage paths, dimen-
car forces air out of the section of shaft below the car sionless;
and into the section of shaft above the car, as illustrated A, = effective flow area per floor batween the shaft and
in Figure 5.15. Klote and Taniura (1986) developed the
the outside, (m2);
following analytical equation for the pressure differ-
ence, 41,,,due to elevator piston effect from the outside c, = flow coefficient for flow around the car, dimen-
to the elevator shaft above the car: sionless;
Chapter 5-Effective Areas and Smoke Movement

A, = free flow area in shaft around car, or cross-sec- bp,- = pressure difference h m the building to the
tional area of shaft less cross-sectional area of the lobby, in. H20 (Pa);
car, I?
(m2); dp,, = pressure difference from the outside to the shaft,
Kpe= 1.66x10~(1.00). in. H20 @a);

The coefficient C, was cvzluated at 6.94 for a two- A, = effective flow area between shaft and the outside,
car shaft with only one car moving and at 0.83 for a two- ti? (m2);
car shaft with both cars traveling side-by-side together. Air = leakage area between the building and the lobby,
The value for the two cars moving together is believed
to be appropriate for obtaining approximations of pres- rt' (m2).
sures produced by the motion of a car in a single car This series flow path analysis does not include the
shaft. For the sake of simplicity in the analysis leading effects of other shafts, such as stairwells and dumbwait-
to Equation (5.41), buoyancy, nind, stack effect, and ers. Provided that the leakage of these other shafts is rel-
effects of the heating and ventilating system were omit- atively small compared to AOi, Equation (5.42) is
ted. Omitting stack effect is equivalent to stipulating appropriate for evaluation of A, for buildings with open
floor plans. Further, Equation (5.43) is appropriate for
that the building air temperature and the outside air tem-
closed floor plans, provided all the flow paths are in
perature are equal.
series and there is negligible vertical flow in the build-
For the system of three series flow paths from the
ing outside the elevator shaft. The complicated flow
shaft to the outside illustrated in Figure 5.15, the effec- path systems probably require case-by-case evaluation,
tive flow area, A,, per floor is which can be done by using the effective area tech-
niques presented later in this manual.
To test the above theory, experiments were con-
ducted in a hotel in Toronto, Ontario, Canada. Figure
5.16 shows measured pressure differences across the top
where floor elevator lobby while a car was descending. Also
A, = effective flow area, ft2(m2): shown is the calculated pressure difference, which is in
good ageement with the measurements. This experi-
A,, = leakage area behveen the lobby and the shaft, li? ment is described in detail by Klote and Tarnura (1986).

A;,. = leakage area between the building and the lobby,

A,, = leakage area between the outside and the building,


ft' (m').
A detailed discussion of effective flow areas is pro-
vided later in this text. in a similar manner to the devel-
opment for stack effect, the pressure difference from the
lobby to building interior can be expressed as

Time (S)
Figure 5.16 Pressure difference across elevator- lobby
where of a Toronto hotel due to piston effect.
Principles of Smoke Management

Example 5.12 Pressures Due to Moving Elevator Car


What pressure differences are produced by a downward-moving elevator car with a velocity of 600 fpm (3.05 mls) in a single-
shaft? The shaft is 20 stories high and the car is on the 18th floor (No= 2 and Nb = 17). The areas are
f? (m2)
A, area between lobby and shaft 1.60 (0.149)
Ai,, area between building and lobby 0.42 (0.039)
AOi,area between outside and building 0.54 (0.050)
As, cross-sectional area of shaft 60.4 (5.61)
19.4 (1.80)
A,, free flow area around car
Use C = 0.65, C, = 0.83, and p = 0.075 1b/ft3@/m3). From Equation (5.42), the effective area is 0.325 f?(0.302 m'). From Equation
(5.26), the pressure difference kom the outside to the shaft, Apso,is 0.30 in. H20 (75 Pa). From Equation (5.43), the pressure differ-
ence kom the building to the lobby is 0.18 in. H20 (45 Pa).

The pressure difference, Mri, cannot exceed the = 1.66 X I O -(1.00).


~
K,,
upper limit of
This relation is for unvented tlevator shafts o r
shafts for which the vents are closed. The pressure dif-
ference, (M,;),,,is strongly dependent upon U, As, and
A,. For example, Figure 5.17 shows the calculated rela-
where '
tionship k t w e e n (W,;),, and U due to one car moving in
(Wri), = upper limit of the pressure difference from the a single-car shaft, a double-car shaft, and a quadruple-
building to the lobby, in. HzO (Pa); car shaft. As expected, (Wri), is much greater for the
single-car shaft. It follows that the potential for smoke
= air density within the shaft, 1b/ft3(kg/m3); a

problems due to piston effect in single-car shafts is


2
= cross-sectional area of shaft, ft' (m ); h h i g r e X e X i ~ i ~ E ~ a r S L f t J . C b X p a r i s O nf <
= effective flow area per floor between the shaft stack effect induced pressure differences indicates that
-- can be larger than those of other driving forces-
they
and the outside, ft2 (m2): (Tables 5.1 and 5.2).
= velocity of elevator car, fpm (rnls);
Operation of- .elevators
-- --by the fire ssrvice during a
= free flow area in shaft around car, or cross-sec- -, --. .
fire can result in smoke being pulled into the elevator
tional area of shaft less cross-sectional area of
the car, ft2 (m2);
-
shaft by piston effect.
--
It seems a safe rzconmendation
that fire fighters should favor the-"se o f elevators in
= leakage area between the building and the mult~ple-carshafts over ones in singe-car shafts. Klote
Air
-(1988) developed another analysis of piston effect.
lobby, ft2 (m2); including the influence of elevator smoke control, and
= flow coefficient for flow around the car, dimen- experiments conducted by Klote and Tamura (1987)
sionless; were in good agreement with this theop.
Chapter 5-Effective Areas and Smoke Movement

Example 5.13 Upper Limit of Pressure Due to Elevator Motion


1. What is the uppcr limit of the pressure difference produced by the moving elevator car in a singlecar shaft fiom Example 5.6? The
values used in #is calculation a&
U, car velocity 600 fpm (3.05 d s )
C, flow coefficient for flow around elevator car 0.83
p, air density in shaft 0.075 lb/@ (1.20 kg/m3)
A,, effective area between shaft and outside 0.325 ft2(0.0302 m2)
Ai, area between building and lobby 0.42 ft2 (0.039 m2)
A,, cross-sectional area of shaft 60.4 ft2 (5.6 1 m2)
A,, free flow area around car 19.4 ft2 (1.80 m2)
From Equation (5.44), the upper limit of pressure difference fi-orn the building to the lobby is 0.19 in. H20(47 Pa).

2. What would be the upper limits of pressure difference if the car were in a double-car shall ora quadruple-carshaft? For multiple-car
shafts, C, = 0.94 is used. The areas for these shafts are:
For double-car shaft
A,, cross-sectional area of shaft 120.8 ft2 (1 1.22 m')
A,, free flow area around car 79.8 ft2 (7.41 m')
For cyadru~le-carshaft
A,, cross-sectional area of shaft 241.5 ft2 (22.44 m'
)
A,, free flow area around car 200.5 ft2 (1 8.63 m')
From Equation (5.44), tlie upper limits of pressure difference from the building to the lobby are:
For the double-car shaft: 0.035 in. H20(9.0 Pa).
For the quadruple-car shaft: 0.022 in. H20 (5.5 Pa).

IPressure differences, (Ap,.;),,,for other car velocities are sho\vn on Figure 5.17.
these neutral plane locations, the flowv rates and pres-
Car Velocity (rnls)
sures can be evaluated.
o.2or ; , , 2
1
3
l, 1 ,
4
l 5, 50
40
-
g Shaft with a Continuous Opening
Single Car
Shaft Quadruple
The flow and pressures of normal stack effect for a
Car Shaft single shaft connected to the outside by a continuous
Double Car opening of constant width from the top to the bottom of
Shaft the shaft is illustrated in Figure 5.18. The following
Q analysis of this flow, and the resulting location of the
n neutral plane, \\.as developed by McGuire and Taniura
ioo 200 400 600 800 1000
Car Velocity (fpm) (1975). The pressure difference from the shaft to the
Figure 5.17 Calczclated q m r limit ofpi-esszcre d(lj^er- outside is expressed by Equation 15.25). The mass flow\.
etice,Ji.onithe elevator lobby to the build- rate, dm;,,, through tlie differential section, tlh. of the
i~lgdue to pis1017effect. shaft below the neutral plane is

LOCATION OF NEUTRAL PLANE

In this section, methods of determining the location where


o f the neutral plane arc described for a single shaft con-
nected to the outside only. The methods of effective area
can be uscd to extend this analysis to buildings. Using A' = area ot'thc opening pcr unit hcight
Principles of Smoke Management

Example 5.14 Location of Neutral Plane II


with Uniform Leaka~e
Calculate the location of the neutral plane for a 100 tt
(30.5 m) tall building of uniform floor-to-floor leakage. The
inside temperature is 72°F (22"C), and the outside temperature
is 0°F (-18°C).

From Equation (5.48), the neutral plane is located at a height of


48.8 ft (14.9 m) above the bottom of the building. This is
slightly different from the generally accepted approximation of

Figure 5.18 Nonnal stack effect betweett a single shaft


comected to the olrtside a continzro~rs
opening.

To obtain the mass tlow rate into the shaft, this


equation can be integrated from the neutral plane ( h = 0)
to the bottom of the shaft (It = - H,,).

Pressure ~iiference.Ap,,
Figure 5.19 Nortnal stack G e c t for a single shafr with
In a similar manner. an expression for the mass hvo openings.
flow sate from the shaft can be developed, where H is
the total height of the shaft.
Shaft With Two Vents

Normal stack effect for a shaft with two openings is


illustrated in Figure 5.19. The pressure difference from
For steady flow, the mass flow rate into the shaft
the shaft to the outside is expressed by Equations (5.25)
equals that leaving it. Equating Equations (5.46) and
and (5.26). To simplify analysis, the distance? H ,
(5.47), canceling like terms, rcarrangins. and substitut- between the openings is considered much greater than
ing, Equation (5.24) yields the hzight of either opening. Thus, the variation of pres-
sure \\h11 height for the openings can be neglected, and
H,,
-- I
(5.45) the mass flow rate into the shaft can be expressed as
H l+(T~/T0)""

where

H,, = distance from the bolt0111of the shaft to the neutral and the mass flow rate out of is
plane, ft (m);
H = height ol'shal't. fi (m); ~ j l ~= c,~,,I-)
, , ~ (5.50)
Ty = absolute temperature of air in shaft. "R ( K ) ;
where A , and A* are the areas above and below the neutral
r, = absoluw temperature ofoutside air. "R ( K ) . plane. Equating these two flo\vs as was done above yields
Chapter 5-Effective Areas and Smoke Movement

small in comparison to the shaft height, H. Thus, a con-


stant pressure difference can be used to describe the
flow through the vent. The mass flow out of the shaft is
the sum of the flow out of the continuous opening,
where
expressed as Equation (5.47), plus the flow out of the
H, = distance from the bottom of the shaft to the neutral vent of area A, located a t an elevation of H, above the
plane, ft (m); shaft bottom. .
H = heightofshaft, ft(m);
T, = absolute temperature of air in shaft, "R (K);
T, = absolute temperature of outside air, "R (K);
A, = area above neutral plane, ft2 (m2); The conservation of mass equation for the shaft can
be written as
Ab - = area below neutral plane, ft2(m2).
The location of the neutral plane is highly depen-
dent on the ratio Ab/A,. For Ab /A, that approaches zero,
H, approaches H. This means that if the area at the b d -
tom is very small compared to the area at the top, then
the neutral plane is at or near the top area. Equation
(5.5 l ) is a strong function of the flow areas and a weak
Canceling like terms and incorporating Equation
function of temperature. (5.53) results in
Example 5.15 Location of Neutral Plane
with Two Equal Openings
1
What is the location of the neutral plane in a l00 ft (30.5 m) tall
shaft with hvo equal leakase areas (Ah = A,) at the shaft top
and bottom? The inside temperature is 72°F (22°C). and the
outside temperature is 0°F (-l 8°C). As would be expected, this equation reduces to
Equation (5.48) for A,. = 0. Equation (5.54) can be rear-
From Equation (5.5 l), the neutral plane is located 46.4 fi (14.1
ranged as
m) above the bolton1area. This is only a little less than Example
5.14 with the continuous opening (48.8 ft [14.9 m]).

1 Example 5.16 Location of Neutral Plane


with Two Unequal Openings
l
What is the location of the neutral plane in a 100 ft (30.5 nil
tall shafi with a 4 ft2 (0.37 m') opening at the top and a I ft2
(0.093 m') opening at the bononl? The inside temperature is
72'F (22"C),and the outside temperature is 0°F (-I 8°C).

From Equation (5.5 1). the neutral plane is located 93.3 fi (28.4
m) above the bottoni area. This illustrates the extent to which
nonunifoml leakaze areas can cause the ncutral plane to be h r
1 f r m the building's mid-height.

Vented Shaft Neutral Plane -

The flow and pressures of normal stack ellkcl lor a


shaft connected to the outside by a vent and a continu-
o u s opening are shown in Figure 5.20. The tbllowing
analysis is tor a vcnt above the neutral plane, but a simi-
lar one can be madc I'or a vent below the neutral plane.
This analysis is an extension of one by McGuire and
Talnura (1 975) for a top \-ented shali. The mass flow
into the shaft is expressed by Equation (5.46). For sim-
plicity of analysis, the height of the vent is considered
Principles of Smoke Management

much greater than the area of the continuous opening


(A'H). As with Equatibn (5.51), the above equation is a
strong function of the flow areas and a weak function of
temperature.
Regardless of whether the vent is above or below
For relatively large vents, the ratio A1H/AV the neutral plane, the neutral plane will be located
approaches zero. As A1H/AVapproaches zero, the first between the height described by Equation (5.4) for an
and third terms in the above equation approach zero, and unvented shaft and the vent elevation, H , Further, the
the equation is reduced to H, = H , Thus, the neutral larger the value of A, /ArH, the closer the neutral plane
plane is at or near the vent elevation, for a vent area very will be to H ,
CHAPTER 6

Principles of Smoke Management

he term "sn~okemanagement," as used in this criteria for the construction of smoke barriers, including
manual, includes all methods that can be used doors and smoke dampers in these barriers. The extent
singly or in combination to modify smoke move- to which smoke leaks through such barriers depends on
ment for the benefit of occupants or firefighters or for the size and shape of the leakage paths in the barriers
the reduction of property damage. The use of barriers, and the pressure difference across the paths. Hazard
smoke vents, and smoke shafts are traditional methods analysis (chapter 9) can be used to evaluate the perfor-
of smoke management. The effectiveness of barriers is mance of con~partmentation.
limited to the extent to which they are free of leakage
paths. The effectiveness of atrium smoke vents and Dilution Remote From a Fire
smoke shafts is limited to the extent that smoke must be Dilution of smoke is sometimes referred to as
sufficiently buoyant to overcome any other driving smoke purging, smoke removal, smoke exhaust, or
forces that could be present. smoke extraction. Dilution can be used to maintain
Fans are used with the intent of providing smoke acceptable gas and particulate concentrations in a room
protection by means of pressurization. The mechanisms subject to smoke infiltration through leakage paths from
of compartmentation, dilution, pressurization, airflow, an adjacent space. Tliis can be effective if the rate of
and buoyancy are used by themselves or in combination smoke leakage is small compared to either tlie total vol-
to manage. smoke conditions in fire situations. These ume of the safeguarded space or the rate of purging air
mechanisms are discussed in the sections below. supplied to and removed from the space. Also, dilution
can be beneficial to the fire service for removing smoke
SMOKE MANAGEMENT
after a fire has been estinguished. Sometimes, when
doors are opened, smoke will flow into areas irltended to
Compartmentation
be protected. Ideally, such occurrences of open doors
Barriers with sufficient fire endurance to remain will only happen for short periods of time during evacu-
effective throughout a fire exposure have a long history
ation. Smoke that has entered spaces remote from the
of providing protection against fire spread. In such fire fire can be purged by supplying outside air to dilute the
compartmentation, the walls, partitions, floors, doors, smoke.
and other barriers provide some level of smoke protec-
The following is a simple analysis of smoke dilu-
tion to spaces remote from tlie fire. Tliis section dis- tion for spaces in which there is no tire. At time zero (I =
cusses the use of passive compartmentation, while tlie 0), a compartment is contaniinated with some concen-
use of compartmentation in co~ijunctionwith pressuriza- tration of smoke and no additional smoke flows into the
tion is discussed later. Many codes, such as the NFPA compartment or is generated within it. Also, the contam-
10 1 L{/b S a j i ! ~Code (NFPA 2000c), provide specific inant is considered uni fol-mly distributed throughout tlie
Chapter 6 -Principles of Smoke Management

space. The concentration of contaminant in the space Example 6.2 Smoke Dilution in a
can be expressed as
Space Remote from the Elre
A space is isolated from a fire by smoke barriers and selfclos-
ing doors so that no smoke enters the compartment when the
doors are closed. However, when a door is opened, smoke
This equation can be solved for the dilution rate and flows through the open doonvay into the space. If the door is
the time. closed when the contaminate in the space is 20% of the bum
room, what dilution rate is required so that six minutes later the
concentration will be I% of the bum room?

The time, r, is 6 minutes, and C& is 20. From Equation (6.2),


the dilution rate is about 0.5 changes per minute or 30 ach.

Caution About Dilution Near a Fire


where Many people have unrealistic expectations about
CO = initial concentration o f contaminant what dilution can accomplish in the fire space. The anal-
C = concentration of contaminant at time, t ysis of the previous section is not applicable to spaces in
a = dilution rate in number of air changes per minute which there is a fire. There is no theoretical or experi-
I = time after smoke stops entering space or time after mental evidence that using a building's heating, ventilat-
which smoke production has stopped, minutes ing, or air-conditioning (HVAC) system for smoke
e = constant, approximately 2.7 1S dilution will result in any significant improvement in
The concentrations COand C must be ekpressed in tenable conditions within the fire space. I t is well kno\vn
the same units, and they can be any units appropriate for that HVAC systems promote a considerable degree of
the particular contaminant being considered. In reality, it sir mixing within the spaces they serve. Because of this
is impossible to ensure that the concentration of the con- and the fact that very large quantities of smoke can be
taminant is uniform throughout the compartment. produced by building fires, it is generally believed that
Because of buoyancy, it is likely that higher concentra-
dilution o f smoke by an HVAC system in the fire space
tions would tend to be near the ceiling. Therefore,
exhausting smoke near the ceiling and supplying air will not result in any practical improvement in the tena-
near the floor will probably dilute smoke even faster ble conditions in that space. Thus, it is recommended
than indicated by Equations (6.2) and (6.3). Caution that smoke purging systems intended to improve hazard
should be exercised in the location of the supply and conditions within the fire space or in spaces connected
exhaust points to prevent the supply air from blowing to the lire space by large openings not be used.
into the exhaust inlet and, thus, short-circuiting the dilu-
tion operation
Pressurization
Esnmple 6.1 Smoke Purgin!: After the Fire is Extin~uished Systems using pressurization produced by mechani-
1. After the fire department puts out a fire, they want to clear cal fans are referred to as stnoke contt-01 in this book and
the smoke quickly so that they can make an inspection to deter- in NFPA 92A (NFPA 2000a). A pressure difference
mine if the fire is completely out. If the HVAC system is capa- across a barrier call control smoke movement, as illus-
ble of a dilution rate of 6 ach. how long will it take to reduce trated in Fizure
- 6.1. Within the barrier is a door. The
the smoke concentration to I% of the initial value? high-pressure side of the door can be either a refuge area
or an egress route. The low-pressure side is exposed to
The dilution rate, a, is 0.1 changes per minute, and C, /C is smoke from the fire. Airflow through the gaps around
100. From Equation (6.3), the time to get the concentration to the door and through construction cracks prevents
I% is 46 minutes. Considering the desire of the tire department smoke infiltration to t!le high-pressure side. When the
to quickly inspect the area, such a long purging time will prob- door in the barrier is opened, airflow through the open
ably be excessive.
door results. When the air velocity is low, smoke can
2. If the tire department wants the space to be purged in 10 flow against the airflow into the refuge area or egress
minutes. what dilution rate is needed? route. as shown in Figure 6.2. This smoke backflow can
be pre\mted if the air velocity is sufficiently lar,ne, as
The time, I, is 10 minutes, and C,/C is 100. From Equation shown in Figure 6.3. The magnitude of velocity neces-
(6.2).the dilution rate is 0.46 changes per minute. or about 28 sary to preLrent backflow depends on the energy release
changes per hour. rate ofthe fire. as discussed in the liext scction.
Principles of Smoke ManagemeM

In this case, the appiopAate physical quantity is' pressure


difference. Consideration of the two mechanisms as sep-
arate has the added advantage that it emphasizes them dif-
ferent considerations that need to be given for opened
High Pressure Side and closed doors.
Low Pressure Side
To ensure that expansion pressures are not a rob-
lem, pressurization systems should be designed so that a
c u d patin exists for smoke movement to the outside. This
'\ \\Y\'; \\\\ ..\\\\\.\\\\ \ \\\ ... path could be as simple as relying on a top-vented eleva-
Figure 6.1 Pressure differ-ence across a barrier of a -
tor shaft, f
e -
exhaust. It is important that some
i ed
snzoke control system can prevelir smoke
be provided. The pressurization systems most com-
infiltratioii to r/7e high-press~rreside of h e
=ly used are pressurized stairwells and zoned smoke
barriei
control. Elevator smoke control is less common.
Detailed design analysis and general considerations
about these pressurization systems are discussed later in
this manual.

,\ \,,\'...\
.. Relatively
Low Air
Velocity

....' \,\,,\~.
+

..
-
---+
+
%\ ..,\\.;.., ...%, \...~,, .>.,,\V:\,\\... . ...., .
Airflow
Airflow has been used extensively to manage
smoke from fires in subway, railroad, and highway tun-
nels. Large flow rates of air are needed to control smoke
floiv. and these flow rates can supply additional oxygen
to the fire. Because of tlie need for complex controls,
Figure 6.2 Smoke bnc&fIowagainst low air veloci/J! airflow is not used so extensively in buildings. The con-
tlirorrgli an open door-rvny. trol problem consists of having very small flows when a
door is closed and then having those flows increase sig-
nificantly when that door opens. Further, it is a major
concern that the airflow supplies oxygen to the fire. This
. . ,.
"-., . . ,:~'\ ..'..\'..'\"\.',. .' '\,\\\\\.\'\\
, .-.,....................
'.
. . . . . . . . . . . . ........................
................
\,\
section presents the basics of smoke control by airflow,
....;:..........................................
>
, . . . ., . . . . , ...... , . . . . . . . . . .
Caution: ..................................
Because it supplies
which demonstrate why this technique is not recom-
oxygen to the fire. mended, except wlien the fire is suppressed or, in the
airflow needs to be ----+ rare cases, when fuel can be restricted with confidence.
used with great care.
High Air
Thonias (1970) determined that airflow in a corri-
BVelocity dor in which there is a fire can almost totally prevent
smoke from flowing upstream of tlie fire. As illustrated
in Fizure 6.4, the smoke forms a surface sloped into the
direction of tlie oncoming airflow. Molecular diffusion
Figure 6.3 High uir- velocip 1171arrghan open door- is believed to result in transfer of trace amounts of
way pre1~er71s
suioke backj7ow. smoke, producing no hazard but just the smell of smoke
upstream. There is a minimum velocity below in which
smoke will flow upstrezm, and Thomas developed tlie
Pressurization results in airflows of high velocity in follon.ing enlpirical relation for this critical velocity:
the small gaps a,ound closed doors and in construction
cracks, thereby preventing smoke backflows through
these openings. Therefore, in a strict physical sense, the
pressurization is equivalent to tlie mechanism ofairfiow
that is discussed in the nest section. However, consider-
ing these mechanisms as separate is advantageous for critical air velocity to prevent smoke backflow,
discussing smoke management systems.
enerzy release rate into corridor,
For a barricr with one or more large openings, air corridor width,
velocity is the appropriate physical quantity for both
density of upstream air,
design and nicasurcment. Ho\i.cver. wlien there are only
small cracks, such as thosc around closcd doors, design- specific heat of downstream gases,
ing to and measurement of ail- velocities is impractical. absolute temperature of downstream gases.
Chapter 6- Principles of Smoke Management

K = constant on the order of 1, through an open doorway, air transfer grille, or other
g = acceleration of gravity. opening. The critical velocities calculated from Equa-
The units are not given for Equation (6.4), as it is tions (6.4) and (6.5) are approximate because an approx-
valid for any homogenous system of units (Appendix imate value of K was used. However, the critical
A). The downstream properties are considered to be suf- velocities from this relation are indicative of the kind of
ficiently far downstream of the fire for the properties to air velocities required to prevent smoke backflow from
be uciforrn across the section. Note that_ T is for the fires of different sizes. As ca: be see;; from Figure 6.5,
downstream gases, and p is for upstream gases. This the critical velocity is less for wider corridors. Examples
means that p is not calculated from T. The critical air 6.3 and 6.4 illustrate the flows needed for different fires.
The equation of Thoinas can be used to estimate the
velocity can be evaluated at p = 0.08 1 lb/ft3 (1.3 kg/m3),
Cp= 0.24 Btullb OF (1.005 W k g 'C), T = 81°F (27OC), airflow rate necessary to prevent smoke backflow
and K = I. through an open door in a boundary of a smoke control
system. Rilling (1980) developed another equation for
calculation of the critical velocity, and Tamura (1991)
conducted fire experiments to determine the critical
velocity for snioke flow through an open doonvay.
where
Uk = critical air velocity to prevent sn:.?ke backflow, fpm While the critical velocity can be calculated, the
oxygen supplied is a concern. Huggett (1980) evaluated
(m/?);
the oxygen consunied for combustion of numerous natu-
0 = energy release rate into cor~idor,Btu's (kW); ral and synthetic solids. He found that for most materials
W = corridor width, fi (m); that are involved in building fires, the energy released
K, = 86.3 (0.292). per unit of mass of oxygen consunied is approximately
Equation (6.5) can be used when the fire is located 5630 Btu/lb (13.1 MJJkg). Air is 23.3% oxygen by
in the corridor 01 when the snioke enters the corr~dor weight. Thus, if all the oxygen in a pound of air is con-

Airflow y;j;j!;;;;;j:;.;?;i;j:;,:

Because it supplies oxygen


to the fire, airflow needs to
be used with areat care. L

Heat Release Rale (MW)


0 0.4 0.8 1.2 1.6 2.0 2.4
800 -
-g 600 -
-
X
.-
0
400 -
9
-
m
-
.-
.-
6 200 -
Principles of Smoke Management

sumed, 1300 Btu of heat is liberated. Stated in the S1 suppression is evaluated to be acceptable. The methods
system, if all the oxygen in a kg of air is consumed, 3.0 of tenability analysis discussed in Chapter 9 can be used
MJ of heat is liberated. As can be seen from Example to evaluate the consequences of such failures.
6.3, the air needed to prevent smoke backflow can sup-
port an extremely large fire. In most locations of com- Example 6.3 Airflow to Prevent Smoke
mercial and residential buildings, sufficient fuel (paper, Backflow from a Small Fire
cardboard, furniture, etc.) is present to support very An energy release rate of 142 Btu/s (150 kW) can be thought
of as the size of a large wastebasket fire. What flow rate of air
large fires. Even when the amount of fuel is normally
is needed to prevent smoke backflow h m such a fire in a cor-
very small, short-term or transient fuel loads (during ridor 4 ft (1.22 m) wide and 9 ft (2.74 m) high?
building renovation, material delivery, etc.) can be sig-
nificant. From Equation (6.5), the critical velocity is 286 fpm (1.45 m/
S). The cross-sectional area of the corridor is 4 x 9 = 36 ft2
Because of the concern about supplying combus-
(1.22 x 2.74 = 3.34 m2). The flow rate is .the cross-sectional
tion air to the fire, caution is recommended when air-
flow is used for smoke protection. The common use of
airflow to manage smoke movement in conjunction with
fuel restriction in rail and highway tunnels is probably 11 Example 6.4 Airflow to Prevent Smoke 1 1
Backflow from a Large Fire
justifi-kd by the lack of appropriate smoke management An energy release rate of 1420 Btu/s (1.5 MW) would result in
alternatives. The use of fuel restriction or fire suppres- a large portion of the corridor beins completely involved in
sion t6 limit the size ofthe fire for a smoke mana,oement fire. What flow rate of air is needed to prevent smoke backflow
system relying on airflow has the potential for cata- from such a fire in the corridor of Example 6.3?
strophic failure. Therefore, the use of airflow is not rec-
From Equation (6.5), the critical velocity is 616 fpm (3.13 nds).
ommended for smoke management in buildings except
The flow rate is about 22,200 c h (10.5 m3/s).
when the potential for failure of fuel restriction or fire
chapter 6-Principles of Smoke Management

Example 6.5 Airflow Through a Doorway and Fire Growth


1. Thomas indicated that his relation for critical velocity can be used to obtain a roughestimate for doorways. A m m fully involved ir

I fire could have an energy release rate on the order of2270 Btu/s (2.4 MW). what estimate of critical v&city is obtaded 6bm the
Thomas equation for a door 3 ft (0.9 m) wide?

From Equation (6.5), the critical ve1ocity.k about 793 @m (4.03 mk). If the door has an of area 20 ft2 (1.9 m2), this would amount tc
a fiow of 15,900 cfin (7.48 m3/s).

II/l 2. Consideration of a smaller fire, such as the wastebasket tire of Example 6.5, may be appropriate for many situations. What flow rate
does the Thomas relation indicate is needed to prevent backflow for the above door?

Q = 142 Btu/s(lSO kW), W=3 ft(O.9m)

l From Equation (6.5), the critical velocity is about 300 fpm (1.5 mls). For a door area of 20 ft? (1.9 m2), this would amount to a flow 01
6000 cfin (2.8 m3/s).

IIl 3. What size fues can this airflow support? Consider that all of the oxygen in the air is consumed, and that the air density is 0.075 lb/$
(1 -2 kg/m3).

Approximately 1300 Btu of energy is released when the oxygen in a pound of air is consumed, 15,900 c& can support the following
size fire:

y ) ( E J ( l 3
ft
::fiy) = 25,800 BWs (27.2 MW)

For 6290 c h , the energy release rate would be 10,200 BWs (1 0.8 MW). These fires are very large. Airflow intended to prevent smoke
backflow can cause a fire to grow significantly if there is sufficient material to bum. Therefore, the use of airflow for smoke control is
1 not recommended except when the fire is suppressed or in the rare cases when fuel can be restricted with confidence.

Buoyancy doors. The flow through a crack or other opening can be


represented by the general function,
Buoyancy of hot combustion gases is employed in
both fan-powered and non-powered venting systems.
Such fan-powered venting for large spaces is commonly
employed for atriums and covered s h o p p i ~malls_A
~~ where
-
concern with atrium smoke management systems is that
the sprinkler flow will cool the smoke, r e d u c w y - V = volumetric flow rate through the path,
<
- _-
%cy and, thus, system effectiveness. There is no ques-
tion that spr~fiklETlow does-cooi..mke. but i t j s
Ap = pressure difference across path,
f = general functional relation.
unknown to what extent that cooling reduces&&.ve- .
ness of fan-powered venting. Further research is needed The particular form of the functionf depends on the
i z t h i s area. However, the existing information can be geometry of the opening and Reynolds number. The
used to develop new design information for fan-pow- Reynolds number is
ered venting systems. NFPA 92B (NFPA 2000b) pro-
vides methods of design analysis for smoke
management systems in large spaces, such as a t r i u m
and shopping malls. where
R, = Reynolds number, dimensionless;
AIRFLOW AND PRESSURE DIFFERENCE
D, = hydraulic diameter of flow path, in. (m);
For a crack, gap, or other opening with a pressure
difference across it, a flow will result from the higher U = average velocity in flow path, fpm (mts);
pressure to thc lower pressure. Many different equations 2
v = kinematic viscosity, $/S (m /s);
have been used to express the relation between fluid
KR = 1.39 X 1U3(1 .OO).
flow rate and pressure difference with regard to air and
smoke flow in buildings. This section contains a discus- Values of kinematic viscosity are listed in Tables
sion of some of the more common equations, as well as A.8 and A.9 of Appendix A. The hydraulic diameter is
a detailed discussion of flows through the gaps around four times the cross-sectional area of the path divided by
Principles of Smoke Management

the "wetted perimeter" of the path. For example, the p density gas in path, lb/ft3 (kg/m3);
=
hydraulic diameter of a circle is the diameter of the cir-
KO = 776. (1.00);
cle, and the hydraulic diameter of a square is the side of
the square. For the long rectangular gaps around doors, KO? = 12.9 (1.00).
the hydraulic diameter is the gap thickness (D,, = 2a, Dynamic forces dominate flow with Reynolds num-
where (z is the gap thickness). The Reynolds number is bers greater than about 2000 or 4000, depending on path
usually thought of as the ratio of-kinetic forces to vis- geometry. At these Reynolds numbers, the flow
cous forces. Later sections discuss different approaches becomes turbulent. For turbulent flow, the velocity at a
that apply for flow dominated by viscous forces, kinetic given point fluctuates rapidly in an apparent random
forces, or both. manner.
The pressure difference above can be expressed as Equation (6.10) is similar to Equation (6.9) except
that it has been multiplied by.density (remembering that
r i ~= pi'). Equation (6.9) has been applied so exten-
where sively to orifice flow meters that it is often referred to as
pi = pressure at path inlet, the orifice equalion, and Equation (6.10) also is referred
to by the same name.
p, = pressure at path outlet,
The orifice equation is also commonly used for
p = den& gas in path, analysis of airflow in buildings and for analysis of
Zi= elevaiion of the path inlet, smoke management systems. Because the orifice equa-
Z,= elevation of the path outlet, tion is based on Bernoulli's equation, it strictly applies to
steady, frictionless, incompressible flows. However, the
g =acceleration of gravity. flon. coefricient was introduced to account for friction
Equation (6.8) is for constant density in the flow losses due to viscosity and for dynamic losses. The flow
path and for flows where the values of the inlet pressure, coefficient depends on the Reynolds number and the
outlet pressure, inlet elevation, and outlet elevation are
all constants. This representation is not appropriate for
inlet and outlet pressures that vary considerably with the
elevation, as is often the case for flows of hot firs gases.
However, for smoke control design, analysis of flows is
limited to normal building and outside temperatures.
-
geometry of the flow path. For flows through gaps
around doors and through construction cracks, the coef-
ficient is generally in the range of 0.6 to 0.7, but the
presence of stationary vortices in larger openlngs such
as stain\.ell doorways can reduce the flow coefficient to
about O . j j . Flow areas are discussed later.
Thus, this representation is appropriate for smoke con- For standard air density of p = 0.075 1b/ft3(1.20 kg/
trol analysis, as well as general considerations of airflow
111') and for C = 0.65, Equation (6.9) can be expressed as
in buildings.
i/ = K+A& ( G. I I )
Orifice Equation
For large Reynolds numbers- flow is directly pro- where
portional to the square root of the pressure difference
across the path: k = volumetric flow rate through the path, cfm (m3/s);
A = flow area (also called leakage area), ft2 (m2);
Ap = pressure difference across path, in. H 2 0 (Pa);
hj = 26 10 (0.839).
Equation (6. I l ) gives flow at standard temperature
70°F (21°C) and standard atnlospheric pressure of
.7 psi (l 0 l kPa).
where Frequently, volunietric flows are adjusted to stan-
i/ = volumetric flow rate through the path, cfni (ni3/s); dard \,ol~~metric llow rates. The mass flow rate is
divided by the standard density to obtain the standard
ti~ = mass flow rate through the path, Ibis (kgls): volunietric tlow rate. This is convenient because it
C = dimensionless flow coetlicient; allon.s engineers to think in terms of the familiar volu-
1 ? metric flow rates. Further. these standard flows can be
A = flow area (or leakage area). fi- (m-): treatsd as mass flow rates because they only deviate
4 3 = pressure dillerence across path, in. H 2 0 (Pn): fi-om mass Ilow ratcs by a constant.
Chapter 6-Principles of Smoke Management

Equations (6.9), (6.10), and (6.1 1) are extensively


used for analysis of smoke control systems in this man-
ual. For normally constructed buildings, these equations
are recommended for all smoke control calculations. By
a normally constructed building, it is meant to be one
that has at least tight wall and floor leakage and that
does not have gasketed or sealed interior doors. Tight
leakage of walls and floors is discussed in the section on
flow areas. The rest of the flow equations presented in
' , 1
:\\\\\\\\L\\\\\\\\\\\\\\\\\\\S\\\\\\\\'
VHQ r

this section are included for the unusual cases of very


tight construction. Figure 6.6 Parabolic velociy profile for- Poiseuille
flow between two parallel plates.
Example 6.6 Flow Calculated bv the Orifice Equation
1. Calculate the volumetric flow through a path by the orifice
equation for the following values:
A = 1 (0.0929 m2)
C = 0.65
Ap = 0.05 in. H20 (12.4 Pa)
p = 0.081 lblf? (1.30 kgfm3)

From Equation (6 g), the flow rate is 560 cfm (0.26 rn'ls).

2. Calculate the above flow for standard density of 0.075 lb/ft3


(1.20 kg/m3). Fully Developed Lammar ~ i o w

Figure 6.7 Developnzenf oflaminar-flow in a gap.


Usins Equation (6.9),theflow is 580 cfm (0.026 m3/s).This flow
is at p = 0.08 1 lb/$ (1.30 kg/m3)and not standard c h i (or rn3/s).
there are inlet and outlet losses due to flows just outside
Plane Poiseuille Flow the gap. These deviations from plane Poiseuille flow can
be significant and are accounted for in methods of anal-
For low Reynolds numbers, flow is directly propor- ysis presented later.
tional to the pressure loss. Viscous forces dominate flow
with Reynolds numbers below about 100 to 1000, Exponential Flow Equation
depending on particular path geometry. Plane Poiseuille
flow is an exact solution to the Navier-Stokes equations In order to accommodate the flows, which are
for the flow of a viscous fluid between nvo parallel and between viscous dominated and kinetic dominated, the
infinitely long plates. The velocity distribution between following exponential relation has been used exten-
the plates is parabolic, as illustrated in Figure 6.6. The sively in analysis of airflows through buildings:
fluid velocity varies only in the dirxtion perpendicular
to the flow, and this type o f flow is referred to as larninar
flow. The average velocity, U, for plane Poiseuille flow where
is proportional to pressure loss (dp/d~).
V = volumetric flow, cfni (rn3/s);
C, - flow coefficient for exponential flow equation, ft3
min-' (in. H20)-"(m3 S-' Pa-");
where
Ap = pressure difference across the path, in. H,O (Pa);
a = distance benveen plates (gap thickness);
11 flow exponent, dimensionless.
=
p = dynamic viscosity;
For a flow exponent of n = 0.5, Equation (6.13) is
p = pressure essentially the same as the orifice equation. For 11 = I ,
Real gaps in buildings are not infinitely long, and Equation (6.13) describes viscous dominated tlow. As
some distance is needed for the parabolic flow profile to \vould be expected from the above discussion, the flow
become established, as illustrated in Figure 6.7. The exponent n varies from 0.5 to l .
pressure losses (dphh) over this inlet length are greater Equation (6.13) only approximates the relation
than those of f ~ l l ydeveloped parabolic flow. Further, bettveen flow and pressure difl'erence, and the values of
Principles of Smoke Management

C, and n depend on the range of Ap. This equation has 4 = pressure difference across gap, in. H20 (Pa);
proven useful for the evaluation of flows through many Dh = hydraulic diameter, in. (m), Dh = 2a,
small cracks in buildings at low levels of pressure differ-
ence. However, this equation is not directly related to p = density of gas in gap, lb/@ (kg/m3);
the geometry of the flow path, and the values of C, for v =kinematic viscosity, f?/s (m2/s);
particular flow paths must be determined empirically. KNp = 0.108(1 .OO)
F_or analysis of buiiding airfiow, h e exponents of inte-
rior paths are often taken at 0.5, and exponents of exte- Gross and Haberman used an analytical method of
rior walls often are considered to be about 0.6 or 0.65. Miller and Han (197 1) to account for the pressure losses
in the entrance region before fully developed flow is
Gap Method achieved in a straight-through slot. Their relation for
Gross and Haberman (1988) developed a general- flow versus pressure difference is shown in Figure 6.8.
ized approach, the gap method, for determining the Three regions of flow through the straight-through slot
leakage through gaps of different geometry such as were identified, and equations for these regions are:
those of door assemblies. They developed a functional
relationship between the dinlensionless variables NQ Region 1 (%scous dominated region-for NPs250):
and NP.

Region 2 (Transition region-for 250<NP<106):

and NQ = 0 . 0 1 6 9 8 4 ~ ~ ~ (6.1 7)
where a = 1.0 1746 -0.044181 Log, o(NP)

Region 3 (Kinetic dominated region-for NP>106):


where
NQ = dimensionless flow rate;
NP = dimensionless pressure diference;
The equations for regions I and 3 were developed
R, = Reynolds number, dimensionless(Equation (6.7)); by Gross and Haberman, and the exponents are as
a = thickness ofgap in direction perpendicular to flow, expected; considering that region 1 is dominated by vis-
in. (m); cous forces and region 3 is dominated by kinetic forces.
s = depth of gap in flow direction, in. (m); Region 2 is a transition between the other two regions,

-- Region l
+- Region 2
-- . Region 3

10 1o2 1o3 104 105 106 10'


NondimensionalPressure Difference, NP

Figurc 6.8 Flow and pressure ~rlo/io~ishi,u


for S//-aigI71-/lit-o~lgli
gaps
Gross atid Habernimli [ l 9881).
(aclc~p/ecl~fiut~t
Chapter 6-Principles of Smoke Management

Table 6.1:
Flow F a c t o r s f o r Single- and Double-Bend G a p s
Dimensionless Flow Factor Flow Factor
Pressure for Single-Bend for Double-Bend
Difference, NP Slot, F, Slot, F2
Less than or
equal to 4,000 1.OOO 1 .OOO
7,000 0.98 1 0.939
10,000 0.972 0.908
10' 10' 106 10'
15,000 0.960 0.880
NondimensionalPressureDifference. NP
20,000 0.952 0.862
40,000 0.935 0.826 Figure 6.9 Flow factors for gaps (adapted from
100,000 0.9 10 0.793 Gross and Habeman [1988]).
200,000 0.890 0.772
400,000 0.872 0.742 Gap T h i i n e s s , a (mm)
1,000,OOCi 0.848 0.720 0 1 2 3 4 5 7
2,000,000 0.827 0.700
- straqht Gap

and an approximation developed by Forney (1989) is


- 0.8
0
C
1 Bend
.g 0.6 -
used for this region. This approximation is particularly E 2 Bends
attractive for computer applications because it is contin- 6 0.4 - Note: Gap depth in flow direclion
uous with the expressions for the other two regions. -6 - is 2 in.(50.8 mm). pressure difference
U- 0.2 is 0.04 in. H>O(10Pa). and air
Equations (6.7) and (6.14) can be combined to temperatureis 7OSF(21 .C).
obtain a relation for volumetric flow rate through a
straight-through slot. "0 0.05 0.10 0.15 0.20 0.25 0.30
Gap Thickness, a (in)

(6. 9) Figure 6.1 0 Flow coefficients for stl-aigl~tgaps atld


gaps with be17o's.

where
ness, n, and the f l o is
~ less for gaps with bends than for
voltrmetric flow rate, cfin (m3/s); straight gaps.
I r is espected rhat the gap model predictions for a
dimensionless tlow;
relatively \ride gap (and relatively large Reynolds num-
depth of gap in flow direction, in. (111): ber) would be closer to those of the orifice equation than
hydraulic diameter, in. (m),(D,, = 20); predictions of the exponential tlon equation. Figure
6.1 1 compares predictions of the orifice equation, the
length of gap, fi (m); exponential flow equation, and the gap method far a 0.5
in. (1 2.7 mm) wide gap, and it can be seen that the pre-
kinematic viscosity, ft2/sec (~i?/s); dictions of the orifice equation are almost identical with
60 (1 .OO). those of the gap method. As might be expected for a 0.1
- ~

in. (2.54 mm) gap, the predictions of the exponential


Frequently, slots around doors have one or more flow equation with 11 = 0.65 are much closer to those of
bends. For single- a:ld double-bend slots, the dimen- the gap method (Figure 6.12).
sionless flow, NP, can be obtaincd by multiplying values
The design book of Klote and Fothergill (1983)
for a straiphr-thr.ough slot by flow [actors F , and 1;2
used Equations (6.9) and (6.1 1) for all smoke control
(where F , is for single-bend slots, and F2 is for a dou- analysis because it was bcliwcd that the orifice equation
ble-bend slors). These flow factors are presented in was sufficiently.accurate for design analysis. Klote and
Table 6. I and Figure 6.9. Bodart (1985) reevaluated this use of the orifice equa-
Figure 6.10 shows the tlow predicted by the gap tion and the exponential flow equation. They experi-
mctliod fol- a stmight gap and gaps with bends. As mentally determined flow coefficients and exponents for
would bc cspcctcd, rhc Ilow incrcascs \\.it11 gap tliick- the leakage paths of' rhc French Firz Research Tower
Principles of Smoke Management

using regression analysis. Computer flow simulations For most of the applications of this book, flows are
using the exponential flow equation with experimentally represented and calculated by the orifice equation. Two
determined exponents were in good agreement with approaches to prescribing values for C and A are:

I
simulations using the orifice equation. It can be con-
cluded that use of the orifice equation for all flow paths 1- Use the cross-sectional area for A, and C is chosen to
in normally constructed buildings yields acceptable obtain the desired value of the CA product.
for pressurization smoke control design pur- 2- Arbitrarily choose C, and choose A to obtain the
poses, No similar study was conducted for smoke man-
desired value of the CA product.
agement systems without pressurization.
The first approach is used with orifice flow meters
Example 6.7 Gap Method for FIGThrough Door caps and many other flow paths for which the cross-sectional
A door has the dimensions shown in Figure 6.13. What is the area can readily be determined and for which C values
flow through the gaps between the door and the door frame at a are available. For flow coefficients of many items. read-
pressure difference of 0.15 in. H 2 0 (37.3 Pa)? Use the follow-
ers are referred to Idelcnik (1 986)-
ing properties of air at 70°F (21 "C):
The geometry of construction cracks in walls and
floors is complicated and for these cracks, measurement
of cross-sectional areas is impractical. The second
approach above is used for these cracks with the flow
For the slot at the door bottom: areas listed in Table 6.2 for C = 0.65. It is believed that
actual leakage values for walls and floors are primarily
a = 0.50 h'(0.0 127 m) Dh = 2a = l .OO in. (0.0254 m)
dependent on workmanship rather than construction
L=3ft(O.914m) x = 1.75 in. (0.0445 m) -
Ap = 0.15 in. H20 (37.3 Pa)
From Equation (6.1 S), NP = 28.2~106. Pressure Difference (Pa)
From Equation (6.18);NQ = 2950.
From Equation (6.19), V = 152 cfni (0.0718 m3/s) flow -
a
through slot at door bottom. .Ei 6 0 -
0

For slots at to^ and sides: - 0 . 6 -c


E
a = 0.12 in. (0.00305 m) D,,
= 2a = 0.24 in. Note: Gap thickness is 0.5 in (12.7 mm). 3
gap depth in flow direction is 2 in. (50.8 mm).- 0.3 2
(0.006 10 m) pressure difference is 0.04 in. H,O (10 Pa).
and air lemwrature is 70 'F(21
1 . 'Cl..
L= 17 ft(5.18 m) s = 2 3 7 in. (0.0602 m) 0l l I
0 0.1 0.2 o! S
Ap = 0.15 in. H 2 0 (37.3 Pa)
Pressure Difference (in H,O)
From Equation (6.1S), NP = 51000.
From Equation (6.17), NQ = 109.8. Figure 6.11 Co~npnrisonof val-ioz/sflo\c~jln~ctio,u
for
From Equation (6.19), V = 181 cfin (0.0855 m3/s) if the slot a 0.5 ill. (12.7 mm) wide goy.
had been straight.
From Figure 6.9, F , = 0.93 for a single-bend slot. Pressure Difference (Pa)

V = 181 ,(0.93) = 168 cfin (0.0792 m3/s) flow through slots at Exponential
top and sides.
Total flcw: 152 + 168 = 320 c h (0.15 l ni3/s) m
with n = 0.65 ..
0.015
Gap Method of Gmss
and Haberman,(1988)
F L O W AREAS AND C O E F F I C I E N T S 0.010

In the design of smoke control systems, airflow Y

paths must be identified and evaluated. Some leakage


0
paths are obvious, such as gaps around closed doors, 0 -0.1 0.2 0.3

open doors, elevator doors, windo\vs, and air transfer Pressure Difference (in H,O)
grilles. Construction cracks in building walls and floors Figure 6.12 Co~qmrisonof var-iousJ o I ~ . ~ ~ / I Ifolc ~ ~ o I ~ s
arc less obvious but no less important. n 0.I ill. (2.54 nrnl) wide gnp.
Chapter 6 -Principles of Smoke Management

0.62 in

Door
(b)
T
F;Y~725;7k
(C)

(a)
Figure 6.13 Dimensions for Example 6.7: (a) front of door;
(b) gap at top and sides, and (c) gap at bottom.

Table 6.2: materials, and, in some cases, the flow areas in particu-
Typical Leakage Areas of Walls a n d Floors of lar buildings may vary from the values listed. The sec-
s ' C = 0.65
Commercial ~ u i l d i n ~ for ond approach above also was used for the flow areas of
elevator doors listed in Table 6.3.
Area
' ~atio~ The gap method can be used to determine values of
Construction Element C and A for flow through gaps around doors. Tables 6.4
Exterior Building Walls and 6.5 provide this flow information using approaches
(includes construction cracks, 1 and 2, respectively. The flows ca!culated by these
cracks around windows and doors) tables are equivalent to each other, and users can select
the approach convenient to their application.
Additional data concerning building components
Stairwell Walls lTight 0.14 X lo4 are also provided in Chapter 25, "Ventilation and Infil-
(includes construction cracks but tration, of the 1997 ASHRAE Handbook-Fundamen-
not cracks around windows or doors ~~~~~g~ 0. I I 1 0-3
tals." The leakage flow rates of door assemblies can be
Loose 0.35 X Io-~ measured and rated at ambient temperature and elevated
Elevator Shaft Walls Tight 0.18 X IO-~ temperatures in accordance with UL 1784 (1990).
(includes construction cracks but
Average 0.84 X Io-~ For open stairwell doorways, Cresci (1973) found
not cracks around doors)
loose 3.18 X 1o -~ that stationary vortices form in the doorways a;~dthat
AIAj the resulting flow through those doorways was about
Floors half of that which would be expected without such vorti-
Tight3 0.66 X 1 o - ~
(includes construction cracks and ces. Using approach I , Table 6.6 lists flow areas of open
gaps around penetrations) Average 0.52 X 104 stairwell doorways for C = 0.35. Alternatively, approach
I loose3 1 0. I 7 X Io -~ 2 can be used where C = 0.65 and the flow area is about
l . Flow area ratios for C = 0.65 at 0.3 in. H20(75 Pa). half the cross-sectional area.
2. A I S flow area. A , is wall area, and A,is floor area. Values a f
area ratios based on pressurization measuremenls in buildings The determination of the flow area of a vent is not
by Tamura and aiilson(1966). Tamura and Shaw (1976a: always straightforward because the vent surface is usu-
1976b; 1978) and Sham et al. (1993).
3. Values exlrapolated from average floor lightness based on rangs ally covered by a louver and screen. Thus, the flow area
o f tiglitness o f other constructionelements.
is less than the vent area (vent height times width).
Because the slats in louvers are frequently slanted, cal-
culation of the flow area is further complicated.
Principles of Smoke Management

Table 6.1:
Typical Flow Areas for Elevator ~ o o r s with
' C = 0.65
Door Width Flow ~ r e a '

ft m Tightness ft2 m2

Closed Doors 3.0 0.914 Tight 0.34 0.032


Average 0.48 0.045
Loose 0.60 0.056

Tight
Average
Loose

Tight . 0.37 0.035


Average 0.53 0.049
Loose 0.66 0.06 1

Tight
Average
Loose

Tight
Average
Loose

Opened Doors 3.5 1.07 Avcrage 6.0 0.56


I. This table is for clc\-atordoors 7 fi (2.13 m) Ihiph. Flow areas t'or C = 0.65 at 0.1 in. H1O( 2 5 Pa).
2. Values of flow area based on pressurizxi~n mea~~~rcn~cn~s in building by Tamura and Shaw (1976b).

Table 6.2:
Flow Coefficients for Gaps Around ~ o o r s '
Gap Thickness Cap Thickness
Width at Top and Sides at Bottom Cross-Sectional Area Flow Coeflicient
in. m in. nlni in. nlm ft' m'

36 0.9 14 0.02 0.50s 0.25 6.36 0.090 0.0084 0.57


Chapter 6-Principles of Smoke Management

Table 6.3:
Flow Areas of Gaps Around ~ o o r s Using
' a Flow Coefficient of 0.65
Cap Thickness Cap Thickness
Width at Top and Sides at ~ o t t o m Flow ~ r e a '
in. m m. mm in. mm fi? m2
36 0.914 0.02 0.508 0.25 6.35 0.079 0.0073

l . This table is for doors 7 ft(2.13 m)l~lgh,1.75 in. (44.5 mm) thick. and with a door slop protruding 0.61 in. (15.7 111111) iron1 the liamc.
2. The flow area should not be confused tvith the cross-sectional area o f the gaps. The flow area is for uss in Ilic orilicc. cq~lstic~ii
will1 C = 0.65. The tlow

the gap method.

Table 6.4:
Areas and Flow Coefficients for Open Stairwell ~ o o r s '
Door Width Flow Area Flow Coefficient
Condition of Door in. m ft2 m' C
Propped Fully Open 36 0.9 14 21.0 1.95 0.35
Person in Doonvay2 36 0.914 10.5 0.78 0.35
Propped Fully Open 44 1.118 25.7 2.3s 0.35
Person in Doonvay' 44 1.118 12.5 1.19 0.35
1. This [able i s for a door hsighr o f 7 li (2.13 m).
2. The llow arca ir !alien as halfofthc arca ofthe fully opsn door. allon.ing for the door hcing only partly opcr. 3nd a person
hlocLing p m ol'ths dooncay.
Principles of Smoke Management ,

Example 6.8 Flow Area of Stair


Pari I. What is the leakage area between an interior stairwell and the building if the stairwell walls are of average tightness? The stair
well door is 7 ft (2. i3 m) by 3 ft (0.914 m), with a 0.08 in. (0.00203 m) gap on the sides and top and with a 0.25 in. (.Cl0635 m) gap a
the bottom. The stairwell is 8 ft (2.44 m) by 18 ft (5.49 m) with a floor to ceiling height of 10 ft (3.05 m).

For the stairwell walls:

Wall area is 2(8+18)10 = 520 ft2 (48.3 m3). From Table 6.2 for a stairwell wall of average tightness, the ratio of the leakage area thc
wall area is 0.1 I X 1 03. The leakage area of the wall is 0.1 1X 1o 3 (520) = 0.057 fr2 (0.0053 m2).

For the naps around the door:

From Table 6.5, the flow area of this door is 0.169 (.O 157 m2).

Total flow area:


Because these flow areas are in parallel (Chapter S), the total flow area is the sum of the individual areas: 0.057 + 0.169 = 0.226
(0.0210 m2) flow area between the stairwell and the building on a per floor basis.

Part 2. What would the flow area be if the construction tighmess were loose and the door undercur 0.75 in. (0.0 19 1 m)?

For the stairwell walls:

From Table 6.2 for a stairwell wall of loose tightness, the ratio of the leakage area to wall area is 0 . 3 5 10-'.
~ The leakage area of the
~ (520) = 0.182 ft2 (0.0169m*). .
wall is 0 . 3 5 10'

For the g a p around the door:

From Table 6.5, the flow area of this door is 0.320 ft2 (.0297 m2).

Total flow area:

The flow area between the stairwell and the building on a per floor basis is 0.182 + 0.320 = 0.502 f; (0.0166 m'). This is about double
the flow area of the first part, illustrating the extent to which flow areas can vary.

PRESSURE LOSS OF SHAFTS AND DUCTS

Straight Ducts and Shafts


where
The pressure losses due to friction in ducts and
shafts is represented by A = area of the duct or shaft, f;(m 2 );
P = perimeter of duct or shaft, fi (m).
Equation (6.20) is ths Darcy-Weisbach equation for
pressure loss in ducts and pipes.
For ducts and pipss, the friction factor can be
where
obtained from the traditional Moody diagram (Figure
& J= pressure loss in shaft or duct due to friction, in. 6.14), or it can be calculated from the Colebrook equa-
H20 (Pa); tion.
f = dimensionless friction factor of shaft or duct;
L = shaft or duct length, ft (m);
D,, = hydraulic diameter of shaft or duct, ft (m);
where
p = density of gas inside shaft or duct. 1b/ft3 (kdm3); E = roughness of the inside surface of the duct. fi (m);
U = average velocity inside shaft or duct, fpm (rids); R, = Reynolds number (sse Equation (6.7)).
Kyl, = 1 .G6 X 10-"(1.00). Some categories of duct roughness, E, are listed in
The hydraulic diameter of shaft or duct is Table 6.7. Equation (6.77) can be solved numerically f0r.f
Chapter 6-Principles of Smoke Management

Reynolds Number. R&

Figure 6.14 1\4oo41~


diagram fo1-jktiot7 fnciot- forflow in ducts andpipes.

Table 6.7:
Duct Roughness categories1
Roughness Roughness, e
Duct Material Category ft mm
Uncoated carbon steel, clean Smooth 0.000 1 0.03
PVC plastic pipe
Aluminurn
Galvanized steel, longitudinal seams, 50 in. (1200 mm) joints Medium Smooth 0.0003 0.09
Galvanized steel, continuously rolled, spiral seams, 120 in. (3000 m m )
joints
Galvanized steel, spiral scan with l, 2, and 3 ribs, 144 in. (3600 nim)
joints
Galvanized steel, longitudinal seams, 30 in. (760 mm)joints Average 0.0005 0.15
Fibrous glass duct, rigid Medium Rough 0.003 0.90
Fibrous glass duct liner, air side n-it11 facing material
Fibrous glass duct liner, air side spray coated Rough 0.0 1 3.0
Flexible duct. metallic
Flexible duct.
-.
all types of Ljbric and wire

by the Ne\vton Raphson r i ~ e t h o dFor


. ~ the,fitlll,~a!g/rflow
regio17(Figure 6.14), the friction factor can be calculated
from

6. As suggcstcd by Gcorze \\falton of tlic National In network computer flow tnodels (Chapter 8), it
Institute of Standards and Teclinolopy. the clliciency
oTthis numerical solution is signilicnrltly i ~ n p r o ~byd can be useful to use the equivalent orificc area for a duct
suhstitutinp.~= /-I1' and solving Ibr .v. or shaft. This is the area of an orifice that has the z n l c
Principles of Smoke Managemeqt

pressure loss as a section of duct. The flow through the The flow also can be expressed as

-
orifice.is
P = A,U

where
where As = cross sectional area of the duct or shaft, ft2(m2);
P = volumetric flow rate through the duct or shaft, cfm U = average velocity in the duct or shaft, @m, (mls).
3
(m /s); Considering Ap = Apj and combining Equations
C = dimensionless flow coefficient; (6.20), (6.24), and (6.25) results in
A, = equivalent area, ft2 (m2);
Ap = pressure difference across path, in. H20 (Pa);

p = density gas in duct or shaff lb/ft) (kg/m3);


K, = 776. (1.00).
Figures 6.15 to 6.19 show area ratios (A, /As) for
fully rough flow for the duct roughness categories listed
in Table 6.7.

Hydraulic Diameter, D, (m) Example 6.9 Equivalent Area of A Shaft


Calculate the equivalent area of a concrete shaft 8.6 ft (2.62 m)
I( ft 11
by 12 ft (3.66 ni) with a length equal to the floor height of 12

From Table 6.7, the roughness category of a concrete duct is


rough. This indicates that the AdAScanbe obtained from Figure
6.19. From this figure, AdAS = 12.5, and
2 2
A, = 12.5AS = 12.5(8.6 x 12) = 1290 ft (120m ) .
This large equivalent area is indicative of a duct section with a
"0 20 40 60 80 100
small pressure lossdue to friction.
Hydraulic Diameter, Dh(R)
Figure 6.15 Area ratiofor-sniooth ducts.

Hydraulic Diameter, D,, (m)

.
_ . .
. ..:....
~..
. I .
.
.
0 20 40 60 80 100
Hydraulic Diameter. D,, (ft)
Chapter 6 -Principles of Smoke Management

Hydraulic Diameter, D,, (m) p = density o f gas inside stairwell, l b / P (kg/m3);

U = average velocity inside stairwell, @m ( d s ) ;

K,, = 1.66 X 104 (1.00).

A relationship for the equivalent orifice area for the


stairwell can be obtained in the same manner as was
done for the duct.

Hydraulic Diameter, Dh(ft)


Values of K, are listed in Table 6.8.
Figure 6.18 Area ratio for medium rough ducts.

Calculate the equivalent area of a stairwell 8 ft (2.44 m) by


18 ft (5.49 m) with a length equal to the floor height of 12 ft
(3.66 m). There are no people in the stairs and the treads are
Hydraulic Diameter. D, (m)

l. FromTable6.8, the friction factor, K,, is32,andA,.JA, = 0.28.

The equivalent area is A, = 0.28 (8 X 18) = 40 ?f (3.72 m').

2. An alternate approach is below.


From Equation (6.2 l),
D =-
4A = ------
4(S X 'g' = 11.1 ft (3.38 m).
h P 2(8+ 18)
From Equation (6.28),

Hydraulic Diameter, D, (R)


Figure 6.19 Area ratio for rough ducts.

SYMMETRY
The concept of symmetry can be used to simplify
Stairwells flon networks, thereby simplifying analysis. While
Tamura and Shaw (1976b) showed that the pressure advances in network modeling (Chapter 8) have reduced
losses due to friction in stainvells is similar to that of the need for such simplifications, symmetry can srill be
shafts, and this pressure loss is useful. Figure 6.20 illustrates the floor plan of a multi-
story building that can be divided in half by a plane of
symmetry. Flow areas on one side o f this plane are squal
to corresponding areas on the other side. If the flon s and
where pressures are solved for one side, those on the other side
AQ= pressure loss in stairwell due to friction, in. H20 are also known. To apply symmetry to a building, ei.ery
floor must be such that it can be divided in the same
(Pal;
manner by the plane of symmetry. If wind effecrs are
K,,, = dimensionless friction factor of stairwell;
included in the analysis, the wind direction must be par-
L = height of section of stairwell, ft (m); allel to the plane of symmstry. It is not necessac that
D,, = hydraulic diameter of stairwell, ft (m); the building bc geometrically synlnletric, as shown In
principles of Smoke Management '

Table 6.8:
Typical Friction Factors and Area Ratios for ~taiwells'
Stairwell Type Floor Height Friction AdAs
ft m Tread people 2
Factors, K, Per Floor
Conventional 12 3.6 Open None 29 0.30
Conventional 12 3.6 Closed None 32 - 0.28
Conventional 8.5 2.6 Open None 61 0.24
Conventional 8.5 2.6 Open High 104 0.19
Conventional 8.5 2.6 Closed None 71 0.22
Conventional 8.5 2.6 Closed High 170 0.15
Scissor 14 4.3 Closed . None 15 0.32
I. Based on data from Tamura and Shaw (1976a) and Achakj~and Tamura (1988).
2. "High" is high density of 0.18 person/ft2(2.0 persodm2).

F LWPressure
I
Knob, 1 Side

k.4
Figure 6.2 1 Diagrari7 of f01,ces 0 1 7 n cloor- in a p/-essur--

where
F = total door openins force, Ib (N);
Figurc 6.20 Building floor- plan illzafr~arirz,o
sjx1n7efr?.
A(. = moment of the door closer and other friction, Ib fi
concepf.
PJm);
Fff = door width, ft (m);
Figure 6.20; it must be synimetric only with respect to
flow. A = door area, f? (ni2);
= pressure difference across the door, in. H 2 0 (Pa);
DOOR-OPENING FORCES n = distance froni the doorknob to the knob side of the
The door-opening forces due to the pressure differ- door. fi (m);
ences produced by a smoke control system must be con- Kd = 5.20 (1.OO).
sidered in any design. Unreasonably high door-opening The moment to overcome the door closer and fric-
forces can'result in occupants having difficulty or being tion consists of all moments about the hinge due to the
unable to open doors to refuge areas or escape routes. door closer or friction forces such as friction in the
This is addressed in the next section. The following hinges or rubbing of the door against the door frame.
analysis is for a door hinged at the edge with a door The force at the knob needed to overcome hinge friction
knob, as shown in Figure 6.2 1. Users need to adapt the is about 0.5 to 2 Ib (2.3 to 9 N). Some poorly fitted doors
analysis to fit other conditions, such as pi\ ots inset froni rub against the frames, resulting in extremely high door-
tlie edge. opening forces. Ideally, such poor workmanship will be
The forces on a door in a smokc control system are identified and corrected during building commissioning.
illustrated in Figure 6.21, and tlie sum of the nionients The component fo,rce, F,, at the knob to overcome the
about the hinge is door closer and other friction is
Chapter 6 - Principles of Smoke Management

This can be substituted into Equation (6.29) to obtain d = distance from the doorknob to the knob side of the
door, ft (m);
Kd = 5.20 (1 .OO).
This relation assumes that the door-opening force is
and this can be solved for the pressure difference as applied at the knob. This force to overcome the door
closer is usually greater than 3 Ib (13 N) and, in some
2 ( W - d ) (F- F,) cases, can be as large as 20 Ib (90 N). Caution should be
@ = KdWA exercised in evaluating the door closer force because the
force produced by the closer when the door is closing is
where often different from the force required to overcome the
F = total door opening force, Ib (N); closer when opening the door. Many door closers
require less force in the initial portions of the opening
F, = force to overcome the door closer and other fric- cycle than that required to bring the door to the full open
tion, Ib (N); position. For this discussion, the force to overcome the
door closer and other friction is that force at the very
W = door width, ft (m); beginning of the opening process. The pressure differ-
A = door area, (m2); ence component of the door-opening force can be deter-
mined from Figure 6.22 for a door 7 ft (2.13 m) high
Ap = pressure difference across the door, in. H20 (Pa); with a knob located 3 in. (0.076 m) from the edge.

Pressure Difference (in H20)

Figure 6.22 Doo,--oj>enit7gforces due lo pressure diflerence.


Principles of Smoke Management ;

Example 6.11 ~oor-openingForce. females is only 2 0 Ib (91 N). The five percentile push
force of healthy male adults is 4 5 Ib (200 N). These
1. What is the door-opening force for a door 7 ft by 3 ft
forces are gradually applied, and a 'jerk" method o f
(2.13 m by 0.9 1 m) subject to a pressure difference of 0.25
suddenly applying the force results in a peak force o f
in. H 2 0 (62 Pa)? The force to overcome the door closer and
175 Ib (780 N). These push forces are one handed, and
other friction is 10 Ib (44 N), and the knob is 3 in from the the subjects are not leaning forward; the push force
door edge. increases to 146 Ib (652 N) for a forward leaning two-
W= 3 ft(2.13 m) d = 0.25 ft (0.076 m) handed push.
Ap = 0.25 in. H 2 0 (62 Pa) F, = lOlb(44N) The Life Safe@ Code (NFPA 2000c) states that the
force required to open a n y door in a means o f egress
A = 3 X 7 = 21 ft2 (1.95 m2) Kd = 5.2 (1.00) shall not exceed 30 Ib (133 N). Based o n the data of
From Equation (6.3 l), the door-opening force is 25 Ib ( l l 0 Read and Shipp, it seems that this 30-lb (133 N ) limiting
N). Alternately, Figure 6.22 gives 15 Ib (66 N), and adding force is appropriate for most occupancies, but care
this to the door closer force gives 25 Ib ( l l0 N). should be exercised when building occupants are likely
2. What is the pressure difference across a door that has a 30 Ib to have low levels of pushing and pulling strength. For a
(133 N) door-opening force and a frictional and door closer 30-lb (133 N) limitation on door-opening force with a
force of 5 Ib (22 N)? The door is the same size as in part 1 side-hinged door with a singe knob, the maximum
above. +
allowable pressure differences are listed in Table 6.1 I .
The fire effect of buoyancy of "hot" smoke can be
F=3O lb(l33 N) incorporated in the selection of the minimum design
F,= 5 Ib (22 N) pressure difference. Unless otherwise stated, the mini-
From Equation (6.32), Ap is 0.42 in. HzO (104 Pa). mum design pressure differences used in this manual
incorporate buoyancy and are based on the idealization
DESIGN PRESSURE DIFFERENCES that the mass flo\v through the leakage paths is constant
It is appropriate to consider both a maximum and a for the duration of the fire. A method for handling vari-
minimum allowable pressure difference across a barrier able mass flon. through these paths is presented in
o f a smoke control system. The values discussed in this Chapter 9.
section are based on the recommendations in NFPA 92A The smoke control system should be designed to
(NFPA 2000a). The maximum allowable pressure dif- maintain this minimum value under likely conditions of
ference should be a value that does not result in exces- stack effect and wind and when there is no building fire
sive door-opening forces. The force that a particular (such as during acceptance o r routine testing). NFPA
person can exert to open a door depends on that person's 92A (NFPA 20COa) suggests minimum design pressure
strength, the location of the knob, the coefficient of fric- differences, and these values are listed in Table 6.12.
tion between floor and shoe, and whether the door The values for nonsprinklered spaces are those that will
requires a push or a pull. not be overcome by the buoyancy forces of hot gases.
Read and Shipp (1979) studied door-opening These values for sprinklered buildings were calculated
forces, and they present strength data for the very young from the equation for buoyancy of combustion gases
(age S to 6 years) and the elderly (age 6 0 to 75 years). (Chapter 5) for a gas temperature of 1700°F (927OC),
From Tables 6.9 and 6.10, the five perce~tilepushing for a neutral plane located at a height of two-thirds of
force for the very young females is only 6.5 Ib (29 N), the ceiling height below the ceiling and with a safety
and the five percentile pushing force for the elderly factor of 0.03 in. H 2 0 (7.5 Pa).

Table 6.9:
Functional S t r e n g t h Values f o r A g e G r o u p 5 t o 6 y e a r s '
Fifth
Mean, Maximum, minimum, Percentile,
Function . Gender Ib (N) Ib (N) Ib (N) lb (NI
Push M 20 (90) 26 (155) 7.2 (32) 8. l (36)-
F I6 (73) 2s (l 26) 10 (46) 6.5 (29)
Pull M 27 (120) 41 (184) 18 (82) 17 (77)
F l9 (86) 32 (141) l l (48) 8.7 (39)
I. Note: :\dapkd ( i o l i ~llcad and Shipp (1979). Sul$cctr i w d only one I l n ~ i d .Suddsnly applied ''jerk" ~ u s h e and
s pulls o r two-handed forward lesning
puilics would h a w rcsultcd ill frcnwr Ibrccs.
Chapter 6-Principles of Smoke Management

Table 6.10:
Functional Strength Values for Age Group 60 to 75 years1
Fifth
Mean, Maximum, Minimum, Percentile,
Function Gender 111 (NI (N) Ib 0 lb 0
Push M 53 (237) 121 (540) 2 1 (92) 23 (101)

Pull
F 45 (201) 9 1 (407) 22 (100) 21 (95)
I. Note: Adapted fror. Read and Shipp (1979). Subjects used only one hand. Suddenly applieda'jerk" pushes and pulls or hvo-handed forward-leaning
pushes would have resulted in gearer forces.

Table 6.11:
Maximum Allowable Pressure Difference Across Doors, in. H 2 0 ( ~ a ) l

Door Closer
Force, Door Width, in. (m)

S (35.6) 0.41 (102.) 0.37 (92.1) 0.34 (84.5) 0.3 1 (77. l) 0.28 (69.7)
10 (14.5) 0.37 (92. l) 0.31 (81.5) 0.30 (74.6) 0.28 (69.7) 0.26 (64.7)
I2 (53.4) 0.34 (84.5) 0.30 (74.6) 0.27 (67.2) 0.25 (62.2) 0.23 (57.2)
14 (62.3) 0.30 (74.6) 0.27 (67.2) 0.24 (59.7) 0.22 (45.7) 0.21 (52.2)
I. i\:olc: Adnpted from NFPA (2000al. Total door opening force is 30 lb (133 N). and the door l~eiehris 7 fi (2.13 m).

Table 6.12:
Suggested Minimum Pressure Design ~ifference'
Design
Building Ceiling Pressure
~~~e~ Height, ~ifference?
ft (m) in. HzO (Pa)
AS Any 0.05 (12.4)
NS 9 (2.7) 0.10 (24.9)
NS l 5 (4.6) 0.1; (34.8)

I. Adnpted from NFPA (2000n). For d a i g n purposes, a sn~okscontrol systeni should maintain these mini-
mum pressurs diflerences under likely conditions of stack elfect or wind.
2. AS for sprinklered and NS for nonsprinklsrsd.
3. TIis prsssurc dilference mcasurcd ber\\ven [hc smoke zone 2nd adjacent spnces while the afictsd areas
arc in the smokc control mods.
Principles of Smoke ~ a n a ~ e & n t

Pressure differences produced by smoke control WEATHER DATA


systems tend to fluctuate due to the wind, fan pulsations,
The indoor to outdoor temperature difference has
doors opening, doors closing, and other factors. Short-
an impact on building airflows and pressures. For some
term deviations from the suggested minimum design analyses, wind data may be needed. The 1997 ASHRAE
pressure difference may not have a serious effect on the Handbook-Fundamentals, Chapter 26, "Climatic
protection provided by a smoke control system. There is Design Information," provides weather data for loca-
no clear cut allowable value of this deviation. It depends tions throughout the world. NFPA 92A and NFPA 92B
on tightness of doors, tightness of construction, toxicity suggest that the 99.6% heating dry-bulb (DB) tempera-
of smoke, airflow rates, and on the volumes of spaces. ture and the 0.4% cooling DB temperature be used as
Intermittent deviations up to 50% of the suggested mini- the winter and summer design conditions. NFPA 92A
mum design pressure difference are considered tolerable and NFPA 92B also suggest that the 1% extreme wind
in most cases. velocity be used as the design condition.
CHAPTER 7

Air Moving Equipment and Systems

he National Board of Fire Underwriters exam- SMACNA 1990, 1987; Handbook of HVAC Design

T ined the NFPA fire data from January 1936 to


April 1938 to determine the extent of the smoke
hazards due to heating, ventilating, and air-con-
1990).
The simplest systenls consist of a fan in a housing,
such as a roof-mounted atrium exhaust fan. Most sys-
ditioning (HVAC) systenls (NBFU 1939). Of 25 fires tems are more complicated, with ductwork and some of
recorded, 19 had conlbustion of parts of the air-moving the following components: supply air outlets, return air
system. Ducts, duct linings, and filters bunied. In five inlets, fresh air intakes, humidifiers, filters, heating and
cases of no fire in the HVAC system, smoke was distrib- cooling coils, preheat coils, and dampers. Ductwork is
uted by the system. This report has had a nlajor impact constmcted of a variety of materials, including steel,
,on tlie materials and consiruction of modern HVAC sys- aluminum, concrete, and masonry. Duchvork of fiber-
tems, as is apparent from examination of current codes glass, gypsum board, and fabrics is used with some
and standards. The report recommended that HVAC sys- restrictions. Discussions of fans and dampers are pro-
tems be shut d o h during fire situations to prevent them vided later. The air-moving systems that are discussed
from spreading smoke and supplying combustion air to later are primarily intended for maintaining comfort
the fire. System shutdown became the standard response conditions. Exhaust systems for toilets, laboratories, and
to fire. However, operation of the HVAC system In a kitchens are not discussed, but they are generally less
s G k e controi mode has become a common alternative complicated and use many of the same conlponents.
in recent years, as discussed in later chapters. -
--

The information in this chapter is provided as a HVAC LAYOUT


broad and general background on air-moving systems. In large buildings, the heating and cooling loads
The material was selected to aid in tlie understanding of often vary considerably from one location to another.
the smoke control systems discussed in later chapters. Heat is transferred to or from the spaces near the exte-
This information should help tire protection engineers, rior walls depending on outdoor weather conditions.
firefighters, and code oRicials to commur.icate with Solar radiation affects each of the exterior zones differ-
HVAC designers and to recognize and understand ently. It is common to divide a building into four perim-
HVAC equipment. Because energy conservation is a
major concern, energy efficiency of systems and equip-
-
eter zones and a core zone as shown in Figure 7. l a. The
heating and cooling capacities cd the perimeter zones
ment is addressed in this chapter. This chapter is not an agdesigned to accommodate outside temperatures and
exhaustive treatment of the firc safety requirements of solar loads. Because of the heat produced by occupants,
HVAC systems, and the design or such systems should hghting, and equipment. the core zones often need cool-
be done by experienced professionals. Many publica- ing even in the winter.
tions provide more detailed information about these sys- The perimeter zones can be conditioned by a vari-
tems and equipment (for example, ASHRAE 2000a; ety of means, including fan coil units, air conditioners,
and heat pumps. Generally, fan coil units are supplied
Chapter 7-Air Moving Equipment and Systems

with hot and cold water to allow both heating and cool-
Floor
ing. Often, air conditioners and heat pumps are located 29
through-the-wall. Both fan coil units and through-the-
wall equipment can receive ventilation air directly from
the outside or from a ducted ventilation systeni. In large 25
commerc~albuildings, ventilation air is needed to con-
(a) Perimeter and Core Zones
trol the odors due to cooking, smoking, perspiration, and
other processes.7 The perimeter zones may be served by 20
ducted forced air systems, and the core zone is usually
served by such forced air systems. Some types of forced
air systems are capable of satisfying a wide range of 15
needs simultaneously and are used to serve both perime- (b) Ducied Supply and Plenum Return
ter and core zones. The different types of forced air sys-
tems are discussed later.
~istributionon a floor is often through ducts R e q - i
Duct
- /
I Mechanical Penlhouse 1
Dun
located above a suspended ceiling. Return air is often - 1 1 1
pulled through the plenum space above the ceiling, as
shown in Figure 7.1 b. The return may be ducted above
the ceiling as well. Mechanical equipment of a forced
air system may be located on each floor (Figure 7. I b), (C ) Central System in Penlhouse (d) MuRipleMechanical Floors
on one floor (Figure 7.lc), or on several floors (Figure
7. l d). Figure 7.1 Some HVAC a1-1-or7ge11zazts.
The arrangements above are but a few of those pos-
sible. There may be several forced air systems on each
floor. There may be several units located in a penthouse, system economics but little effect on airflow. The forced
each serving its own vertical portion of the building. air systems discussed in the following sections can be
Sometimes, several air systems are used and the areas completely built in the field, factory-fabricated subsec-
served are selected on the basis of having similar heat- tions can be field assembled, or completely factory fab-
ing and cooling demands. These demands depend on ricated systems can be installed.
occupancy, the presence of heat-releasing equipment,
electrical lighting levels, and heat transferred to or from Constant Volume, Single Zone
the outside. For a complicated building (such as hospi-
tals, laboratories, and hotels), the duct systems can be Figure 7.2 is a representation of a single-fan, con-
intertnrined to such a level that considerable study is stant volume system. The term "constant volume" is
needed to understand which systems serve which areas. used in the HVAC industry to indicate that the system
pcoduces a constant or nearly constant volumetric flow
FORCED AIR SYSTEMS rate of air. This system is used in residences and some
small commercial applications. In this esample. return
Four common types of forced air systems are air from the living quarters is drawn i n at one location,
flows through filter, fan, and coils, and is distributed
constant volume, single-zone systems,
back to the residence. This system does not ha\.e the
constant volume systems with terminal reheat,
capability of providing fresh outside air. These systems
variable air volume (VAV) systems, and are intended for applications where there is sufficient
dual-duct systems. natural air leakage through cracks i n walls and around
windows and doors for odor control.
There are numerous variations on these systems.
Generally; the heat source for heating coils is hot water. Single-zone systems are so called because they
However, other sources, such as steam or electrical serve only one HVAC control zone, For esample: a resi-
resistance heating, are possible. Cooling coils can be dential system is controlled by a thermostat to maintain
supplied with chilled water or with refrigerant. The the temperature in the l i v i n ~quarters. Generally, the res-
source of heating or cooling has signiticant effects on idential system has a two-position control system,
allowing only "on" and "olt" operation to maintain tem-
pel-ature and humidity conditions.
7. In small b u i l d i n g and residences, such odor col]-
trol is achie\.cd by ~iaturallyoccurring air inl?l[ration Frequently in commercial buildings. constant vol-
through construclion gaps and cracks. ume systems have two fans and are capable ol'pro\.iding
Principles of Smoke Management

ventilation air as illustrated in Figure 7.3a. The return Constant Volume, Terminal Reheat
fan permits lower supply fan speeds and quieter opera-
The constant volume, terminal reheat system is
tion. The return air fan provides positive return and intended to serve many HVAC control zones, as illus-
exhaust from the conditioned space. During cold trated in Figure 7.4. This system can have an econo-
weather, many large commercial buildings have so mizer as can all the following systems. The supply fan
much heat generated by equipment and people that cool- provides cooled air to each zone, where it is reheated to
ing is required. To save energy, cold outside air can be the temperature required to maintain comfort conditions
used for this cooling. The system of dampers and con- within that zone. The airflow rate through the system is
trols that maximizes the use of outdoor air for cooling is constant, and control is achieved by varying the heat
called an economizer. input to each reheat coil. This system is capable of
achieving a high level of temperature and humidity con-
For systems with an economizer, the humidifier and trol for each zone. However, terminal reheat is not very
cooling coils need to be protected from freezing. Thus, energy efficient.
the preheat coil is used to temper the outside air to 38°F
to 45°F (3°C to 7°C) when the outside air is below freez- Variable Air Volume
ing. The preheat coil and reheat coil can be used when The variable air volume system varies the supply
heating is required. The reheat coil used with the cool- rate of conditioned air to the space to maintain comfort
ing coil allows precise humidity control. conditions. Additionally, the temperature of the supply
the supply fan and fan have the same a'' may be varied. There are many a ~ ~ r o a c h efor s
achieving variable flow. In the system depicted in Fig-
flow rate, the system is said to be in a "balanced condi-
ure 7.5, flow to each zone is controlled by a damper or
tion." Many designers size the exhaust fan at about 80%
other flow control device in the VAV unit. This unit is
or- 90% o W flow of th fan provide 'light sometimes referred to as the VAV terminal box. Gener-
building
-- - pressurization (about 0-05 in. H70 [ l 2 PA1). ally, the supply and return fans are capable of variable
The intent is to prevent normal infiltration of airborne flow rates alld are controlled by the static pressure sen-
dirt, odors, and pollen from the outside into the building. sors. Some of the approaches that are used to achieve
Figure 7.3b is a line diagram illustrating the same sys- variable flow rates through fans are variable pitch inlet
tem as that of Figure 7.3a. In the rest of this chapter, line
diagrams will be used to illustrate systems. The compo- Exhaust
Louver Damper
nents of the following systems are the same as those \ / Relum Fan
Exhaust
shown in Figure 7.3a and 7.3b. Air +- C C

(a) Diagram showing duct thickness

Exhaust
,,n
/ Return Fan
Exhaust
AirC

Return Air Humidifier


/ ,Cooling Coil
/
Outside
Bu113ing
Spaces
Outside Filter Preheat
\Reheat
Air Damper . , Coil Coil

(b) Diagram with line representation of duct

Figurc 7.2 Si17gIe-jbr7sysletr,. Figure 7.3 Corisfanfvolume, single-zone sysfem.


Chapter 7-Air Moving Equipment and Systems

Exhaust Reheat Coil -


1

Exhaust
Air

Humidifier
Retum Air
Damper I -U

Outsid
Air

Supply Fan
Outside Air Filter Preheat
Damper Coil

Figure 7.4 Constant volrtnze systern with terminal reheat.

Figure 7.5 Vasiable-air-volunze(VAC.') system


vanes, discharge dampers, variable pitch motor sheaves, systems have been used in multi-room buildings to
eddy current couplings, variable speed DC motors, and accommodate highly variable heating and cooling loads.
variable frequency AC motor speed controllers. As with A dual-duct system can be constant volume or VAV.
constant voiunie systems, VAV systems can be designed Operating costs of VAV dual-duct systems are less than
to provide building pressurization. those of the constant volume systems.

Dual Duct FANS


The dual-duct system co~lditionsall the air at a cen- ANSIIASHRAE Standard 149 (ASHRAE 2000b)
tral location and distributes it to the conditioned spaces establishes methods o f laboratory testing and documen-
t h r o ~ ~ gtwo
h supply ducts. One duct conveys cold air, tation for fans used for smoke exhaust.
and thc other warm air (Figure 7.6). A mixing box sup- There are two general fan classifications-xentrifu-
plying each zone combines the two airstreams in the gal and axial. Figure 7.7 illustrates the basic parts of a
proper proportions to achieve comfort conditions. These centrifi~galfan. Flow within a centrifugal fan is prima-
Principles of Smoke Management

rily in a radial directiod to the impeller. Figure 7.8 illus- these fans have single-width impeller blades and
trates the basic parts of an axial fan. Flow within an straightening vanes to direct air parallel to the shaft.
axial fan is parallel to the shaft. Tubular centrifugal fans are primarily used for low-pres-
sure HVAC applications, particularly as return air fans.
Centrifugal Fans These fans have significant space savings over other
Centrifugal fans used in the HVAC industry are centrifugal fans.
generally classified by impeller design as forward - Backward impeller rotation is a common problem
, ., curved, backward curved, and airfoil (Figure 7.9). with systems with centrifugal fans. It is important to
Forward-curved centrifugal fans rotate at a rela-
note that backward rotation of centrihgal fans results in
tively low speed and are generally used to produce high reduced flow in the normal direction. This problem is
flow rates and low static pressures. Backward-curved often not recognized because of the mistaken belief that
fans rotate at about twice the speed of forward-curved backward rotation of these fans results in backn.ard
fans and have a higher
- efficiency. The higher rotational flow. The normal direction of airflow and the direction
speed requires more expensive fan construction. Both of rotation of centrifugal fans is shown on Figure 7.7.
forward- and backward-curved impeller blades are sin-
gle width, stamped from sheet metal. Airfoil fans are
Axial Fans
basically backward-curved fans with blades of varying
thickness to improve fan efficiency. Airfoil blades are The common types of axial fans used in buildings
designed using the same airfoil technology that is used are propeller fans, tubeaxial fans, and vaneaxial fans
to design airplane wings. (Figure 7.11).
Required performance and economics are major For propeller fans, a variety of impeller designs are
factors in the selection of a fan type for a particular employed with the intent of achieving high flow rates at
application. However, the following generalizations can low pressures. The irnpellers of propeller fans have two
or more blades and are usually of inexpensive construc-
be made concerning application. Fonvard-curved fans
tion (for example, these blades are often stamped from
are used for low-pressure HVAC applications, including
sheet metal). Propeller fans are used for low-pressure,
c l ?
high flow rate applications, including kitchen exhaust,
equipment. Airfoii and backward-curved fans are used toilet exhaust, stairwell pressurization, and space venti-
for general purpose HVAC applications, and airfoil fans lation.
are usually limited to large systems where the enernv Tubeaxial fans have a higher efficiency and can
savings are significant. operate at higher pressures than propeller fans. Vaneas-
~ubuGcentrifugalfans (Figure 7.10) are an excep- ial fans have still higher et'ficiencies and operating pres-
tion to the classification by impeller type. Generally, sures. Blades of tubeasial and vaneaxial fans can be

Exhaust Return

Exhau
Air

Outsid
Air

Damper Coil

Figure 7.6 D~~nl-dz/cl


sysletil.
Chapter 7-Air Moving Equipment and Systems

Direclionof Bla

Rim Impeller

Figure 7.7 Centrifugal fan components.

Guide Vane Inlet Bell,

Head
Figure 7.8 Axial fan coniponetits.

Forward-Curved Backward-Curved Airfoil


24 to 64 Blades 10 to 16 Blades -10 to 16 Blades
About 65% Efficiency About 75% Efficiency About 80% Efficiency

Figure 7.9 Itq~eller-


types for centrifugal fans.
-. . .

Principles of Smoke Management

single thickness or airfoil design. Adjustable pitch Multi-blade dampers operated by electric motors or
blades' are used on some vaneaxial fans to obtain high pneumatic pistons to vary the flow rate are called con-
efficiency. Both tubeaxial and vaneaxial fans have the trol dampers. Dampers used to resist the passage of fire
advantages of straight-through flow and compact instal- are called fire dampers, and these can be multi-blade
lation. Tubeaxial fans are used for low- to medium-pres- dampers (Figure 7.13) or curtain dampers (Figure 7.14).
sure HVAC applications, and vaneaxial fans are used for Dampers used to resist the passage of smoke are called
low- to high-pressure HVAC applications. smnke dampers, and these can also be either multi-blade
Unlike centrifugal fans, backward rotation of an or curtain. Combination dampers can be used to balance
axial fan normally results in backward flow. This back- airflow, control airflow, resist the passage of fire, and
ward flow is at a reduced airflow rate. More information resist the passage of smoke.
about both centrifugal and axial fans is provided by Jor-
gensen (1 953), ASHRAE (2000~)and AMCA (1 990a, Fire Dampers
1987). Generally, multi-blade fire dampers are held open
by a fusible link and are spring loaded. In a fire situa-
DAMPERS tion, hot gases cause the link to come apart, allowing a
In air-moving systems, dampers are used to spring to slam the blades shut. In place of fusible links,
some manufacturers use other heat responsive devices.
balance airflow, In the United States, fire dampers are usually con-
control airflow, structed and labeled in accordance with standard UL
resist the passage of fire, or 555 (UL 1999). Prasad (1995) tested the ability of fire
resist the passage of smoke. dampers to close under conditions of still air, airflow,
ambient temperature, and elevated temperature. In
Balancing dampers are used in supply ducts and response to the findings of Prasad's findings, the 1999
return ducts to adjust the airflow to the design values. version of UL 555 includes closure tests for static sys-
These dampers can be of simple construction (Figure tem (with no airflow) and dynamic systems (with air-
7.12) or of multi-blade construction (Figure 7.13). flow). The dynamic tests can be at ambient temperature,
250°F (120°C) or 350°F (180°C).

Centrifugal
Impeller
J r . < Propeller -
Propeller Roof 1 I Wall Fan

Exhaust Fan Propeller Roof


Fan Emciency
About 25% Efficiency
About 25% Efficiency

Straightening Vanes
Tubeaxial Fan
About 55% Efficiency
- am- Vaneaxial Fan
About 70°h Efficiency

Figure 7.10 fitbular cent/-ifugalfan. Figure 7.11 Types of axial fans.

. Duct

Arm' c
Splitter Damper Round Damper Rectangular Damper

Figurc 7.12 Dotuper types ~tscdjorDalot~ci/~g.


Chapter 7-Aii Moving Equipment and Systems

Channel Frame
\
Shafl

Shafl
7 /
Extension Blade

Angle
stop \
J
l
Opposed Action Damper Section
Note: Horizontal (floor) type
,Channel Frame curtain dampers must have
spring closure, but vertical
(wall) type curtain dampers
can have either spring or
gravity closure.
--

Shafl
7=
Extension

Parallel Action Damper Seclion


Figure 7.14 Cut-~ait~fii-e
cinirpet:

Ill, and the maximum leakaze rarts are listed in Table


7.1.
Thc particular class of damper specified -should
- --
be
--
selected based on the requirements of the __-._..
application.
Smoke Dampers - - ~-

For example, the dampers i n the supply and return ducts


In the United States, smoke dampers are usually
can have some leakage v-ithour adversely afi'scring-
constructed and classified for leakage in accordance Y

smoke control system performance. Thus, a de&a-cr


with standard UL 555s (LL 1999a, 1999b). The stan-
dard includes construction requil-ements and tests for might select class I I or I l l smoke dampers for such an
2__

cycling, temperature degradation, dust loading espo- application. However, a designer might choose clzss 1
sure, salt-spray exposure. air leakage, and operation dampers for applications that require a - very right
under airflow. These dampers are classified as I, 11, or damper, such as a return air damper (Figure 7.3).
-
Table 7.1:
L e a k a g e Classifications f o r S m o k e D a m p e r s
(Adapted from UL 5558 [UL 19991)
At 1.0 in. H 2 0 (250 Pa) At 4.0 in. H 2 0 (1000 Pa)
Classification cfmtft? m3s-'m-' cim/it2 ,,,Zs-l n,-3
I 4 0.020. S 0.04 I
II 10 0.05 1 20 0.102
II 40 0.203 SO 0.406
At 8 in. H20 (2000 Pa) (3000 Pa)
A t 1.2 in. 1{20

I II 0.056 14 0.07 1
II 2s 0.147 7
>. 3
- 0 . L 7s
II 112 0.569 130 0.7 1 1
CHAPTER 8

Computer Modeling

moke management applications of computer mod- BUILDING AIR AND

S eling have increased dramatically in the last few


decades. Many computer models have been devel-
oped for fire science and fire protection engineering
SMOKE n o w MODELS
Computer programs that simulate building airflow
can be useful for analysis of pressurization smoke con-
applications by a number of organizations. Many of trol systems. Airflow programs that can simulate con-
these applications are very useful for smoke manage-
taminates or smoke concentrations throughout a
ment design. The Smoke 1\4anagen7ent Progt-am CD
building can be useful tools for hazard analysis. The
that accompanies this book contains a number of com-
CONTAM program that is on the CD accompanying the
puter applications that can be useful for smoke manage-
book has air and contaminate flow capabilities. and it
nlent (Table 8.1). Most of these programs were
also is used for some of the examples of this book.
developed at the National Institute of Standards and
A discussion of the earlier models provides a back-
Technology (NIST). The NIST computer applications
ground for CONTAM. All of the airflow programs also
are in the public domain, \\ hich means that they are not
calculate the pressures throughout the building. The
covered by copyright protection and they can be freely
National Research Council of Canada (NRCC) devel-
copied and used by anyone.
oped airflow programs (Sander 1974; Sander and
The computer applications on the CD can be classi-
Tamura 1973). The ASCOS program (Klote 1982) sim-
fied as building airflow models, zone fire models, detec-
tor actuation models, CFD models, elevator evacuation ulated airflow and was specifically developed as a
model, 2nd collections of engineering tools. This chap- research too1 for analysis of.smoke control systems.
ter is a g e ~ e r a ldiscussion of these classes of models ASCOS was extensively used for smoke control design
except for CFD models and the elevator evacuation for much of the 1980s and 1990s. Yoshida et al. (1979);
model. The CFD models are dealt with in Chapter 17. Butcher et al. (1969); Barrett and Locklin (1969): Evers
The elevator evacuation model, ELVAC, is discussed in and Waterhouse (1978); and Wakamatsu (1977) devel-
Appendix C. oped programs that also simulate smoke concentrat;ms.
The treatment in this chapter is of a general nature.
For details and equations of particular models, readers Network Models
should see the documentation for the model. The equa- These models represent a building by a net\vork of
tions in this chapter are only intended to describe some spaces or nodes, each at a specific pressure and tempera-
of the more important concepts of computer modeling, ture. The stairwells and other shafts can be niodeled by a
and these equations are not ~ntendedto be used for cal- vertical series of spaces-ne for each floor. Air flows
culations. Accordingly, units are not given for variables through leakage paths from regions of high pressure to
of this chapter. However, all of these equations are valid regions of low pressure. These leakage paths are doors
for S1 units or any other I~omogcneousunit system (see and windows that may be opened or closed. Leakage
Appendix A) can also occur through partitions, floors, and exterior
Chapter 8- Computer Modeling

Table 8.1:
innthe
Computer Software ~ ~ ~ l i c a t i o s Smoke ~anagement
progmms CD'
Software Name Comments
Classification
Building Air and CONTAM Airflow analysis including contaminants
Smoke Flow
Zone Fire ASET-C Available Safe Egress Time - C++ Language Version is part ofthe ASMET package
of engineering tools.
CFAST Consolidated Fire and Smoke Transpott Model
LAVENT Model for the Prediction of Detector Activation and Gas Temperature in the Presence
ofa Smoke Layer
JET Model for the Prediction of Detector Activation and Gas Temperature in the Presence
3f a Smoke Layer
AZONE Atrium zone fire model includes plugholir,g anddelayed smoke exhaust fan activation
(Cha~ter141
Detector Actuation DETACT-QS Detector Actuation - Quasi Steady
DETACT-T2 Detector Actuation - Time squared
CFAST Detector actuation is one feature ofthis zone fire model
LAVENT Detector actuation is one feature of this zone fire model
JET Detector actuation is one feature of this zone fire model
Elevator Evacua- ELVAC Elevator Evacuation
tion
Collection 01' Engi- ASMET Atria Smoke Management Engineering Tools
neering Tools
FAST A collection of equations and fire protection engineering tools including CFAST
I. Note: All progmnis liskd in this table are public donlain sofiwnre developed by NIST, except for AZONE. which was devslopcd by John H. Klote. Inc.

walls and roofs. The airflow through a leakage path is a 3. The net air supplied by the air-handling system or by
function of the pressure difference across the leakage the pressurization system is assumed to be constant
path. and independent of building pressure.
In this model, air from outside the building can be 4. The outside air temperature is assumed to be con-
introduced by a pressurization system into any level of a stant.
shaft or even into other building spaces. This allows S. The barometric pressure at yound level is assumed
simulation of stainvell pressurization, elevator shaft to be standard atmospheric pressure (1 0 1325 Pa).
pressurization, stairwell vestibule pressurization, and The results of the program are not very sensitive to
pressurization of any other building space. In addition, changes in atmospheric pressure. For altitudes consider-
any building space can be exhausted. This allows analy- ably different from sea level, a more accurate value of
sis of zoned smoke control systems where the fire zone barometric pressure can be substituted by changing a
is exhausted and other zones are pressurized. The pres- statement in the subroutine INPUT and one in the sub-
sures throughout the building and steady floii7 rates routine CORR.
through all the flow paths are obtained by solving the The following is a simple overview of a nehvork
airflow network, including the driving forces, such as model. This overview only considers one flow path
wind, the pressurization system, and inside-to-outside between any two nodes, but mar,y network models
teniperature difference. allow a number of flow paths between the same two
The assu~nptionsof the ASCOS model are similar points. The mass flow in a path between two nodes can
to other network nlodels, and these assumptions are: be represented as

1. Each space is considered to be at one specific pres-


sure and one specific temperature.
where
2. Thc flows and leakagc paths are assumed to occur at
midheight o f e a c l ~Icvcl. d, - . =
'1
mass tlow from node i to nod? j,
Principles of Smoke Management

functional relationships appropriate for a path The solutior?to this set of equations is the pressures @,,P*,
between nodes i and j, ... pN) for which all the right-hand side is zero. From these
pressure difference fiom node i to node j. pressures, all of the pressure differences and flows through-
out the building can be calculated.
A number of functional relationships for flow are
discussed in Chapter 6. Possibly the orifice equation and Because of the difficulty in solving these equations,
the exponential equation are the most ccmmon such the numerical routines of many of the above models
functions. A function can also be used to represent the were slow and would sometimes fail to converge to a
I ~,
flow of a fan, which is an exception in that fan flow is solution. Such convergence failures seemed to happen
more often with large and complicated networks.
from a node of lower pressure to a node of higher pres-
sure. An ASHRAE-funded research project (Wray and
The pressure difference can be expressed as Yuill 1993) evaluated several algorithms to find the
most appropriate one for analysis of smoke control sys-
'J = p;-p.+p;g(Z;-Zj)
Ap.. J (8.2) tems. They selected the AIRNET routine developed by
Walton (1989) as the best algorithm based on successful
where convergence, computational speed, and use of computer
pi = pressure at node i, memory. None.of the routines of this study take advan-
tage of the repetitive nature of building flow networks,
fi = pressure at node j, so data entry for these routines is difficult and time con-
pi = density gas at node i, suming.
Z, = elevation of node i,
CONTAM Model
5 = elevation of node j,
g = acceleration of gravity. There are two versions of this model: CONTAM96
(Walton 1997) for use with the DOS operating system
For steady flow, conservation of mass at node i can and CONTAMW (Dols et al. 2000) for use with the
be stated as the sum of the mass flows leaving node i are Windows 95, 98, or NT operating systems. The techni-
zero. In equation form, this is cal aspects of these models are the same, and they are
,it referred to in this section simply as CONTAM. A simple
user guide for getting started with CONTAM is pro-
C/;J(A~rJ)= 0 (8.3) vided in Appendix D.
/ = I
CONTAM uses an improved version of the AlR-
where M is the number of flow paths between node i and NET algorithm that was selected as the best algorithm in
other spaces. The mass flows entering node i have negative the study mentioned above. Further, CONTAM has a
values. Writing the conservation of mass equations for each method of graphical data input that reduces both learn-
node in the building results in ing time and the likelihood of input errors.
CONTAM was developed for indoor air quality
f l ~ ( & ~+fl2(Aplr)
l) + ... +fl.\i(A~lh.)
= 0 1

applications, but it has been extensively used for smoke


f2,(&2l) +f22(Ap22)+ ... +J;v(@2,v) = 0 , management applications. This model simulates con-
(8.4)
taminant flow, as well as airflow throughout a building.
For smoke management applications, the contaminants
can be the products of combustion.
Substituting Equation (8.2) into the ab0L.e set of The CONTAM documentation considers the model
equations yields to be a multi~rrnemodel where the zones would be
rooms or floors of a shaft. The CONTAM model does
not include an energy equation, and so the temperature
of zones needs to be designated by the user.
CONTAM is like the network models above except
that it treats pressures and flow paths in a more general
way.
where Fiis the functional relationship for flows into node i. The pressure in room i is considered hydrostatic,
Equation (8.5) is a set of sin~ultaneousnonlinear equations. and it can be represented as
Chapter 8-Computer Modeling

0
Pressure
(a) Sketch of a room fire
Figure 8.1 Bidirectional flow through an opening
between two zones.

P i = pio- Pig' (8-6)


where
pi = pressure in zone i at elevation z,
N I
C

-
Airflow
C
pi, = pressure at the floor (z= 0) of zone i,
Fire
g = acceleration of gravity, \ \ \ \ \ \ \ \ \ \ \ L
pi = density of air in zone i, (b) Zone model idealization of a room fire
z = elevation above the floor of zone i. Figure 8.2 Rooni .fire(a) sketch m d (b) zone model
The representation of pressure allows for simula- ideafiinrion.
tion of bidirectional flows between two zones connected
by a flow path. Such bidirectional flow can occur when
two zones at different temperatures are connected by a room fire, hot gases rise above the fire, forming a smoke
flow path (Figure 8.1). This is not relevant for smoke plume. As the plunle rises, it entrains air from the room
control systems that rely on pressurization, but it could so that the diameter and mass flow rate of the plume
be significant for simulations of smoke transport that increase with elevation. Accordingly, the plume temper-
does not include pressurization. For flow paths specified ature decreases with elevation. The fire gases from the
at midheight of the floor, airflows, and pressures calcu- plume flow up to the ceiling and f o m ~a hot stratified
lated by CONTAM are the same as, those of ASCOS layer under the ceiling. The hot gases can flow through
within the limits of numerical convergence. openings in walls to othcr spaces, and such flow is
referred to as a doo~jer.The doorjet is similar to a plume
ZONE FIRE MODELS in that air is entrained and the mass flow rate and cross-
Zone fire models have proven utility for many fire sectional area of the jet increase with elevation, and the
protection applications, including hazard analysis. The jet temperature decreases with elevation. Ths difference
concepts behind this type of fire model are the basis of is that the doorjet is tlowing through an opening in a
most of the engineering approaches to smoke manage- wall. Figure 8.21 is a sketch of a room fire.
ment design for atria.
The concept of zone modeling is an idealization of
Early zone fire models include the Harvard Code
the room fire conditions, as illustrated in Figure 8.2b.
(Mitler and Emmons l98 l), ASET (Cooper 1985), the
BR1 Model (Tanaka 1983), and CCFM (Cooper and For this idealization, the temperature of rhs hot upper
Forney. 1990). The University of Maryland has made layer of the room is unifonn and the temperature of the
modifications to CCFM specifically for atrium smoke lower layer of this room is also uniform. The height of
management design (Milke and Mower 1994). The the discontinuity bttween thcse layers is the same
models ASET-C, CFAST, LAVENT, and JET are dis- everywhere. This discontinuity i s called the smoke layer
cussed below. inre&ce. In the idealized modcl, at an infinitesimal dis-
Because zone models were originally developed for tance above the intcrfice, the temperature and contami-
room fires, this discussion will start with room fires. In a nant concentrations are thosc of the smoke laver. At an
Principles of Smoke Management

infinitesimal distance below the interface, the tempera- m, =mass in the upper layer,
ture and contaminant concentrations are those of the
lower layer. However, in real fires, there is a gradual m,,.,, =mass flow rate into the upper layer,
transition rather than an interface.
r ; i , out =mass flow rate out of the upper layer.
The dynamic effects on pressure are considered
negligible, so that the pressures are treated as hydro- The mass flow rates in Equation (8.7) depend on
static. Other properties are considered uniform for each the specific computer model. ASET-B only simulates
layer. Algebraic equations are used to calculate the mass the plume flow into the upper layer with no allowance
flows due to plumes and doorjets. for mass flow out of the upper layer. For this model,
Many zone computer models allow exhaust from ril,. in is the mass flow of the plume and ril,, is zero.
the upper layer, and this capability is essential for simu- For more complex multi-room zone models, nz,, is the
lation of atrium smoke exhaust systems. Many of the
computer zone models estimate heat transfer by meth-
ods ranging from a simple allowance as a fraction of the
sum of all mass flows into the upper layer (plume, door-
jet from another room, HVAC flow, etc.) and ril,, ..,
is
:he sum of all mass flows out of the upper layer (doorjet
heat released by the fire to complicated sin~ulation, from another room, HVAC flow, etc.).
including the effects of conduction, convection, and
radiation. Zone model application to an atrium fire is The conservation of energy equation is also known
illustrated in Figures 8.3a and 8.3b. as the first law of thermodynamics. Because potential
Rockett et al. (1987) compared measured data with energy and kinetic energy are relatively small, they are
data computed by the Harvard Code for a series of fires neglected, and the energy equation for the upper layer is
at the NIST Annex. The temperatures for one of those
fires are shown in Figure 8.4. It can be obseryed that the
temperature for the bum room is well represented by the
zone fire model idealization. However, the temperatures
in the corridor and lobby are only very roughly approxi-
mated by the zone fire model. This supports the opinion
that zone model predictions are less realistic for spaces
away from the fire room.
For more general information about zone fire mod-
els, readers are referred to Karlsson and Quintiere (2000),
Friedman (1992), Jones (19S3), Mitler and Rockett
(1986), and Mitler (1984) and Quintiere (l989a).

Mathematical Description
Many of the early zone fire models were quasi-
steady systems of algebraic equations, and the atrium I
zone fire model, AZONE, discussed in Chapter 14, is (a) Sketch of a n atrium fire
based on this approach. Other models are differential
equation-based, and this section is intended to provide
some idea.of the theory behind these differential equa-
tion-based.models.
The upper and lower layers of a one-room zone fire
model form control volumes, as illustrated in Figure 8.5.
In general, the approach to zone modeling is to write the
conservation equations for the upper and lower layers. Plume
ASET-B is an exception in that equations are only writ-
ten for the upper layer.
The equation of conservation of mass for the upper
layer is
(b) Zone model idealization of atrium fire

Figure 8.3 All-izrm snzoke e.~hatrst (a) sketch a ~ i dfb)


zone model idealizutioti.
Chapter 8-Computer Modeliig

Temperature Rise CC)

F Corridor
i A

E
'8

6 -
25
I
50
I
75
1 .
100
1 .

Note: 8 indicates thermocouple tree. I Burn Room

Temperature Profiles:
+Measured
- - - Calculated from
Zone Fire Model
0 50 100 150 200
Temperature Rise CF)

Temperatxe Rise)C"( Temperature Rise ("C)

Temperature Rise ('F) Temperature Rise CFj

Figure 8.4 ~lfecrswedtriid corrrpli~ecl/et,rpera/u~epru/iles d i e /oa 100 kW/;/-e 200 sccoi~ci.~ ipition (rrdapted
j a i n RocA-et/ et al. [I 98 71).

Control Volume Boundaries

..............................
I
I
I
l Upper Layer
I
I
I
I
I L, hUJeu
I Opening
I
in Room
TJ h,, e,
Lower Layer

Figure 8.5 C'orlrinl ~ ~ ) h i i i m , /zq)pei.


u i aiitl loi\vi- 1cg~er.suj'ci
.viirplc inoirl zoilejiir iiroclcl.
Principles of Smoke Management

where
Q = heat transferred to the upper layer,

W = work done by the smoke layer on the surround- For an ideal gas, Cp, C, R, and y are constants
ings, (Appendix A). The time derivative of Equation (8.10) is

h , i,, = enthalpy of the mass flow into the upper layer,

h, , = enthalpy of the mass flow out of the upper


layer, Substituting this into Equation (8.9) results in
ell = internal energy of the upper layer.

The heat transfer term, Q,, should not be confused


with the heat release rate of a fire. The heat transfer term
is for thermal energy that flows into the upper layer due Combining Equations ( 8 4 , (8.1 l), (8.12), (8.13):
to a temperature difference. Because the upper layer is and (8.16) yields
relatively hot, the term Q,, is generally negative (for
example, heat conduction and thermal radiation from
hot smoke to the walls).
Work is the product of a force, F, acting through a
displace~ent,I (in differential form, work is dW = Fdl).
The displacement for the upper layer is the moving
smoke interface, which is also the surface of the control Equation (8.17) is a form of the conservatix of
volume. The force is the product of absolute.pressure at energy equation for the upper layer. The following con-
the interface and the area of this surface ( F = pAJ. The servation of mass and energy equations for the lower
work term is layer can be developed in a similar manner:

Most zone fire models consider the gases to behave and


as ideal gases, and an ideal gas is one that has the fol-
lowing equation of state:
PI.' = I I I R T (S. 10)
where
p = absolute pressure, The conservation equations can be rearranged as
V = volumeofgas,
m = mass of gas,
R = gas constant,
T = absolute temperature of gas.
The enthalpy of an ideal gas can be expressed as

where C,, is the constant pressure specific heat. The inter-


nal energy of an ideal gas can be expressed as

where C,.is the constant volume spzcific heat. The gas con-
stant of an ideal gas is

The ratio ofspecitic heats, y. is E,, = net energy release rate for the upper layer,
Chapter 8-Compuier Modeling

8, = net energy release rate for the lower layer, controlling vents and sprinklers in compartments
bounded by walls, draft curtains, or combinations of
V = room volume (V = V, + 5). walls and draft curtains. The JET model incorporates the
Equations (8.20) through (8.23) were developed by conductivity factor to account for the effects of heat
Jones (1983), and readers should see that reference for a conduction from the sprinkler head.
detailed description of the net energy release rate terms.
Information about solution of such systems of differen- DETECTOR ACTUATION MODELS
tial equations can be found in many texts on numerical
Fire-driven ceiling jets can have a significant
methods (for example, Burden et al. 198 1).
impact on the performance of ceiling-mounted detection
Equations for plume mass flow and temperature are hardware. The plume rises above the fire. As it impinges
discussed in Chapter 13. Flow through doors and other on the ceiling, the plume gases turn and form a rela-
openings in walls or partitions are calculated in much tively high temperature, high velocity, turbulent ceiling
the same way as horizontal flow through an opening, jet, which flows radially outward (Figure 8.6). The tem-
which is treated in Chapter 13, except that the pressures perature and velocity of the ceiling jet are described by
are complicated by the possibility of both air and smoke Albert's (1972) correlations.
on both sides of the opening.
The detector actuation model, DETACT-QS, calcu-
lates the actuation time of thermal devices below uncon-
ASET-C Model
fined ceilings (Evans and Stroup.1986). The unconfined
ASET-C (Available Safe Egress Time-C Language) ceiling assumption is appropriate for large spaces, such
is a program for calculating the temperature and posi- as open plan ofice spaces, but it does not account for
tion of the hot smoke layer in a single room without the effects of the smoke layer on the ceiling jet in a con-
smoke flow to other spaces. ASET-C is one of the sim- fined space.
plest and easy to run zone fire models. As stated above, Figure 8.7 is a sketch of a ceiling jet in a room with
ASET-C only simulates the upper or smoke layer. The a smoke layer. For a fire below the smoke layer, the
lower layer is considered to remain smoke free and at plume penetrates thc smoke interface, continues to rise
ambient temperature. ASET-C is an adaptation of the toward the ceiling, and entrains smoke from the smoke
ASET-B (Walton 1985), and it is one of the engineering
tools in the ASMET package. Documentation is pro-
vided in Appendix E.
.....

CFAST Model
CFAST is a multi-room zone model that predicts Detector
the effect of a specified fire on temperatures, various gas
concentrations, and smoke layer heights in a multi-com-
partment structure. CFAST has many features, including
forced ventilation, detector activation, and conductive
heat transfer. CFAST is the primary engineering tool in
the FAST package (Peacock et al. 2000). For a technical (a) Sketch of ceiling jet and detector
description of CFAST see Jones et al. (2000).

LAVENT Model Ceiling


LAVENT (Davis and Cooper 1989) is a single room
zone fire model that predicts plume centerline tempera-
ture, ceiling jet temperature, and ceiling jet velocity.
LAVENT can determine activation times of fusible links
controlling vents and sprinklers in compartments
bounded by walls, draft curtains, or combinations of
walls and draft curtains. Note: The ceiling jet flows
radially from the point wher
the plume impinges on the
JET Model ceiling.
Like LAVEKT, JET (Davis 1999) is a s~ngleroom (b) Idealized ceiling jet flow
7one fire model that prcdicts plume centerline tempera-
ture, ceiling jet temperature, and ceiling jet velocity. Figure 8.6 Ceilitig j e ~rmder a Jar ceilirrg. (0) skerch
SET also can determine activation times of fusible links mid (b) idealizcd,/lo~c!.
Principles of Smoke Management,

Geiector
Ceiling Jet
smoke Layer

I 1
Figure 8.7 Sketch of roomfire showing ceiling jet and
smoke layei:

layer. When the ceiling jet reaches the walls, the flow The detector models account for the thermal lag of
turns downward. The effects of the smoke layer on ;he detectors by use of the response time index (RTI), as
ceiling jet'are taken into account in the zone fire models discussed in Chapter 2. The RTI approach is appropriate
CFAST and JET. for the fusible links of sprinklers and smoke and heat
I I
vents.
CHAPTER 9

Hazard Analysis

ost smoke management systems provide HAZARD ANALYSIS CONCEPT

M smoke protection by minimizing people's


contact with smoke or by keeping smoke
completely away from people. As the name implies, ten-
For a particular fire, smoke moves through the
building. As people evacuate the building, they are
exposed to this smoke, which has the potential to impair
ability systems provide smoke protection by maintain- vision and cause incapacitation or fatality. A hazard
ing tenable conditions. Tenability svstems allow smoke analysis can be used to calculate such smoke flow and
contact, but the systems are designed such that the tem- the consequences for building occupants. A hazard anal-
peratures and concentrations of combustion products are ysis can consist of one or a number of fire scenarios.
limited. Hazard analysis consists of the following compo-
An analysis of these systems is called a hazard anal- nents: (I) fire scenario, (2) smoke transport, (3) people
ysis, in that the level of hazard to life is evaluated. Tech- movement, and (4) tenability.
nological advances have made hazard analyses feasible, Fire Scenario. As stated in Chapter 2, a fire sce-
and tenability systems based on such analyses have nario can be thought of as the outline of events and con-
gained a level of acceptance in the last decade. NIST ditions that are critical to determining the outcome of
developed a group of computer programs, HAZARD I alternative designs. In addition to the fire location and
(Peacock et al. 199l), for hazard analysis in spaces con- heat release rate, 0 , the fire scenario includes the status
sisting of a relatively few rooms, such as residences. of the doors, the HVAC systems, the smoke manage-
Bukowski and Spetzler (1992) used HAZARD I to ment system, and other systems. For details about
reconstruct the fire at the Happyland club in the Bronx, design fires, see Chapter 2.
New York, that killed 87 persons. Klote et al. (1992) Species ( 0 2 , N2, CO, CO2, etc.) generation can be
extended hazard analysis to large multi-story buildings included in the fire scenario. The scenario may also
for the study of staging areas for persons with mobility include specifics about the fuel, ignition of multiple fuel
limitations. packages, and the effect of fire suppression. The selec-
Hazard analysis is a powerful fire protection tool tion of the fire scenario can be based on professional
that has application beyond smoke management. This judgement, analysis of historical fire data, or cods
tool can be used to evaluate alternative building materi- requirements.
als and furnishings. The most common smoke manage- Smoke Transport. Smoke can flolv far from a fire
ment applications are compartmentation and atrium and threaten life. The major driving forces that cause
protection. Compartmentation systems can be with or smoke movement are naturally occurring stack effect,
without pressurization. The atrium systems can have buoyancy of combustion gases, expansion of combus-
any con~binationof smoke filling, smoke exhaust, or tion gases, the wind effect, fan-powered ventilation sys-
natural smoke venting. Jt is also possible to use hazard tems, and elevator piston effect. These driving forces are
analysis to evaluate the efect of component failure. discussed in Chapter 5 .
Chapter 9-Hazard Analysis

As discussed in Chapter 3, smoke consists of the Eliminate Heat Exposure Calculation. Detailed
airborne particulates and gases evolved when a material heat exposure calculations are not needed if the maxi-
undergoes pyrolysis or combustion, together with the mum temperature is relatively low. For exposure times,
quantity of air that is entrained or otherwise mixed into Figure 3.7 can be used to make such an estimate.
the mass. The evolved gases are part of the species men- Eliminate Radiation Calculation. If exposure to
tioned above. Generally, when smoke flows away from heat does not cause incapacitation, exposure to thermal
a fire, the concentrations of particulates and evolved radiation wiil not cause incapacitation. Exposure to heat
gases decrease. Conlputer models for smoke transport consists of direct bodily contact with hot smoke, and
analysis are discussed later. exposure to thermal radiation consists of receiving the
People Movement. People movement in fire situa- radiant flux from hot smoke. If the smoke temperature is
tions is complicated. Some people will fight the fire. insufficient for heat exposure to be an issue, the smoke
Others move against the flow of evacuating people in an temperature is also insufficient for thermal radiation
attempt to find or rescue loved ones. Scme computer- exposure to be an issue.
based evacuation models are capable of simulating the Eliminate Toxic Gas Exposure Calculation. For
movement of individual people. As people move many hazard analyses, visibility is the controlling tena-
through the building, they are exposed to smoke. This bility condition. The method described in Chapter 3,
time-integrated exposure can be used in tenability calcu-
"Tenability and Perfect Dilution," can be used to deter-
lations. For iilfonnatioii about calculation of building
mine if exposure to toxic gases is of concern for particu-
evacuation 'time and a discussion of computer-based
lar tenability criteria. (This same method can also be
evacuation models, see Chapter 4.
used to help determine if heat exposure is of concern for
In many applications, consideration is made for particular tenability criteria.) Alternatively, toxic gas
people who are immobilized due to an accident or phys- exposures can be estimated by a simple method, such as
ical disability. Such a person would need to wait to be the FED approach, to denionstrate that exposure to toxic
rescued, and the wait could exceed the-time needed for
gases is of concern.
evacuation of the rest of the building.
Tenability. Tenability calculations estimate the SMOKE TRANSPORT
hazard to life of a scenario. Tenability calculations
address one or more of the following: exposure to toxic For niost applications, smoke transport calculations
gases, exposure to heat, exposure to thernlal radiation, are done by computer. A wide range of computer mod-
and visibility through smoke. For calculation of expo- els can be used, including ( l ) zone fire models, (2) net-
sures and visibility, see Chapter 3. work flow models, and (3) computational fluid dynamic
The exposures are time-integrated doses of toxic (CFD) models. The choice of the model depends on the
gases, heat, and thermal radiation. These doses can be specific application. Smoke transport can also be evalu-
based on the smoke concentrations at several locations ated by physical modeling (Chapter 15).
as people move out of the building. Alternatively, the Use of the zone tire model FAST and the network
doses can be based on the smoke concentrations at one flow model CONTAM for hazard analysis is discussed
locatio~~
while an inlmobilized person waits for rescue. later. For general information about zone fire models
and building air and network tlow model:, see Chapter
Level of Complexity S. For general information about CFD n~odeling,sze
The level of complexity of a hazard analysis Chapter lb.
depends on the particular application. Analysis of sowe Many of theses ~nodelscan simulate production and
of these co~ilponentscan consist of straightfonvard rea- transport orspecific gases (02, N2, CO1COZ,NO2, HCI.
soning, and others require detailed c.~lculations. A feu. HCN, HBr, etc.), but sinlulation of specific gases is not
ways that a hazard analysis can be simplified are dis- generally necessary for design applications. Generation
cussed below. of the specific gases requires detailed knowledge of rhe
Elinii~~ate Evacuation Simulation. For esposures fuel, \\.hicl1 is usually not available in desig~iapplica-
based on proteering immobilized people, the need for a tions. The approach presented in this chapter is one of
detailed c\:acuarion simulation can be eliminated, pro- many possible zeneral methods of calculating tcnabilir?..
vided thar the design wiring is sullicie~ltlylong. This is The mass of fuel consumed by the fire is
because the esposurc iimc considered Ibr a \vaitirlg per-
son \\.ould be rnuch greater than that fbr a person e\.acu-
aring the buildi~lg. While a detailed eixuation
sinlulation may 1101 be ~iecdcd.an csrimatc of' building
evaciraricm time may srill be dcsircd.
Principles of Smoke Management

mass of fuel consumed, lb (g); to 02, N2,CO2, and CO. This allows simulation of the
synergistic effects of CO production and O2 depletion
total heat release rate Btuls (kW); on toxicity of CO. Considering this and that the CO is
chemical heat of combustion BtuAb (kJ/kg); the dominant toxic gas in building fires, limiting the
1 (1000). gases to 02, N2,CO2, and CO is appropriate for many
applications. For information about CO production in
The heat release rate, Q , and the mass of fuel con- fires, see Table 2.1.
sumed, riz , are entered into the computer smoke trans-
For any instant, the visibility can be calculated from
port model, which calculates the concentrations of
material burned, C, at every location and each time
interval in the simulation.

TENABILITY CALCULATIONS where


The following is one of many approaches to tena- Si = visibility at the end of interval i, ft (m);
bility calculations, and more extensive information can
be found in Chapter 3. Tenability analysis addresses vis- K = proportionalityconstant (8 for illuminated signs,
ibility, gas exposure, and heat exposure. and 2 for non-illuminated signs);
The mass concentration of material burned, Ci, can 6, = mass optical density, &lb (m21g);
be obtained from zone fire models. The fractional effec-
tive dose.(FED) can be used to obtain an approximation Ci = concentration ofmaterial burned in interval i, lb/ft3
of the effects of exposure to toxic gases. (s/m3>.
Generally, contact with dry air of temperatures
greater than 250°F (1 2 1°C) can be expected to result in
skin barns. Also, contact with dry air at a temperatuse
less than approximately 250°F (121°C) leads to hyper-
thern~ia.For hyperthermia, heat exposure can be esti-
where mated from
FED = fractional effective dose at the end of interval i
(dimensionless);
ci = concentration of material burned at interval i, Ibl
fi? (gm3);
At = time interval, rnin (min); where
LCt,, = lethal exposure dose from test data, lb ft-' min F,,/,= total cumulative dose (dimensionless);
(g min). At = time interval, minutes;
This equation is for unifornl time intervals, as cal- ?;. = temperature of air in interval i, "F ("C);
culated by computer models, and it evaluates the FED C, = 5.670 (5.185);
for the exposure time at the end of interval i (expos~ve
time is nAt). An FED greater than or equal to one indi- C2 = 0.0152 (0.0273).
cates fatality. The concentration, C;, is the denisty of Incapacitation due to heat exposure would be
materials that started as fuel that have accumulated at a expect<d..fo,r.F"4""?"ig~~:Fr..I~6
.*.:,>.... . rr,.o:f e,qital.to ,one.-
location during the interval i. The concentration has ,,,,: ,,..,:.*.w;..!r!~~.II~...
.. , - g ->:,
I ,
;.,:,
. .: .,,,,,. ..
If contact with gases does 'not r e d t in incaoacita-
units of mass of the material burned per unit volunle. tion due to 11qdt-exposure,thermal radiation from those
The lethal exposure dose, LCtjO, is the product of the
g&.es &&id not result in incapacitation for the same
LCjO and the exposure time. The LCjO is the concentra-
exposure time. Generally, exposure to thermal radiation
tion of airborne con~bustionproducts that is lethal to
50% of the subjects exposed for a specified time. An is not- ,,a! jsye ,[or <atja, app,licatip,n>. For situations
FED of 0.5 can be considered a rough indication of inca- where thermal radiation is significant, see Chapter 3.
pacitation.
When a more accurate evaluation of tosic effects is
desired, the methods discussed in Chapter 3 can be used.
EXAMPLE HAZARD ANALYSIS
i
Hazard analysis has a wide range of applications,
The fractional incapacitating dose (Fl,\;) method is gen- and this example was selected to illustrate some of the
erally considered to be more accurate. Unlike the FED capabilities and limitations of this technology. The
method. calculation of F,,-,! requires calculation of spe- example is a six-story hotel (Figure 9.1) with a lire in
cies concentration. The gases considcsed can be lilnited one ol' the guest rooms on the ground 1100s. Both tlic
Chapter9- Hazard Analysis

VEND HK -
R01 R02 R03 R04 R05 R06 R07 R08 R09 R10
-
-
(b) Plan for Floors 2 6
d
STG ..
- +
ELEV
-
Note: Floor 7 has mechanical penthouse (not shown).

. . . . . .

.. f c o R HK Rm
R01 R02 R03 R04 R05 OFF ;W , Fire R07 R08 R09 R10
- - .-JCSL -
.:ELEV Room
(a) Ground Floor Plan Symbols:
Rn Room Number n STG Stoiage Room HK House Keeping N
COR Corridor
ELEV Elevator
-....-C Office Window
W
-
OFF
-
Ofke
Window
Door
VEND Vending
t
Table 9.1:
Roughly Steady Temperatures for First Floor Rooms of Hotel Fire Example Based o n FAST Simulation
Temperature
Location ~oum' OF "C
Fire Room R06 1700 927
Corridor Section Open to Fire Room COR l 470 243
Corridor Section Open to the Oflice COR 280 138
West Section of Corridor COR2 180 82
1. See F~gure9.1 for location ofroorns.

window and door of the fire room are opened. The open The flow areas used in the CONTAM simulation
window is large enough to allow combustion air to sup- are listed in Table 9.2. For this simulation, the integrity
port a fully developed 5 MW fire (Chapter ?). of the door is considered to be maintained, and warping
CONTAM was used for the smoke transport analy- of the door is considered. When subjected to elevated
sis. The temperature of most of the building locations is temperatures, some doors experience burn-through at
73°F (23OC), and the outside temperature is 20°F (- the edges or may warp to increase the flow area of the
6.7OC). Becausc CONTAM does not include encrgy gaps around he, door edges. Con~bustibledoors are sus-
equations, the temperature of the tire room and that of ceptible to burn-through, and warping is more pro-
nounced with steel doors.
the other spaces open to the burn room needs to bc spcc-
itied. FAST was used to calculate these tc~i~pcmturcs The parar-ncters for the tenability calculations are
(Table 9. l). listed in Table 9.3. The valuesof AH,,,, K, and 6,,,are con-
Principles of Smoke Management

Table 9.2:
Building Flow rea as' Used for Hazard Analysis of Hotel Fire Example
Flow Path Path Name ft?lf? (0, m2/m2\

Hoistway Wall WALL-EL


Other Walls WALL
Floor FLOOR
-

Other Openings: ft2 m2


Double Ext. DOO? DOOR-DB 0.30 0.028
Single DOO? DOOR-SG 0.17 0.016
Open Single DOO? DOOR-WP 21.0 1.95
Elevator Doo? DOOR-E 0.60 0.056
Office window' WIN-OF 20.0 1.86
Open window2 WINLOP 36.0 3.33
~lev%orVent VENT E 4.0 0.37
Equivalent Areas of Shafts: ft2
Elevator Shaft FLOOR-EL 770 72
Stairwells FLOOR-SW 32 3.0
I. Flow areas arc for a loose or relatively leaky building. and for further flow areas. see Chapter 6. Flow coefficient, C. o f 0.65 was used for all tlow paths
except Tor the open door, which was 0.35.
2. These leakage paths are distribuwd uniformly over the height ofthe door or window.

Table 9.3:
Parameters Used for Hazard Analysis of Hotel Fire Example

Steady heat release rate, 0 5000 Btds (4220 kW)


Chemical heat of con~bustion,MC,, 10,700 Btu/lb (25,000 kJlkg)
Proportionality constant, K 8 (for illuminated signs)
Mass optical density, 4, 1600 ft2/lb (0.33 m21g)
Lethal exposure dose for a fully developed fire, LCI 50 0.033 ~b~3 min (530 g m-3 min)
Exposure time 30 minutes

servative for most applications, and the value of LCf is exposure to 180°F (82°C) for about 15 minutes, and a
applicable for fully developed fires. For appropriate person could withstand exposure to 280°F (138°C) for
parameters for other fires, see Chapter 3. An esposure about 4 minutes, after which they would suffer skin
time of 3 0 minutes was used for this example. For a spe- bums. Tolerance to higher temperatures would be much
cific application, the exposure time would depend on a less. From this, it can be seen that the probability o f
number of factors. If this were a failure analysis for a fatality due to heat exposure is high for many spaces on
design study, evacuation time might be taken as 15 to 30 the first floor during this fire.
minutes, including the time before people ssx moving. The results of the tenability calculations are listed
For a fire reconstruction, the exposure time might be in Table 9.4. Graphic presentation of tenability results
taken from the estimates of the people movement based can be useful. The results of the toxicity calculations are
on the fire event time line developed as part of the fire shown graphically in Figure 9.2, and it can be seen that
investigation. the FED exceeds one for many spaces on the first floor.
For these spaces; the probability of fatality is very high.
The tenability calculations did not explicitly The visibility is shown in Figure 9.3, and i t can be seen
include heat exposure. Because the temperatures in this that the visibility is less than 25 ft (7.6 m) throughout
exalnple are consrant, the effects of temperature can be the ground floor. On all floors, the visibility in the stairs
obtained from Figure 3.7. A person could withstand an is less than 25 ft (7.6 m).
Chapter 9 -Hazard Analysis

Table 9.4: Table 9.5:


Summary of Tenability Calculations for Hotel Fire Summary of Tenability Calculations for Hotel Fire
Example Example-Continued
Time (minutes) to Visibility of FED for Time (minutes) to Visibility of FED for
30 min 30 min
Floor Room 200 ft (61.0 m) 25 ft (7.6 m) Exposure Floor Room 200 ft (61.0 m) 25 ft (7.6 m) Exposure
G COR 1.9 2.0 2.497 R13 19.5 . NA 0.008
G CORl R14
G COR2 R15
G ELEV R16
G HK R1 7
G OFF R18
G R01 R19
G R02 R20
G R03 R2 1
G R04 R22
G R05 R23
G -- - R06 STG
G . . R07 SW1
G R08 SW2
G R09 VEND
G R10 CO R
G R11 ELEV
G R12 HK
G R13 R06
G R14 STG
G RI5 SW1
G RIG SW2
G R17 VEND
G R18 COR
G R19 ELEV
G R20 R1 1
G R21 R73
G R22 SW I
G R23 SW2
G S\VI CO It
G SW2 ELEV
2 COR R1 I
2 ELEV R23
2 HK s\v I
2 ROI SW2
2 R02 COR
2 R03 ELEV
2 R01 14 K
2 R05 RI I
2 ROG R25
2 R07 SW1
2 R08 SW2
2 R09 SW2
2 RIO MECH
2 R11 . ELEV I .O 3.6 0.503
Principles of Smoke Managemerit

STG

3rd - 6th Floors Cross hatching indicates FED


between 0.5 and 1.O.

2nd Floor

Ground Floor
UShading indicates FED of 1.0 or more
Figure 9.2 Toxicity for- 30-nzitzufeexposure of ho fel fire exanzple.
Chapter 9 -Hazard Analysis

4th Floor

Cross hatching indicates visibility between 25 ft (7.6 m)


A and 200 ft (61.0 m).

2nd Floor
Shading indicates visibility less than 25 R(7.6 m)
A

Ground Floor
Principles of Smoke Management

This analysis needed to use both a zone fire model prediction. The combined approach above works around
(FAST) and the network flow model (CONTAM). Using these limitations to produce meaningful results. Hope-
a zone fire model to simulate smoke transport for such a fully, a combined zone fire and network flow model'will
large building would be impractical, and a network flow be developed to produce even more realistic predictions
model lacks the desired fire simulation and temperature in the future.
CHAPTER 10

Stairwell Pressurization

any pressurized stainvells are designed and The flow through other shafts (elevators, mail
built with the goal of providing a smoke-free chutes, etc.) is negligible.
escape route in the event of a building fire. A The friction pressure losses in the stairwell are neg-
secondary objective is to provide a smoke-free staging ligible.
area for fire fighters. On the tire tloor, the design objec-
tive is to maintain a pressure difference across a closed The development and analysis of equations provide
stairwell door to prevent smoke infiltration into the considerable insight into stair pressurization. For most
stairwell. practical designs, these idealized conditions are not
Stai~wellsare often pressurized by a single dedi- achieved, but analysis can be done with a computer net-
cated fan, but more than one dedicated fan can be used. work model, such as CONTAM (Chapter 8). The use of
Also, a fan normally used for some other purpose can be such computer methods is discussed at the end of this
used to pressurize a stairwell in a fire situation. HVAC chapter.
system fans have been so used U ith modulating dampers ,When other pressurization systems are present, the
controlled by differential pressure sensors. However, total building flow network, including all of the pressur-
many smoke control designers feel that the same fans ization systenis, must be analyzed. For example, con-
should not be used for both the HVAC system and stair- sider a building with two pressurized stainvells and a
well pressurization because the dampers and controls zoned smoke control system where all three of these
needed only for the stairwell pressurizat~onsystem may smoke control systems are intended to operate at the
be damaged during HVAC system maintenance or mod- same time during a fire. Analysis of these systems niust
ification. Accordingly, it is not surprising that most consist of analysis of all of the systems operating at the
stairwell pressurization systems'have dedicated fans. In same time. Designs for the separate systems operating
this chapter, only systems with dedicated fans will be alone cannot be ''just added" together to get a realistic
discussed. However, this material can be adapted by the design for the three systems operating together. Later
designer who must design a system without dedicated chapters present example calculations of multiple sys-
fans. tems operating together.
The equations presented i! this chapter are for the
idealized conditions listed below. PRESSURIZATION SYSTEMS
It is in~possibleto provide detailed design methods
The only pressurization system is the pressurized for the almost infinite number of possible stairwell pres-
stairwell. surization systems. The intent of this book is to discuss,
The flow areas of the building are the same from in general, some .systemic considerations and alterna-
floor to tloor. tives and to provide detailed analyses of a few systems.
The leakage between tloors is negligible. For the analysis of other systems, designers can, in
Chapter 10-Stairwell Pressurization

Caution: This system should not be used for


tall stairwells (see text).

Fan m'-
Centrifugal

Roof Level
Duct
/ Shaft

7
r / Duct

Figure 10.2 Sfair-wellpressurizafion by mulfiple injec-


tion \*pith the fa17 located at the ground
level.
Figure 10.1 Stairwell presszirizafion bv top injec!!otl.

many cases, use the same principles employed in this


manual to perfonn their own analyses.

Single and Multiple Injection


A single injection system is one that has pressuriza-
tion air supplied to the stairwell at one location. The
most common injection point is at the top, as illustrated
in Figure 10.1. With this system, there is the potential
for smoke feedback into the pressurized stainvell
through the pressurization fan intake. Therefore, the
capability of automatic shutdown in such an event
should be considered.
For tall stairwells, single injection systems can fail
when a few doors near the air supply injection point are
open. All of the pressurization air can be lost through by multiple injec-
Figure 10.3 Sfair-\i,ellpt-essuriiafiot~
these open doors, and the system will then fail to main- tiou ~ i t roof-mountedfan.
h
tain positive pressures across doors farther from the
injection point. To preve1.t this, some smoke control
designers limit the height of single injection stainvells
to eight stories; however, other designers feel this limit Figures 10.2 and 10.3 are two examples of many
can be extended to twelve stories. Careful design is rec- possible multiple i~~jection systems that can be used to
ommended for single injection stainvells in excess of overcome the iirnitations of single injection systems. In
eight stories. Figures 10.2 and 10.3, the supply duct is shown in a sep-
There is the potential for failure of a bottom injec- arate shaft. However, systems have been built that have
tion system when the exterior door is opened. Some of eliminated the expense of a separate duct shaft by locat-
the supply air can short-circuit the system by floning ing the supply duct in the stairwell itself. If the duct is
directly out the opened doorway. It is recommended that located inside the stainvell, care must be taken that the
supply inlets be at least one floor above or below eute- duct does noi become an obstruction to orderly building
rior doors. evacuation.
P ~ C i p l of
e ~Smoke Management

i
l
Many multiple injection systems have been built lated buildings, where total building evacuation by the
with supply air injection points on each floor. These rep- stairwell is planned in the event of a fire. Compartmen-
1 resent the ultimate in preventing loss of pressurization tation can be an effective means of providing stairwell
air through a few open doors; however, that many injec- pressurization for very tall buildings, when a staged
1 tion points may not be necessary. There is some differ-
ence of opinion as to how far apart injection points can
evacuation plan is used and when the system is designed
to operate successfully when the maximum number of
be safely located. Some designers feel that injection doors between compartments are open. This maximum
points should not be more than three floors apart, while number of doors open between compartments would
others feel that a distance of eight stories is acceptable. need to be determined by an evacuation analysis. Com-
! For designs with injection points more than three stories partmentation does have a disadvantage from an archi-
apart, the designer should determine by computer analy- tectural standpoint in that it probably cannot be
sis that loss of pressurization air through a few open achieved without increased stairwell landing space at
doors does not lead to loss of stairwell pressurization. the location of the compartmentation doors.

Compartmentation Vestibules
An alternative to multiple injection is compartmen- A number of pressurized stairwells have been built
tation of the stairwell into a number of sections, as illus- with vestibules, which can be either pressurized or not
trated in'-~igure10.4. The stairwell is divided into a pressurized. Vestibules provide an additional barrier
number of sections or compartments, each compartment around a stairwell and, to some extent, a vestibule can
being from one to about eight floors high. The compart- reduce the possibility of an open-door connection exist-
ments are separated by walls with normally closed ing between the stainvell and the building. An evacua-
doors. Each compartment has at least one supply air tion analysis can be performed to determine the extent
injection point. The main advantage of compartmenta- to which both vestibule doors are likely to be opened
tion is that it allows satisfactory pressurization of stair- simultaneously.
wells that are otherwise too tall for satisfactory Analysis of a pressurized stairwell with an unpres-
pressurization. A disadvantage is the increase in floor surized vestibule can be performed using the same
area needed for the walls and doors that separate the methods employed for analyzing a system without a
stairwell sections. vestibule except that the effective leakage areas from the
When the doors between compartments are open, stainvell to the building would be used. These effective
the effect of compartmentation is lost. For this reason, areas can be detemiined by methods presented in Chap-
compartmentation is inappropriate for densely POPLI- ter 5. No formal method of design analysis has been
developed for pressurized stairwells with pressurized
vestibules, and this topic is beyond the scope of this
Roof Level manual.

Supply Air Intakes


In the pressurization systems illustrated in Figures
10.1, 10.2, and 10.3, centrifugal fans supply pressuriza-
tion air to the stainvell. A shield around the intake
should be considered to reduce adverse effects of wind
on the fan performance. This is especially important for
propeller fans, which are more susceptible to wind
effects than are other types of fan. Roof-mounted pro-
peller fans should have wind shields as illustrated in
Figure 10.5. Because the horizontal component of wind
I 1 1/ Each com~artment
is genei-ally about ten times greater than the vertical
has at least one
supply injection component, wall-mounted propeller fans are estreniely
point. susceptible to wind effects. If wall-mounted propeller
fans are to be used, design analysis should address wind
effects to minimize the probability of these fans being
overpowered by the wind.
I 1 I Ground Level
Outdoor s~iloke movement that might result in
smoke feedback into supply air inlets depends on the
location of the tire. location of points of smoke leakage
Chapter 10 -Stairwell Pressurization

from the building, wind speed and direction, and on the


temperature difference between the smoke and the out-
side air. At present, no formal method of analysis has
been developed for this complex outdoor airflow. How-
ever, some general recommendations can be made. The
supply air intake should be separated from exhausts,
outlets from smoke shafts and roof smoke and 'neat
vents, or open vents from elevator shafts or other build-
ing openings that might expel smoke during a fire.
These smoke outlets include the outlets from a zoned
smoke control system. Ideally, this separation should be
as great as is practically possible. Because hot smoke
rises, consideration should be given to locating supply
air intakes below such critical openings. A commonly
used approach is to have all of the supply air intakes
near the bottom of the building and smoke outlets above
roof level. Another approach is to have the supply air
intakes on one side of the building and the smoke outlets
Figure 10.5 Stairwell pressurization by roof-nzo~cnred
on the other side and on the roof.
propeller fan.

PRESSURE PROFILES
The pressure differences across a stain\;ell normally Top of Stairwell
vary over the height of the stainvell. Analysis of the

P"
Building N t h VerticalLeakage
pressure profiles of unpressurized shafts was presented Between Flmffi (Except at the ends.
in Chapter 5. The analysis of pressure differences in this curve is the same as that for a
building without vertical leakage Building
stairwells presented in this chapter is slightly more com- behveen floors.) Wlthout
Vertical

/\
plicated in that pressurization is incorporated. Leakage
Between
To facilitate analysis, the following discussion is Building Wfih Vertical
Floors
limited to buildings that have the same leakage areas on
each floor. Figure 10.6 shows pressure profiles for pres-
Leakage Through
,
an Elevator Shafl

surized stainvells located in three buildings with difer-


ent leakage characteristics, all of which have the same
stairwell and outside temperatures. These profiles repre-
sent winter conditions; that is, an outside temperature
I I
, Bottom of Stairwell

less than the inside temperature. Pressure Difference


Figure 10.6 PI-essrtr-eprofile for presszlrized stai~x.e/ls
In a building without vertical leakage between
iti three buildings wit17 different leakage
floors or througli shafts other than the stainve!l, the
characteristics.
pressure profile of a pressurized stain\-ell is a straight
line. The slope of that straight line depends on the tem-
perature difference between the stairwell and the outside The pressure difference across a stairwell at one
and on the building leakage areas. This relation is dis- height can be much larger than at another height. There-
cussed later in this chapter. fore, in addition to being concerned with the average
Figure 10.6 shows typical pressure profiles of pres- pressure difference across a stain\dl, a designer should
surized stainvells in a building with leakage between the also be concerned with both the minimum and the ma.xi-
floors a~idin a building without leakage between tloors mum pressure differences.
that are similal-except at the top and the bottom of the
buildings. The extent of the deviation depends on the STAIRWELL ANALYSIS
magnitude of the leakage area between floors. The pres-
sure profiles depend on the leakage areas of the stair- In this section, a method of analysis is presented for
well, the elevalor shati. and the exterior \\-ails, as well as a pressurized stairwell in a building without vertical
the tcmpcratures of tlie building, the stairwell, and tlie leakage between floors. This is the same zero floor leak-
outside air. Analysis of sucl~a building is cornplicatcd age idealizatiou that was used for the analysis of stack
and 1s generally kasible only with the aid of a computer. eft'ect in Chapter 5. The performance of pressurized
Principles of Smoke Management

stairwells in buildings without elevators may be closely


approximated by the method of analysis developed in
this section. where
Neglecting the effects of leakage through floors and Apso = pressure difference at elevation y, in. H,O (Pa);
other shafts increases the spread between the minimum
and maximum pressure differences. In this sense, the ApSob = pressure difference at the bottom of the stair-
analysis is conservative. This analysis considers only well, in. H20 (Pa).
one pressurized stairwell in a building; however, it can The above analysis assumes no change in densities,
be extended to any number of stairwells by use of the ps and po, with elevation resulting in a slight overpre-
concept of symmetry, as discussed in Chapter 6 . The ini- diction of pressure difference. The magnitude of this
tial analysis does not include consideration of open overprediction increases with elevation and, for a 100-
stairwell doors, but they are addressed later in this chap- story building, the resulting error would be less than
ter. 4%. For purposes of this book, this overprediction is
This analysis is for buildings where the leakage 'insignificant. By substituting the ideal gas law into
areas are the same for each floor of the building and Equation (10.3), bpSocan be expressed as a function of
where the only significant driving forces are the stair- temperature.
well pressurization system and the temperature differ-
ence between the indoors and outdoors.

Pressures
For many applications of pressurized stairwells, the
vertical flows within the stair shaft are low. enough so
that friction losses can be neglected. This is particularly
true of the simple stairwell system, which has closed and where
doors. Therefore, the absolute pressure in thestairwell is b = temperature factor, in. H20/ft (Palm);
considered hydrostatic and can be represented as
To = absolute temperature of outside air, O R (K);
Ps = p,, - li',Ps Y (10.1)
Ts = absolute temperature of stainvell air, O R (K);
where K, = 7.64 (3460).
ps = absolute air pressure in stairwell at elevation y, in.
The effective flow area from the stainvell through
H20 (Pal; thc building to the outside is expressed on a per floor
psb = absolute air pressure in stairwell at stainvell bot- basis as
tom, in. H20 (Pa);

ps = air density within the stainvell, 1b/ft3(kg/m3);


y = elevation above stainvell bottom, ft (m);
Kp = constant, O.lg2 (9.8). \vhzre
For the case where the wind velocity is essentially Asso, = effective flow area behveen the stairwell and the
zero, the outside air pressure, po, is also hydrostatic and outside, @f (m2);
can be expressed in the same manner.
ASB = flow area behveen the stainvell and the building,
fi!(m2);
where ifBO = flow area behveen the building and the outside,
p. = absolute air pressure at elevation y, in. H 2 0 (Pa); ft'(m').
pob = absolute air pressure at stainvell bottom, in. H 2 0 The areas in this equation are those of the entire - ~

floor. In such a case, the pressure difference, ApsB,


W ;
bcnveen the stairwell and the building can be expressed
p. = air density outside the stainvell, 1blft3(kg/rn3). as
The pressure difference from the stainvell to the
outside can be expressed as ApSO = p, - p O , and substi-
tuting Equations (10. I) and ( l 0.2) this is
Chapter l 0 -Stairwell Pressurization

The pressure differences Apm and ApsB are related


as follows: = Kq-(
NASB
J;;
APZ
- APSB~
Qs~t-Qs~b
'1 (10.14)

Qso
QSB
= (10.8) where
1 + ( A ~ B / A ~ ~' ) *
VsB = volumetric flow rate of air frorn~stairwellto
which can be rewritten as
building, cfm (m3/s);

AsB = flow area between the stainvell and the building


Pressurization Air
per floor when stairwell doors are closed, ft2
For the case where a stairwell is positively pressur-
(m2);
ized throughout (i.e., the direction of air flow is from the
stairwell to the outside ovzr the entire stairwell height), N = number of floors;
the flow from the stairwell to the outside can be written &I = pressure difference between the stairwell and
in differential form as the building at the stairwell top when all the
stairwell doors are closed, in. H20 (Pa);
d~ = CA,, J -
y d y . hBb
= pressure difference between the stairwell and
the building at the stairwell bottom when all the
The term AI,, is the distributed effective flow area stairwell doors are closed, in. H20 (Pa);
per unit height, which is uniform vertically. This distrib-
uted flow area is expressed as
Because there is no vertical flow in the building,
ifsB = vsBO. This is the flow rate of supply air to the
stairwell necessary to maintain the pressure differences,
where bSBb at the stairwell bottom and &l at the top.
In a building with vertical air leakage, the exact
AI,, = distributed effective flow area per unit heieht, fi
evaluation of the system would require that the effect of
(m); three or more colunlns of air at different temperatures be
H = stairwell height, fi (m); included. Such an analysis is cumbersome and, for prac-
h' = number of floors. tical purposes, a computer is needed. For this reason, the
Substituting Equations (10.4) and (10.1 1) into method of analysis presented in this section is based on
Equation (10.10) gives a building without vertical leakage. In order to make this
analysis conservative when applied to buildings with
vertical leakage, the stakwell temperature is replaced by
the building temperature. Thus, Equation (10.5)
becomes
This can be integrated from y = 0 to = H ro give
the total flow, vsBo, from the stairwell to the building
and to the outside:

where
( l 0.13)
To = absolute temperature of outside air, "R (K);
TB = absolute temperature of t!?e air in the building, "R
where ApSo, is the pressure difference between the stair-
well and the outside at the stainvell top (1. = H). Because (K);
the Apss is a linear function of Apso as expressed in Equa- K~ = 7.64 (3461)).
tion (1OX), Equation (10.13) can be writtsn in temls of the For a building temperature of 70°F (2 1°C) and for
pressure from the stairwell to the building. For C = 0.65, \\.inter conditions, the temperature factor b can be
this becomes obtained from Figure 10.7.
Principles of Smoke Management

The subscripts SB and SO have been eliminated fiom this


equation because it is applicable to flow f h m the stairwell
to either the building or the outside. When applying Equa-
tions (10.16) and (10.17) to flow from the stairwell to the
building, A, = Ass, Apb = ApSBb, and bp, = h,. When
applying these equations to flow from the stairwell to the
outside, A, = ADO,, b p b = PpSob,and Apt = ApSOp Equa-
tion (l 0.17) can be approximated by

The maximum error in this relation is approxi-


mately 6 % and occurs when Apb = 0.

HEIGHT LIMIT
- 20 - 10 0 10 20 SO 40 50

Cutside Temperature, To CF)


As stated before, two problems with pressurized
.............................
stainvells are that the minimum pressure difference may
:SO - 25 -20 -I5 -10 -5 0 5 10 be too low to prevent smoke infiltration and that the
. .:..
,I Outside Temperature, To PC) maximum pressure difference may be too high, making
Figure 10.7 Ternper.ntur-efnctor: door-opening forces difficult. These problems are most
likely to exist in tall buildings during periods of extreme
outside temperature.
In some cases, satisfactory pressurization of a stair-
Average Pressure Difference well can be impossible even when all the stairwell doors
The average pressure difference can be defined as a are closed. By satisfactory pressurization, it is meant
pressure difference unifomi over the stairwell height that no\vhere over the stainvell height is the pressure
that would result in the same total flow as a nonuniforni difference greater than the maximum allowable pressure
pressure profile. The flow from the stairwell can be difference or less than the minimum allowable pressure
expressed as difference.
For a building without vertical leakage, Equation
(10.5) can be substituted into Equation (10.7) and
solved for the height limit, H,,,, below which satisfactory
where pressurization is possible:
N = number of floors;
A, = erective flow area, ft2 (m2);
C = flow coefficient;
4'011 = average pressure direrence across the effective where
flow area, in. H 2 0 (Fa); H,,, = height limit, ft (m);
P = density of air, lb/ft' ( k g h 3 ) ; &, = niaximuni allowable pressure difference
K, = 776 (1 .OO). between the stainvell and the building, in. H 2 0
The effective area can be either the area between (Pal;
the stairwell and the building or between the building Ap,,,;,, = minimum allo\vable pressure difference
and the outside. The average pressure difference needs between the stairwell and the building, in. H 2 0
to be consistent with the effective area. (.P a,l
Equations (10.13) and (l 0.16) can be combined and To = outside design temperature, "R (K);
solved for Apa,.to give
TB = building temperature, "R (K);

A,).,. = ;[ /\/), - A/),, (10.17)


AsB = flow area between the stainvell and the build-
. 7 7
Ing. fi- (m-);
cnapter 10-Stairwell Pressurization

ABO = flow area between the building and the outside, SIMPLE STAIRWELL SYSTEMS
(m2); A simple stairwell system is one for which no
'
Km = 0.13 1 (0.000289). design provisions have been made to overcome the diop
in pressurization when one or more stairwell doors are
Ts was replaced by TB in Equation (10.1 g), s o that opened. Analysis of the simple stairwell system forms a
the equation would yield conservative values of H, for foundation for the analysis of systems with open docrs.
buildings with'vertical leakage. 1n' such buildings, the Some of the stainvell doors must be opened during
actual pressure profiles depend on three or more col- evacuation if the stainvell is being used. No consensus
umns of air at different temperatures. If the stairwell exists concerning appropriate applications of simple
temperature is between the outside temperature and the stairwell systems. A possible criterion for such an appli-
building temperature, then Equation (10.19) will yield cation is that smoke leakage during times of low pres-
conservative results. surization will not adversely affect the use of the
The absolute value of the temperature term is used stairwell during evacuation. In a lightly populated build-
in Equation (10.19) s o that the equation will apply to ing (for example, telephone exchanges, luxury apart-
both winter conditions (TB > To) and summer condi- ments), the stairwell doors may only be open for a few
tions (To > TB). In many cases, ASB is much smaller short intervals during a fire evacuation. Applications of
than AB*, and, in such cases, Equation (10.19) can be the simple stairwell have so far been based on engineer-
~implified~to ins judgment because no formal method of analysis has
been developed for evaluation of effects of intermittent
smoke infiltration. Such an analysis would need to con-
sider tenability conditions, evacuation analysis, and
flow analysis.
The units for this equation are the same as those for The simple stairwell system can use single or multi-
Equation (10.19). For a building temperature of 70°F ple injection. One or more fans are used, which can be
(21°C) and for winter conditions, the height limit, H,,,, centrifu~al,axial, or propeller. When all the stainvell
can be obtained from Figure 10.8. Example 10.1 illus- doors are closed, the system must maintain satisfacton'
trates the use of height limit. pressurization. When stainvell doors are open, the pres-
sure difference across closed stairwell doors usually
drops to low levels [in the range of 0.01 in. H 2 0 (3 Pa)].
180 These low levels are not sufficient to prevent smoke
infiltration into the stairwell, and simple stairwell sj-S-
165 -
tenis are only appropriats for applications for which
150 - stairwell doors are closed for almost all of the time dur-
ing fire evacuation.
-
E
v
135 -
.
..
Example 10.2 is for two 20-story stairwells in the
E 120- same building. Syn1nietr~-is used so that calculations are
-
I
E
.-
'05
:-
:
needed for only one s~ainvell.The same approach can
be used for three or more stainifells.
A .
Em 90 The flow rate of pressurization air is highly depen-
.- ..
a , .
dant on the leakage area. Because these areas can only
= 75.
be roughly estimated in most situations, the fan needs to
60 be sized conservatively so that the fan flow can be
adjcsted tc xceptable It\-els of pressurization during
45
system commissioning. This fan sizing can be by choice
30 of high values of building leakags or of safety factors.
- 20 - 10 0 10 20 30 40 50

Outside Temperature. To(OF)


The calculations of Ezample 10.2 are based on win-
ter design teniperaturcs. This is appropriate when the
inside-to-oulside dcsign temperature difference for win-
Outside Temperature. To(T) ter is greater than thc outsidc-to-inside design tempera-
ture difference I'or summc'r. Othenvise, summer design
data should be uscd.
Principles of Smoke Management

Example 10.1 Evaluate the Possibility of Stair Pressurization .,


Is it possible to pressurize a 150 ft (46 m) stairwell if the outside design temperature is 0 "F (-18 "C)? The minimum and
maximum allowable pressure differences are:

Lipmin= 0.05 in. H20 (12.4 Pa)


Ap,, = 0.40 in. H20 (100. Pa)

Then Apma -Qmin


= 0.40 - 0.05 = 0.35 in. H20 (87 Pa).

From Equation (10.20) for To = 0°F 4 1 8 "C), H, = 160 ft (49 m).

Because H, is greater than the height of the stairwell, satisfactory pressurization of the stairwell is possible. If H, had
been less than the stairwell height, it would not necessarily mean that satisfactory pressurization is impossible, because
the estimate of H, from Equation (10.20) (Figure 10.8) is conservative. (Note that this example has nothing to do with
single or multiple injection.)

Example 10.2 Simple Stairwell Pressurization


Caution: The simple system does not take into account the effect of pressurization drop when stainve11 doors are
opened. The design parameters for this simple system are: Ass = 0.32 ft'(0.030 m2), N = 20, H = 200 ft (61 m), TG
= 14°F (-10°C) or 474"R (263 K), TB = 70°F (21°C) or 530 "R (294 K), Ap,,, = 0.40 in. H20 (100 Pa), and Q,,;,,
= 0.05 in. H20 (12.4 Pa). This analysis is of two stainvells in a building, and the concept of symmetry is used so that
analysis of only one is necessary. Therefore, the flow area, ABO, used in these calculations is half the esriniated value
for the whole building. The leakage area from the building to the outside is estimated at 2.54 ft2 (0.236 rn2). Therefore,
ABo = 2.5412 = 1.27 fi? (0.1 18 m2).

Calculate the height limit from Equation (10.19).

H = 3 1 - 0.05 [ l
0'4 + ('6)
2] = 2 19 ft (67 m).
474 530
1 I
The height limit is greater than the height of the stainvell, so the equations presented in this chapter can be used for anal-
ysis. Calculate the temperakre factor from Equazion (10.5).

., ..
Set = 0.05 in. H 2 0 (1 2.4 Pa), and calculate the pressure difference at the top of the stairwell from Equation (1 0.7).

ApSB, = 0.05 + 0'00170(200) = 0.37 in H 2 0 (92 Pa).

Calculate the flow from the stairwell to the building liorn Equation (lO.l4), using p = 0.075 lb/ft3 (l 2 0 k s ni3).
Chapter 10-Stairwell Pressurization

SYSTEMS WITH OPEN DOORS Note: Canadian system can be single or


multiple injection.
As discussed in the preceding section, when any
stair door opens in the simple stairwell pressurization n
systems, the pressure differences across closed doors
Roof Level
drop significantly. However, opening the exterior stair-
well door results in the largest pressure drop. This is
because the airflow through the exterior doorway goes
directly to the outside, while airflow through other open
doorways must also go through other building paths to
reach the outside. The increased flow resistance of the
building means that less air flows through other door-
ways than flows through the open exterior doorway. The
flow through the exterior doorway can be three to ten
times that through other doorways, and the relative flow
through the exterior doorway is greatest for tightly con-
structed buildings. Thus, the exterior stairwell door is
the greatest cause of pressure fluctuations due to door Figure 10.9 Canadiarz system has exterior door that
opening and closing. operzs automatically on system activation.
For densely populated buildings, it can be expected
that many stairwell doors will be open during fire evac- small. However, the term constant-supply is used to dif-
uation. Accordingly, stairwell pressurization systems in ferentiate this system from the ones with variable-sup-
such buildings should be designed to operate with some ply air rates, but constant supply systems actually have
number of open doors. This design number of open some variation in flow due to the pressure-flow charac-
doors depends heavily on the evacuation plan, and spe- teristics of the fan. Supply air can be introduced at one
cific guidance about this number is beyond the scope of location, or the system can be multiple injection, as
this manual. illustrated in Figure 10.9.
Four types of systzms intended to maintain accept- By eliminating opening and closing of the exterior
able levels of pressurization with all doors closed and stairwell door during system operation, the Canadian
with some doors opened are discussed in this section. system eliminates the major source of pressure fluctua-
tions. This system is simple to design and relatively
System with constant-supply air rate and an exterior inexpensive. Accordingly, thissystem is recommended
stairwell door that opens automaticall!. upon system whenever it can meet the design requirements.
activation (Canadian System).
System with constant-supply air rate and a baromet- Systems with Barometric Dampers
ric damper.
This system has sufficient supply air when a design
System with variable-supply air rate. number of doors are open. When all the doors are
System using stairwell pressurization i n combina- closed, part of the supply air is relieved through a vent
tion with either fire floor venting or fire floor to prevent excessive pressure buildup. Barometric
exhaust. dampers that close when the pressure drops below a
specified value can, be used to minimize air losses
The following is a discussion of these systems. through a vent when doors are open.
Field tests of these and other systems for stairwell pres- There are two approaches to location of barometric
surization were conducted by Butchzr et al. (!97 1); Dias dampers: (I) in exterior stairwell walls or (2) in walls to
(1978); and Taniura (1994, 1990a. l99Ob, and I'BOc). other building spaces. Venting to the building has the
advantage that the barometric dampers maintain the
Canadian System pressure difference of interest, which is from the stair-
The system with constant-supply ;ir rate and an well to the building. However, venting to the building
exterior stairwell door that opens automatically upon has the disadvantage that the air vented can supply oxy-
system activation is essentially the same a, that in the gen to the fire. Exterior venting eliminates this disad-
ro the Nrrriontrl Bliilcli~rgCnrlc. of C m a h
Sirppler~~er~r vantage, but exterior vents can be subjected to adverse
(1985). The supply air rate is not actually constant. but it effects of the wind.
varies to some extent \\.ith the pressure across the fan. Figure 10.10 illustrates a system vented to the
For centrifugal fans this variation in flou. rate can be building at each Iloor. In systcms built with vents
Principles of Smoke Management

between the stairwell and the building, the vents typi- used to vary the flow rate of supp!y air to the stairwell.
cally have one or more fire dampers in series with the The variable-flow fans are controlled by one or more
barometric damper. As an energy conservation feature, static pressure sensors that sense the pressure difference
these fire dampers are normally closed and open when between the stairwell and the building. When doors are
the pressurization system is activated. This arrangement opened, the stairwell pressure drops and the flow rate of
can reduce the possibility of annoying damper chatter supply air is increased to achieve at least the minimum
that frequently occurs with barometric dampers. design pressurization. When all the doors are closed, the
stair pressure increases and the flow rate is reduced to
Systems with Variable-Supply Air Rate prevent excessive pressure differences.
Systems with variable-supply air can be used to In the bypass system, the flow rate of air into the
provide overpressure relief. The variable flow rate can stairwell is varied by modulating bypass dampers,
be achieved by using one of the many fans commer- which also are controlled by one or more static pressure
cially available for a variable flow rate. Alternatively, a sensors that sense the pressure difference between the
fan bypass arrangement of ducts and dampers can be stairwell and the building. The system operates in essen-
tially the same way as the variable-flow fan systems to
prevent excessive pressure differences and provide at
Notes: least the minimum design pressure.
1. Vents have barometricdamperand one or two fire
dampers in series. The response times of these sysiems depend on the
2 A system with vents can be single or multiple injection particular components used for the pressurization sys-
tem including the feedback controls. Figures 10.1 1 and
10.12 show response times of systems tested at the
Roof Level experinxntal fire tower of the National Research Coun-
cil of Canada (Tamura 1990b).

I ' Building
vent to
System with Fire Floor Venting and Exhaust
Smoke venting and smoke exhaust of the fire floor
can improve the performance of a pressurized stairwell.
This smoke removal may or may not be part of a zoned
smoke control system (Chapter 12). Smoke removal can
be accomplished by exterior wall vents, smoke shafts,
and fan-powered exhaust.
Besides providing a path for smoke removal, exte-
rior wall vents allow an increased pressure difference
across the closed staincell door on the fire floor. Venting
Figure 10.10 Stair-~.ell pressurization system with
the fire floor can also aid tire fighters in smoke purging
vents to the bziildi17gat eachjlooc
after the fire has been put out.

II Peak Pressure

Time (minutes)
1.47 in H,O
(365 Pa)

"
0 5 10 15 20 25 30 35 40 45 50 55 60
Time (minutes)
Figure 10.12 Response time of staitwell pressurization system with bjpass system.

Smoke shafts are similar to external wall vents the open doorway drops significantly, as illustraied in
except that smoke from the fire floor is vented through a Figures 1 0 . 1 3 ~and 10.13d. However, the flow through
shaft. The venting is aided by buoyancy forces of hot the large area of an opened doorway can be very large,
smoke. Smoke shafts should be constructed in accor- as can be seen from the examples discussed later.
dance with local codes; specific. engineering data In the winter, the pressure difference across opened
regarding sizing of smoke shafts are available from doors increases \i.ith elevation. The greatest amount of
Tamura and Shaw (1973). pressurization air is needed when the design number o f
opened doors are located in a section at the top of the
ANALYSIS OF SYSTEMS stairwell, as illustrated in Figure 10.13e. This forms a
WITH OPEN DOORS conservative winter design condition. The conservative
summer design condition is for the opened doors to
The analytical approach developed for simple stair- form a section at the bottom of the stairwell, as in Figure
well syste~nscan be extended to pressurized stairwells 10.13f.
with open doors, provided that the frictian losses due to
Equation (10.14) applies when the effective tlo\v
airflow in the stainvell are negligible. Friction losses
area between the stairwell and the building is thesams for
can be minimized by having a multiple injection system
each floor. When some doors are opened and others
designed to minimize vertical airflow in the stairwell.
closed, this flow area varies from floor to floor. Equation
Because the pressure losses due to friction are consid- (1 0.14)'can be applied piecewise to vertical stairwell sec-
ered insignificant for this analysis, the pressure differ-
tions, where the values of Ass and the values of ABO are
ences described by Equations (1 O.4), (1O.7), (1 0.8), and
the same at each floor. Both of these areas are used to cal-
10.15 apply for both summer and winter conditions, as
culate the pressure differences and the effective flow area.
is illustrated in Figure 10.13. Equation ( l 0.14) can be written in a general form for C =
When all of the doors are closed, the pressure dif- 0.65 and p = 0.075 Ib/ft (l .20 km/m) as
ferences are linear, as illustrated in Figures 10.13a and
10.13b. As expected, the pressure differences increase
with eievation in winter and decrease with elevation in
summer. When a door to the outside is opened, the pres- where
sure difference across it increases, as shown in Figures
1 0 . 1 3 ~and 10.1 3d. This means that the flow through i n p = volumetric flow rate from the section, cfni (mds):
open esterior doorway can bc very large. This is espe- N = number of floors in section;
cially true during the summer when the pressure differ- A, = effective flow area per tloor from stairwell, R-7 (m-):
7

ence is greatest at the shaft bottom where most exterior


doors are located (Figure 10.13d). When doors are G = the tlow factor, fpm (m/s).
opened to the buildins, the pressurc difference across The flow factor is
Principles of Smoke Management

(a) Winter With All Doors Closed (b) Summer With All Doors Closed

Doors
Opened

Opened

AP AP
(c) Winter With Some Doors Opened (d) Summer With Some Doors Opened

H H

Y Y

0 0
AP AP
(e) Winter With Design Condition (f) Summer With Design Condition
of Opened Doors of Opened Doors
Figure 10.13 PI-essur-edifjersr7ces with closed at7d opetied stairweN doors.
- Chapter 10-Stairwell Pressurization ' -

dimensionless flow coefficient

flow area between stainvell and outside, fi?


(m2);
pressure difference f?om stairwell to outside, in.
H20 ( W ;

density of gas in path, 1b/ft3(kg/m3);

776. (1.OO).

The pressure diEerence is not always constant ovcr the


opening; therefore, the pressure difference, &So. should
be evaluated at the midheight of the opening.
Design calculations for a ten-story Canadian system
are presented as Examples 10.3 and 10.4. Analys~sin
Ap, - Ap, (in:H,O) these examples is only of one stainvell, but it can be
Figure 10.14 Flol~,
fnctol: thought of as being applicable to any number by appli-
cation of symmetry. The flow area, ABO, is on a per
stairwell basis. Example 10.3 and Example 10.4 show
calculations of the pressurization air for a winter design
temperature of 14°F (-10°C) and a summer design rem-
perature of 94OF (34OC). It is an unusual occurrence that
the total pressurization air calculated for both design
where
temperatures is the same [17.500 cfm (5.26 m3/s)]. As
Qb = pressure difference at the bottom of the section, in.
f

expected from observation of Figure 10.13e and 10.13f,


H 2 0 (Pal; the tlow through the open exterior doonvay is greater
Ap, = pressure difference at the top ofthe section, in. for summer than \\inter (9.200 cfm [4.3 rn3/s] in sum-
H 1 0 (Pal; mer and 6,800 cfni [3.2 m3/s] in winter). For a taller
Kg = 1740 (0.559). stairwell, the flou through the exterior doonvay in sum-
mer would be even greater.
Equations (10.7 1) and (1 0.22) can be used to calcu-
late either irSBor lfsBO.where i s BO is the flow from As with the simple stain\.ell system, safety facrors
the stairwell through the building to the outside. When are needed to size the supply air fan or fans-the fan
calculating vss,A, and the two pressure differences are needs to be sized conservati\.ely so that the fan flow can
from the stainvell to the building. When calculati~ig be adjusted to acceptable levcls of pressurization during
i?,, , A, is the effective flow area from the stairwell system commissioning. This fan sizing can be by choice
through the building to the outside, and the two pressure of high values of building leakage or of safety factors.
differences are from the stairwell to tlie outside. The
flow factor, G, can be obtained from Figure 10.14.
Flows directly to the outside are handled differently NONUNIFORM BUILDING FLOW AREAS
from those through the building. For the exterior doors,
exterior vents, or other openings directly to the outside, Flow areas that differ from tloor to floor can result
the flow can be espressed as in significant challenges to stairwell pressurization. Pos-
sibly the most dramatic example of this is a building
with open parking garages on soliie floors. Figure 10.15
shows the pressure differences for a building with an
where open garage. lt can be seen that the pressure differences
are much greater on the floors where tlie stainvell opens
vso = volun~el~-ic
flow rate from stairwell to outside, into the parking garage. This figure is based on the same
cfi11 (n?/s); assumptions as in rhs preceding discussions except rhat
Principles of Smoke Management

Example 103 Winter Analysis Stairwell With Opened Doors


A Canadian stairwell pressurization system (see text for description) is to be designed for interior doors on 8 of its 10
11 floors. The other design parameters are: ABo = 1.27 f? (0.1 18 m'), Ass = 0.32 f? (0.030 rn2) with stairwell door closed,
11 A , = 10.5 ?f (0.975 m2) with stainvell door opened, To = 14 "F-(-IO°C) or 474OR (263 K), TB= 70°F (21°C) or 530%
11 (294 K), Apm, = 0.40 in. H20 (100 Pa), and hi, = 0.05 in. H20 (12.4 Pa). Because the design temperatures are the
11 same as for Example 10.2, the temperature factor is 0.00170 in. H20/ft (1.39 Palm). In order to ensure that the stainvell

I is adequately pressurized at all levels, the pressure difference at the bottom of the stairwell door to the building is selected
as 0.05 in. H20 (12.4 Pa), when that door is closed. Symmetry can be used to extend this analysis for any number of

I1 stairwells in a building. As with Example 10.2, ABO is estimated on a per stairwell basis.

Closed Door Section

The winter design condition consists of a section of opened doors from the stairwell top down, with the rest of the doors
forming a section of closed doors near the bottom of the stairwell. For the section of closed doors, the flow from the stair-
well to the building will be evaluated, and the following values are used: N = 2, Apb = ApsB at y = 0, Apt = ApsB at y =
20 ft (6.1 m), and A, = ASB.AS selected, Apb is 0.05 in. H 2 0(12.4 Pa). From Equation (10.7), Ap,= 0.05 + (0.001 7 X 20)l
(1 + (0.3211.27)') = 0.082 in. H20 (20.4 Pa). From Equation (10.22), G = 1740[(0.082~~
- 0.05~')!(0.082 - 0.05)] = 669

I1 @m (3.40 nds). From Equation (10.71),

Opened Door Section


G NAsB = 669 (2) (0.32) = 400 cfin (0.2 m3/s).

For the section of opened doors, the flow from the stairwell to the outside will be evaluated, and the following values are
used: N = 8, Apb = A I Jat~ y~= 20 ft (6.1 m), Ap, = Apso at y = 100 ft (30.5 m), and A, = AsBOr First, hB must be
evaluated. From Equation (10.9), bSOb = 0.05 [l+(0.32/1.2712]= 0.053 in. H20 (13.2 Pa). The pressure differences, Apb
and Ap,, are calculated from Equation (10.4) as follows: Apb = 0.053 + 0.0017(20) = 0.087 in. H20(2 1.7 Pa) and Ap, =
0.053 + 0.0017(100)= 0.223 in. H 2 0 (55.5 Pa). From Equation (l0.22), G = 1740 [(0.223'12 - 0.087'~)/(0.223 - 0.057) =
1020 fpm (5.18 nds). From Equation (1 O.6), AsBOe = [10.5(1.27)/(1 0.5' + 1.27')"] = 1.26 fi (0.11 7 m'). From Equation

)II1
3
(10.21), PS, = GNAsBoe= 1020 ( S ) 1.26 = 10,300 cfm (4.9 m /s).

Exterior Stairwell Door

Estimate the flow through the cpened exterior doonvay with air density ofO.075 lwft3(1.20 kglm3)~ n aty
d = 5 ft (1 3 m).
The pressure difference is calculated from Equation (10.9) as @so= 0.053 + 0.00 17(5) = 0.062 in. H20 (1 5.4 Pa). From

l11 Equation (1 O.23), i/sO = 776(0.6j)(l0.5)[2(0.062)/0.075]"


Total Flow Needed During Winter
= 6800 cfm (3.2 m3/s).

The total flow needed to pressurize the stainveli in winter is the sum of these separate flows: 400+10,300+6800 = 17,500
cfm (8.26 m3/s).
Chapter 10-Stairwell Pressurization

Example 10.4 Computer Analysis of a Pressurized Stairwell


This is an example of a building with two stairwells and an elevator shafi with two elevator cars. The building and both
stairwells are 15 stories. Each stairwell is pressurized by a centrihgal fan supplying air at the second story. The s t a b e l l
systems are of the Canadian design, which has an exterior door open automatically upon system activation The design
condition is for fovroper! dmrs between the stairwell and the building. The minimum and maximum design pressure dif-
ferences are 0.05 in. H 2 0 (1 2.4 Pa) and 0.30 in. H 2 0 (74.6 Pa).

The computer program CONTAM was used for this analysis. Appendix G has a detailed list of the design parameters,
flow areas, computer runs, and computer output. The CONTAM runs are listed be!ow.
Building
Run Season Leakage Stair Doors Open on Floors
1 Summer Loose CS 2 , 3 , 4 , 5
2 Summer Loose G
3 Winter Loose G, 12, 13, 14, 15
4 Winter Loose G
5 Summer Tight G 2,3,4,5
6 Wintcr Tight G
11 The data and computer o u t i t for this example are provided in Appendix G It is usually inipractical to determine the val-

l ues of flow areas in buildings, but design calculations can bracket building leakage conditions. Loose and tight building
leakage values are listed in Table G2. It is expected that the building leakage will be between these extremes.

As this is a Canadian system, the door to the outside on the ground floor (G) opens on system activation. The other open
doors were selected because the expected airflows and pressures with these doors open represent the worst case (or near
tlie worst case) conditions as illustrated in Figures 10.13e and 10.13f.

Results of the CONTAM runs are listed below.


Minimum Ap Maximum Ap
Across Closed Across Closed
Stairwell Supply Stairwell Doors Stairwell Doors
cfm m3/s in. HIO Fa in. HzO Pa
20,500 9.67 0.053 13.2 0.6 1 15.2

1 13,900 6.56
For run I, the stairwell supply
0.107 26.6 0.110 27.4
.. - air was adjusted until a value was found such that all of the pressure differences across the
closed stainvell doors would be at least the minimum design value of 0.05 in. H 2 0 (1 2.4 Pa). This same amount of supplq
air was used for the other runs with loose building leakage (runs 2,3, and 4). With doors closed, the highest pressure dig
ferences in the loose building occur during \\.inter, and it can be seen that the maxinium design value of 0.30 in. H2C
(74.6 Pa) is not exceeded.

Run 5 is similar to run 1, except that it is for a tight building. Again the stairwell supply air was adjusted to so that tht
pressure diiTerence would not be less than 0.05 in. HzO (1 2.4 Pa) across closed stairwell doors. This flow rate was used ir
run 6 to verify that the mastmum design value would not be exceeded when tlie stainvell doors are closed.

Nore: If the loose building leakage values have been selected such that they a n be considered limits that are highl)
unlikely to be exceeded, then the highest supply air rate calculated in the CONTAM simulations \vould be a reasonablc
flo\v rate for the supply fans. For this example, this reasoning would result in using stainvell pressurization fans sized a
20,500 cfiii (9.67 mqs).
Principles of Smoke Management,

the floor-to-floor flow areas are not the same on each COMPUTER ANALYSIS
floor. USING A NETWORK MODEL
When these floor-to-floor pressure difference varia- Except as noted otherwise, the preceding sections
tions are unacceptable, approaches to dealing with them were based on the simplifying assumptions of(1) the only
include: pressurization system being the pressurized stairwell, (2)
the flow areas of the building being the same from floor
modify the building flow network (possibly by use to floor, (3) the leakage between floors being negligible,
of partitions or pressure relief vents), (4) the flow through other shafts (elevators, mail chutes,
etc.) being negligible, and (5) the friction pressure losses
eliminate the doors into stairwell on garage floors,
in the stairwell being negligible. Network computer mod-
and use other stairwells for the open garage,
els (Chapter 8) can be used to account for all of these and
use hardware on the stairwell doors to the garage many others. Example 10.4 uses the computer model
floors that assists door opening by reducing door CONTAM to analyze two pressurized stairwells in a 15-
opening forces. story building with elevators.

Office Floor

Office Floor

Office Floor

Office Floor

Ofice Floor

Open Garage

Open Garage

AP
(a) Building Elevation (b) Wlnter (c) Summer
CHAPTER 11

Elevator Smoke Control

his chapter addresses two very different kinds of sive pressures at the top of the hoistway due to a rising

T elevator smoke control systems. One has the


objective of providing smoke protection for the
elevator system so that it can be used for fire evacuation.
elevator car.
The idea that the vents are needed to prevent exces-
sive pressures is doubtful for two reasons. First, vents
Most elevators worldwide do not have smoke protec- would also be needed at the bottom of the hoistway if
tion, fire protection. and other features necessary for the pressures from moving elevator cars needed to be
them to be considered as a means of fire evacuation. relieved. Second, the pressures produced by moving ele-
Elevator systems not specifically designed and built for vator cars are very small, as described in the following
fire evacuation should not be used in fire situations section about piston effect.
(Sumka 1988). Honever, the use of elevators for fire To understand the of idea venting odorous gases, a
evacuation is a topic that has received considerable historical perspective is needed. In 1853, an elevator
attention in recent years. Because the concept of eleva- safety device to prevent elevator cars from falling was
tor evacuation is so new, this chapter provides a general developed by Elisha Otis. By the 1880s, elevators
overview of the topic in addition to the smoke control gained wide acceptance in many large cities. During the
considerations. 19th and early 20th centuries, the standards of sanitation
The other kind of elevator smoke control system were not as advanced as those of today, and it is likely
addressed in this chapter is intended to prevent smoke that open elevator hoistways were used as trash chutes
flow to other floors by way of the hoistway (elevator by some people. Further, it is possible that vents were
shaft). The problems that can result from snioke migra- required at the top of elevator hoistways to relieve some
tion through lioist\vays are illustrated by the fire at the of the malodorous gases emanating from garbage at the
MGM Grand Hotel (Best and Demers 1982). The fire bottom of the hoistway.
occurred on the ground floor, but smoke migrated to the Regardless of the original purpose for these vents,
upper floors where the majority of the fatalities the idea that they can significantly improve smoke con-
occurred. The hoist~..aysat this hotel did not have any ditions during a building fire has gained wide accep-
special smoke protection, and they were one of the tance even in the absence of supporting data. Research
is needed to evaluate the effect of vents on the hazard of
major paths of smoke ~nigrationto the upper floors.
smoke exposure during building fires.
TOP VEhT For most of the elevator pressurization systems dis-
cussed in this chapter, there is either no top vent or the
The requirement for vents at the top of the hoist- top vent is closed. For energy conservation, these-top
ways has been in codes for so many decades that the vents are often normally closed. Such normally closed
original intent of the vents is uncertain. The most com- vents should remain closed during elevator pressuriza-
mon reasons that the authors of this book have heard for tion unless the open vent is part of the pressurization
these vents are that they ( l ) vent smoke during a build- system design. The capability of remote operation of top
ing fire, (2) vent odorous gases, and (3) prevent esces- vents may be desired by the fire service.
*
Chapter 11-Elevator Smoke Control

(a) System Wlth Lobby (a) System Whotit Lobby


Used for Elevator Evacuation Used for P ~ t i o ofn Smoke
Migration

Figure 11.1 Elevator cc]-motion ardpistotz effect in pressurized ele-


vator- shafis.

PISTON EFFECT the measured curves by only about 0.004 in. H 2 0 (I Pa)
and, for the 15th floor, the extremes deviate by about
The transient pressures produced when an elevator 0.03 in. H 2 0 (S Pa).
car moves in a shaft are a concern for elevator smoke
From the analysis by Klote, equations wers devel-
control. Such piston effect can pull smoke into a nor-
oped for the critical pressure difference at which piston
mally pressurized elevator lobby or hoistway. Analysis
cfrect cannot overcome the elevator pressurization sgs-
of the airflows and pressures produced by elevator car
ten1 both for systenis intsnded to prevent smoke migra-
nlotion in a pressurized lioistway was developed by
tion tlirougli the hoistway and for systems intended for
Klote (1988), based on the continuity equation for the
elevator evacuation.
contracting control volume in the hoistway above a
moving elevator car. The elevator system can be with or
Without Enclosed Lobby
without enclosed lobbies (Figure 1 1. l).
This section is limited to elevators without enclosed
Piston effect experiments (Klote and Taniura 1987)
lobbies, and the elevator pressurization systems dis-
were conducted on an elevator of a hotel in Mississauga,
cussed in this section are intended to prevent smoke
Ontario, Canada. This elevator served each floor of the
migration through the hoistway. The critical pressure
15-story building, and the hoistway was pressurized by
difference, 43,,.,,, is froni the shaft to tlie building:
a vane axial fan. Figure 11.2 is a coniparison of mea-
sured and calculated pressure differences due to an ele-
vator car ascending fro111 tlie ground floor to the top
floor. The general trends of the calculations are in agree-
ment with the measurements. On the ground floor, pis- where
ton effect causes a rapid drop in pressure followsd by a = critical pressure difference, in. H 2 0 (Pa):
gradual pressure increase as the car moves away from
the ground floor. A reduction in pressure is expected = air density in hoistway, lb/ft3 ( k g & j;
below an ascending car. This pressure reduction = cross-sectional area of hoistway, fi (m2):
decreases as the car moves away due to the etYect of
increasing leakage area of the shaft below the car. On = leakage area betiveen lobby and building. f~?
the top floor, piston effect due to the ascendin,0 car
(m2);
causes a gradual pressure increase with distance traveled
until the car gets close to that floor. On a middle floor = free area around the elevator car, ft-7 (m-):
7 . .

(the 8th) the pressure increases as tlie car approaches, = stfective area bet\vsen hoistway and outside, fi
drops suddenly as the car passes, and increases after it
travels away. For thc ground and Stli floors, tlie (m2);
extremes of the calculated c u n e s deviate from those of = elcvalor car velocity. Ipm (m/s);
Principles of Smoke Management

C, = flow coefficient for flow around car, dimension- where Aio is the leakage area between the building and the
less; outside in fr2 (m2). Example 11.1 illustrates calculation of
= 1.66 X lo6 (1.00). the critical pressure difference for an elevator pressuriza-
Kp,
tion system without enclosed lobbies.
The flow coefficient, C,, was determined experimen-
tally (Klote and Tamura 1986a) at about 0.94 for a mul-
tiple-car hoistway and 0.83 for a singlecar hoistway. The
With Enclosed Lobby
effective area from the elevator to the outside is For elevator pressurization systems intended for
fire evacuation, the elevator lobby is enclosed to help
protect people waiting for the elevator during a fire
emergency. The critical pressure difference, ApCri,, is
from the shaft elevator lobby to the building:

where Air is the leakage area between the building and the
.-
;0.20 - -50 a8 lobby in ft2 (m2). Equation (11.3) is the same as that for the
0 C
upper limit of pressure difference due to p~istoneffect in an
-40 g
G G unpressurized hoistway in Chapter 5 even though the two
a 15th Floor
0.10 - a
equations were derived differently. The effective area
-20 0 between the hoistway and the outside is
l
V)

0.05 - a
-10 .
0
0 10 20 30
lime (seconds)
0.30r .
where
A,, = leakage area between lobby and hoistway, ft2 (m'),
A,, = leakage area between the building and the outside,
(U
Z 0.15- -40 f ft2 (m2).
a Example 11.2 illustrates calculation of the critical
a, 0.10- 3
pressure difference for an elevator pressurization system
V) -20 2
,$! 0.05 -
8th Floor with enclosed lobbies.
-10
0 S M O K E CONTROL FOR PREVENTION O F
0 10 20 30
lime (seconds)
S M O K E MIGRATION
0.30r These systems consist of supplying air to the hoist-
way with the intent of producing a pressure difference
sufficient to prevent smoke flow into the hoistway in the
event of a fire. Upon fire detection, the general proce-
dure is for elevator cars to be taken out of nornial ser-
vice and automatically recalled to the ground floor.
Some elevators also have the capability for recall to an
alternate floor in the event of a fire on the ground floor.
In some localities, the elevator doors remain open after
the car reaches the ground floor or the alternate floor. In
0 - 0
0 10 20 30 other localities, the elevator doors are closed after suffi-
lime (seconds) cient time to allow passengers to leave the car. The fire
Measured service has elevator keys allowing them to operate cle-
- - - - - - Calculated
vators for rescue and for transportation of personnel and
Figure 11.2 Meas~tr-eda17dcalc~tla~ed p~sss~rt-e
differ- equipment to fight the fire.
El7CES d ~ t eto the pis~ot~ e/f^ect of an As stated for stairwell pressurization, the flow rate
-
ascet7dit7g eleiator cat-. of air is highly dependant on the leakage area. Because
Chapter 11-Elevator Smoke Control

Example 11.1 Piston Effect and Pressurization to Prevent Smoke Migration


A hoistway with two cars is pressurized to a minimum of 0.05 in. H20 (12.4 Pa) &orn the hoistway to the building. This
system is to prevent smoke movement through the elevator shaft, and there is no enclosed elevator lobby. Will the pres-
sure difference due to elevator piston effect be a problem? The parameters are: Asi = 1.52 f? (0.141 m2), Aio = 2.26 ft2

From Equation (l l.2), A, = 1.26 ft2 (0.117 m2).


From Equation ( l l. l), Q,&= 0.028 in. H20 (6.9 Pa).
The hoistway is pressurized at a level above Therefore, piston effect will not pull smoke into the hoistway.

Example 11.2 Piston Effect and Elevator Evacuation


I . A hoistway has two cars and is pressurized to a minimum of 0.05 in. H20 (12.4 Pa) t?om the elevator lobby to
the building. Wll the pressure difference due to elevator piston effect be a problem? The parameters are: A,, = 1.60
fi? (0.149 m2), A,, = 0.42 fi? (0.039 m'), A, = 0.54 (0.0502 m2), As = 121 'rf (1 1.2 m2), A, = 79.8 ft' (7.43 m2),
p = 0.075 lblf? ((1.20 ksjm3), U = 500 filmin (2.54 mk), C, = 0.94.

From Equation (l 1.4),A, = 0.325 fi2 (0.0302 m2).


From Equation (1 1.3),41,,.~,= 0.024 in. H 2 0(6.0 Pa).
The hois~\vayis pressurized at a level above 42,,i,. Therefore, piston effect will not pull smoke into the elevator lobby.

li 2. If the hoistway i n the esan~pleabo5.e is for a sinzle car, will piston effect be a problem? The parameters are the same
as above, cscept A, = 60.4 ft2 (5.6 1 ni'). A, = 19.4

From Equation (l 1.3),&C,.I,= 0.13 in. H 2 0 (33 Pa).


ft2 (13 0 In2),and Cc = 0.83. The effective area is the same.

The Iio~stwayis pressurized at a level below Q,,.;,.Therefore, piston effect may pull smoke into the elevator lobby. Pos-
sible solutions include a slower car speed. use of another elevator with multiple cars in the hoistway, and a higher level of
hoishvay pressul-ization.Also, Apt,., is an uppennost value, and a more detailed analysis might show that piston effect is
still not a problem. Fu~ther,piston etkcr lasts only a few seconds, and a hazard analysis could be used to evaluate the
ef'fect on life safety
Principles of Smoke Management

40 AP
(a) Wlnter (b) Summer
Figure 11.3 . Pressure dfference profile for pressurized elevator shaft
in idealized building with outside ester-ioi-doors oper7.

these areas can only be roughly estimated in most situa- the simple equations. This approach has the advantage
tions, the fan needs to be sized conservatively so that the of being able to account for complicated building flow
fan flow can be adjusted to acceptable levels of pressur- networks. Network models including CONTAM are dis-
ization during system commissioning. This fan sizing cussed in Chapter 8.
can be by choice of high values of building leakage or of Example 11.4 illustrates the use of CONTAM for
safety factors. analysis of a pressurized elevator system. Because of
flow through the stai~~vells and floors and friction losses
Analysis by Simple Equations in the shafts, the pressure profiles for this example (Fig-
The equations for analysis of pressurizzd stairwells ure 1 1.4) differ froni those for the ideal building without
presented in Chapter 10 can be adapted for use with vertical leakage (Figure 1 1.3).
pressurized elevators by redefining the subscript S in the
analysis from stairwell to hoistway. Such an analysis is With Stair Pressurization
then applicable for the idealized conditions listed below. Often elevator hoistways are pressurized in con-
junction with stairwell pressurization, as in Example
The only pressurization system is the pressurized I 1.5. This example is the same as Example 1 1.4 except
elevator. for the pressurized stairwells. Because of stainvell pres-
The flow areas of the building are the same from surization, the pressure profiles of Example 1 1.5 (Figure
floor to floor. ! 1.5) are closer to those of the ideal building without
The leakage between floors is negligible. vertical leakage (Figure l l 3 ) than those of Example
The flow through other shafts (stainvells, mail I 1.4 (Figure 1 1.4).
chutes, etc.) is negligible.
T h e friction pressure losses in the hoistway are ueg- SMOKE CONTROL FOR
ligible. ELEVATOR EVACUATION
Throughout most of the world, there are signs next
Figure 11.3 shows the pressure difference profiles to elevators indicating that they should not be used in
of a pressurized elevator in a building \\.ith exterior fire situations and that stairwells should be used for fire
ground floor doors open With the exterior doors open, e\.acuation. Thcse elevators are not i~dendedas means
the pressure on the ground floor is nearly the same as of fire egress, and they should not be used for fire evac-
that outdoors. For a mathematical description of this, uation. However, some peopie cannot- use stainvells
readers should see the section on effective flow areas in because of physical disabilities, and for these people,
Chapter 5. Example 1 1.3 is based on the pressure differ- fire evacuation is a serious problem (Pauls 1988; Pauls
ence, from the elevator to the buildinp being equal and Juillet 1989).
to the pressure difference, froni the elevator to the This section discusses smoke control systems that
outside at the ground floor. can be used to proyide smoke protection for elevators as
a part of an ovcl-all elevator protection scheme to allow
Analysis by Network Model fire evacuation by elevators. The information in this
Network computer programs can be used for analy- chapter is based on a joint project of the National Insti-
sis of systems without thesc simplifying conditions ol' tute of Standards and Technology (NIST) in the United
Chapter 11-Elevator Smoke Control

Example 113 Pressurized Elevator


hlculate
- - the
- -
air needed to oresswiz a h o h v durine summer conditiom with the mound floor exterior doon o f the building
f the shaft (or the vent is &tly closed), and the flow area of the vent can be neglecied.
. There is no vent at the too
-open.
i

m
I

Elevators

Stairwell
Typical Floor Plan
h e parameters are:
lumber o f stories 6
{eight per story 12 R (3.66 m)
lumber of ears i n hoistway 2
htside summer design temperame 89 OF (32OC)
3uilding design temperature 73 "F (23OC)
dinimum design pressure difference 0.05 in. H20(12.4 Pa)

'low Areas on Ground Floor


3ehveen elevator and building per floor 1.20 ft'
(0. lll m2)
3ehveenbuilding and outside per floor 42 (3.9 m')
:low Areas on Other Floors
3etween elevator and building per floor 1.20 (0.1 11 m2)
lehveen building and outside per floor 0.80 R? (0.074 m2)

h e analysis is done piecewise as described Tor pressurized elevators in Chapter 10. The two pieces are ( l ) the ground tloor and (1)the rest ot'!he hoisnvay.

Ihoose ,$gs8,
= 0.05 in. H 2 0 ( 1 2.4 Pa).

'ram Equation (l0.9).


2
ipsoi = MsB,[~ + (AsB/ABo)] = O.OS[I + (1.2/0.8) 1 = O.IW in. H20(40.6 pa)

ro= S9 + 460 = 549"R; Ts= 73 + 460 = 5 3 "R.

'ram Equation (10.5).

5 =K (L- = 7.64(&-A) = -0.000418 in. H201m.


"0

Ihe heightaltlieshaR is H = 6x12 = 72 fi(21.9 m).


t e a r n n ~ eEquation (10.4) to g t
bpsob = NSO,- hH = G.163 -(-0.000418)(72) = 0.193 in. H 2 0 ( 4 8 Pa). .

41 die ground floor, the en'ecdve area is

3 3
Ihe density ofair in the building is p = 0.075(530/533) = 0.0746 lblfi (Ililm ).

'ram Equation (10.16). the flow rrom the ground lloor is


V = N K ~ A ~ C J w b = (1)(776)(1.2)(0.65)J(2)(0.193)/(0.0746) = I380 cfm

The height orthe rest orthe hoistway is H = 5 X 12 = 6 0 ft(18.3 m).


Rearrange Equation (10.4) to get the pressure difkence at the second floor
ApSob = 4 ~- b H~= 0.163
~ -,(-0.000118)(60) = 0.188 in. H 2 0 (47 Pa)
i71caverage pressure dimerehce for this section is
Q', + '.lG3 0'188
- --
+

4'o,, = = 0.176 in. H 2 0 (43.8 Pa).


2 2
For tliese upper Iloors. the effective area is

Froni Equation (10.16). the llow in this section oTclcvator is


i' = N R ~ . - ~ ~ c JwP = (5)(776)(0.66h)(0.65)~(2)(0.17b)/(O.~746)
= 3650 ctin
Principles of Smoke Management

Example 11.4 Elevator Pressurization to Control Smoke Migration


Calculate the supply air needed for summer and winter design conditions to pressurize a 14-story hoistway.

Elevators

Stairwell
Typical Floor Plan
The design parameters and flow areas are:
Number of floors sewed by elevator 14
Location of hoistway supply air inlet Penthouse (l 5th floor)
Height between floors 12.0 ft 3.66 m
Outside winter design temperature 14°F -10°C
Outside summer design temperature 93OF 34°C
Building design temperature 70°F 2I0C
Winter stairwell temperature 45°F 7°C
Summer stairwell temperature 82°F 28°C
Minimum design pressure difference 0.05 in. H 2 0 12.4 Pa
Areas: ft2 m2
Leakage area of exterior building walls per floor 2.26 0.210
Flow area of two open exterior ground floor double doors 84 7.8
Leakage area between floors of the building 0.850 0.0790
Leakage area of stainvell walls to the building 0.11 0.0 102
Leakage area of stainvell walls to the outside 0.1 1 0.0102
Leakage area around closed single doors 0.25 2.32
Leakage area of hoistway walls to the building 0.074 0.00687
Leakage area around closed elevator doors 0.63 0.0585
Leakage area of closed vent at top of hoistway 0.20 0.0 186
Equivalent orifice area for friction losses in stain<-ell(see Chapter 6) 40 3.72
Equivalent orifice area for friction losses in hoist~r-ay(see Chapter 6) 1360 126
Flow coeficients for all leakage and flow areas, except open doorways 0.65
Flow coefficients for open doonvays 0.36
The program CONTAM was used for this analysis (output not shown). This program calculates pressures and flows throughout the
building. Because flow rates were needed that would result in a minimum pressure difference, the supply flow rate had to be changed
and the prograln renm a number of times until a supply rate was found that resulted in the desired minimum pressure difference. The
flow rates are:

cfm Lls
Winter 18.0SO 8530
Chapter 11-Elevator Smoke Conhol

II Exsmple 1 1 5 Elevator Ila'ktway and SraimeU Prr~suriration


Cakulate h e supply air nwded hrummet and winkr design m n h m to prcssurizc a 14-story hoistway and two Y.rwcllr The design parametem
and flow areas & those used i n Example 11.4.
I
As with h e previous example, flaws and pressures w e n calculated using CONAM. The flow rates are:
Elevator Hoistway Each Stairwell
cfm Us cfm Us
Winter 15,900 7.500 5,853 2760
Summer 13;380 6.320 4,660 7-200

€\ample 11.6 Elevator Pressurization for Elevator Evacuation with P w u r e Relief


Calculate the supply air needed for summer and winter design conditions to pressurize a 14-story elevator used for emergency fire evacuation. A vent
at the top o f the elevator is used for pressure relief.
I i
Elevators

Lobby Stairwell

III( Typical Floor Plan


The d e s i g paramerer; and flow areas are:
Number o f floors sensed by elevator I1
Location o f hoisruay supply air inlet Penthouse (15th floor)
Hsiglit between tloors 12.0 ft 3.66 m
Outside winter design telnpcrature 14'F -10°C
Outsidc sunimer design trinperature 93°F 54:C
Building design tenipcra~ure 70°F 21°C
Winter slainvcll Iemperature 4j°F 7°C
Sumnlcr stainvell tcmperaturc 82°F ?SzC
hlasimum design pressure dilkrsnce xross IL,r\by doors 0.52 in. H?O 79.6 Pa

Mininwndesi~n
pressure dilkrcnce across k5h!. doors 0.05 in. H?O 12.4 Pa

Design doors open durmp w i n m Ground floor and floors 12, 13, and 14
Dcsign doors opcn during summcr Ground floor and lloors 2. 5. and 4
Arus ft' m-
Lcakase area o f cslcrior building u-alls psr 2.26 0.110
I-Ion area oi1u.o open es~cnorground Iloor 2cuhlc doors S4 7.8
I.cakagc area bcrwen tloorr o f ~ l ~
buildinp
s 0.850 0.0790
Lcakagr area dslainvcll aalls to the bulldm< 0.11 0.0102
Leakage area orst:tinwll u i l s m h c outside 0.11 0.0102
I.cakagc arc8 aroul~dclosed iingls doors 0.25 0.0Zj
L c a h g c area ol'lwistu%iy walls 10 ~ h building
c 0.071 0.00657
L c l a g c area arou~idclosed clcvauir dams 0.63 0.0585
Leakage are;^ around closed lobby doors 0.50 0.0465
I'low area o f o p n lohby d w r s 42 3-90
Leakqe arca orclused vent at top ol'li0ist\\.3! 0.20 0.0186
Equivalent arilicc arm fbr iric~ionlosscs i n r:~in\.r.ll (*cc i-li:~ptcrh ] 40 5.72
Equivalcnl orilice arsa Tor friction losses i n bc:r.laa). (\cc Ci~:~plcr
6) l .X0 126
F l o cocllicicnt.;
~ (or all Ic3La:s and h v ; ~ r c s c\ccpt
. o l x n do\m\ays 0.65 .

Flow cocllicicnts for open doonra).s 0.16


The progran CONTAhl was used forthis anal! >is (output II~II s l a ~ u i ~As) . with li3n1plcs 11.4 arid 11.5. COWAM had to h:run aad rerun a numher
ofti~ncsu1obtain the suppl!. rat t l m rcsul~cdin 5 s desired i ~ i i n i ~ i ~pressure
wrl diltcrcnec. l l l e !h\\~ I\vith
LY lhc d c s i y nun1h.r ul'doors open are:
cfm IS!.
\\-in~c: 16.200 7.h5U
Summcr I(l.000 5.WlO
For I 6.200 cliii. tlic ~ ~ t i n i ~ nand
u n i111;1~inwni
pr:?wrc dill>renccs 3cross tllc Ik>hb>J<I(TS:re:
3linin1u111 Xl;lsin~a~r~
ill. H . 0 I'a in. 1 1 > 0 Pa
5 . .
\Vinicr 0 . lL:. .,>., ll..XJl 74.9
S111111iicr 0.25: 02.7 1l.luS 74.1
c s tt i11li11
IIIL.C prcwlrc d i l k r c ~ ~ c :ire t l ~ cd c s i :~: i ~i i i t ~ w ~;d
~ III:~~~
KIU~
CI. ~ I I I I
-
164
Principles of Smoke Management

L 2 U

0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4


Pressure Difference (in. H,O) Pressure Difference (in. H,O)
(a) Elevator pressure differences in winter (a) Elevator pressure differences in winter

Pressure Difference (in. H,O) Pressure Difference (in. H,O)


(b) Elevator pressure'differences in summer (b) Elevator pressure differences in summer

Note: Stairwells
are not pressurized.

L
I I
G] I
~;0'12 -0.06 0 0.06 0 0.05 0.10 0.15 0.20
Pressure Difference (in. H,O) 'Pressure Difference (in. H,O)
(c) Pressure difference from stairwell to building (c) Pressure difference from stairwell to building

Figure 11.4 P~zssuredifference profiles calculated by Figure 11.S Pressure difference profiles calculated
CONTAM for a presszrrized elevator in by CONTAM for a presszrrized elevator
bzrilding with outside exterior doors open in building with outside exterior- doors
(Esa/nple 11.4). open (Examnple I 1 S).

States and the Natio~ial Research Council of Canada scheme for fire evacuation, and fire evacuation by these
(NRCC) to evaluate the feasibility of using elevators for conventional elevator systems is not recommended.
the evacuation of the handicapped during a fire (Klote
and Tarnura 1987, 1986a, 1986b; Tamura and Klote Concerns about Elevator Evacuation
1988, 1987a, 1987b). Before this joint project, Klote
(1984, 1983) conducted field tests of several elevator This section provides a description of many con-
pressurization systems. It shoi~ld be emphasized that cerns about elevator evacuation, and the nest section
conventional elevators do nor ha\-e any protection discusses these concerns along with one approach to
Chapter 11-Elevator Smoke Control

deal with them. The 1976 edition of the Life Safity Code Elevator lobbies, hoistway, a i d elevator machinery
(NFPA 1976) listed the following "problems" involved room must be protected against fire and smoke.
with the use of elevators as fire exits:* Elevator equipment and electrical power must be
protected from the water exposure of sprinklers and
Persons seeking to escape from a fire by means of fire hoses.
an elevator may have to wait at the elevator door for Elevator machine room must be protected from
some time, during which they may be exposed to overheating.
fire, smoke, or developing panic. Reliable electric power must be supplied.
Automatic elevators respond to the pressing of but- In areas of high seismic activity, elevator equipment
tons in such a way that it would be quite possible must be protected from earthquakes.
for an elevator descending from floors above a fire The likelihood that elevators will be available dur-
to stop automatically at the floor involved in the fire ing fires needs to be ensured by use ofmultiple cars
and open automatically, exposing occupants to fire or by quick response maintenance contracts.
and smoke. Elevator control must ensure safe and efficient
Modern elevators cannot start until doors are fully evacuation.
closed. A large number of people seeking to crowd Communications capabilities are needed between
into an elevator in case of emergency might make it people waiting for elevators and the fire service or
impossible to start. appropriate building personnel.
Any power failure, such as the burning out of elec- Evacuation capacity of the elevator system must be
tric supply cables during a fire, may render the ele- adequate for the number of people intended to use
vatois~noperativeor might result in trapping the system.
persons in elevators stopped between floors. Under
fire conditions, there might not be time to'pemiit As previously stated, elevator cars are controlled so
rescue of trapped occupants through emergency that they go to the ground floor in the event of a fire
escape hatches or doors. alarm. In the event of fire on the ground floor, the eleva-
tor cars go to an alternate floor. The fire department or
I t is common for elevators serving more than three other authorized personnel can then use the elevators for
floors to dsscend autoniatically to the ground floor in evacuation. Firefighters, police, and uniformed guards
the e\;ent of a fire. Fire fighters have keys to control ele- have positions of authority in our society. With the ele-
vators rnaiiually during building evacuation and fire vators controlled by such authority figures, the likeli-
fighting. However, smoke infiltration into hoistways hood of a large number of people crowding into the
frequently threatens lives and hinders use of elevators elevator and making it impossible to close the doors will
by fire fighters. probably be reduced. Of course, there may be other
approaches to elevator control that could allow orderly .
In addition, there are three other concerns. First,
evacuation by elevators.
water from sprinklers or fire hoses could short out or
Reliability of electric power consists of ensuring a
cause other problenls with electrical power and control
source of power and ensuring continued distribution of
wiring for the elevator. Second, shah pressurization
power to where it is used. Considerable experience
cocld result in elevator doors jammhg open, limiting
exists in ensuring the supply of electrical power for crit-
movement of the car. Third, piston effict could pull
ical functions in hospitals, communication facilities,
smoke into the elevator lobby or thc hoistway, and a
computer facilities, and the like. For these applications,
method of preventing this has already been presented in
a major concern is providing backuppower when power
this chapter.
supplied by the local utility is interrupted. These appli-
cations operate most or all of the time, and they need
Conceptual Solution for Elevator Evacuation highly reliable power for all the time that they operate.
The feasibility of elevator evacuation for office Fire evacuation by elevators is different in that this
buildings and air traffic control towers is discussed by mode of elevator operation is only needed during a
Klote et al. ( 1992. 1994). In order to overcome the con- building fire. At most, the fire evacuation mode of an
cerns discussed in the preceding section, an elevator elevator would be expected to operate for a few hours
system used as a lire exit needs to have the following per year. Thus, the probability of simultaneously having
a fire and having the utility company's power inter-
rupted is relatively small. However, the probability of
S. This c'dition ol'the L i f i So/./,r Code \\as the last having a power distribution failure during a fire is rela-
cditioii to list rhcsc "problen~s." tively high. This is because lire frequently damages
Principles of Smoke Management

electrical distribution within- buildings. Therefore, the tests conducted by Klote (1 984), no door jamming was
power distribution to the elevator and associated smoke encountered at pressure differences as high as 0.3 in.
control fans should be such that it is highly unlikely that H 2 0 (75 Pa). When door jamming was encountered in
a fire could interrupt electrical power to this equipment. an elevator without smoke control, it was found that
T h x e are numerous applications of electric power only a small additional force applied by the palms of the
and electronic systems being designed and built to func- hands was sufficient to prevent jamming. Fire fighters
tion when in contact with water. Street lighting and traf- can be taught to overcome door jamming this way, and
fic lights operate during rain, -and swimming pool elevator doors could be fitted with grips or handles to
lighting operates underwater. In fact, some elevators aid in this effort.
operate on building exteriors where they are subjected
to rain and the other elements. It is beyond the scope o f Smoke Control Considerations
this manual to examine specific approaches to making Smoke control systems for elevator evacuation
these systems resistant to water; however, it is obvious n u t provide smoke protection for elevator lobbies,
that the technology exists to make elevator systems hoistways, and machinery rooms. Protection of lobbies
function when they are subjected to water. is essential so that people will have a safe place to wait
Considerable information is available concerning for the elevator. Protection of the machinery room is
the fire resistance of walls, partitions, floors, doors, etc. important to prevent damage to elevator machinery. Fig-
The ability to design and build elevator lobbies and lire 11.6 illustrates a system that pressurizes the hoist-
hoistways-that can withstand severe building fire has way directly and indirectly pressurizes the elevator
existed for years. Smoke protection for elevator systems lobby and the machinery room.
is the topic of the next section. As stated for other pressurization systems, the flow
Elevator doors jam open wl:x the force of the door rate of air is highly dependant on the leaka,oe area.
opener is insufficient to overcome the force .of friction. Because these areas can only be roughly estimated in
The friction force increases with the pressure difference most situations, the fan needs to be sized conservatively
from the hoistway to the lobby. In tall buildings, eleva- so that the fan flow can be adjusted to acceptable levels
tor doors frequently jam open during extremely cold of pressurization during system commissioning. This
weather. This is caused by stack effect induced pressure fan sizing can be by choice of high values of building
differences. Elevator mechanics commonly adjust the leakage or of safety factors.
door-closing forces to prevent door jamming. During Pressurization air can also be supplied to the eleva-
elevator smoke control operation. the possibilih of door tor lobbies. Examination of the relative leakage areas of
jamming may decrease or increase. If the leaka,me area the elevator system provides insight into both hoistway
of the elevator lobby doors is less than that of the eleva- and lobby approaches to pressurization. Considering the
tor doors, the pressure difference across the elevator leakage from the elevator lobby to the outside to be neg-
doors can be less than that normally occurring. In field ligible,

Machinev
Room I
,,Lobby
Machinery
Room I
/Lobby
/
Building
P
Space

$
.-
I
3:
4-
-I-
Car
-
1' 1: I
Fan

, ,;, ;'
, 4.. ,., ':.:.
,,;/ ,<', ,,y A
5 Note: The supply duct on
Pit
- the.floor needs to be in a fire
rated enclosure.
(a) Shaft Pressuriiation (b) Lobby Pressurization
Chapter 11 -Elevator smoke Control

in a pressure drop from 0.13 in. H 2 0 (32. Paj to 0.03 in.


H20 (7 Pa) for a system without features to resist pres-
sure fluctuation. . .

where During a fire, it is expected that several exterior


pressure difference from hoistway to lobby, in. doors will be propped open, and the elevator doors will
&sr =
open and close as elevators are used for evacuation. Fur-
H20 (Pal; ther, stairwell doors are likely to be opened and closed
&.
11-
= pressure difference from lobby to building, in. as people use them for evacuation. It is envisioned that
H20 (Pal; lobby doors will close automatically upon smoke con-
trol system activation. However, lobby doors can be
A, = leakage area between the building and the lobby
inadvertently blocked and the closing mechanism can
in f? (m2); fail. It is anticipated that occupants will close any such
A,. = leakage area between the lobby and the shaft, ft2 opened lobby doors to prevent being exposed to smoke.
Doors may not be closed on floors where there is no
(m2). smoke danger or there are no people waiting in the ele-
For elevator doors with wide gaps that are common vator lobby. The smoke control system should be
in most buildings, the gap areas are generally in the designed to maintain pressurization when some elevator
range of 0.34 to 0.72 ft2 (0.032 to 0.067 m2), as shown lobby doors are open on floors away from the iire. The
in Table 6.3,. Based on general experience with building examples presented later deal with pressure fluctuations
leakages, Ai,/A, is about 0.4 for construction of average due to doors opening and closing.
tightness and about 0.1 for tight construction. From
Equation (1 1 .S),&&,.l Api;.is therefore 0.16 and 0.01 for Smoke Control Systems
average and tight construction. Thus, the pressure in the Elevator smoke control systems can incorporate
elevator lobby can be expected to be close to the pres- features to deal with pressure fluctuations due to open-
sure in the hoistway, provided that the construction is ing and closing doors. An alternative is a pressurized
not unusually leaky. Pressurization air can be supplied hoistway without provisions for these fluctuations such
to the elevator lobbics (Figure 11.6b). However, from that smoke transport through the hoistway is insignifi-
the above discussion, it seems that this direct lobby cant when evaluated by a hazard analysis. Features for
pressurization does not resuh in any significant dealing with pressure fluctuations include pressure relief
improvement in pressurization over supplying the air vents, vents with barometric dampers, variable-supply
into the hoistway, as illustrated in Figure I I h a . air fans, fire floor venting, and fire floor exhaust.
Direct lobby pressurization has some advantage
over direct hoistway pressurization in purging small Pressure Relief Vent System
amounts of smoke from the lobby. Part of the pressur- This system has a constant supply air rate fan and a
ization air to an elevator smoke control system goes pressure relief vent to the outside, as illustrated in Fig-
from the hoistway to the outside, and the rest goes from ure 11.7. The area of this vent is fixed and sized for
the lobby through the building to the outside. With operation in the smoke control system. The vent can be
direct lobby pressurization, both of these amounts flow fitted with automatic dampers if it is desired for it to be
through the lobby. Such an increased tlow rate tends to normally closed. The supply rate varies to some extent
better purge any small amounts of smoke that would get with the pressure across the fan, but the term "constant
into the lobby-before smoke control activation or when a supply" is used to differentiate this fan from one that has
person is entering the lobby. The relative benefit of this a variable supply rate. The vent must be large enough
improved purging con~paredto its cost has not been that the maximum allowable pressure difference is not
evaluated. The following discussions have been focused exceeded when all doors are closed. When paths to the
arbitrarily on the hoistway pressurization systems. outside are opened, air flows through then1 and the
hoistway pressure drops. This system must maintain at
Pressure Fluctuations Due to Open Doors least the minimum allowable pressure difference when
Elevator systems must be designed to maintain some design combination of paths is open.
design pressure differences under the likely conditions
of opened and closed doors. Klote and Tamura (198Ga) Baromekic Damper System
showed that opening a large flow path from the pressur- This system is siniilar to the one above except that
ized spaces to the outside can result in a signiticant loss the vent has a barometric dampcr that closes when the
in pressurization. For example, opening the elevator pressure drops bcloiv a specified value. The use of these
doors, elevator lobby doors. and exterior doors resulted dampers mini~iiizcs air losscs when paths from the
Principles of Smoke Management

Machinery
Room I
~ e l i eVent
f
(This vent can
be an open
vent or a
barometric
damper.)

be in a fire rated shaft.


., ,

Figure 11,7 Elevator smoke control with a pressure Figure 11.8 Elevator smoke control with fire jloor
relief vent. exhaust.

hoistway are opened, and the pressurization fan can be


sized smaller than for the above system. A normally System with Fire Floor Venting or Exhaust
closed automatic damper in parallel with the barometric
Smoke venting and smoke exhaust of the fire floor
damper can prevent damper chatter caused by the wind.
can improve system performance. The venting or
exhaust increases the pressure difference from the lobby
Variable Supply Air System to the fire floor. The vents can be exterior wall vents or
nonpowered smoke shafts. Figure 11.8 shows a fan-duct
Variable supply air can be achieved by using one of system intended to exhaust the fire floor. Upon detection
many fans commercially available for variable flow of fire or smoke, the damper opens on the fire floor and
rate. Alternatively, a fan bypass arrangement of ducts the exhaust fan is activated. The detection system must
and dampers can be used to vary the flow rate of supply be configured to identify the fire floor.
air to the hoistway. The variable flow fans are controlled
by one or more static pressure sensors that sense the
pressure difference between the lobby and the building. Design Analysis
There are two approaches for use of the sensors. The air-
There are many different approaches that can be
flow rate can be controlled by the average of all signals
taken to the design of the systems discussed above. The
from the sensors or it can be controlled by the signal design of an elevator smoke control system includes the
from the fire floor.
selection of a system for dealing with pressure fluctua-
Using the average of all the signals has the advan- tions, determining appropriate values for leakage areas
tage that no information is required about where the fire andother parameters, as well as calculating the perfor-
is located. u s i n g the fire floor sensor signal requires mance of the smoke control system. The objective of the
intormation about the fire location. This information can design analysis is to determine a flow rate of pressuriza-
come from smoke detectors, heat detectors, or sprinkler tion air that ~ v i l lresult in acceptable pressurization with
water flow indicators. Using the fire floor signal has the a minimum and a maximum design number of large
advantage that the system maintains a set pressure dif- open paths from the hoistway to the outside. Example
ference at this most critical location. 11.6 illustrates design of a pressure relief system.
CHAPTER 12

Zoned Smoke Control


he stairwell systems and elevator systems dis- Open and Divided Spaces
cussed in previous chapters were primarily Zoned smoke control works for smoke zones that
intended to prevent smoke infiltration into these are open spaces, such as open plan offices. Zoned smoke
shafts. However, smoke can flow through cracks in control is more complex for potential smoke zones that
floors and through unpressurized shafts to damage prop- are divided into a number of separate spaies. Examples
erty and threaten life at locations remote from the fire.
The concept of zoned smoke control discussed in this
chapter is intended to limit this type of smoke move-
ment within a building.

SMOKE CONTROL ZONES


A building can be divided into a number of smoke
zones, each separated from the others by partitions and
floors. In the event of a fire, pressure differences pro-
Note: In these figures.
duced by mechanical fans are used to limit the smoke the smoke zone is indicated
spread to the zone in which the fire initiated. The con- by a minus sign and
pressurized spaces are
centration of smoke in this zone goes unchecked. indicated by a plus sign. A
smoke zone can consist of
Accordingly, in zoned smoke control systems, it is one floor as in (a) and (b) or
intended that occupants evacuate the smoke zone as of more than one floor as in
(c). All the nonsmoke
soon as possible after fire detection. zones in a building may be
pressurized as in (a).
Frequently, each floor of a building is chosen to be or only the nonsmoke zones
a separate smoke control zone. However, a smoke con- adjacent to the smoke zone
may be pressurized as in (b).
trol zone can consist of more than one floor, or a floor (c) and (d). A smoke zone
may be part of a floor as in (d)
can consist of more than one smoke control zone. Some
arrangements of smoke control zones are illustrated in
Figure 12.1. When a fire occurs, all of the nonsmoke
zones in the building, or only zones adjacent to the
smoke zone, may be pressurized. When the fire floor is
exhausted and only adjacent floors are pressurized, as in
Figure 12. I b, the system is sometimes called a "pressure (d)
sandwich." Figure 12.1 Some at-rangernenrs of smoke conlr-01
rones.
Chapter 12-Zoned Smoke Control

of divided spaces are nursing wings of a hospital, a floor


of a hotel, and a floor of an apartment building. There
has been some success with using divided spaces as
smoke zones when the entire divided space is also an
HVAC zone, but care needs to be taken to ensure even
distribution ofsupply and exhaust air within each smoke
zone during smoke control system operation.
Zoned smoke control has also been used in con-
junction with passivi smoke control (compartmenta-
tion). An example is an office building where the
occupant floors are the smoke zones of a zoned smoke
control system, and the mechanical room and elevator
machine room are protected by compartmentation. Figure 12.2 Flowpattern due to smoke venting.
Another example is a hotel where the corridors are the
smoke zones, and the guest rooms are protected by corn-
partmentation. Analysis of compartmentation smoke all of the exterior walls. For buildings that are much
management is discussed in Chapter 9. longer than wide, the vents can evenly be divided
between the two long sides. Exterior wall venting is
SMOKE ZONE VENTING most appropriate for buildings with open floor plans and
vent& of smoke from the smoke zone is important least suitable when the floor plan is divided into man!.
because it. prevents significant overpressures due to compartments. Because the flow of hot gases through a
thermal expansion of gases as a result of the fire. Vent- wall vent can be substantial, precautions should be taken
ing can be accomplished in the folloning three ways: in the design of exterior walls to minimize the possibil-
ity of exterior fire spread to floors above the vent.
exterior wall vents,
smoke shafts, and Vent Areas
mechanical venting (or exhaust).
The following is a method for evaluating the size of
exterior wall vents presented in essentially the sams
When the first two methods of venting are used, it
form as originally developed by Tamura (1978a). In this
is essential that adjacent zones (or all nonsmoke zones)
analysis, each floor consists of a smoke zone. For the
be pressurized in ordzr to maintain prsssure differences
analyses presented in this and the following section, ths
at the boundaries of the smoke zone. Mechanical
effects of fire are indirectly incorporated in the selection
exhaust by itself call result in sufficient pressure differ-
of minimum design pressure difference (Chapter 6). For
ences for smoke control. However, in the event of win- this system, the fire floor (smoke zone) is vented to thc
dow breakage or a large opening to the outside from the outside, supply and exhaust fans serving the fire floor
smoke zone, n~echanicalexhaust might not be able to are shut off, and the floors above and below the firc
ensure favorable pressure differences. floor are pressurized.
Smoke purging, consisting of equal air supply and Air flows from floors adjacent to the fire iloor and
exhaust rates, is not considered here because it does not through the vent to the outside, as illustrated in Figurs
produce pressure differences that control smoke move- 12.2. Because the supply and exhaust fans are shut off
ment. It is generally believed that such purging at the on the fire floor, the total airflow rate through the wall
airflows available with HVAC systems cannot signifi- vents equals the total flow rate into the vented floor
cantly reduce smoke concentrations in a zone where a from the surrounding smoke control zones:
large tire is located. Dilution away from the fire is dis-
cussed in Chapter 6.

EXTERIOR WALL VENTS where


Exterior wall vents can consist of n.indo\vs or pan- A,, = flow area of the exterior vent, ft2 (m');
els that open automatically when the smoke control sys-
tem is activated. The system considered here consists of A, = etrective
..
flow area of the enclosure of the smoke
a ventcd smoke zone without any mechanical exhaust zone to the other zones, ft2 (m');
and zdjacent zoncs that are pressurized. p~ = smoke zone pressure, in. H20 (Pa);
In order to minimize adverse efkcts of wind, the
area of wall vents should be evenly distribirted among p. = outside pressure, in. H 2 0 (Pa);
Principles of Smoke Management

pB = building pressure on nonsmoke zcnes, in. H20 Substituting the above into Equation (12.2) and
(W. rearranging yields
The effective flow area, A,, includes the flow areas
of the walls of vertical shafts, floor constructions, and
duct openings (return and exhaust) of the smoke zone.
Effective flow areas are discussed in detail in Chapter 6.
Rearranging Equation (l 2.1) yields A plot of Equation (12.3) is shown in Figure 12.3.
\ This shows that for particular values of ApBo and A,, the
pressure difference, ApBA across the boundary of the
smoke zone increases as the vent area, A, increases. For
large values of A , ApBFapproaches ApBO.
Opening a stairwell door on a floor of a nonsmoke
zone increases the pressure difference across the closed
stairwell door on the fire floor (smoke zone). This can
be explained by use of the concept of the effective flow
area (Chapter 6), and it is left to the reader as an exer-
cise. Opening doors in a stairwell on both a nonsmoke
zone floor and the smoke zone floor results in consider-
able airflow to the smoke zone, which is accompanied
where by reduced pressure difference across the boundary of
the smoke zone.
ApBo = pressure difference fi-om rhe nonsmoke zones
to the outside, in. H20 (Pa): Esample 12.1 Vent Areas and Pressure Differences
ApBF = pressure difference from the nonsmoke zones 11 I . If the ratio of A,/& is 1, what is the ratio of
to the smoke zone, in. H20 (Pa); Q g ~ / A p g o?

ApFo = pressure difference from rhe smoke zone to the From Equation (12.3), A p B F / ~ p B O= 0.5. Thus, the
outside, in. H20 (Pa). pressure difference across the boundary of the smoke
zone is only half that from the building to the outside.
Then

l 2. If A,/A, is 2, what is Q B F / A p B o ?
From Equation (12.3), A p g F / A p B 0 = 0.8. This is much

I 3. IfAJA, is 3, how does A p B r / ~ p B ochange?


From Equation (12.3), ApBF/ApBo = 0.9.

Pressurization Airflow Rates


The effective flow area, A,, of the enclosure of the
smoke zone to the other zones usually consists of the
sum of the flow areas between the smoke zone and
many other nonsmoke zones. This is expressed as

wlltre
4, = etfective flow area of the enclosure of the smoke
zone to the other zones, ft2(m2);
11 = number of adjacent nonsmoke zones;
Chapter 12-Zoned Smoke Control

ABFi = flow area between nonsmoke zone i and the For an unsprinklered fir.e, the gases leaving the
smoke zone, f? (m2). smoke zone are likely to be relatively hot. However, the
flows in question are both from the nonsmoke zones,
Considering steady flow conditions, the mass flow rate of
which are probably near building temperature. Consid-
pressurization air entering a nonsmoke zone equals the
ering the very approximate nature of flow area esti-
flow rate of air leaving the zone: mates, the errors involved in using volumetric flow rates
( 12.5) at standard conditions are not significant. Such equa-
tn,; = rizBFi + rilBOi tions are
where
mti = mass flow rate of pressurization air into zone
i, Ibis (kg$); where

mass flow rate from zone i to the smoke vBFi = volumetric flow rate from zone i to the smoke
mBFi =
zone, lb/s (kg/s); zone, cfm (m3/s);
ABF; = flow area between nonsmoke zone i and the
z B 0 = mass flow rate from zone i to the outside, Ib/s
smoke zone, ft2 (m2); '
(kds).
= pressure difference from the nonsmoke zones to
QBF
The flow rate from zone i to the smoke zone can be
expressed in the fomi of the orifice equation, the smoke zone, in. H20 (Pa);
Kj. = coefficient, 2610 (0.839);
and
where
'GFi- mass flow rate from zone i to tlie smoke zone,
where
Ibis (kds);
C = dimensionless flow coefficient; ifBOi = volunietric flow rate from zone i to the outside,

A ~ ~
= flow
i area between nonsmoke zone i and the cfm (n13/s);
smoke zone, ft2 (m2); ABOi = flow area between nonsmoke zone i and the

density of air in flow path, lb/ft3 (kglm3); outside, ft2 (m');


P =
= pressure difference froni the nonsmoke zones
WBF= pressure difference from the nonsmoke zones to
to the outside, in. H 2 0 (Pa);
tlie smoke zone, in. H 2 0 (Pa);
= coefficient, 2610 (0.839).
K,,, = coefficient, 12.9 (1.00).
The pressure difference from the nonsrnoke zones
Similarly, tlie mass flow rate to the outside is can be obtained from Equation (12.3) as

where
IilBOi - mass flow rate from zone i to t11eoutside, Ibis where
(kds); qjBo = pressure differencefrom the nonsmoke zones to
C = dimensionless flow coefficient; the outside, in HzO(Pa);
ABO~ - flow area between nonsmoke zone i and the MsF = pressure ditkrence from the nonsmoke zones to
outside, ft 2 (m2) the smokc zone, in. H 2 0 (Pa);

P - density of air in flow path, lb/ft3 (kglm3); A,. = tlow arcn of the exterior vent ofthe fire floor, ft2
4 ~ 0
= pressure difference from the nonsmoke zones (111~);
to tlie outside, in. H 2 0 (Pa); 4 , = efl'cctive flow area of the enclosure of the
K,,, = coeflicient, 12.9 (1.00). snioks zonc to the other zones. ft2 (ni 2).
. .

Principles of Smoke Management

Example 12.2 Supply Air and Exterior Wall Vents MECHANICAL EXHAUST
The smoke zone of a zoned smoke control system is to have Mechanical exhaust 'of the smoke zone can be
exterior wall vents and two adjacent nonsmoke zones. Supply accomplished by either a dedicated exhaust system or
and return are shut off to the smoke zone, and the adjacent
by the exhaust fans of the HVAC system. Generally,
zones are pressurized. The nonsmoke zones have the same
flow areas: such exhaust is done in conjunction with pressurization
ABol = ABO2 = 4.5 f:(0.42 m2) and ABFl = ABR = 3.0 of nonsmoke zones. These systems can also include _
stairwell pressurization.
(0.28 m2). Use AJA, = 2, and ApBF = 0.10 in. H20 (25 Pa).
Mechanical exhaust by itself can result in sufficient
How much pressurization air is needed?
pressure differences to control smoke. However, in the
From Equation (12.4), 4 = 3.0 + 3.0 = 6.0 ft2 (0.56 m2). event of window breakage or another large opening to
the outside from the smoke zone, the pressure differ-
From Equation (12.10), ApBo = 0.10((1+2~)/(2~))
= 0.13 in
ences can decrease significantly. For this reason,
H20(3 1 Pa).
mechanical exhaust alone does not constitute an ade-
From Equation (12.8), GFI= 26!0(3)(0.1)% = 2500 c h quate smoke control system when there is a significant
(1.2 m3/s). probability of window breakage or an opening from the
smoke zone to the outside.
From Equation (12.9), = 2610(4.5)(0.13)" = 4200
In the smoke zone, the location of the exhaust inlets
cfm (2.0 m3/s).
is important. These inlets should be located away from
The supply air is 2500 + 4200 = 6700 cfm (3260 Us).
exit stairs so that smoke in the vicinity of the shaft inlet
SMOKE SHARS does not pose an increased hazard during evacuation or
fire fighting. Because hot smoke frequently stratifies
A smoke shaft is a vertical shaft intended to be a near the ceiling, it is recommended that smoke exhaust
path for smoke movement from the fire floor to above inlets be located in or near the ceiling.
the level of the roof. Generally, the driving force of Exhausting air from the smoke zone results in air
smoke movement is buoyancy, although the flow from the outside and from other zones being pulled into
through some smoke shafts is aided by mechanical fans. the smoke zone. This air flowing into the smoke zone
This mechanical exhaust is addressed in the next sec- can provide oxygen to the fire. Most commercial air-
tion. A smoke shaft can serve one floor, a group of conditioning systems are capable of moving about four
floors, or all the floors in a building. Smoke shafts have to six air changes per hour, which probably accounts for
openings above the roof level and on the floors they the popularity of six air changes in smoke control appli-
serve. These openings are fitted with dampers that are cations. Current designs are based on the assumption
nomially closed. In a fire situation, only the damper on that the adverse effect of supplying oxygen at six air
the fire floor and the top outside damper open to vent changes per hour is insignificant in comparison with the
smoke outside. Smoke shafts should be constructed in benefit of maintaining tenable conditions in zones away
accordance with local codes. Tamura and Shaw (1973) from the fire. Thus, six air changes is recommended as
provide inforniation concerning sizing of smoke shafts. the upper iimit for exhaust airflow.
Smoke shafts used in conjunction with pressurization of In any analysis of a smoke control system, the fire
nonsmoke zones can produce pressure differences to effects of buoyancy and expansion need to be addressed.
restrict smoke to the smoke zone. This can be done directly as part of the analysis or indi-
rectly. As discussed in Chapter 4, the indirect approach
Smoke shafts lend themselves to use in buildings consists of establishing a minimum design pressure dif-
with open floor plans. The air movement caused by ference that will not be overcome by buoyancy pres-
smoke shafts operating during normal siack effect tends sures resultilig from smoke at design temperatures. This
to pull smoke toward the smoke shaft inlet on the fire indirect approach is much simpler, and so human errors
floor. It is recommended that smoke shafts be located as in analysis, other aspects of design, construction, and
far as possible from exit stairwells, so that smoke in the commissioning are less likely. The following sections
vicinity of the shaft inlet does not pose an increased haz- present both methods.
ard during evacuation or fire fighting. Because hot When the temperatures on both sides of the bound-.
smoke frequently stratifies near the ceiling, it is recom- ary of the smoke zone are the same, the pressure differ-
mended that snioke shaft inlets be located in or near the ence across the boundary is the same over the height of
ceiling. the barrier. This is the condition under which smoke
Chapter 12-Zoned Smoke Control

Exhaust Fan Temperature


The mass flow through the exhaust fan is

where
= mass flow rate of exhaust fan, Ibls &$S);
me

p& = density of gases in exhaust fan, 1b1@ (kg/m3);


(a) Pressure Difference Without Fire =volumetric flow rate of exhaust fan, ft3lmin
(m3/s);
K, = 0.01667 (1.00).
The effect of fan temperature on smoke control sys-
tem performance is of concern. Fans are approximately
constant volumetric flow rate devices. Using the ideal
gas equation, the mass flow rate through the fan is a
function of the absolute temperature of the gases in the
fan.

(b) smoke Control System Prevents Infiltration


lnto Protected Spaces
where TI, is the absolute temperature of the gases in the
exhaust fan in "R (K). Increased fan temperature decreases
the mass flow rate of the exhaust fan, resulting in a reduc-
tion in smoke control system pressurization.
The maximum allowable fan temperature can be
calculated as

where
qa,= absolute temperature of the gases in the exhaust
fan, "R (K);
T,. = absolute temperature of the gases in fan under
nonnal conditions, "R (K);
= allowable fraction reduction in mass flow rate
through fan.
control systems are almost always tested. When the Example 12.3 Fan Temperature
., in the snioke zone are '%ot," the buoyancy of the
oases If a reduction of 20% in the mass flow rate is acceptable,
hot gases results in a nonuniform pressurz difference. what is the maximum allowable fan temperature?
Figure 12.4a is a uniform pressure difference at the niin- I

imuni design value. This minimum value is selected The paraiiieters are: T,. = 70 "F + 460 = 530 "R, b= 0.2.
such that positive pressurization of the snioke zone con- From Equation (1 2.13), G,,, = 5301(1 - 0.2) = 663 OR or

tinues, pro\.ided that the mass flow from nonsmoke


zones to the srnoke zone remains unchanged and that the
From Example 12.3, a 20% reduction in mass flow
snioke zone does not exceed its design temperature through the fan occurs at 203°F (95OC). When man!.
(Figure 12.4b). However, if this mass flow decreases. I-IVAC systems are used for smoke control, they exhaust
srnoke may flow into the "protected" spaces, as illus- air from all or niost of the rooms on a floor. Thus, hot
tire gases and lower temperature air from remote rooms
trated in Figurc 1 2 . 4 ~Thc
. method of anal\.sis presented
are mixed, and the fan temperature is much lower than
i n the follon'ing scction allows evaluation ol' this that of the firc gases. Also, heat transfer. from the
decreased Inass ilow rate. exhaust duct lowers the fan temperature.
Principles of Smoke Management

Table 12-1: ~ ~ ' 7 ' ~ A fire space is a room or a corridor that is filly
Typical Gas Temperatures and Densities for involved in fire. A communicating space is one that is
Severe Building Fires connected to the fire space by an open door or other
Temperature Density
large opening. A removed space is a room or other space
connected to a communicating space by an open door or
"F "C lblfe kg/m3
other large opening. The removed space is not con-
Fire space is a room or corridor 1700 927 0.01 84 0.294
h l l y involved in fire nected to the fire space or is only connected to it by very
Communicatirrgspace is a 800 427 0.0315 0.504 small cracks or gaps. A separated space is a space that is
room or other space connected not connected to any of the three spaces above, or it is
to the fire space by an open door only connected to them by very small cracks or gaps.
or other large opening
Removedspace is a room or 400 204 0.0462 0.739 To determine the extent of each type of space, a
other space connected to a com- floor plan should be evaluated in light of likely locations
municating space by an open of fires, doors likely to be opened, and doors likely to be
door or other large opening. The
removed space is not connected
closed. ~ r o mExample 12.4, it can be seen that cool air
to the fire space or is only con- from the separated spaces mixes with the hot gases from
nected to it by very small cracks other spaces and cools them. If the fan temperature is
or gaps too high, the zone can be increased in size so that air
Separated space is a room or 80 27 0.0736 1.18 from separated spaces will further dilute the hot gases.
other space not connected to any
ofthe three spaces above, or it is
Example 12.4 Fan Temperature and
only connected to them by very
Smoke Control Zone Size
small cracks or gaps
A smoke control system has exhaust rates from the following
The temperature of the gases in the fan can be con- spaces:
servatively estimated by considering dilution of hot cfrn
gases with cooler gases and neglecting heat transfer. Fire space: 400 0.189
Considering constant specific heat, the fan temperature Communicating S00 0.378
can be expressed as space:
Removed space: l SO0 0.850
Separated space: 6000 2.83
Table 12.1 provides descriptionsof these spaces, gas temperatures, and
densities. Will the fan tempsrature have a signiticant adverse cKec~of
the performance of the system'?

From Example 12.3, the fan temperature can be 203°F (95°C) or less
and the effect on system performance will be acceptable.
where
Tbr7= temperature of the gases in the exhaust fan, "F
("C);
pi = density of gases in space j, lblfi? (kg/m3);

V = volumetric flow rate of exhamt from space j, cfm

= temperature of gases in space j, "F ("C);


n = number ofspaces. From Equation (l 2.14).
Table 12.1 lists typical gas temperatures and densi-
ties for severe building fires that can be used in Equa-
tion (1 2.14). The following idealized types of spaces are C
addressed in this table: '= I
Tfan = ' ,, - 101300- IS2 "F (83 "C)
557.3
Fire space
Communicating space
Removed space
Separated space Fan temperature will not adwsely affect sysrcin pcrlorninnce.
Chapter 12-Zoned Smoke Control

USE OF W A C SYSTEM recirculatidg air within building spaces. During normal


In many buildings, the HVAC system serves many HVAC operation, the return damper .is completely or
zones, as illustrated in Figure 12.5. For such a system, partially open to allow air from building spaces to be
smoke control is achieved by the following sequence mixed with outside air. This mixture is conditioned and
upon fire detection: supplied to building spaces to maintain desired tempera-
ture and humidity. This process is shown in Figure 12.6.
The smoke damper in the supply duct to the smoke During smoke control operation, the return damper must
zone is closed. be tightly closed to prevent smoke feedback into the
The smoke dampers in the return duct to nonsmoke supply air, as is illustrated in Figure 12.7.
zones are closed.
As discussed in Chapter 7, smoke dampers are sup-
If the system has a return air damper, it is closed.
plied in several leakage classifications. The particular
Precautions must be taken to minimize the proba- class of damper specified should be selected based on
bility of smoke feedback into the supply air system. the requirements of the application. For example, the
Exhaust air outlets must be located away from outside dampers in the supply and return ducts shown in Figure
air intakes. To conserve energy, most HVAC systems in 12.5 can have some leakage without adversely affecting
modem commercial buildings have the capability of smoke 'control system performance. Thus, a designer

k.::.'..::
~xhaust Mechanical L Outside .-.,:
4::::-
L..'....
Mechanical Stside
-:--
Air- 2
-- Penthouse Air Penthouse :Air
c<

- I -T-,Supply Duct

-T/i
A-

Return
Duct\- I Smoke
p
am er

lP
-+
'
\
- I
-- LT
&
,/Damper
- - .
C

L
_1 L - L
(a) Normal HVAC Operation (a) Normal HVAC Operation

Notes:
I.Smoke control is achieved by closing the smoke damper in the supply duct to the smoke zone
and closing the smoke dampers in the return duct to the other zones. Return air damper (not shown)
must be closed to prevent smoke from being pulled into the supply air.
2. For simplicity, distribution ducts on each floor and equipment in the penthouse are not shown.
Figurc 12.5 Sclie17iatico f Z O I I s~iioke
~ ~ conr1.01system ~rsinga11 HVAC qare17i tliat selves

-
Return Fan

Exhaust
Air
/

/ I I-{ From Return

1' I Supply Fan . T


,.

Outside Equipment: Filters,


~ir- Heating & Cooling
Coils, etc.

Figurc 12.6 HKAC .v~rrerri~c'itlr~.ecit~crrlario~i


capahilir~iri tlrc rio1-17~al
HKAC
lllotic.
Principles of Smoke Management

Return Fan

Outside
c -
/

~ir- Heating & Cooling

Figure 12.7 HIQC syslenz with r.ecirculation capability it7 the sinoke control
mode.

might select class 11, 111, or IV smoke dampers for such


:I?;.:.<
an app!ication. Further, a designer might choose class I Mechanical Outside
dampers for applications that require a very tight .i Penthouse
....................
....................
... . . . . .. .. .. .. .. .. .. .. . :..*-. . . . ...
damper (for example, the return damper illustrated in
Figures 12.6 and 12.7).
Some designers have eliminated the smoke damp- Return
Duct \ ...
ers from the return air system in the mistaken belief that
-7-
A

Smoke
the resulting system would still be effectke. This idea '
\
-
I /Darnper
consists of shutting a smoke damper in the supply to the -C

smoke zone and relying on the return air beins pulled


from the zone to produce a significant pressure differ- Supply-Damper Only System
ence. However, shutting the supply to the smoke zone -- CAUTION ---
lowers the pressure there and. for these supply-damper- THIS SYSTEM MAY NOT WORK AND
only systems, the return airflo\v from the smoke zone is SHOULD NOT BE USED.
Notes:
also reduced. Field tests on such systems sponsored by 1. This system is not recommended because
the U.S. Veterans Administration have indicated that it generally does not achieve satisfactory
these supply-damper-only systems produce insignificant pressure differences to control smoke
movement.
pressure differences (Klote 1986). Thus supply-damper- 1. For simplicity, distribution ducts on each
only systems are not recommended. In a fire situation, floor and equipment in the penthouse are not
these small pressure differences can be overcome by shown.
buoyancy of hot smoke, stack effect, or other nomially Figure 12.8 Sche~naticof Jailwe /o achieve sinoke
occurring building airflows. Figure 12.8 illustrates the co/i/rolLy only shutting a smoke damper
failure of a supply-damper-only system to control in /he supply duct lo the snioke z o ~ e .
smoke movement with resulting smoke flo\v to the floor
above the fire floor due to buoyancy or stack effect.
This kind of system was tested at two new Veterans
For systems where the HVAC system serves only Administration hospitals (Klote 1986), where each floor
one smoke control zone, smoke contrd can be achieved of each wing was a smoke control zone supplied by a
by putting the KVAC systems in t!le modes below. separate HVAC system. This performed well, was espe-
cially simple, and required no expensive dedicated -
Srrioke Zorze: return fan-on, supply fan off, return equipment.
damper closed, and exllaust damper open (option-
ally the outside air damper may bc closed). ANALYTICAL APPROACHES
A1onsrriokeZoric: return fan 011; suppl!- f i n on, Most zoned smoke control systems are in buildings
return damper closcd, and outsidc air dnmpcr open with a numbcr of floors, shafts, and rooms per floor. As
(optionally tllc exhaust air dmlpel. Ins! bc closed). with other smoke control systems, the flows at the
Chapter 13 -Fundamental Concepts for Atria

Mass Flow with Virtual Origin Correction Q = heat release of the fire, Btuls @W);
Heskestad's equation for the mass flow of an axi- D/ = diameter of fire, ft (m);
symmetric plume is
C
,, = 0.278 (0.083).
. 113 5/3 2/3
tiz = C,, QC (Z - L,) [ l + cU2Qc (z - Z J - ~ ' ~ ] In Figure 13.1, the virtual origin is shown above the
,(13.1),
for ( z 2 z l ) cop of the fuel, but it can also be below the fuel. The
sign convention is: for the virtual origin above the top of
the fuel, zo is positive, and for the virtual origin below
the top of the fuel, zo is negative. The convective portion
of the heat release rate, Q,, can be expressed as

mass flow in axisymmetric plume at height, Ibls where X, is the convective fraction of heat release as dis-
(kds); cussed in Chapter 2. The convective fraction varies fiom
about 0.15 to 0.9, and using a value of G = 0.7 is common
convective heat release rate offire. B d s (kW);
for most design applications.
height above fuel, ft (m);
virtual orisin correction of the axis~mimetric Flame Height
plume, fi (m);
Equations (13.1) and (13.2) are dependant on the
mean tlame height, ft (m); mean flame height of the fire. The flanie height depends
0.022 (0.07 1 ); on the fire geometry, the ambient conditions, the heat of
0.19 (0.026); combustion, and the stoichiometric ratio. A relationship
(Heskestad 2002) for flame height that can be used for
0.0 126 (0.0051); many fuels is
0.556 (0.1663.
Because smoke was defined to include the air that is
entrained \\.it11 the producls of combustion, all of the
mass flow in the axisymmetric plume is defined as where
being smoke. It follows that these equations can be z, = mean flame height, ft (m);
thought of as equations for the production of smoke
from a tire. Simplitied axisymmetric plume mass equa- C,, = 0.788 (0.235).
tions will be presented later, and the same comments This flame height is the elevation where tlie niaxi-
also apply to them. mum plume temperature is 900°R (500 K). The ceiling
A condition of the axisym~netricplums is that the heights of atria are relatively higli, and it is tlie nature of
tire is circular, and many experimental studks have used atria smoke management that the elevations, z, of inter-
liquid pool fil-es in round pans. However, a tire need not est are much greater than either virtual origin, zo, or the
be round for the axisymmetric plume equations to be flanie height, zl.
:. useful. The effective fire diameter can be expressed as
,r : ...

D,- = 2 J z (13.3) C e n t e r l i n e Plume T e m p e r a t u r e


The temperature varies over the plume cross sec-
where A is tl~carea of the fire. At some distance above the tion, and the plume temperature is greatest at the center-
f?e, the plume li-om fire that is not round will bz nearly the line of the plume, as shown in Figure 13.2. The
same as that from a round fire. centerline temperature is of interest in the unusual cases
when atria are tested by real fires. The centerline tem-
Virtual Origin perature equation (Heskestad 1986) is
Heskestad's (1953) relationship for the \.irtual ori-
gin is
. l /j
S,, = C , jQ- - I .OZ D,. ( 1 3.4)
Principles of Smoke ~ a n a ~ e r n e i t

D -L (13.9)
P - CPD

where
D,, = diameter of visible axisymmetric plume, ft (m);
z = height above fuel, ft (m);
CpD = plume diameter coefficient ranging from 2 to 4.
This equation indicates that the axisymmetric
plume diameter can vary significantly, and it is sug-
gested that the value of CpDbe chosen s o that the results
Y

Radial ~adiai of calculations are conservative.


Distance Distance
Figure 13.2 Radial temperature variation of axisym- AXISYMMETRIC PLUME WITHOUT z,
meb-ic plume. Axisymmetric plume equations that neglect the vir-
tual origin are often used for atria applications because z
is much greater than z,. The siniplified equations are
absolutecenterlineaxisynllnetricplume tempera- listed below, and the consequences of this simplification
are discussed later.
ture at elevation z, "R (K);
absolute ambient temperature, OR (K); Mass Flow:
density of ambient air, l b / g (kg/m3); . I ,'3 5/3
riz = ColQCI + c , ~ Q ~ f o r r r z , (13.10)
acceleration of gravity, WS' (nds2);
and

For the conditions of 529"R (294 K), p, of 0.075 Ib/


ft3 (1.2 kg/m3), g of 32.2 ft/s2 (9.8 rn/s2), and Cpof 0.24
Btu/lb OF (I .OO kJ/kg "C), Equation (13.7) becomes
tir = mass tlow in axisymmetric plume at heightz, Ib/s
(kgls):
Q, = convectijt heat release rate of fire, B t d s (kW);
where = height above fuel, ft (m);
Tp = centerline axisymlnetric plume ten-iperature at I: = mean flame height, ft (m);
elevation z , OF ('C); C,, = 0.022 (0.071);
To = ambient temperature, OF ('C); C, = 0.0042 (0.00 18);
C,, = 338 (25). C,,, = 0.020s (0.032).
It should be noted that when 1 is less than z l , the
Plume Diameter condition of z being much greater than zo may not be
Coosidering that a fire plcnie is composed of met. However, the separate equation above for z < z, is
eddies, determination of the plume diameter is difficult. included tbr conlpleteness.
The plume diameter has been based on both visual
observations and temperatures. One definition of this Flame Height:
diameter is that it is the position at which the plume
temperature has decreased to some fraction of the cen-
terline temperature. The following equation is a way of
expressing the expected range of diameter (Figure 13.1 ) where
of an axisymmetric plume: ,
C , , = 0.533 (0.166)
Chapter 13-Fundamental Concepts for Atria

Centerline Temperature: Discussion


The purpose of this section is to evaluate the impact
of neglecting the virtual origin correction. To do this, the
fire diameter needs to be addressed. The heat release
density of a fire is q = Q / A . Substituting this relation
where into Equation (13.3) results in the following equation for
Tcp = absolute centerline axisymmetric plume tempera- the effective diameter:
ture at elevation z, OR (K);
To = absolute ambient temperature, OR (K);

p, = density of ambient air, lb/fi? (kg/m3);


Table 2.2 lists heat release densities for some ware-
g = acceleration of gravity. ft/s2(m/s2); house materials and pool fires. In this table, q ranges
C,, = 0.0067 (9.1). from 8 S t d s ?f (90 kw/m2) to 1250 Btds ft2 (14,000
kw/m2). The low value is for a proprietary silicone
For the conditions of 529"R (294 K), p, of 0.075 Ib/ transformer fluid, and the upper value is for polystyrene
ft3 (1.2 kg/m3), g of 32.2 ft/s2 (9.8 m/s2), and C p of 0.24 jars in compartmented cartons stacked 15 ft (4.57 m)
Btu/lb "F (I .OO kJ/kg "C), Equation (13.13) becomes high. These extreme fuel arrangements are not likely to
be found in atria, and eliminating them results in a range
of 35 Btds ft2 (400 kw/rn2) to 900 Btds ft2 (10,000
kw/m2).

where Figure 13.3 shows the effect of heat release density,


g , on the location of the virtual origin. For 35 Btuls ft2
TV, = centerline axisymmetric plume temperature at ele- (400 kw/m2), zo is about -2.6 ft (-0.8 m) at i) of 1900
vatio!] z, "F ("C); Btds (2000 kW) and -14 ft (-4.3 m) at Q of 24,000
To = ambient temperature, O F ("C); Btds (25000 kW). The negative values of zo indicated
that the virtual origin is below the fire surface. For 880
C,, = 338 (25). Btuls ft2 (10,000 kw/m2), zo is about 3.9 ft (1.2 m) at
Q of 1900 Btuls (2000 kW) and 10 ft (3 m) at i) of
Example 13.1 Simple Plume Calculations 24,000 Btu/s (25000 kW).
For a 4000 Btds (4220 kW) firr. \r.hat is the mass flow and
centerline temperature of the plume at z of 35 fi (10.7 m) above Figure 13.4 shows the impact of the virtual origin
the fucl'? Use a con\wtive fraction of0.7, and the ambient tem- correction on plume mass flow for q = 35 Btuls ft2 (400
pcrature is 72°F. kw/m2) and y = 880 Btu/s ft2 (1 0,000 k ~ / m ' ) .Neglect-
ing the virtual origin correction results in overprediction
From Equation (13.5). thc convecti\.e heat release rate is for q = 35 Btuls ft2 (400 kw/m2) and underprediction
DC = = 0.7(4000) = 2800 Btds (2950 kW!.

From Equation (13.12). the mean flame height is Heat Release Rate. i)(1000s kW)
.2/5
z l = 0.533& = 0.533(2800)"-" = 12.8 ft (3.9 m). 0 5 10 15 20 25

Because 2, < z, the plume mass tlow is calculated from Equa-


tion (13.10):
. 1 /3 j/3
NI = 0.022Qc z + ~.00420~

From Equation (13.14). the centerline plume temperature is

Heal Release Rate. Q (1000s Btuls)


Figure 13.3 The @ecr o f hear release densirj: q, 011
117e wir11ud origitz.
Principles of Smoke Management

Elevation. z(m)

Virtual Origin Correction: - 4000


-
- - - - - ~t q = a80 B~UIS ft' (1 0.000 kwtml)
a000 - - - ~t q = 35 B~UIS f? (400 kwlm')

.- None

-
5-2
-E
6000 -
Heat Release Rate
-
U
d
$ 4000 -
S
2000 -

0
0 75 225 300
Elevation, z (ft)
..-.
Figure 13.4 Comparison of nzass ~ I O Wpredictions u ~ t hand rvithozrt
correction for virtual origin.

for q = SS0 Btuls ft2 (10,000 k ~ l m ' ) . These over- and


Heat Release Rate, Q (1000s kW)
underpredictions are with reference to Equation (13. I).
An estimate of the uncertainty of Equation (1 3.1) is
not available, b u ~it should be noted that the state of
plume technology is such that the above ranges may be
within the uncertainty of Equation (13.1). Further, fire
spread by radiation can result in a number of nearby
fires with separate plumes joining together as they rise.
Theories have yet to be developed for such multiple fire
plumes. There is no question that both Equations (13.1)
and (13.10) reflect the important trends of mass flow Heat Release Rate. Q (1000s Btuls)
being a strong function of eleva~ion,z, and a weak func- Figure 13.5 Cotr~pat-isonof 117eatzf i m e height u.itll
tion of the convective heat release rate, Q,. However, and without virtzral origin correctiotl.
when using Equation (13.10), it is suggested that the
location o f the fire surface be conservatively selected.
For example, if fires may be possible anywhere from the
floor level' to 3 m (10 ft) above the floor, conservative WALL AND CORNER PLUMES
selection af the fire surface would be at the floor. A fire that is located next to a wall will entrain air
over a smaller perimeter than a fire located far from a
Figure 13.5 compares the predicted flame heights
wall. The fire and wall plume may be considered half
from Equation (1 3.6) tvith the approximaLe relation of
that of the axisymmetric plume (Figure 13.6). Thus, the
Equation (1 3.13). Again, the approsimate relation is in mass flow rate of a wall plume is half that of an axis)m-
the middls of the range of predicted values. It is appar- metric plume, but the analogous fire for the axisymmet-
ent that flame height, z l l increases with q. In atria smoke ric plume is twice that of the "real" fire creating the \\.all -
management design, flame height is primarily used to plume. The mass flow rate of a wall plume can be esti-
ensure that the plume mass flow equations are appropri- mated as follo\vs:
ate. The flame height, z l , rangcs from about 8 ft (2.4 m)
I. Let Q be twice that of the wall plume.
to 14 ft (1.1m ) aL 1000 Btuls (2 100 kW) and from about
14 fi (4.3 m) to 39 ft (12 111) at 25,000 Btuls (26,000 2. Apply a plume equation to calculate the mass floi~.,
kW).
Chapter 13- Fundamental Concepts for Atria

Example 13.2 Wall and Corner Plumes


For the Q , z, and X, of Example 13.1, what is the mass flow
of a plume for each fire location shown in Figure 13.7?

The mass flows are calculated from Equation (13.16) as fol-


lows:
Mass Flow,

n Ibls kg/s
Fire far from walls 1 121 55
Fire at exterior corner 4!3 101 46
Fire at wall 2 78 35
Fire at interior corner 4 51 23
Figure 13.6 Wall pl~ime.
As expected, the mass flow from Equation (13.16) for a fire
far from walls has the same mass flow as the axisymmetric
plume in .Example 13.1. The presence of a wall or corner
reduces air entrainment into the plume, so the mass flow of
Fuel@,,,,, = j Fuelw = these plumes is less than that of the axisymmetric plume.

BALCONY SPILL PLUMES


(a) Fire far from walls (b) Fire at wall
A balcony spill plume originates from a fire when
the smoke flows under a balcony and spills into the
atriu~u(Figure 13.8). When tlie fire is in a room that
opens onto the balcony, the mass tlow rate can be
approximated as (Law 19%; CIBSE 1995):
7 l :3
i = C, ( l ) (z,, + 0.25H) (13.17)

wlicrc
(c) Fire at interior (d) Fire at exterior
corner corner ~ i r = mass tlow in p l u m at height zb, Ib!s (kgs);
= Iicat release of thc lire, Btuls (Id\;):
W = plume width as it spills undcr balcony, ft (m);
q, = Iieiglit above balcony, ft (m);
3. The smoke production of the wall plume is fir 12. H = height of balcony above top of fuel, ft (m);
C,, = 0.12 (0.36).
This approach is very rough for heights in tlie range
of the flame height, but the flow becomes more realistic When draft curtains are used (Figure 13.S), the
for higher elevations, z. The approach can be applied to width, W, of tlie spill plume is the distance between the
any axisymrnetric plume model, but for consistency curtains. In the absence of draft curtains, the following
with the information above, 1i1 would be calculated approximation can be used.
from Equation (1 3.10).
Thc above approach can be extended to plumes
from tires in corners. For relatively high elevations
above the fuel, Equation (13.10) can be cstended to W = width ol'rhc opening liom the tire room, ft (m);
become the ge17ernl 11.crllpllcnle ey~rcr/ion.
h distance from the opcning to tlie balcony edge, ft (m).
=
Equations (l 3.17) and ( l 3. IS) onl) apply when the
height of the opening to the firc room is suficiently
below thc ccili~lgsuch that tlic inomenturn of the ceiling
jet in tlic lire room Jocs not directly contribute to the
where 1 1 is a lire location factor that is sho\\w in F i y r e flow out ol'thc ol~c~ling. Tlic thickness of the ceiling jet
13.7. For li~rthcr information about \w11 and corner is in the range ol' 10% to 20% of the height from the
plumes, sec Mowrcr and Willinnison (1987). base o f thc lil-c to tlic ceiling. Bascd on this. i t can be
Principles o f Smoke Management

stated that the momentum of the ceiling jet is not a con- rather than HRR. Those using the correlations of Mor-
tributing factor when the top of the opening is not gan et al. will need to convert HRR to fire perimeter.
greater than 80% of the distance from the base of the Equation (l 3.17) is extensively used for design anal-
fire to the ceiling. . ysis, but there is controversy about the extent of its appli-
For spill plumes not consistent with the conditions cability. ASHRAE Technical Committee 5.6 is planning
of Equation (13.18), scale modeling (Chapter 15), CFD a research project consisting of large-scale fire experi-
modeling (Chapter 16), or other correlations can be ments to check the applicability of Equation (13.17) and
used. While Morgan et al. (1999) is a source of other to develop information for some spill plumes not consis-
correlations, most of these are in terms of fire perimeter tent with the conditions of Equation (13.18).

Example 133 Balcony Spill Plume


What is the mass flow of a balcony spill plume with the parameters listed below?
Heat release rate, Q - 500 Btds (528 kW)
Height above balcony, zb - 20ft(6.10 m)
Height of balcony above top of fuel, H - 10 ft (3.05 m)
Width of the opening from the tire room, W - 6 ft (1.83 m)
Distance from the opening to the balcony edge, b - 12 ft (3.66 m)
From ~ 4 a t i o n(l3.18), plume width as it spills under the balcony is approximated as
W = w + b = 6 + 1 2 = Igft(5.49m).
From Equation (13.17), the mass flow of the spill plume is

I
a Doorway

Front View With


Draft Curtains

Section View
Front View Without
Draft Curtains
Figure 13.8 B n l c o y spill plzrtne
Chapter 13- Fundamental Concepts for Atria

Inserting this factor into Equation (13.10) yields an


equation for axisymmetric plumes:
1/3 5/3
fit = c a l Q c (2, +a) + ca9~c (13.2 1)

Inserting the same factor into Equation (13.16) I


yields a general equation that incorporates the presence
of walls:

where z, is the height above the top of the window. Substi-


tuting Equations (13.19) and (13.20) into Equation (13.22)
Figure 13.9 M'irdow plume. results in

WINDOW PLUMES
A window plume is one that flows through an open-
ing such as a window or door to a room with a fully where
involved fire (Figure 13.9). As described in Chapter 2, a C,+.2 = 0.077 (0.68);
fully involved fire is one where all, the combustible
C,,, = 0.18 (1 59).
materials in the room are burning. The high intensity of
such a fire explains why window plumes are not nor- Equations (13.22) and (13.23) apply to wall and cor-
mally considered design fires in sprinklered buildings. ner plumes, and with i 7 = I , these equations become the
In such a fire, fuel \.olatilized in the room would burn same as the window plume equations. I t may be noticed
outside the opening. The heat release rate of a fully that Equation (13.23) does not contain a heat release rate
developed tire is constrained by the combustion air that tenn, and this can be so because the fire is ventilation con-
can reach the lire. and such a fire is referred to as being trolled such that the heat release rate depends on the ven-
Accordingly, the size of the fire
ve~ililnfioti.cot~~rolleti. tilation opening. This analysis of window plumes is based
depcnds on the size and shape of the opening to the on the assumptions concerning entrainment and the
room and the material burning. Based on experin~ental adaptability of the general wall plume model, but it has
fire data Ior \irood and polyurethane in a room with a not been experimentally verified.
single rectangular opening, the average heat release rate
iS A V E R A G E PLUME TEMPERATURE
The average temperature of the plume can be
obtained from a first law of thermodynamics analysis of
the plume. Consider the plume as a steady flow process
where
with the control volume shown in Figure 13.10.
0 = heat release of the fire, Btuls (kW): Neglecting the small amount of mass added to the
plume flow due to combustion, the first law for the
A,,. = area of ventilation opcning, ft2 (m'); plcme is
H!,. = height of ventilation opening, ft (m):
C,,., = 61.2 (1260).
0,
Q ~ + = til(lte- 11;+ AKE + A P E ) + l~ ((13.21)

Thc equations for tlle axisymmetric plume or wzll where


and corner plumes can be adapted for the window
Q~ = heat generated within the control volume, Btuk
plume. This is accon~plisliedby determining the entrain-
ment ratc at the tip of the flames coming out of the (kW),;
opening and determining the height in the axisymmetric Q, = heat transferred liom surroundings into the con-
plume cquation that would yield the same entrainment. trol volume. Blu/s (kW):
The height m m in the asisyrnmetric plume equation
nceds to bc adjusted by the follo\ving factor: tit = Inass flow rate. Ib/s (kds);
Principles of Smoke Managemen't

hi = enthalpy of flow entering the control volume,


Btu/s Ib (kWkg);
h, = enthalpy of flow leaving the control volume,
where
Btu/s Ib (kWikg);
z,, = maximum height at which plume is considered
AkE = change in kinetic energy, Btu/s.lb (kwikg); buoyant, ft (m);
ME = change in potential energy, Btuls Ib (kW/kg);
QC = convective heat release rate, Btu/s (kW);
W = work done by system on its surroundings, Btu/s
ATnlit7 = minimum temperature rise of plume above
(kW). ambient, "F ("C);
For the steady plume, the work is zero and the
c,,,,, = 189 (14);
changes in kinetic and potential energy are negligible.
The heat generated is the heat release .of the fire CznLr2= 19.1 (0.0254).
(Qg = Q ). Heat is transferred from the plume by con- Equation (13.26) applies to axisymmetric plumes.
duction and radiation to the surroundings (Q, = -Q, The idea of minimum temperature rise is that it is the
where Q, is the radiated heat), so that (Qc = i), + Q ,). smallest temperature rise at which the plume has suffi-
Specific heat can be considered constant (h = Cpr). The cient buoyancy to continue to be a strongly buoyant
first law leads to an equation for the plume temperature. plume.
No research has been conducted to determine the
appropriate value of the niinimuni temperature rise for
vario;; applications and velocities of air currents. Until
better information is available, 3.G°F (2°C) is suggested.
where Figure 13.1 I shows the maxi~numplume height for this
minimum temperature rise.
Tp = average plume temperature at elevation z, "F ("C);
For a total heat release rate of 500 Btuls (530 kW).
To = ambient temperature, "F ("C); the masiniuni plume height from Figure 13.11 is about
130 ft (40 m). For a total heat release rate of 2000 Btuls
Cp = specific heat of plume gases, BtuAb "F (kJkg "C).
(2100 kW), the maximum plume height is about 220 ft

,
Fire plumes consist primarily of air mixed with the (67 m). While the maximum plume height is not a con-
products of combustion, and the specific heat of plume cern for most designs, it needs to be considered for atria

i
gases is generally taken to be the same as air [Cp = 0.24 with high ceilings.
Btullb "F (1.00 kJ/kg "C)].

Examplc 13.4 Average Plun~eTemperature l ~ i ~ p l zo w


atp Elevation
What is the average temperature of the plume in Example

From Equation (13.25),tlle average plume temperature is

As expected, the average plume temperature is less than the


centerline plume temperature.

MAXIMUM PLUME HEIGHT


The plume mass flow eql-!ations were developed-for
strongly buoyant plumes. When smoke is not hot
enough to rise, it will stagnate or be carried away by
existing air currents. Combining Equations (13.10) and
(13.25) yields the following espression for the maxi-
mum plume height at \\:l~ichthe plume can be consid-
ered strongly buoyant.
Chapter 13 -Fundamental Concepts for Atria

Heat Release Rate. Q (1000s kW) outside Air T, p,


-
5
70oO.~~
10
I
15
1
20
I
25
I -
E
A.. 4liL
2 600 -
N

m
E- 500 -
.- - 135 'G
I

.-
2
2 l00 -
1. Maximum plume height is for an
axisymmebicplume with a minimum
temperature rise of 3.6 OF (2.'~).
2. This figure is for Q, = 0.7Q.
- 45
a
5
.g
S
L rni Floor To,p,
I
0 I l I I - 0 (a) Sketch of Gravity
0.5 5 10 15 20 25 (b) Pressure Profile of
Heat Release Rate. (1000s Btuls)
Smoke Venting Smoke Layer
Figure 13.11 Maximum plunze height. Figure 13.12 GI-avivsmoke venting.

VOLUMETRIC FLOW
The volumetric flow of a plume is
where
p,. = reference density, 1bnl/ft3 (kghn3);
T, = absolute reference temperature, OR (OK).
where
There are an infinite number of possible p,; T,. pairs.
tit = mass flow in plume at height z, Ibls (kgls); and one that can be used for such calculations is p,. =
0.075 lbm/ft3 ( 1 2 0 k g h 3 ) , T,. = 530 OR (294 K).
V = volumetric smoke flow at elevation z, cfnl (rn3/s);
= density ofplume gases at elevationz, lb/ft3 (kg/m3); CONFINED FLOW
PP
C/, = GO(1). As already noted, the diameter of a plume increases
with height. For a tall narrow atrium, the plume may
AIR AND PLUME DENSITY contact all of the atrium walls before the plunie reaches
the ceiling. Where a plume contacts a wall, i t cannot
The density of air and plunie gases is calculated
entrain air. For smoke management purposes, the smoke
from the perfect gas law:
layer interface should be considered the elevation where
the smoke contacts all or most of the atrium walls.

NATURAL VENTING
where
Natural vents consist of openings in the ceiling
p = density of air or plume gases, lbm/f$ (kg/m3); through which smoke flows due to buoyancy. The hot
smoke layer under the ceil~ngacts to force smoke out of
p = absolute pressure, lbf/ft2 (Pa);
the vent and to pull makeup air through other openings
R = zas constant, ft Ibfllbm OR ( J k g K); into the atrium. The analysis of natural \,enting that fol-
T = absol~itetemperature, OR (K). lows is adapted from an analysis by Thomas et al.
The absolute pressure is often taken to be standard (1963), and it is illustrated in Figure 13.1 2. The ternper-
atmosplleric pressul-e of 21 16 lbflft' (101,325 Pa), and ature in the atrium below the smoke layer is considered
the gas constant is generally taken to be that of air, the same as that outside.
which is 53.3 ft IbVlb~nOR (287 Jlkg K). As discussed in Chapter G, the mass f l o w out of the
At most localions, atmospheric pressure call be con- vent and in the inlet opening can be expressed by the
sidered constant for purposes of calculating air and orifice equalion as
smoke density. This means that p/R can be considered
constanl. and dcrisily can h e n be calculated.
Principles of Smoke Management

l
and A, = vent area, fi? (m2);

nzi = K , ~ c ~ A ~ J E i (13.3 1) A~ = inlet opening area, fi? (m2);


, . p, = outside air density, 1blft3(kg/m3);
where
ril" = mass flow rate through the vent, Ibls (kgts); g = acceleration of gravity, 32.2 ft/s2 (9.80 m, s2);
db = depth of smoke layer below the smoke vent, R (m);
i /izi = mass flow rate through the inlet opening, Ibls (kg/
To = absolute temperature of outside air, "R (K);
S);
c, = flow coefficient of the vent (dimensionless); T, = absolute temperature of smoke, "R (K);
c. = flow coefficient ofthe inlet opening (dimension- TRANSPORT TIME LAG
less);
A plume takes time to rise to the ceiling, and a ceil-
A, = area of the vent, fi? (m2); ing jet takes time to form a smoke layer under the ceil-
ing (Figure 13.13). The idealized zone fire model
A; = area of the inlet opening, fi2 (m2);
considers that (l) the smoke from the plume reaches the
- upper layer at the instant of combustion and (2) a uni-
PS - density of the smoke, 1b/fi3(kglnm3);

P0 - density of the outside air, 1b/fi3(kg/m3);


-
PS - pressure of smoke layer at the ceiling, in. H 2 0

-
P0 -

Pb =

K,, =

layer is (a) Growing plume

P, -P!, = K,,,gdh(~o - P,) (13.32)

where
y = acceleration of gravity, ft/s2(nds2);
db = depth of smoke layer below the smoke vent, ft
(14;
G= 370 ( 1 .OO).
The flow coetticients are considered to be equal ( C
= C,, = C;). The mass flow out the vent equals that (b) Fully developed plume
through the inlet opening (/ill: = ii~,).The smoke density and growing ceiling jet
can be ex~ressedas
. .. .. . .. . . . . . . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ... .. ..
...................................................
................................

To
P, = PO-. (13.33)
?Y

Combining Equations (13.30) through (1 3.33) yields


I /2
C A , . P , , [ ~ P / , , ( T . ~T<,)(T<,/TJl
-
rir ,. = (13.34)
7 l/?
[T,. + (."/A;)- T,,l

where (c) Fully developed plume


and ceiling jet extending
/ir,, = mass llow rate tl~roughthe vent, Ibls (kys); under entire ceiling
C = discharge cocllicient (dimcnsionless); Figure 13.13 Develop~l~enf of pl~rri~eat7d ceiling jet.
Chapter 13- Fundamental Concepts for Atria

= transport time lag of plume, s (S);


Ceiling Height (m) tpl
= growth time, S (S);
tg
H = height ofceiling above top of hel, ft (m);
Q = heat release rate, Btuls (kW);

In general, t-squared fires have greater plume time


lags than steady fires as can be seen from Figure 13.14.
This is to be expected in that the t-squared fires start at
an HRR of zero and increase uith the square o f time,
while the steady fir& start at their full fire size much
like a gas burner.
For a slow t-squared fire with a ceiling height o f
Ceiling Height (ft) l00 ft (30 m) above the base of the fuel, the plume lag is
only about 18 S. Such a small t m e lag illustrates the
Figure 13.14 Pllrn7e trampor? fag.
rational for neglecting the plume lag.

form smoke layer forms at the first instant that any Ceiling Jet Lag
smoke reaches the ceilinp. The zone fire models were Newman and Mon.rer also developed the follon.ing
developed for fire rooms such as ordina~ybedrooms, relationships for the time lag of ceiling jets resulting
dining rooms, and recreation rooms. In such rooms, the from steady and unsteady fires.
errors resulting from these simplitications were insignif- Foi- a steady fire,
icant. Atria are much larssr, and the following sections
provide a means of evaluating the errors rcsulring from
neglecting time lag.
Neglecting these lag times results in ~mdcrcstimat-
ing atrium tilling time and detector activation time. For For a t-squared fire.
atrium filling applications. this is consel~valivci n that
occupants will have more time before smoke rcaches a
articular level. For srnoke detector calculations.
neglecting these lag time results in underpredicling thc
activation time, which is not conservalive i n ha1 pcoplc where
will have less time to act than indicated by h c prcdic-
= transport time lag of ceiling jet, s (S);
tion. 1 ,
1. = radius or Ii~rizontaldistawe from centerline of
Plume Lag plume, fi (m);
Nenmian (1 9S8) and Mowrer ( 1990) devclopcd C,, = 0. l GS (0.833)
relationships for the time lag of plumes I'rom s m d y and C$ = 0.278 (0.72)
unsteady fires. Steady and unsrcady tires are discusxd
in Chapter 2. Because il is h e nature of tires to g~-o\v. As with the plume lag, the t-squared fires ha\.s the
the time lag time for steady l i r a is probably less signifi- greatest ceiling jet lag as shown in Figure 13.15. The
cant. ceiling jct of a fast t-squared fire with H = 20 fr (6 m )
For a steady fire, takes about 25 s to spread out to a circle with a 25 i't 17.6
m) radius.

Total Transport Lag Time


The total transpon lag time is [lie sun1 of the plume
For a t-squared tire. lag rime and thc ceiling jet lag time,

\vherc whcrc I, is the tolal tralispon Iag rime (S).


Principles of Smoke Management

!
I PLUGHOLING maximum mass rate of exhaust without plughol-
il ing, lbls W s ) ;
i When the smoke layer depth below an exhaust inlet
i is relatively shallow, a high exhaust rate can lead to absolute temperature of the smoke layer, "R, (K)
l
entrainment of cold air from the clear layer (Figure absolute ambient temperature, "R, (K)
I 13.16). This phenomenon is called plugholing.
depth ofsmoke layer below bottom of exhaust
I
inlet, %(m);
Number of Exhaust Inlets exhaust location factor (dimensionless);
To prevent plugholing, more than one exhaust point 0.354 (3.13).
may be needed. The maximum mass flow rate that can In the context of plugholing, the smoke layer depth
be effkiently extracted using a single exhaust inlet is is always a distance from the smoke interface to the bot-
given as [CIBSE 19951 tom of the exhaust inlet. For an exhaust inlet located in a
wall, the depth of the smoke layer below the bottom of.
exhaust inlet is illustrated in Figure 13.17.
Based on limited information, suggested values of P
are 2.0 for a ceiling exhaust inlet near a wall, 2.0 for a
where
wall exhaust inlet near the ceiling, and 2.8 for a ceiling
exhaust inlet far from any walls (Figure 13.18). It is sug-
Radial Distance (m) gested that d/D be greater than 2, where D is the diame-
0 5 10 15 20 ter of the inlet. For exhaust inlets, use D = 2ab/(a + b),
100 i I I I

"
0 25 50 75
Figure 13.17 Depth, d, ofsrnoke layer- beloiv bottorl~
Radial Distance (ft) ofe.dmtst inlet.
Figure 13.15 Ceiling jet tra1lsp01-f
lug.

Ceiling Inlet
I1
t 0=2.8
(a) Ceiling lnlet Away From Walls

Ceiling Ceiling Inlet,


1
-+lnlet
n
p = 2.0 p = 2.0

(b) Wall lnlet Near (c) Ceiling lnlet Near


The Ceiling :
The Wall
Figure 13.16 Pl11gholi11gof air- iuto smoke e.rhartst Figure 13.18 Vulltes o f exhaltst location,fbcto/:p
i111etsc017 r-es~iltin /bilro.e oj' a snloke
esl~atists~ste~rl.
Clmpter 13-Fundamental Concepts for Atria

(a) Exhausting air when there is no smoke layer. where the fimction ROUND indicates that the value in the
parentheses is to be rounded up to the nearest whole num-
ber.

Separation Between Inlets


When the exhaust at an inlet is near this maximum
flow rate, adequate separation between exhaust inlets
needs to be maintained to minimize interaction between
the flows near the inlets. One criterion for the separation
between inlets is that it be at least the distance from a
(b) Exhausting air witil smoke when the depth of single inlet that would result in arbitrarily small velocity
the smoke layer, d, is less than the minimum depth. based on sink flow. Using 40 fpm as the arbitrary veloc-
d,, needed to prevent plugholing. ity, the minimum separation distance for inlets located
in a wall near the ceiling (or in the ceiling near the wall)
is

where
S,,,il, = minimum separation between inlets, ft, (m);

i/, = volumetric flow rate, cfin (m3/s);


(c) Exhausting only smoke when the depth of 13 = exhaust location factor (diniensionless)
the smoke layer, d, is greater than the minimum depth.
d,, needed to prevent plugholing.

Figure 13.19 Smoke n17dnir exhazat conditions.


Minimum Depth to Prevent Plugholing
The approach of the previous section to plugholing
considered the number of exhaust inlets needed to pre-
vent plugholing, but the issue can also be viewed as the
smoke layer depth needed to prevent plugholing for a
where n and b are the length arid width of the inlet. The specific eshaust flow rats.
results of experiments conducted at the National The minimum smoke layer depth to prevent plugh-
Research Council of Capada are consistent with this oling can be expressed as
approach for dealing with plugholing (Lougheed and
Hadjisophocleous 199'1, 2000; Lougheed et al. 1999;
Hadjisophocleous et al. 1999).
Equation ( l 3.40) can be expressed in terms of volu- where
metric flow rate as = minimum smoke layer depth to prevent plughol-
d,,,
ing, ft (m);
,,C
, = 0.78 (0.15).
Satisfying Equation (13.44) is equivalent to meet-
wherc
ing the criterion for the number of exhaust inlets of the
v,,~,,,= maximil~nvolumetric tlow rate at 7;, cfin (m3/ previous section. This means that meeting either of
these criteria is sufticient to ensure that the other crite-
S); rion will be met.
C,),! = 0.537(0.00887).
Plugholing Flow
Considering cqual Ilows at each eshaust inlet, the Conditions of air eshaust and smoke exhaust are
numbcr. 1VjI,,,,,. ol'cshaust irilcts is illustrated in Figure 13.19. The mass flow ofcxhaust air
Principles of Smoke Management'

is the sum of the exhausted smoke and the exhausted air. 13.20. Larger exponents result in predictions of lower
For an exhaust inlet, the exhaust can be written as smoke layers.

me = nze0+ mes (13.45)


MINIMUM SMOKE LAYER DEPTH
where
The previous section addressed the smoke layer
= total exhaust from an inlet, Ib/s (kgls); depth needed to prevent plugholing. This section
me, = lower layer air exhausted from an inlet, Ibls (kgls); addresses the minimum design smoke layer depth
needed to accommodate the physical flow of smoke
me, = smoke exhausted from aninlet, Ib/s (kgls). under the ceiling. Readers are cautioned about differ-
When the smoke layer depth is at or below the min- ences in the definitions of smoke layer depth. For plugh-
imum smoke layer depth to prevent plugholing (d 2 d,,), oling applications, the smoke layer depth is the distance
the mass flows are i e=o0 and riles = where li, from the smoke layer interface to the bottom of the
exhaust inlet. In this section, the smoke layer depth is
is the total volumetric exhaust in ft3/s (m3/s) and p, is the distance from the smoke layer interface to the ceil-
smoke density, lb/ft3 (kglm3). When the smoke layer ing.
depth is zero (d = O), the mass flows are ~iz,, = p,& When a smoke pll!me reaches a flat ceiling, the
and riles = 0 where p, is the ambient or lower layer air smoke is deflected into a ceiling jet flowing radially
from where the plume impacts the ceiling. As already
density, lb/ft3 (kg/m3).
In the absence of a formally developed and experi- stated, the depth of the ceiling jet ranges from about
mentally verified theory of plugholing flow, the follow- 10% to 20% of the distance from the base of the fuel to
ing relations are suggested: the ceiling. This is an oversin~plificationin that the ceil-
ing jet is about 10%, but at the walls, the jet turns back
d
V ( -) and flows under itself. For an enclosed room or atria
'0
i s= p for 0 5 d < d,,! (13.46)
" '"1 where there are no openings near the ceiling, the smoke
layer forms as illustrated in Figure 13.21. For suc11 an
enclosed room or atrium, it is recommended that the
rile = p for d,,, 5 d design smoke layer depth be at least 20% of the distance
from the base of the fire to the ceiling.
and
For rooms and atria with an opening at the ceiling,
the smoke of the ceiling jet flows out through the open-
ing (Figure 13.22). Provided that the opening is wide
enough, the minimum smoke layer depth is about 10%
of the distance from the base of the tire to ihe ceiling.
rile, = 0 for d,,, -< d For further information about ceiling jets, sec Beyler
(I 986).
where a is the plugholing exponent. The effect of this
exponent on 'the smoke layer height is illustrated in Figure HORIZONTAL SMOKE FLOW
Horizontal flow out of a smoke layer and the corre-
sponding pressures are shown in Figure 13.23. These
Atrium height: 40 fl(12.2 m) - l2 pressures are hydrostatic, that is, the dynamic pressure
Atrium Areq: 10,000 ff(929 m')
Exhaust: 120.000 c h (57 m'k) components are negligibly small. The opening 1s consid-
Exhaust Location Factor: 2
g
- 11 .P ered sufl-iciently large that the pressure below the smoke
No, of Exhaust Inlets: 1 Z
Heat Release Rate: 5000 B t u k ;
layer can be considered the same as that outside the
(5300 kw) z opening. Further, the top of tlle opening is sufficiently
-102
a
Y below the ceiling that the ~nomentumof the ceiling jet
E
0 has no impact on the smoke flow through the opening.
9
The smoke Flow through tlie opening depends on
30 60 90 120 150 180 tlie buoyancy of tlie smoke and the thickness of the
rime (S)
smoke layer under tlle first level.
Figure 13.20 Effecr of' /lie pl~igl~oli17ge.vponenl 017
smoke IO>TI. 11cigI71OS sinrrilnled bv //W
AZONE 111ocle1.
Chapter 13-Fundamental Concepts for Atria

(a) Development of ceiling jet


, Ceiling Jet
Figure 13.22 Smoke layer in a roorn or atria n.irh a17
opetiirg at the ceilirg.

where
rii = mass flow from smoke layer through opening, Ib/s
(Ws);
W = width of opening, ft (m);
(b) Ceiling jet reaching a wall and turning back h = depth of smoke layer, ft (m);
,Ceiling Jet, g = acceleration of gravity [approsin~ately32 ~ s (9.8
'
ds2)1;

minimum smoke

20% of the distance


M
layer depth is about ::s::"~':

from the base of the .....


p, = ambient air density, lb/ft3(kgm3);
p, = smoke density, lb/ft3(kg/m3):
C = flow coefficient (dimensionless - approximattl>.
0.7);
fuel to the ceiling. Fire
c/fl= I .oo (l).
(c) Formation of minimum smoke layer depth for enclosed room

Figure 13.21 Developrno~toj' tr~iniri71unsriroke layer- Equation (13.48) can also be witten as
in an enclosed root11or-atria.

where
Cl@ = 3.74 (2.07)

(a) Sketch of horizontal flow


Pressure
(b) Pressure profiles
Figure 13.23 Hor~izor~tulflo~~~
t/ir.o~~gIi
a17 o/m~ingor. 1 w 7 / .
Principles of Smoke Management

Airflow for Fire in Communicating Space


Airflow can be used to prevent or mitigate smoke
Atrium Smoke,
Space
originating in a communicating space from migrating to
the atrium (Figure 13.24). This can be accomplished by
exhausting the communicating space and supplying air
to the atrium such that the velocity at the opening of the
communicating space is sufficient to prevent smoke
Figure 13.24 Airflow used to prevenl or n~ifigafe from flowing to the atrium. As discussed in Chapter 6,
smoke originaling in a cornrnzlnicaling to prevent such smoke migration, the average velocity in
space from tnigmting lo [he afriurn. the opening must equal or exceed the limiting velocity
(Heskestad 1989).

Substituting the ideal gas equation into Equation


(13.49) and rearranging yields

where

v = limiting average air velocity, fpln (mls);


where
g = acceleration of gravity, tVs2 (nl/s2);
T, = smoke temperature, 'R (K);
H = height of the opening, ft (m);
To = ambient temperature, OR (K);
Ts = absolute temperature of the fire space, OR (K);

As stated above, this analysis applies when (I) the r, = absolute ambient temperature, OR (K);
opening is considered sufficiently large that the pressure
below the smoke layer can be considered the same as cafl = 38 (0.64).
that outside the opening, and (2) the ~uomentuniof the
ceiling jet has no inipact on the smoke flow through the Airflow for Fire in Atrium
opening. The comments concerning the momentum of
the ceiling jet in the above section about the balcony Airfloxv also can be used to prevent smoke originat-
spill plume also apply here. When these conditions are ing in the atrium from flowing into a com~nunicating
not met, the horizontal smoke flo\\. can be analyzed by space. The limiting air velocity can be calculated from
physical nlodeling or computational tluid dynamics.
The above analysis is a subset of the approach used
by the multiroom zone fire models. The multirooln
models allow for the possibility of smoke flowing into a
room that has a smoke layer that has descended below
the top of the opening. The pressure differences and
limiting average air velocity, fpm (~n/s);
flows at openings between compartments of a zone
model can b; complex, as is explained by Jones and
heat rclease rate, Btds (kW);
Bodart ( 1 956).
distance above the base of the fire to the bottom of
COMMUNICATING SPACES the opening, ft (m);
Communicating spaces are spaces within a building 17 (0.057).
that have an open pathway to an atrium such that smoke
from a tire in the communicating space can move unim- Equation (13.52) is not applicable when z is less
peded into the atrium. Smoke from a tire in an atrium than 10 ft (3 m). Further, v, should not exceed 200
can also move unimpeded into the communicating fpm (I m/s). If the opening to the comn~unicatingspace
space. Communicating spaces can open directly to the is above the smoke interface, the limiting air velocity
atrium or can be connected through othsr open spaces. sliould be calculated from Equation (13.51).
CHAPTER 14

Atrium Systems

t is well known that the ability of sprinklers to sup- AZONE are on the CD that accompanies this book.

I press fires in spaces with ceilings higher than 35 to


50 ft ( l l to 15 m) is limited. Because the tempera-
ture of smoke decreases as it rises (due to entrainment of
AZONE is a model specifically written for analysis of
smoke movement in atria, and a detailed description of
AZONE is provided at the end of this chapter.
ambient air), smoke may not be hot enough to activate
sprinklers mounted under the ceiling of an atrium. Even
if such sprinklers activate, the delay can allow fire
growth to an extent beyond the suppression ability of
ordinary sprinklers. Considering the limitations of com-
partmentation and sprinklers for atriums, the importance
of atrium smoke management is not surprising.

SYSTEMS
Approaches that can be used to manage smoke in
atria are ( l ) smoke filling, (2) mechanical exhaust, (3)
natural venting, and (4) tenability systems. Most of
these approaches have the goal of not exposing occu-
pants to smoke during evacuation except for the tenabil- (a) Fire in atrium space producing an axisymmetric plume
ity systems. The goal of the tenability systems is not to
subject occupants to untenable conditions.
For all of these approaches, the design fire can be
steady or unsteady. For information about design fires,
see Chapter 2. Fire location is an important factor; for
example, a fire in the atrium space may produce an axi-
symmetric plume while a fire in a space open to an
atrium may produce a balcony spill plume (Figure 14.1).
In North America, systems are usually designed for fires
in the atrium. In Australia, the United Kingdom, and
other parts of Europe, design fires are often in spaces
open to the atrium, such as shops and offices.
Analysis of these approaches can be done by use of
(b) Fire in space open to atrium producing a balcony spill plume
equations or computer zone fire models. For general
information about these computer models, see Chapter Figure 14.1 Locatioti of fire cat7 deter-mine the k i d
8. Computer zone fil-e models CFAST, ASET-C, and o~jlllit~le.
Chapter 14- Atrium Systems

-
I ailical level above the highesl -pied

Figure 14.2 Afr-i~rm


smokefilling.
Figure 14.3 Illustration ofsmoke layer-s for empirical
SMOKE FILLING filling equations and zone fir2 nrodels.
This approac!i consists of having occupants evacu-
ate the atrium or through the atrium as smoke fills the where
atrium space (Figure 14.2). Smoke filling applies only 2 = height of the first indicationof smoke above the
to very large volume spaces where the filling time is suf- fire surface, fi (m);
ficient for evacuation, including the time it takes to H = ceiling height above the fire, ft (m);
become aware of the fire and to prepare for movement
1 = time, s (S);
to an exit. Chapter 4 addresses people movement and
fire evacuation. Smoke filling calculations can be done Q = heat release rate from steady fire, Btds (kW);
by the computer zone fire models or by application of
A = cross-sectional area of the atrium, ft2 (m');
the empirical filling equations presented below.
cc,l = 0.67 (1.1 1).
Empirical Filling Equations Equation (14.1) is based on a plume that has no
contact with the walls. Because wall contact reduces
The empirical filling equations are based on smoke entrainment of air, this condition is conservative.
filling tests (Heskestad and Delichatsios 1977; Nowler Equation (14.1) is Tor a constant cross-sectional
1987; Mulholland et al. 198 I ; Cooper et al. 198 1 1 Hag- area with respect to height. For other atrium shapes, the
glund et al. 1985). zone fire model AZONE, physical modeling, or CFD
Because of the difficulty in determining the bottom can be used.The equation is appropriate for A/H' from
of the smoke layer from experimental data, the correla- 0.9 to 14 and for values oTz greater than or equal to 20%
tions below use tile unique concept of the first indication of ff. A value of z/H greater than one means that the
of smoke (Figure 14.3). In the idealized zone model, the smoke layer under the ceiling has not yet begun to
smoke interface is considered to be a height where there descend. These conditions can be expressed as
is smoke above and none below. In actual fires, there is
A = Constant with resepct to r , (14.2)
a gradual transition zone between the lower cool layer
and upper hot layer. The first indication of smoke can be
thought of as the bottom of the transition zone. As might
be expected, predictions using the equations of this sec-
tion differ froill predictions using a zone fire model. and

Filling by a Steady Fire


For a steady fire. the smoke filling can be appmsi-
When Equation (14.1) is solved for ?/H, the user
mated as
\\,ill find that ?/H is often outside the acceptable range.
The steady filliiig equation can bc solved for time.
Principles of Smoke Management

Figures 14.4 and 14.5 show the time predicted from


Equation (14.5) for the top 80% of the atrium to fill with
smoke (z/H = 0.2). Considering that evacuation times
are often in the range of 15 to 30 minutes, it can be seen where
from these figures that smoke filling is only appropriate
z = height of the first indication of smoke above the
for very large atria. The dashed lines on these figures
fire surface, ft (m);
show the range .of applicability (0.09 A / H 5~ 14) of
the steady filling equation. H = ceiling height above the fire, ft (ft);
I = time, s (S);
Example 14.1 Smoke Filling by a Steady Fire
tg = growth time, s (S);
How long does it take for a 5000 Btds (5280 kW) fire to fill
the top 70 ft (21.3 m) of an atrium with smoke? The height A = cross-sectional area of the atrium, ft2 (m2);
and area are 100 ft (30.5 m) and 100,000 ft2 (9.290 m2). Cd = 0.23 (0.9 1).
The height of the firs: indication of smoke abo\.e the fire As with Equation (l4.1), Equation (14.6) is conservative in
surface, z, is 100 - 70 = 30 ft (9.1 m), and z/H = 30/100 = that it estimates the height of the first indication of smoke
0.3. and is for a plume that has no wall contact.
From Equation (14.5), the filling time is Equation (14.6) is also for a constant cross-sec-
tional area with respect to height, and the comments
about atria of other shapes in the section above also
apply to this section. The equation is appropriate for AI
H? from 1.0 to 23 and for values of z greater than or
equal to 20% of H. A value of zlH greater than one also
means that the smoke layer under the ceiling has not yet
begun to descend. These conditions can be expressed as
Filling by an Unsteady Fire
A = Consrant with respect to H, (14.7)
As discussed in Chapter 2, the t-squared fire can be
used as an approximation of the growth stage of fire
development. For the to7steody filling eqrtotiot7 dis-
cussed below, the fire continues to grow throughout the and
filling process. As already stated, evacuation times are
often in the range of 15 to 30 minutes. The fire at the
end of the evacuation can be extremely large, limiting
the applicability of this equation (Table 14.1). However,
the unsteady filling equation is included here for com- The growth time, lS, has already been discussed,
pleteness. and values of it and characteristic fire growths are dis-

Ceiling Height Above Top of Fuel (m)


Figure 14.4 Titne for stuokefi.ot~r0 5000 Btrds (5280
k W)fit.elo./ill 111elop 80% of'11teolriritt~
(zM = 0.2) eslitmled bv tire etrrpit-icol
,/illitrg ecpitiotl (/P ro7i1.s).
Chapter 14- Atrium Systems

Table 14-1:
Heat Release Rate at the End of the Evacuation Time for Unsteady Filling Equation
Evacuation Slow Fire Medium Fire Fast Fire Ultra Fast Fire
Time tg= 600 S fg = 300 S fg = 150 S fg=75s
Minutes Btuls kW Btuls kW Btuls kW Btuls kW
15 2,250 2,370 9,000 9,500 36,000 38,000 144,000 152,000

30 9,000 9,500 36,000 38,000 144,000 152,000 576,000 608,000


Notes: I . tII is the growth rime for a t-squared fire to reach 1000 Btds (1055 kW); see Chapter 2.
v

2. Because ofthe laqe fires at the end ofthe evacuation time, the unsleady filling equation has limited applicability.

- 30 60
H = 100 ft (30.5 m)
A = 50.000 f f (4650 m')
,3O - H = 100 fl(30.5 m)
A = 50.000 ft' (4650 m') /
/
/ --

- 2 5,
Q = 5000 Btuk (5280 k W ) E

Zone Fire Models:

I g Zone F~reModels
CFAST -ASET-C
Steadv -','>. \ 70 - CFAST - 20

0 300 600
1

900
Time ( S )
1200
'
1500
' 10
1800 3 0 ,a0 0
Time (S)
,e ,I 10

Figure 14.6 Conzpar-isoiz of clear heights siti~ulated Figure 14.7 Coinpar-isoil of snzoke 10-vet- tettpv-rr-
by d@wnt rnodels. twes sitnirlated!C d i f f e ~ utuodels.
t

cussed in Chapter 2. As with the steady filling equation, indication of smoke above the fire, as illustrated in Fig-
the unsteady filling equation can be solved for time: ure 14.3. The zone models predict the clear height as the
smoke interface. For these reasons, it is expected that
the empirical steady filling equation would predict
lower clear heights than the zone models.
Heat transfer was calculated differently for each of
where Cej3 is 0.363 (0.937). the zone models. The CFAST simulation calculated heat
transfer to gypsum board walls and ceiling based on the
Computer Modeling temperature difference between smoke layer and the
gypsum board. Both ASET-C and AZONE use factors to
The height of the smoke layer ajove the fuel is
estimate heat transfer.
sometimes referred to as the clear- height, and Figure
14.6 shows a comparison of clear heights predicted by ASET-C estimates heat transfer by the heat loss
different zone fire n~odelsand the steady filling equa- fi-action, &which is the fraction of the heat release rate
tion. These predictions are for a large atrium o f H = 100 of the fire that is lost to the bounding surfaces of the
ft (30.5 m) and A = 50,000 ft2 (5780 m2) with a steady room and its contents (Appendix F). The heat loss frac-
fire of 5000 Btuls (5270 kW). It can be observed that the tion is generally in the range of 0.6 to 0.9. AZONE eval-
predictions of ASET-C and AZONE are close to each uates heat transfer by the convective fraction, X,, and
other. CFAST and the steady filling equation predict the wall heat transfer fraction, v.
The convective Lac-
lower clear heights. tion is the convective portion of the heat release rate; for
The differences in predicted clear height can be more information about this fraction, see Chapters 2 and
attributed to inherent differences in the prediciive tools. 13. The wall heat transfer fraction is the fraction of the
These differences include ( l ) the plume models. (2) the plume enthalpy flowing into the smoke layer that is lost
definition of clear height, and (3) the approach to heat to the walls and ceiling.
transfer. For each of the zone models, the mass flow of The smoke temperatures associated with the clear
the plume is calculated from different plume models. heights of Figure 14.6 are shown in Figure 14.7. For
As previously stated, the empirical equation is con- ASET-C, a value of 2,. = 0.6 was chosen. For AZONE,
scrvativc in that it predicts the clear height as the first X,. = 0.7 and 11 = 0.4 wcrc used. The factors are rclatcd
~ r i n c i ~ lof
e sSmoke Management

as Ac = I - ( I - ?l)and, thus, AZONE was effectively Steady Conditions


simulated with Ac = I - 0.7(1 - 0.4) = 0.58. It is not sur- The method of analysis presented in this section is
prising that the smoke temperatures are almost the same based on the simplieing assumptions below.
for the ASET-C and AZONE simulations (Figure 14.7).
The smoke temperature of the CFAST simulation The only mass flow into the smoke layer is the fire
was higher, but the convection coefficients upon which plume.
the wall heat transfer was based are calculated from The only mass flow from the smoke layer is the
general correlations. No convection coefficients have smoke exhaust.
been developed specifically for fire compartments. The exhaust is removing only smoke and not any
air from below the smoke layer.
MECHANICAL EXHAUST The smoke layer height is constant (Figure 14.10).
j The flows into and out of the smoke layer are at
Mechanical smoke exhaust is probably the most equilibrium.
common form of atrium smoke management in North Heat transfer between the smoke layer and the sur-
America. As with natural venting, mechanical smoke roundings have reached equilibrium.
exhaust can be based on either a steady or an unsteady
design fire. The equations of the next section deal with a Before using this method, designers need to verify
steady fire, and zone fire models can be used to analyze that these assumptions are appropriate for their applica-
smoke flow due to an unsteady fire. tion.

~ i i u r 14.8
e Katzi~-alsmoke volti~ig. Figure 14.10 Mecha17icalsuioke exhalist and cotista~it
clear keiglit.

Velocity Unaffected By Building


C

Note: Because wind


can produce positive,
preSsures at the top L
of an atrium, natural \
smoke venting is not
recommended for an
atrium anached to or
near a tall building in

.... . . .C- > ,. .. .. . . .: . .,,. .. ....


. .

Figure 11.9 Windjlo~c.pattern prodtrci~iga positive p/-esslrr-e011the top o f a n atrizr~ndue to the


prcse~lcc? f a ~ u lbziilding
l ~iear-by.
Chapter 14- Atrium Systems

To calculate the exhaust flow rate, the plume equa- 0, = convective heat release rate offire, ~d~
tions from Chapter 13 are adapted with variables rede-
fined for the following application: Cp = specific heat of plume gases,Btuflb "F (kJAcg "C);
1 /3 5/3
q = wall heat transfer firaction (diiensionless).
~ z = C a I Q cz +C,~Q, forz>zl (14.11) As already stated, the wall heat transfer factor is the
and fraction of the convective heat release rate that is trans-
f e n d to the waiis and ceiling of the atrium. This factor
depends on a number of conditions, including the geom-
etry of the space, the construction materials of the walls
where and ceiling, and the smoke layer temperature.
An atrium with no heat transfer is referred to as an
rir = mass flow exhaust of exhaust air, Ibls (kgts);
adiabatic atrium (v = 0). The adiabatic assumption is
- convective heat release rate of fire, Btuk (kW); conservative in that it results in high predictions of volu-
QC -
= height of the smoke layer interface above the metric smoke exhaust, but it is not conservative with
fuel, fi (m); respect to plugholing. In the absence of research about
the wall heat transfer fraction, values of q are expected
Z1 = mean flame height, ft (m);
to be in the range of 0.3 to 0.7 for walls and ceilings of
c,, = 0.022 (0.071); normal construction materials (brick, concrete, glass,
c,, = 0.0042 (0.00 l S); gypsum board, etc.).
The density of the exhaust gases can be calculated
Col0 = 0.0203 (0.032).
from the perfect gas law,
The mean flame height is

where where
C,, l = 0.533 ( 0166). p, = density of exhaust gases, lbrnlf? (kgh3);
91-ictlyspeaking, Equations (14. I I ) and (14.12) are
for the mass flow rate ofan asisym~nelricplume into the p = atmospheric pressure, lbflfi2 (Pa);
upper layer. M'hen the axisymmctric plume equations
at-e not appropriate, other plunle equations may be used.
R -
gas constant, ft Ibfllbni "R (Jkg K);
7;. = absolute temperature of exhaust gases, "R (K).
For the balcony spill p l u m equations and the window
plu~neequations, see Chapter 13. Alter-natively, the density of the exhaust gases can
l'he convective the heat I-elease rate, G,., is be calculated from

Q,. = x,.Q (14.14)

whcrc
= convcctivc lyaction ol'heal relcasc (see Chap[cr.s2 where
Y,,
and 13); T,. = absolute reference temperature, "R (K);

0 = total Ilcat rcleasc rate, Btuls (kW). p,. = density at reference temperature, lbm/ft3 (k9/m3).
For convei~iciicc,the tarn sr~okclayer Iieighr will There are an infinite number of pairs of T,. and p,.
be used to mean lllc heigh~of the smoke layer inlerfacc. that can be used in Equation (14.17), and one such pair
The term ;is snioke layer licigllt above the luel. is 530°R (294 K) and 0.075 lbrn/ft3 (1.20 kg!m3).
The te~npc~-nlure of the smoke layer can he The volu~netricHow of exhaust gases in plume is
cxprcsscd as

bc(l - 11 )
T , = 7;) + ---
lil c,,
i/ = volumetric tlow ofeshaust gases, cfni (m3/s):
whcr-c
I;. = s~iiokcIaycr temperature. "F ("C): a = mass Ilow ofeshaust air, Ibls (kgs);
7;, = a~nbicntrcmpcraturc, "F ("C): p,, = density ol'csllnust gases, lwft3 (kg/&);
C,,- = 60 ( 1 ).
Principles of Smoke Management,

Example 14.2 Steady Smoke Exhaust


What is the smoke exhaust needed to maintain a smoke layer height of 36 ft (11.0 m) with the design parameters listed below?
Ambient temperature 72.0°F (22OC)
Ceiling height 45 ft (13.7 m)
Convective fraction 0.7
Height of top of fuel Oft(0m)
Heat release rate 2000 Btuk (21 10 kW)
Wall heat transfer fraction 0.4

Note that the smoke layer depth is 45 - 36 = 9 ft (2.7 m), which is 20% of the height of the atrium ceiling above the fuel. This
depth accommodates the formation of the ceiling jet as in the section "Minimum Depth of Smoke Layer" in Chapter 13.
From Equation (14.14), the convective the heat release rate is

Q, = x c =~ 0.7(2000) = 1400 Btds(1480kW).

From Equation (14.13), the mean flame height is


2/5
z, = 0.533Qc = 0 . 5 3 3 ( 1 4 0 0 ) ~=~9.7
~ ft (3.0 m).

The smoke layer height, z, is 36 ft (l 1.0 m).


Because z, is less than z, the mass flow is calculated from Equation (14.1 l):
. 1 /3_5/3 + 0.0042Qc = 0.022(1400'/~)(36~/~)
+ 0.0042(1400) =
1 = 0.220, , 102 Ib!s (46.4 kgls).
From Equation (14.15), the smoke temperature is

1) From Equation (l4.17), the smoke density is

1) From Equation (14.18), the volumetric tlow ofexhaust gases is

'CJnsteadyConditions rt is the nature of fire that it is an unsteady process


Probably the reason that steady fires are used exten-
Unsteady analysis of an atrium exhaust system may
sively is that they lead to the simple steady analyses like
be done to
the one above. While large steady design fires can be
selected to yield conservative designs, these design fires
simulate a combination of smoke filling and snioke are not realistic. See Chapter 2 for information about
exhaust, design fires. Zone fire models such as CFAST and
simulate the effects of an unsteady fire, and AZONE can be used for analysis of atrium smoke
determine the impact of activation time on smoke exhaust systems \\.it11 unsteady fires.
layer depth. Before smoke exhaust fans can be turned on, the
presence o f the fire needs to be detected. There is some ,

A combination of smoke filling and smoke exhaust delay betwzen detection and activation, and it takes
can be used for an atrium that is not large enough to some time for the fans to come up to full speed. Detec-
qualify for smoke protection solel!~ by smoke filling. tion time can be estimated from the inforniation about
For this combination approach, the exhaust fans need to the lag times of plumes and ceiling jets pro\.ided in
be sized so that the smoke filling time is greater than the Chapter 13. When appropriate, detection should takc
evacuation time, including the time it takes to become into account the potential that there could be a stratified
aware of the fire and to prepare for movement to an exit. layer of hot air under the ceiling, as discussed later.
Chapter 14 -Atrium Systems

-9

-2
5- l,=150s 5 - A=l~~)fl~(92.9rn')
1 1
0- 0 0 0
0 60 120 180 240 300 0 60 120 180 240 3M)
l i m e (S) l i m e (S)
(a) Variation of smoke layer with atrium area. A (b) Variation of m o k e layer witJ~exhaust adimtion time. l ,

Notes:
-9 1. The Are is a 1-squared fire up to 2000 Btuk (2110 kWW). after
that the HRR remains mnstanL
2. The exhaust flow rate was seleded so that lhe midness of lhe
-7 - smoke layer would be 6 fl(l.83 m) at a sieady HRR of MOO
Etuk (2110 k W .
.c 3. As wilh other zone fire models, the details of lhe ceiling jet are not
to= 1 9 -5 simulated by AZONE. Thus. onty the portions of these graphs
.... - =
, 300 . '5 where lhe smoke layer is Celow about 24 R(7.3 m) are realistic.
---- f,=600s . 5 4. Other factors are:
Ambient Temperature. T, = 72.0 'F (22.2 %l
l, = 90 S -2 Ceiling Height. H= 30.0 fl(9.1 m)
- A = 1000 ft'(g2.9 m2) -1 Height of top of fuel. H
, = 0 A (0 m)
Exhaust Row rate. V = 49500. h(23.4 m%)
0 0
60 120 180 240 3M) Exhaust location factor. P = 2
Exhaust location Delow ceiling, d. = 0?i (0 m)
lime (S)
Number of exhaust inlets. ,N =6
(C) Variation of smoke layer fire growth. l,
Wall Heat transfer fraction. q = 0.4

Figure 14.11 U17steadylayer- height sit~rdated.by the zonefire model AZONE.

It is possible tliat the snioke layer could descend areas (A/H~> 5 where H is the atrium height), the effect
well below the design smoke layer height based on a of fan activation at 90 s would not be expected to have
steady analysis. To check the effect of activation, an adverse effect on the smoke layer heiglit. For atria
AZONE allows tlie user to specify the acti\.ation time o f with relatively small areas (A/H~< 5 ) , the smoke layer
the smoke exhaust fan. could drop below the design lieight, resulting in smoke
Figure 14.1 1a shows the effect of the atrium area on contact with people. AZONE can be used to analyze the
the smoke layer height as calculated by AZONE for an effects of activation tinie on the smoke layer lieight.
atrium 30 ft (9.14 m) in height with an exhaust activa-
tion tinie of 90 seconds. It can be seen tliat for an atriuni Makeup Air
area, A, of 5000 ft2 (465 ni2) or more, the delay in acti-
vation does not have an adverse effect on smoke layer For steady flow, the mass flow of air or smoke
height for the conditions of the simulations. For A = exhausted from the top of an atrium equals the mass
2000 ft2 (186 ni2) or less, the smoke layer drops well flow of air entering below the smoke layer. This airflow
below the design lieight for the conditions of the simula- entering the atrium is referred to as niakeup ai:, and
tion~. makeup air can be either supplied naturally or by fan
power.
Figure 14.1 1 b shows the effect of exhaust activation
time, to,,, on smoke layer height for a 30 ft (9.14 m) tall Fan-powered niakeup air is often sized at 90% and
atrium with A = 1000 ft2 (92.9 ni2). As expected, the 97% of the exhaust airflow rate, and the balance of the
smaller the activation time, tlie less the effect on smoke air needed to acco~nnlodatethe exhaust naturally flows
layer lieiglit. At t,,, = 30 S, the smoke layer stays above through openings or leakage paths. Natural makeup air
design lieiglit tliroughout the simulation. flows through openings, such as open doorways and
Figure 14.1 1c shows the effect of tlie fire growth vents, and sometimes makeup airflow paths are complex
time, t6" on slnoke layer height for a 30 ft (9.14 m) tall conlbinations of rooms and corridors. Computer net-
atriuni with A = 1000 ft2 (92.9 ni2). As would be work airflow programs, sucli as CONTAM (Chapter 8),
expected, the less the growth tinie (faster the tire), the can be used for analysis of these complex flow systems.
greater the effkct on tlie smoke layer height. The velocity of makeup air should not destroy the
While a study has not been made on the effect of plume structure or significantly deflect the plume at an
the activation time on smoke layer height. some gcneral- angle. It is believed tliat keeping the velocity at or below
izations can be made. For atria with relatively large 200 fpm (l nils) will prevent sucli plume disruption.
Principles of Smoke Management

NATURAL VENTING temperature may be less than the outdoor summer


Natural smoke venting is common in many parts of design temperature.
the world, such as Europe, Australia, and New Zealand. The smoke temperature and mass flow of the plume
As stated in Chapter 1, natural venting was developed in can be calculated from the same equations that are used
response to several fire tragedies in the 19th and early for mechanical exhaust as discussed later.
20th centuries.
Natural venting relies on the buoyancy of hot
Wind
smoke to force smoke out of open vents at or near the For an atrium attached to a tall building or very near
top of the atria (Figure 14.8). Natural venting can be a tall building located in open terrain, wind can produce
based either on a steady or an unsteady design fire. The positive pressures at the top of the atrium, as shown in
equations in the next section are for a steady fire, and Figure 14.9. Because such positive pressures can inter-
zone fire models can be used to analyze smoke flow due fere with natural venting, natural venting is not recom-
to an unsteady fire. mended for atria with s ~ windh conditions.

Steady Conditions Makeup Air


For natural smoke venting described by Equation
The equation developed in Chapter 13 for the mass
(14.19), makeup air flows naturally through the inlet
flow rate through the vent is
opening of area, At Makeup air is generally supplied
through open vents or doorways. A sprawling atrium
can be divided into a number of large spaces with smoke
vents so that the smoke vents in the spaces away from
the fire can be opened for makeup air.
wherc
TENABILITY SYSTEMS
m, = mass flow rate through the vent, Ib/s (kgls);
As already stated, the approaches discussed above
C = discharge coefficient (dimensionless);
have the goal of not exposing occupants to smoke dur-
A,, = vent area, f? (m2); ing evacuation. Tenability systems are designed to
maintain tenable conditions with occupant exposure to
A. = inlet opening area, ft2 (mZ);
smoke.
P0 - outside air density, lb/ft3 (kg/m3); Hazard analysis consists of evaluation of smoke
transport, people movement (evacuation time), and tena-
g = acceleration of gravity, 32.2 ft/s2(9.80 m/ sZ); bility. While smoke transport can be simulated by zone
db = depth of smoke layer below the smoke vent, ft (m); fire models, CFD modeling has the significant advan-
To = absolute temperature of outside air, "R (K); tage of being able to simulate variations of temperature
and concentrations of combustion products in the smoke
r, = absolute temperature of smoke, "R (K); layer.
Because buoyancy of hot smoke is the driving force Evacuation time can be evaluated by the methods of
of natural venting, the mass flow rate, );I,,, through the Chapter 4. Tenability analysis should address visibility,
vent increases with increasing smoke temperature, TT. gas exposure, and heat exposure, and extensive inforrna-
As the size of a fire increases, the mass flow rate of the tion about tenability can be found in Chapter 3. Ths cal-
plume into the upper layer increases and the temperature culation method for tenability described in Chapter 9
of the smoke layer increases. For a fire larger than the can be used for atria.
d-sign fire, the smoke temperature goes above the
design value, and the mass flow rate through the vent STRATIFICATION A N D DETECTION
increases above the design value. This benefit is unique Often, a hot layer of air forms under the ceiling of
to natural venting, and it helps offset the greater amount an atrium as a result of solar radiation on the arriuin
of smoke produced by fires that might exceed ths design roof. While studies have not been made of this stratified
fire. layer, building designers indicate that the temperatures
For air-conditioned atria, it is possible that the of such layers are often in excess of 120°F (50'C).
smoke temperature may be less than the outdoor sum- Temperatures below this layer are controlled by thc
mer design temperature. This can result in doivnward building's heating and cooling system. and the tempera-
outside airflow through the atrium smoke vents. To ture profile can be considered to increase significantly
avoid such downward flow through smoke vents, natu- over a small increase in elevation as shown i n Figure
ral smoke venting should nor be used wlicn the smokc 14.12.
Chapter 14-Atrium Systems

Elevation Abjve Fwl (m)


0 30 60 90
120
Heat Release Rate:
-1w E
5.000 Btul S(5.280kW)
2.OM)Btu1 S(2.110 k W )
-80
f
p
D

-60 cP
W

-40
P
- 20
l I
Q

50 I
0 50 100 150 200 250 300
Temperature Elevation Above Fuel (R)

Figure 14.12 Temperature profile of hot layer- of air Figure 14.14 Average femperature of axisyn~rnetr-ic
zrnder atrium ceiling. pfznne.

For redundancy when using this approach, more


than one beam smoke detector is recommended.
b. Horizontal Beams to Detet the Smoke Layer at
Various Levels
The purpose of this approach is to quickly detect
the development of a smoke layer at whatever tem-
perature condition exists. One or more beam detec-
tors are located at the roof Icvcl. Additional
detectors are located at otlier levels lower in tlie
volume. The exact positioning of the beams is a
Figure 14.13 Smoke stratificatior~under a layer of hot
function of the specific design but sllould include
ail:
beams a1 the bottom of identified unconditioned
spaces and at or near the design smoke level \\it11
When the average temperature of the plume is less several intermediate beam positions at otlier levels.
than that o f t h e hot air layer, tlie smoke will form a strat-
ified layer under it, as shown in Figure 14.13. Average c. I-lorizontal Beanis to Detect tlie Smoke Plume
plume temperatures are sho\vn in Figure 14.14, and it The purpose of this appl-oacli is to detect [he rising
can be observed that the average plume temperature is plume rather than the stnokc layer. For this
often less tliali expected temperatures o f tlie hot air approach, an arrangement ol'beams are installed at
layer. Thus, when there is a hot air layer under the a level below tllc lowest expected stratification
atrium ceiling, smoke cannot be expected to reach the level. These beams need to be close enough to each
ceiling of the atrium; and smoke detectors mounted on other to ensure intersection of the plume, the spac-
that ceiling cannot be expected to go into alami. ing being based on the width of the beam at the
least elevation above a point of fire potential.
Beam smoke detectors can be used to overcome this
detection difficulty. The follo\\:ing are beam detection Tile approaches described above are illustrated in
approaches that can provide prompt detec:ion regardless Figure 14.15, and approach (a) has the advantage that i t
of the temperature of the air under tlie ceiling at the time does not require the location of a number of horizontal
of fire initiation. beams. Some bean1 smoke detectors are subject to false
a ~ t i \ ~ a t i obyn sunlight, and alternative (a) min;niizes the
a. An Upward-Angled Beam to Detect the Smoke possibility of such false activation bp orienting the
Layer rcccivcr at adownward aligle.
The purpose of this approach is to quickly detccl All of the coniponcnts of a beam s~iiokedetector
the development of a smoke layer at whatcver tcm- nccd to bc located so they are accessible for nlainte-
perature condition exists. One or more beams arc nancc. For thc arrangement sliown in Figure l?. 15. a
aimed at an upward angle to intersect thc smoke roof opciiing (not shown) could provide access for
layer I-cgardless of h e Imel of s~nokestratilication. mainte~iancc.
Principles of Smoke Management

Plan View Section


(a) Upward Angled Beams t o Detect the Smoke Layer

Suggested Spacing
of Beams:

Plan View Section


(b) Horizontal Beams t o Detect the Smoke Layer at Various Levels

Suggested Spacing
of Beams:
X = -B
4

Plan View Section


(c) Horizontal Beams to Detect the Smoke Plume
Figure 14.1 A~-ra~~,oe~nerits
of beam smoke detectors.
Chapter 14 -Atrium Systems

M2 = mass of smoke layer at the end of the interval (kg).


Step I: Assign Values to Consfan&
C,, R, P-, U. X,, a The change in energy of the smoke layer can be
expressed as
t
Step 2: Read Data
Z.H.A.H,.,,~,.Q.~.
where
P. 4. N&". L.4".L
AE = change in energy of the smoke layer (H);
t
Step 3: Assign lnitlal Values
Cp = specific heat of smoke ( k l k g K)
wall heat transfer fraction (dimensionless);
=
m, =o; T , =C; p,=p.l(RT,,);
z=H-H,; E=O; M = O ; Tp = absolute temperature of plume gases entering
Q = O ; ~fr,<o.
Q=Q~ smoke layer (K);
T I=
t
Step 4: Calculate for each time step
=
absolute temperature of smoke layer gases at the
beginning of the time interval (K);
absolute ambient temperature (K);
r=r+ar ; c ; Q ; Q<;d ; d-;
To
m,:z,:mp; T,; M ;M ; AE; As with the mass flows, Tp is evaluated at the end
E;<,;p,; < ; X of the interval. The ambient temperature is considered
1 constant throughout the calculations. The smoke tem-
Step 5: Check for perature, T,,, at the beginning of the interval was used
because the value at the end is unknown. Selection of a
Write Oulput small interval makes the resulting error negligible.
The energy in the snioke layer is

Tiue where
El = energy of the smoke layer at the beginning of the
False interval (U);
Slep 6: Output
E2 = energy of the smoke layer at the end of the interval
(kJ).
If I is an
The smoke temperature, at the end of the time
even multlple Write Output
interval is

char-I for- AZONE


Figure 14.17 Sirr7pl~fiedflo~t~
The density of the smoke layer is
the same reason, all equations used to describe AZONE
are also in S1 units.
For simplicity, the mass flows, rii, and nr', , are
evaluated at the end of the interval, and a small interval where
was used to minimize errors. However, these flo\vs
could be evaluated at a time within the interval with the = smoke density at the end of the interval (kg/mZ);
goal of using a larger interval. ,vo = ambient pressure (Pa);
The mass of snioke in the layer at the end of the R = gas constant (Jkg K).
interval can be expressed as
The volume. V?, of the smoke layer is

where
M, = mass of smoke layer at the beginning oftlie inter- For an atrium of constant cross section, the height
val (kg), of the smoke layer above thc top of the fuel is
' Principles of Smoke Management
. -

Variable Atrium Area


Height and area pairs in descending order are pre-
scribed as hiand A i for i to n. For each height, h , the
where atrium volume, Gi, above that height is
z2 = height of the smoke layer above the top ofthe
fuel (m);

, .
H = height of atrium (m);
/

H--l = height of he1 (m); The units of hi,Ai,and V,; are m, m2, and m3. The
terms hi,Ai,and V,; are terms of arrays (sequences of
A = cross-sectional area of atrium (m 2).
numbers), and the subscripts i and j are what is referred
It should be noted that H has a different definition to as dummy variables. For example, h; where i = 3 is
in AZONE than it has for the empirical filling equations. the third value of the height array. Before calculations
The various height temis above are illustrated in Figure are done for the time intervals, the values of Voiare cal-
14.17. Determination of z2 for an atrium of variable area culated for i = to n.
is discussed later. The values at the end of the current The atrium area at any height X is
time step become those at the b2ginning of the next time
step.

Plugholing where
For each time interval, the exhaust from the smoke A(x) = atrium area above height X (m2);
layer, i z , , is calculated, taking into account any plugh- X = height above atrium floor (m).
oling that might be happening. The minimum smoke j = dummy variable such that 17, <X 5 h,- ,.
layer depth to prevent plugholing is
The volume above any heights is

where V(x) is the atrium volume in m3 above height X.


where The height of the smoke layer above the floor is the
d,,, = minimum smoke layer depth to prevent plughol- value ofx, which satisfies the following equation:
ing (m);

V,, = volumetric flow rate per exhaust inlet (m3/s);


The value o f x that satisfies Equation (14.33) can be
p = exhaust location factor (dimensionless); determined by any of a number of root finding methods.
Cph4= 0.15. In AZONE the method of bracketing and bisection was
used (Press et al. 1986).
The volumetric flow per inlet is p,, = I/,/N; The height of the smoke layer above the top of the
where Nin[,, is the number of exhaust inlets. The fuel is
exhaust from the smoke layer is

Time Interval
The time interva~,~ At, needs to be selected so as to
minimize error. Theoretically, errors associated with the
interval size are due to inaccuracies of numbers used -
I = p l . for d,,! I tl from previous iiltervals. In AZONE, TI,M,, and E l are
calculated in the previous interval, and the values of the
where heat release rate and exhaust airflow are each evaluated
d = depth ofsmoke layer bclow bottom ofeshaust inlet,
9. The time interval should not be confused with the
(m); output internal. Calculations are made at each time
a = plugl~olingexponent (dimc~lsionless). interval. but data arconly written at theoutput intervals.
Chapter 14-Atrium Systems

Table 14-2:
The Effect of Time Interval on the Accuracy of AZONE ~imulations'
Time Steady ~ i r e ~ Fast t-squared ~ i r e '
Atrium Cross-sectional Interval,
Height, H Area, A At Simulation Time ~rror' Simulation Time ~rror~
ft m ft2 m2 S S % S YO
Small Atrium
l
30 9.14 1,000 93 0.005 30 0.0 90 0.0
0.0 1 30 0.0 90 0.0
0.05 30 0.2 90 0. I
0.20 30 1.2 90 0.2
0.50 30 3.7 90 0.6
1 .OO 30 7.7 90 1.2
5.00 30 65.0 90 G. I
Small Spread-Out Atrium
30 9.14 12,000 1,110 0.01 240 0.0 300 0.0
0.05 240 0.0 300 0.0
0.20 240 0. I 300 0.1
0.50 240 0. I 300 0.1
1.OO 240 0.3 300 0.3
5.00 240 I .j 300 I .5
20.00 210 6.3 300 6.1
Lnrgc Atrium
I SO 45.7 25,000 2,320 0.0 1 4SO 0.0 300 0.0
0.05 480 0.0 300 0.0
0.20 480 0.0 300 0. I
0.50 480 0. I 300 0. I
1 .OO 4SO 0.3 MO 0.3
5.00 IS0 I .-I .NO I .A
70.00 480 6.0 .300 5.8
Large Spread-Out Atrium

20.00 1 200 0.7 600 0.7


I. Co~iditio~isol'thc si~iiuliltions:( l ) alnhlcnt rsmperaturt ol'7OoF (?IT).( 2 )cnnsl:~ntcross-secliotial ;lrc:u. (31 no sliitAs r.h;~ust.(4) top of filet a[ lloor
level. (5) wall Iieal transfer fraction 01'0.3.
2. The steady lire was 5.000 Rtuls (5275 kW).
3. For the 1-squared tirc, the growth tinw \\-as 150 S.
4. The error, d. is 111~error ol'thc smok In!;cr height, 2. using the equation S = lo0(:,,, -:)/I \\-hersI,,,
is 11icvaluc of: at :>c smallest
tinic intcrval Tor that a t r i i ~ ~size.
ii

at a point in the interval. For calculations made u.ith Table 14.2 lists 21-rors01'~1noke-tilli1lgsinlulations
successively smaller interval sizes. the absolure values for several values ol'Dt fbr rht at[-ium size categories ( I )
o f such theoretical errors also become smaller. small, (2) small spread out. (3) large, and (4) large
In a d d i t ~ o nto tlleoretical errors. round-ofi' errors spread out. Tliese el-rors pertain to the height o f the
also can be associated with interval size. The nature o f smoke layer, and errors (not s h o \ w ) of smoke layer teni-
numerical round-oti'errors is such that predictions made perature were less. For the t i l w intervals used, round-off
for very small interials can have ven. lalgc errors. S o errors due to small il?[cr-valsize did not occur.
the time interval needs to be cvaluartd s o that i t is nei- The largest cl.tor lis~cdin T a b k 14.2 was 65% for a
ther too large o r too stiiall. steady lirc in the snl:rll a~riuln.\\:llich sho\vs that these
Principles of Smoke Management

errors can become so large that results of a simulation For the small atrium, this interval resulted in errors of
can be meaningless. What is desired is an interval size 0.1% and 0.2% for the t-squared fire and the steady fire.
that has acceptable errors for all atria that might be ana- Accordingly, 0.05 S was chosen as the time interval 'for
lyzed. An interval of 0.05 s results in errors less than AZONE.
0.05% for all the atrium size categories except small.
CHAPTER 15

Physical Modeling

ne option when conventional methods of analy-

0 sis are inappropriate is fire testing in a reduced


scale model, and there is considerable experi-
ence with application of physical models to fire
technology. Such scale modeling has been used to R, = Reynolds number,
rsconstruct fires for fire investigations. Two examples
of such fire reconstructions are the King's Cross subway I = length,
station fire in London, United Kingdom (Moodie et al.
1988), and the Hart Albin department store fire (Quin- C' = velocity,
tiere and Dillon 1997). p = density, and
Froude modeling is probably the most common
p = dynamic viscosity.
kind of physical modeling used for smoke transport, and
NFPA 92B recognizes Froude modeling as a method of The above equation is a little dit'ferent from that
analysis of smoke management systems for atria. given in Chapter G for the Reynolds number. In Chapter
This chapter addresses the fundamentals of physical 6: the Reynolds number was expressed in terms of kine-
modeling of smoke movement with special emphasis on matic viscosity, v, where v = p/p. Also, the units of the
Froude modeling. For further information about fire variables are given in Chapter G . Most of the equations
applications of physical modeling, readers are referred in this chapter are intended to describe physical model-
to Arpaci and Aganval (199Q), Quintiere (1989b), ing and not be used for calculations. Accordingly, units
Heskestad (l 972, 1975), and Hottel (l96 1). are not given for variables in most equations of this
chapter. However, all of these equations are valid for SI
DIMENSIONAL ANALYSIS units or any other homogeneous unit system (Appendix
AND SIMILITUDE A).
The idea of dimensional analysis is to express a The Froude number can be thought of as the ratio of
complicated process in terms of a relatively few dirnen- inertial forces to gravity forces. Because the buoyancy
sionless variables. This can simplify the analysis and of hot smoke is a gravity force, the Froude iiun~beris
make physical modeling possible. Many dimensionless very important in physical modeling of smoke movc-
parameters can be viewed as being ratios of fluid forces. ment. The Froude number'' is
Three dimensionless parameters that- are of particular
interest in this chapter are the Reynolds number, Froude 10. An alternate fonn of the Froude number is I;;. =
number, and Prandtl number.
~ l ( ~ 1 ) This
' ' ~ . is 'simply the square root of the Froudc
The Reynolds number can be thought of as the ratio number. which is used in this book, and basic concepts
of the inertial forces to the viscous forces. and this num- concerning the Froude number and thc scaling rcla-
ber distinguishes between flow regimes such as laminar lions for Froudc nodel ling are the same rcgardlcss 01'
and turbulent. The Reynolds number is ~ r h i c hform of the Froudc number is used.
Chapter 15-Physical Modeling

Table 15-1:
Quantities and Associated Dimensional Formulas

where Dimensional
Quantity Formula
F, = Froude number, -Symbol(s)
Length L, X, z L
g =acceleration of gravity, and . Time
U = velocity. Mass
The Prandtl number is a dimensionless number, Temperature
which is the following combination of fluid properties: Force
Pc Heat
P,.= -y Velocity
Acceleration
where
P, = Prandtl number, Work
Pressure
Cp = constant pressure specific heat,
Density
P = dynamic viscosity, and
k = thermal conductivity.
Internal enersy
Enthalpy
Dimensional Formulas Specific heat of a solid
The system o f primary dimensions (or base dimen- Constant pressure specific heat
sions) can be chosen as length L, time t , temperature T,
and mass M. The dimensional formula of a physical Constant volume specific heat
quantity follows from definitions or physical laws. For Dynamic viscosity
example, the dimensional formula oCa doorway width is Kinematic viscosity v=p/p L* /I
[L] by definition. The brackets [ ] indicate that the quan- Thermal conductivity X- ML/$T
tity has the dimensional formula within the brackets.
The dinlensional formula of velocity is [L /t] and that of
acceleration is [L / t 2 ] . Other dimensionless quantities can be evaluated in
For a homogeneous unit system,'' Newton's second the same way.
law is
The n Theorem
The n (pi) theorem (Buckingham 19 15) states that,
where for any phksical application o r process that includes t7
F = force, quantities in which there are m dimensions, the quanti-
m = mass, and ties can be arranged into 17 - m independent dimension-
less parameters. further, some functional relation of
a = acceleration.
these 7t - n7 independent dimensionless parameters
The dimensional forn~ulafor force is the dimen- exists that describes the physical application or process.
sional formula of mass times that of acceleration. This is
Consider an application for which A,, A2, A3, . . ., A,,
[ML / P ] . Work is force acting through a distance, so the
are the essential quantities involved. such as length,
units of work are [ M L ~ / 81. The dimensional formulas
velocity, pressure, mass, etc. A functional relation of
of a number of physical quantities are listed in Table
these quantities can describe the application, and this
15.1.
can be expressed as
A dinlensionless quantity has no dimensions; for
example, the dimensions of the Froude number can be F ( . 4 , , A z ..4;, ..., A , , ) = O. (1 5.6)
evaluated as
The quantities A,. AZ, A3, ..., A,, can be arranged
into dimensionless groupings or parameters n l , n 2 ,
n3,..., n ,,.,,,. The functional relation of these n groups
will also describe the application
I I . For a discussion of Iiornogeneous unit systems,
scc Appendix A .
Principles of Smoke Management

The advantage of using the ll groups is that the and


number of independent variables is reduced from m to m
- n. For a specific application, some of the II groups
may be constants.
~n algebraic technique for determination of the ll p, is the ambient pressure distribution.
groups based on the ll theorem is presented in a number
of texts, such as Kreith (1965) and Streeter and Wylie Conservation of Energy:
(1979). A disadvantage with this technique is that there
is no one unique solution for the Il groups, and a num-
ber of possible combinations of rI groups may need to
be evaluated.

Similitude Equation of State:


Physical modeling has been used in many areas of The equation of state for an ideal gas is
engineering, such as wind tunnel studies of aircraft,
flow in rivers, and smoke transport buildings. The basic
concept is that a scale ,model of a full-scale facility is Variables in the above governing equations are
built, and conditions of the tests are maintained such Cp = specific heat,
that the rI groups are preserved. This means that at a
g = acceleration of gravity,
particular location in the model, each rI group has the
same value as it has at the corresponding location in the k = thernlal conductivity,
full-scale facility. T = temperature,
For perfect similitude, all the rI groups would be y = pressure,
preserved. Fortunately, perfect similitudeis not always p, = ambicnt pressure,
necessary. Useful results can be obtained from preserva-
u = X con~ponentof velocity,
tion of only some of the Ilgroups, provided that the
impact of other Il groups is not significant. This is X = position,
explained later.
Q"' = rate of chemical energy per unit volume,
Development of Dimensionless Groups p, = ambicnt density,
An alternative to the Iltheorem approach for devel- I radiant intensity,
=

oping rI groups is the differential equation approach, K = absorption coefficient,


which is more elegant and provides a high level of o = Stefan-Boltzman constant,
understanding. Further, the differentia; equation p = dynamic viscosity.
approach can be used to develop Ilgroups for physical
For an ideal gas, the gas constant, R, is
modeling of smoke movement in general, and those
groups can then be evaluated for specific modeling
approaches.
For the development of the dimensionless groups of where C,, is the constant volume specific heat.
interest, the governing equations of fluid dynamics that Dimensionless variables are defined below.
are listed below are in a one-dimensional form.

Conservation of Mass:

- ,(
(15.8)
G/ u.1

Conservation of Monlentum in the Vertical Direction:

where
p' = 11 - p , , (15.10)
Chapter 15-Physical Modeling

where the characteristic time, r, is chosen to be [/(I.This


(15.20) means that Ill = 1. Because this ll group is a constant, it is
always preserved and can be ignored.
(15.21) The next ll group is

where p+ is the characteristic pressure defect (p*= p,U 2).


Substituting this definition of p* into Equation (15.29)
results i n r12 = I, so this Fl group also can be ignored.
The third Il group is
where
I geometric length scale,
=
U characteristic velocity,
=
r = characteristic time, 17; is the Froude number, As previously stated, the Froude
To = ambient temperature, nurner can be tliou~htof as the ratio of inertial forces to
p, = ambient pressure, buoyancy gravity forces.
The four-th n group is
p, = ambient density,
3
p+ = characteristic pressure defect (p=p,U-).
By substitutiy the dimensionless variables of
Equations (15.15) to (15.23) into the governing equa- n, is the Reynolds number.
tions, the following nondirnensional form of the govern- The nest ll group is
ing equations can be developed:

Mass:
115is thc Prandtl nu~nber.For ~nanygases including air, the
Prandtl number is nearly constant with respect to tempera-
ture. Smoke is air mised with a relatively sniall amount of
combustion products. and the properties of smoke are gen-
Momentum: erally taken to be the same as those of air. Thus, nj can be
neglected for modeling done in air.
Thc next threc ll groups pertain to heat transfer:

Energy:
7 -
and

+ n3n5n,n,[ 1ic,(0-49
.IZ

+ U + n s21
g
(l 5.26) //I.

n, = ~ , , c , , u v ' (1 5.35)

7 he last n group is
State:
n -I
11, = 5
c,.
( l 5.36)
i= (+--)B? ( 1 5.17)
"any
n , is tlic commonly used ratio of specific heats,
The first n group is and this ratio is a constant Ibr ideal gases. For air, the
mtio of sDecitichears has a ncarly constant value of 1.4.
n = -
l
(15.2s)
Thus, n, can bc neglected for niodeling that is done in
I L'r air.
Principles of Smoke Management
. -

vided later about sizing the model to minimize the


effects of not preserving the Reynolds number.
For Froude modeling, the temperatures from 'the
model are the same as for corresponding places in the
full-scale facility. Because the temperatures are the
same for both, the heat transfer is somewhat similar for
both. However, the heat transfer groups (H6, H7, H,) are
not preserved. For smoke away from the flame, the tem-
perature is low enough so that the heat transfer groups
do not need to be preserved. However, for higher tem-
perature gases, such as those of flames, neglecting these
groups is inappropriate.
There has been considerable experience with
Froude modeling, and the comparison between full-
scale and 117 scale model temperatures (Figure 15. l) by
0 600 1200 1800
Quintiere, McCaffrey, and Kashiwagi (1978) illustrates
Time (S)
the degree of agreement that can be expected. Chow and
Figure 15.1 Froude modeling cornparisor7 of corri-
Lo (1995) used Froude modeling to simulate smoke
dor gas temperature (adapted from
movement and smoke filling in an atrium.
Quintiere, McCafiey, and Kashiwagi
[l 9 781).
Saltwater Modeling
The use of one fluid to model the flow of another is
TYPES O F MODELING called analog modeling, and saltwater has been used
extensively to model smoke movement. The idea of salt-
Froude modeling, saltwater modeling, and pressure
water modeling is to submerge the scale model in a tank
modeling have all been used to simulate smoke move-
of fresh water and inject saltwater to simulate a heat
ment in fire applications. Each of these modeling
source. Because the saltwater has a higher density than
approaches has less than perfect similitude in that no
fresh water, the saltwater tends to flow down, whereas
practical approach can preserve all the H groups. How-
smoke tends to flow upward. This is accommodated by
ever, these modeling approaches have produced good turning the model upside down in the tank.
quantitative results and provided insight into smoke
movement phenomena. Frequently, the models are constructed of a trans-
parent polymer, and the saltwater is dyed blue. This
Froude Modeling helps people to see, photograph, and video the saltwater
flow. The major advantage of saltwater modeling is
As previously stated, Froude modeling is probably
probably that it helps people to visualize smoke P-ow.
the most common approach to the physical modeling of
smoke movement. A scale model of the room, atrium, or Saltwater modeling is similar to Froude modeling
other facility is built. Tests are conducted in the model in that the Froude number is preserved. The concentra-
in air at normal atmospheric conditions. The scaling tiori of salt is adjusted such that the density forces of the
relations are used to size the design fire and any forced saltwater in the model correspond to that of smoke in
air flows, and these relations are used to translate mea- the full-scale facility. Saltwater modeling has no heat
surements from the model to the full-scale facility. transfer, but the saltwater mixes with the fresh water as
These scaling relations are discussed later in detail. it flows in the model. Because of the lack of heat trans-
Because buoyancy is a gravity force and dominates fer, saltwater modeling is not appropriate for simula-
the flow resulting from fires, the Froude number (n3) tions of flow of flames or flow near flames.
must be preserved. For reasons already discussed, i l l , Chow and Siu (1993) conducted smoke filling visu-
,,
H2, n and n9are also preserved. alization experiments on several atria using saltwater
If the size of the model is appropriately chosen, the modeling. Yii (1998) conducted a series of saltwater
flow becomes fully turbulent and the viscous effects at modeling experiments of balcony spill plumes. For gen-
solid surfaces are negligible. For this reason, the Rey- eral information about saltwater modeling, see Steckler,
nolds number (n,) can be ignored. Information is pro- Baum, and Quintiere (1986).
Chapter 15 - ~ h ~ s i cModeling
al

Pressure Modeling T, = temperature of gas in model, "F ("C);


Pressure modeling is included for completeness. Tf = temperature of gas in full-scale facility, "F ("C).
This modeling preserves both the Froude number and Because the model and the hll-scale facility a r e a t
the Reynolds number. The Reynolds number is pre- the same temperature and pressure, the scaling relation
served by changing the ambient pressure. The pressures for density is
can be described as

where
where p,, = density of gas in model, lb/& (kg/m3) and
p,, = pressure of the model, psi (Pa); pf = density of gas in full-scale facility, lb/fi? (kg/m3).
pf = pressure of the hll-scale facility, psi (Pa); Preservation of the Froude number can be
I,,, = length in the model, m (ft); expressed as
If = length in the full-scale facility, m (ft);
The units listed for Equation (15.37) are ones that
might be expected for this application, but this equation
is applicable to a wide range of units provided that both wherc
pressures are in the same units and both lengths are in
U,,, = velocity in the model, Ws ( d s ) ;
the same units. For example, the pressures could be in
atmospheres. and the lengths could be in inches. Uf = velocity in the full-scale facility, Ws (mls); and
A one-eighth-scale model would need to be tested = acceleration of gravity, ft/s2(m/s2).
g
in a pressure vessel at a pressure of about 23 atmo-
It follows from Equation (15.41) that the scaling
spheres. Probably due to the expense of testing in a pres-
relationship for velocity is
sure vessel and the extent to which Reynolds number
effects can be minimized in Froude modeling, pressure
modeling is hardly ever used. Like Froude modeling,
pressure niodeling does not preserve the heat transfer
groups. where
S C A L I N G RELATIONS F O R U,, = velocity in the model, Ws (mls);
FROUDE MODELING Uf = velocity in the full-scale facility, Ws (m/s);
The basic concept of a scalc model is I,,, = length in the model, ft (m); and
length in the full-scale facility, ft (m).
lf =

The units listed above for the variables of Equation


(15.42) were selected as they are ones that might be
where used for an application, but many other units can be
X,,, = position in the niodel, ft (m); used in this equation. The requirements for units are that
XJ = position in the tidl-scale facility, ft (m); U,, and Uf must be in the same units, and I,, and +must
be in the same units. For example, both velocities could
l,,, = length in the model, It (m); be in feet per minute (fpm), and both lengths could be in
lf = length i n the ftlll-scale facility, fi (m). inches. All scaling rela.tionships discussed in this section
The ratio (I,,, / l,) is the scale of the model. For are o f a similar form, and this basic idea about the suit-
example, for a one-tenth-scale niodel, l,,, /lf= 1/10. ability of a wide range of units is also true for all these
As already stated for Froude modeling, the temper- other scaling relationships.
atures liom the model are the same as for corresponding Volumetric flow is velocity multiplied by area, and
places in the Sull-scale facility. Thus, the scaling relation the relation beconles
for temperature is

wlicr-c where
Principles of Smoke Management

V, = volumetric flow in model, @/S (m3/s); where

VJ = volumetric flow in full-scale facility, 91s (m3/s). 4, = pressure difference in model, in. H 2 0 (Pa);

Mass flow rate is volumetric flow multiplied by = pressure difference in full-scale facility, in. H 2 0
density, s o combining Equations (15.40) and (15.43) 0'4.
results in The use of some of the scaling relations is illus-
1 5/2 trated in ~ x a m ~ l 15.1
e s and 15.2. Tsujimoto, Takenou-
h,,, = "l (A) chi, and Uehara (1990) conducted experiments that
J!f verified the above scaling equations for smoke move-
where ment in atria. Quintiere, McCaffrey and Kashiwagi
(1978) conducted smoke movement experiments that
rii, = mass flow in model, lb /S (kg Is); verified these scaling relations for smoke flow in a room .
and corridor.
hJ = mass flow in full-scale facility, Ibls (kgls).
Velocity is length per unit time, and substituting U,,, Example 15.1 The Scaled Fire
= l,/ t , and =U
! $1 t/into Equation (15.43) results in For a 5000 Btds (5280 kW) fire in a full-scale facility, what

l1
is the corresponding fire in a one-sevtnth scale model?
Using Equation (15.46),

where
t, = time in model, s (S);
Q , =
1 5/2
= ~ooo(!)~/~= 38.6 Btuis(40.7 kW).
Il
t/ = time in full-scale facility, s (S).
Consider the convective heat portion of the heat
1.75 ft (0.533 m) above the floor at 42 s after ignition. I-low
release rate as enthalpy flows. Q,,, = riz,,,Cp~T and does this convert to the full-scale facility?
QJ = k/CPAT(AT = AT,,, = ATf), then Equation (15.41)
becomes
I Rearrange Equation (15.45) as

where 11 Rearrange Equation (15.38) as


Q,, = heat release rate in model, Btuk (kW);

QJ = heat release rate in full-scale facility, B t d s (kW).


l1 .vJ = X.(+)
111
= 1.75(8) = 14 ft (4.27 m).

This means that at 119 s the smoke layer would descend to


If the convective fraction of the fire in the model: 14 ft (4,27 m) above the floor in the full-scale facility.
XC,,,,, is the same as that in the full-scale facility, ,ye$ the
scaling relation for the convective heat release rate is
Approximate Heat Transfer Scaling
Even though the heat transfer groups and the Rey-
nolds number were not preserved, some heat transfer
effects can be partially preserved by considering surface
where
heat transfer and solid heat transfer. For a semi-infinite
Q,, l,l = convective heat release rate in model, B t d s (kW); surface, the wall and ceiling materipls can be scaled by
convective heat release rate in full scale facility,
=
B t d s (kW).
The ressure difference due to velocity is
M,,, = p 2 1 2 and A J J ~= and substituting this where
/ '
into Equation (15.42) yields (remember, p = p,,, = p$ (kpc?,,:l,l = thennal inertia of the wall or ceiling material
of the model, ~ t u 'in h-' ft.' "F-' ( k ~ m-'
'
K-~S);
Chapter 15-Physical Modeling

(kmwf
= thermal inertia of the wall or ceiling material suggested that objects less than about 9 in. (0.23 m) can
probably be neglected.
of the hll-scale facility, ~ t in uh-' ~ff5 " F - ~
As already stated, the size of the model needs to be
chosen appropriately SO that the viscous effects are neg-
The thermal inertia (km for several materials is ligible and the Reynolds number (n4)can be ignored.
listed in Appendix A (Tables A10 and A1 I). Example The model needs to be large enough so that the flow is
15.3 illustrates calculation of the scale thermal proper- fully turbulent at locations of interest. The general rule
ties. In this example, Equation (15.49) was used to scale is that the smallest length that can support such turbu-
the thermal inertia of the model to 3.8 ~ t ft" u OF-'
~ h-' lent flow is about 1 ft (0.3 m).
(0.44 k w 2 I ~ ~ K - ~Thermal
S). inertia only needs to be The following example illustrates the selection of
scaled very roughly, and materials ranging from con- the scale for a model. Consider that it is desired to real-
crete to plasterboard would be acceptable (Tables AI0 istically determine flows in openings from the atrium to
and AI I). the communicating spaces. These openings are 8 ft (2.4
m) high and 12 ft (3.7 m) wide. Consider that this height
Example 15.3 Scale Thermal Properties
is the smallest location where fully developed flow is
The walls and ceiling of a full-scale facility are made of con- needed. Then this opening in the scale model should not

I
Crete. How do the thermal properties scale to a one-eighth-
scale model?
be less than about I ft (0.3 m) in the model. Thus, the
model should be one-eighth scale..The scale for each

IFrom Table A10 (Table All fors1 units), kpC'of concrete is


25 Btu' ft-""~' h-' (2.9 kw2 m4 K-~S).
Using Equation I modeling applicat~onshould be determined by consider-
ation of what flows are important to simulate.
While some heat transfer effects can be paitially
preserved by scaling the thermal inertia using Equation

I
1 0.9
3.8 Btu' K4 (1 5.49), only very rough scaling is needed as discussed
above. Glass is often used for some of the walls to make
visualizat~onof smoke flow possible.
l l ~ h ethennal properties only need be scaled very roughly, and 11 As previously stated, Froude modeling is appropri-
a wide range of materials would be acceptable. J ate for smoke temperature away from the flame. Froude
modeling is appropriate for simulation of smoke trans-
PRACTICAL CONSIDERATIONS FOR port of an atrium fire where the flames do not reach the
FROUDE MODELING ceiling. The flames would not be expected to be mod-
Sometimes it is stated that the scale model needs to eled realistically, but the smoke flows away from the
be built such that evely dimension is an exact fraction of flames would be expected to be realistically modeled
the full-scale facility, but not every small detail of the Froude modeling would also be appropriate for
full-scale facility needs to be replicated. Little objects simulation of smoke flows in a building from a fully
such as small light fixtures, light switches, doorknobs, developed room fire. Because of high temperatures, the
rnoldings, smoke detectors, and sprinklers would not be modeling is not appropriate for the fire room or any
expected to impact the gross flow of smoke, and these flames that might be flowing from that room, but realis-
objects can be neglected. In the absence of well-devel- tic modeling of smoke flow away from the fire room
. ,
oped criteria about the size of such little objects, it is would be espected.
CHAPTER 16

Computational Fluid Dynamics

omputational fluid dynamics (CFD) consists of Equations are used in this chapter for the purpose of

C dividing a space into a large number of control


volun~esand using a computer to calculate
approximate solutions to the governing equations for
explaining concepts. Accordingly, units are not given
for variables in this section. However, all of these equa-
tions are valid for S1 units or any other homogeneous
each control volume. These control volumes are often unit system (see Appendix A). For details and equations
called cells. CFD is sometimes called field modeling, of particular CFD models, readers should see the docu-
and a thorough knowledge of this topic requires an mentation for the model.
understanding of graduate level fluid dynamics. The
intent of this chapter is to provide some understanding EXAMPLE APPLICATIONS
of the capabilities and limitations of CFD with respect
The atrium in the Lloyds Building is 1 12 ft by 38.1
to fire applications and smoke flow in atria for readers
ft (34.2 m by 1 1.6 m) and about 240 ft (72 m) high (Fig-
who do not have such an understanding of fluid dynam-
ure 16.1). For this shaft-like atrium, the plume would
ics.
contact the sides of the ztrium, making the conventional
Many computer CFD programs have been devel-
analytical methods for atria (Chapter 14) inappropriate.
oped that are capable of simulation of fire-induced
Waters (1989) used CFD modeling to simulate smoke
flows. Friedman (1992) discusses ten such codes. Sev-
movement in the atrium and design a smoke manage-
eral o f these are general purpose codes that are commer-
ment system.
cially available. For more information about CFD,
readers are referred to Anderson, Tannehill, and Pletcher The Luxor Hotel and Casino is a 30-story pyrami-
(1984); Abbott and Basco (1 989); Hoffmann (1989); dal structure, 200 ft (61 m) high with a 500 ft by 500 ft
Hirsch (1988; 1990); Kumar (1983); and Markatos (150 m by 150 m) base, as shown in Figure 16.2.
(1986). Because of the shape of the structure, the conventional
This chapter addresses the basic concepts of CFD analytical methods are not applicable. CFD modeling
modeling, including the co;:ventional approximations was used to design a unique smoke management system
for the effects of turbulence. A new approach is also dis- for this structure (Evans 2001). The system consisted of
cussed that has the potential to accurately simulate tur- supply fans at the base of the structure and an exhaust at
bulence. This chapter is intended to provide background the top. The supply fans produced an upward spiral flow
information fcr people interested in the possibility of that kept the smoke away from the balconies. Other
using CFD modeling for smoke management applica- atrium smoke management applications of CFD model-
tions. Before using a CFD model, people should, at a ing are presented by Sinclair (2001) and Mills (2001).
minimum, learn about the theory, capabilities, and limi- A fire application of CFD is the NlST analysis of
tations of their specific model. This chapter is not a flame blow-down 'at a Navy tire tighter training facility
con~prehensivetreatment of the subject, and persons (Forney and Davis 1992). The facility was used to recre-
desiring to write a CFD model need to go to other ate the effects ofjet pool fires on the deck of a ship. A
Soul-ces. section of "deck" surface was built o f steel grating
Chapter 16 -Computational Fluid Dynamics

below which there were computer-controlled p p a n e CFD modeling was used by Klote (1999b) to study
burners that simulated the fires. When there was little or the interaction between HVAC airflow and smoke detec-
no wind, the flames would be 2.5 to 3 m (8 to 10 ft) tor activation. A FORTRAN subroutine was written to
high. However, under moderate winds, the smoke and modify a commercially available CFD model to calcu-
flames would be blown down into the space below the
late detector activation time. Figure 16.4 shows the cal-
grating. A commercial CFD model was used to evaluate
culated activation time 2 in. (50 mm) below the cei!i~g
possible alternative solutions to this problem and arrive
at a solution. One alternative was a wall intended to of an open plan office. As expected, the activation time
shield the facility, and the performance of this is shown is delayed in front of the slot diffusers. The surprise was
in Figure 16.3. The wall did not prevent flame blow- that activation time also was delayed near the ceiling
down. The solution consisted of a combination of a return.
fence in place of the wall, plus pressurization of space
Comparisons c f room fire data with CFD simula-
under the grating. When installed, this solution qualita-
tively perfomied as predicted. tions have been conducted by Davis, Forney, and Klotc
(1991) and Morita and Hirota (1989). A CFD analysis
was made as part of the fire reconstruction for the fire ar
the King's Cross train station in London, U.K. (Simcox.
Wilkes, and Jones 1989). CFD modeling has been used
to study smoke detector activation times under beamed
ceilings (Forney, Davis, and Klote 1992; Forne!..
Bukowski, and Davis 1993). and Figure 16.5 is a com-
parison of simulaled and measured temperatures under a
beamed ceiling.

Note: The smoke mana


rnent system consists
of supply fans (see
dashed lines) an
a n exhaust a t

Figurc 16.2 CFD aua!i.sis nns used to d e s i p t17e


(a) Typical Elevation
smoke ~ ~ 1 a 1 7 n g ~sj,stetir
~ ~ 1 ent
~ rthe
t Lrr-vor
Hotel atid Cosit70 (ndoptedfi-otn Evnm
,Toilet Capsule / Elevat0r [200 l]).

Open Boundary ,

Elevator Exterior Elevators

' Flame Blow-down

Figure 16.3 CFD .vitii~tltrtecl


totr/)o.trtro.ccotlto~o~s of'
/]nule h/oic~-clo~i~t~
(11 .Yol:i. ,/its jightit~,o
tt~ririit~g
/it~~i1111~.
Principles of Smoke Management

vector addition shown graphically in Figure 16.6 can be


written as
a + b = i ( a x + b , ) + j ( a , , + b y ) + k ( a z + b z ). (16.1)

The dot product of vectors a and b is

It can be observed that the addition of vectors


results in a vector, and the dot product of two vectors is
a scalar. Further information about vector analysis is
provided in many texts, such as Hay (1953) and
Borisenko and Tarapov (1968).
Figure 16.4 Lines of smoke detector activation time In Cartesian coordinates, the velocity vector, U, is
(seconds) 2 in. (50 mm) below the ceil- expressed as
ing of an open plan ofice (adapedfiom
Klote 19996).
where U,v, and w are velocity components in the X,y, and I
directions, respectively, and i, j, and k are unit vectors in
the X, y, and z directions, respectively.
Experimental Data: 0 R 0 1 The conservation equations are written below.
Simulated Data: 0 U 0

300 62 Mass:

Erne (S)

Figure 16.5 Co~nparisonof CFD-sitnufated temper-


atures and experi~nentaldata for flow
under a bea117edceiling.

FUNDAMENTAL EQUATIONS

This section lists a very general form of the funda-


mental equztions of fluid dynamics with the intent of
giving readers an appreciation of the level of complexity
of this topic. The equations of this section are for
unsteady, compressible, viscous flow with variable vis-
cosity.
Quantities that have only magnitude are scalars.
Examples of scalars are mass, density, area, and temper-
ature. A quantity that has niagniiude and direction is a
vector, provided that it obeys the law of addition of vec-
tors illustrated in Figure 16.6. In Equations (16.5), (16.6), and (16.7), the terms
that include the dynamic viscosity, p, are called the vis-
In Cartesian coordinates, the vector a can be repre- cous terms. Because the dynamic viscosity is on the
sented as a = io, + ja,, + k n l , \\here a, a,, and a, are order of 1 0 . ~Ib ft-l S-' ( 2 ~ 1 0 kg
- ~m-Is-'), the viscous
the magnitudes of vector a in the I,y, and.z directions, terms are relatively small. While these viscous terms
respectively, and i, j, and k are unit vectors in the S, !), can be neglected for many applications, they provide the
and z directions, respccti\.ely. In a similar manner, the mechanism for converting kinetic energy to thermal
vector b can be represented as b = i b , + jb,. + kb, . The energy.
Chapter 16-Computational Fluid Dynamics

Energy:

Equation of State
In addition to the conservation equations, an equa-
tion relating pressure, temperature, and density is
The dissipation function, 0,represents the timerate needed. Such equations are called equations of state.
at which energy is dissipated per unit volume through The perfect gas law is frequently used in CFD applica-
the action of viscosity. The dissipation function tends to tions:
cause flows to go to rest, and this function is expressed
as

where R is the gas constant


Strictly speaking, the conservation of momentum
equations are the Navier-Stokes (N-S) equations, but the
term "N-S equations" is often used in a broader sense to
mean all of the conservation equations plus the equation
Variables in the conservation equations above are
of state, and, in this chapter, the broader meaning is
p = density, used. It is not possible here to discuss all of the assump-
p = pressure, tions involved in the derivation of N-S equations. How-
p = dynamic viscosity, ever, two of the more important ones are the continuum
k = thermal conductivity, assumption and the stress-strain relationship for a New-
tonian fluid. For an exhaustive derivation of the N-S
T = absolute temperature,
equations, readers are referred to Aris (1962).
Cp = constant pressure specific heat, At the level of generality presented above, it is
g"' = heat release rate per unit volume, beyond the state of technology to solve the N-S equa-
tions exactly. Even with the simplifying assun~ptionsof
I =time,
incompressibility or of Boussinesq tlow,'%t is still not
= body force in the j direction, and possible to solve the three-dimensional N-S equations
A>
V = vector differential operator called del. exactly. Exact solutions have been obtained for a lami-
Del is defined as nar flow in simple geometries, and the most notable
application of inexact solutions is to boundary layer
a
V = i-+j,+k-
a a flows. Exact and inexact solutions are discussed in sev-
2.r 0,lJ a2 eral texts (White 1974; Sherman 1990; Schlichting
1960; Schetz 1993). By experimental verification of
If a scalar function, 9, is operated on by del, it is
these solutions, the N-S equations have themselves been
written as
12. Boussinesq flow is an approximation to com-
pressible flow that extends the incompressible flow
equations by considering density as a function of tern-
The material derivative acting on y, is defined as perature (Sherman 1990, p. 83).

Note: The vectors can be added by


moving b so that its "tail" is at the
arrow of a, then Ihe vector from the
Tail" of a to the arrow of b is the

(a) Two vectors a and b (b)Add~tionof vectors a and b

Figure 16.6 Gtq~hicalillccs1rc7fiotiof addiliotl of vecfors.


Principles of Smoke Managemen't

verified for laminar flow. There is no such verification vortices on the flow tilrough open stairwell doors is dis-
for turbulent flow. However, CFD simulations of turbu- cussed. The turbulent nature of fire plumes is apparent
lent flow based on the N-S equations as discussed below to anyone who has seen one.
often correspond well with experimental data. In CFD modeling, the turbulent effects that are
smaller than the cell size cannot be simulated by solving
Boundary Conditions the N-S equations. Turbulence modeling has been devel-
In CFD modeling, the conditions at the boundaries oped to account for these small-scale effects, and turbu-
of the flow field need to be stipulated. Figure 16.4 is an lence modeling is based on Reynolds averaging as
example of such boundaries. In this figure, the boundary discussed below.
conditions consist of (I) solid wall (and ceiling), (2)
plane of symmetry, (3) velocity boundary, and (4) open Reynolds Averaging
boundary. The most common condition for the solid Conventional CFD modeling is based on the
walls is zero velocity at the wall surface. At the solid assumption that the fluctuations associated with turbu-
surfaces (walls, floors, and ceilings), the tangential com- lence are random. There is evidence that this may not be
ponent of velocity is generally considered to be zero. true, but the CFD technology that has been developed
This boundary condition is referred to as the no-slip using this assumption has considerable utility. The time-
condition. averaged quantity is defined as
The symmetry boundary can be compared to a mir-
ror in that it is as if the flow were reflected by this
boundary. As with a solid surface, there is no flow
through a symmetry boundary, but there can be flow at a The "randomly" changing variables are considered
symmetry boundary provided that the direction of such to be made up of a time average plus a fluctuation.
flow is in the plane of the boundary. These are written as
Both velocity and open boundaries can be used
where mass is to enter or leave the domain. The domain
is the region of space for which the simulation is made.
Velocity boundaries are used to define the velocity
entering or leaving the domain. For Figure 16.4, the slot
diffusers are velocity boundaries with velocities stipu- Figure 16.7 illustrates average and fluctuating
lated at an angle such that the flow \vould become velocity in the X-direction. The time average of a fluctu-
attached to the ceiling. ating quantity is zero.
Open boundaries are also called pressure bound-
aries because the pressures outside the domain are stipu-
lated. The CFD model calculates the flows at these
boundaries from the pressures. To improve accuracy, the I t follows that
domain is made larger than the volume of interest so that
the pressure boundaries are away from the volume of
interest. T!:is was done in the simulation shown in Fig-
ure 16.4. The area shown in this figure is a slice through
the volume of interest, but the domain is larger so that
the flow is simulated for some distance beyond the open and
boundaries and the ceiling return.

TURBULENCE While p = 0, the product of two tluctuating quan-


Because air cannot be seen, people are not aware of (3
tities is not equal to zero 0 ).
the turbulent nature of the flow that surrounds them. A CFD with turbulence modeling was initially devel-
cup of coffee can be used as a simple way to illustrate oped for incompressible flow, and so Reynolds a\.erag- -
turbulence. Short!y after coffee has been poured, the ing is illustrated here for that flow. Equation (16.4)
surface of the coffee in the cup appears still and the cof- becomes
fee seems to be completely at rest. However, when
cream is poured into the cup, the eddies and vortices in
the flow becomes obvious.
Most of the flow of smoke management applica- This is the conservation of mass equation, and it is
tions is turbulent. In Chapter 6, the etfect of stationary also called the continuity equation.
Chapter 16 -Computational Fluid Dynamics

Time Time
(a) Steady Flow (b) Unsteady Flow
Figure 16.7 Velocity components in the X-direction

Applying the averaging to Equation (16.20) yields a Turbulence Modeling


time-averaged conservation equation: With the N-S equations, there is one unknown vari-
able for each equation, and this one-to-one correspon-
dence of unknowns and equations is an essential
condition for solution of the system of equations.
Because of Equation (16.15), there are two unknown
variables for each equation with the Reynolds averaged
This is called the Reynolds averaged .continuity N-S equations. In order to solve the Reynolds averaged
equation. Observation of Equation (16.20) shows that it N-S equations, additional equations are necessary. This
is similar to Equation (16.31) except that the variables addition of equations to the Reynolds averaged N-S
are replaced by average terms plus some fluctuating equations is called turbulence modeling.
components. The additional equations are empirical, and they
can be algebraic equations, ordinary differential equa-
For incompressible flow with constant viscosity,
tions (ODE), or partial differential equations (PDE).
Equation (l 6.5) can be written as
Classification of turbulence models is generally based
on the number of PDEs in the model. For example, the
Prandtl algebraic model has no PDEs, and it is referred
to as a zero equation model. The k-E model is a two-
= - ag + -
a~ a X [ 3 ( t r
-
a 2- p ,E-- - -
:; E)] equation model, and Reynolds stress models arc three-
( I~ . ~ ~ )
to five-equation models. Each turbulence model has its
own ad\.antages and proponents who expound on those
advantages. A discussion of the various turbulence mod-
els is be\ond the scope of this chapter.
This is the conservation of momentum equation in Because the k-E model is extensively implemented
the X-direction, and it can be time averaged as was done in CFD codes, a few details of it are given. This model
with the continuity equation. was developed by analogy to incompressible boundary
layer flo\v (Harlow and Nakayama 1968; Launder and
Spalding 1974). The term k is the kinetic energy of tur-
bulence. and it is

k = -I( u ' u f + v'v' + I ~ . ' I I . ' )


a [ [C" + - (16.24)
.
a ; ) -- -- ,-
+-
a p- a), ax l ' p ' l l ' - p u V 1 " - p'11.v
)_

l 2

The rerm E is the turbulence dissipation rate,

Equation (16.23) is the Reynolds averaged x-


mo~nenturnequation for incompressible flow. In the same where
manner, the rest of the governing equations can be aver- I, = dissipation length;
aged to form a set of Reynolds averaged N-S equations. C = constant.
Principles of Smoke ~ a a ~ e m e n ' ;

In addition to the above constant, there are several Governing Equations -


others, and the values in the 1974 paper by Launder and FDS solves the governing equations listed earlier
Spalding are almost the same ones used for most appli- plus conservation of species equations. Species equa-
cations today. Nam and Bill (1993) developed modified
tions are used in many CFD programs to simulate the
k-E coefficients that improve predictions slightly. It is
flow of various gases (such as 02, N2, CO, etc.).
well known that CFD models with k-Eturbulence under-
- In FDS, p in Equation 16.13 is replaced by a con-
predict the diameters of fire plumes. For a discussion
about extension of the k-E model to compressible flow stant "background" pressure, p,. The use of this "back-
and a general presentation about the mathematics of ground" pressure in the equation of state is referred to as
CFD to fire applications, the reader is referred to Kumar the low Mach number assumption. The Mach number is
(1983). the ratio of velocity to the speed of sound, and the low
Mach number assumption filters out the sound waves
that travel at speeds much faster than those of typical
fire applications. This filtering out of sound waves has
For decades, the National Institute of Standards and the advantage that the time step in the numerical solu-
Technology (NIST) in Gaithersburg, Maryland, has
tion is bound only by flow speeds and not the speed of
been conducting basic research in CFD modeling of fire
sound.
and smoke transport. This research provides evidence
that turbulence is not made up of random fluctuations, As already stated, FDS does not rely on Reynolds
but that it has a structure of eddies and vortices that can average equations and turbulence modeling. For FDS
be simulated by computational methods. The CFD simulations where the grid resolution is not fine enough
model, Fire Dynamics Simulator (FDS), is a product of to capture the mixing process at all relevant scales, a
the NIST CFD research effort. FDS is in the public large eddy simulation (LES) can be used. In the LES,
domain, and it can be obtained from NIST at no cost. the viscosity is modeled as
Rehm and Baum (1978) of NIST developed a
unique form of the N-S equations that incorporates com-
pressibility effects sufficient for thermally driven flows
where
of smoke and hot gases generated in a fire. Also at
NIST, the FDS model was developed (McGrattan et al. pLEs= viscosity used in LES,
2000; McGrattan and Fomey 2000), which solves the N- CS = empirical constant,
S equations developed by Rehm and Baum. FDS uses a
A = length on the order of a grid cell, and
numeric solwr based on fast Fourier transforms (FFT)
that reduces calculation timc to a fraction of that with IS1 = magnitude of the stress tensor.
conventional models. Because FDS does not rely on The square of magnitude of the stress tcnsor is
Reynolds average equations and turbulence modsling, it
has the potential to simulate turbulence more realisti-
cally.
The key to the level of flow detail that can be pro-
duced by FDS is the large number of cells that can be
used. in all CFD models, the flow field is divided into a
number of cells. The higher the number of cells, the In Equation (16.26), ~ L E Sis the maximum of p or
greater the flow detail and the greater the running time. p ( ~ ~ ~ ) ' lwhere
5 l p is the dynamic viscosity of the gas
A 100,000 cell applicdtion with a conventional CFD or the \veighted average of the dynamic viscosities of
model \ \ i l l take about the same time as a 1,000.000 cell the constituent gases when individual species are simu-
applicatior, with FDS all other things being equal. It is lated. Use of the viscosity of Equation (16.26) accounts
no wonder that FDS is capable of very realistic flow for viscous effects on a scale smaller than the cell size.
simulations.
A limitation of FDS is that tlie cells n!ust be rectan- u s e and Verification of FDS
gular with aspect ratios that arc not very large. This lin- While it is impossible to completely verify any
itation is due to the FFT solver. This means that the grid CFD model, FDS was used extensively for fire applica-
cells cannot conform to curved shapes or \valls that are tions for several years before i t was released to the pub-
not at righr angles ro the orhcr \valls. Ho\ve\.er. rcla- lic. A sample of these applications of FDS arc
tivcly small cells can be uscd ro form s~t~-faccs \\.irh steps McGrattan, Baum, and Rehm (1996, 1999); Davis,
10 approsi~nalethesc shnpcs. Notarianni, and McGrattan (1996); and Baum. McGral-
Chapter 16-Computational Fluid Dynamics

tan, and Rehm (1996, 1997). These simuiations included in all of the large commercial packages. Pre-processing
room fires, warehouse fires, townhouse fires, airplane software helps the user generate the grid, specifj. the
hangar fires, sprinklered fires, unsprinklered fires, and boundary conditions, and define other input parameters.
fires with draft curtains. For geometries that are somewhat complicated, grid
An FDS simulation of a fire ~ l u m has
e realistic pul- generation capabilities can save significant amounts of
sating eddies, as shown in Figure 16.8, and average user time.
velocities and temperatures of this simulation agree well FDS is an exception in that there is no pre-process- r
with experimental data, as can be seen in Figure 16.9. ing software, and the data are read directly by the pro-
I

cessing software before simulation. Because FDS


SOFTWARE allows only rectangular cells, the lack of pre-processing
Typically, software for CFD applications falls into software is not a disadvantage.
one of the following groups: (I) pre-processing, (2) pro- some CFD codes allow the user to write FOR-
cessing, and (3) post-processing. Not all CFD packages TRAN routines that become part of the processing soft-
have all of these sofnvare groups, but they are available ware. Such routines can define an unusual boundary
condition or the performance of a detector. For some
commercial software packages, user written FORTRAN
routines are essential in order to simulate smoke trans-
port in buildings. While FDS allows the user to modify
the FORTRAN code, this is not normally needed, as
FDS was specifically written for fire applications.
For a three-dimensional simulation, it is not
unusual to divide the space of interest into about 50,000
cells, and some applications have many times more. For
each cell, pressure, density, temperature, three velocity
components, and a number turbulence modeling vari-
ables are calculated several times for each second of
simulated time. To reduce the size of files, data are not
stored for every time step calculated-in some cases, it
is stored every 10 seconds of simulated time. If a fire
simulation has 50,000 cells and saves 10 variables per
cell for each 10 seconds simulated, 20 minutes of simu-
lated fire results in the generation of a file of over 60
Figure 16.8 tye dFDS 01 017 i17.510171it7
Plume s i t i ~ i ~ I ( ~L million numbers. It is not feasible to examine so many
ti117e (Bmrm. A4cG1nttc111, 0 1 7 ~ 4 I ? ~ / I I I I numbers in tabular form to understand the results of a
1997). CFD simulation, and graphical techniques are needed.

2.5 o Correlation of McCaffrey 0.6


- Calculated By FDS Model ko
2 0.5
-
e 2.0
C
'
:0.4
3
1
1.5 .2
-
-
S
'0 1.0
2
0.3

-
0 3 0.2
9 Q)
a
0.5 E 0.1
F
0 0
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
Cross Plume Distance, r/D' Cross Plume Distance, r a g
Figure 16.9 Ratlial p ~ ~ f i l e qc -/ 1.e10c.i~.
otrcl teri7l)emtlo.e f o r a pool f i r e (ndaptedj.oci7
Brizoil. McCr~attctr~, arid Rchrtr [lYY7]).
Principles of Smoke Management

Post-proccssing software is used for graphic display the public domain so that it can be obtained at no cost.
of data from the files. This display can vary fiom simple Because it has been specifically developed for fire
two-dimensional black-and-white contour plots to three- applications, it does not require that the user write cdm-
dimensional color movies where the view can move puter code in order to make routine smoke transport
around the flow field. Smokeview (Forney and McGrat- simulations.
tan 2000) is the post-processing package that was spe-
The general purpose commercial CFD models have
cifically developed for FDS.
typically taken tens or hundreds of person years to
CONSIDERATIONS ABOUT CFD develop. These models are rich in features that allow
them to be used for a wide range of applications, such as
For a CFD simulation, the choice of CFD software,
aircraft design and air movement in rooms.
form of N-S equations, grid, turbulence model, fire
model, boundary conditions, and other factors need to To simulate the buoyant flow associated with fire
be chosen so that they are appropriate for the applica- applications, it is generally accepted that the N-S equa-
tion. For a specific application, there are many CFD tions need to be capable of simulating compressible
approaches that can provide useful information. Suc- flow. This can either be by using the fully compressible
cessfbl CFD modeling requires experience and an version of the N-S equations, or a "partial" compressible
understanding of the technology. However, a few com- form as is done in FDS. Because the Boussinesq approx-
ments can be made. imation does not accurately simulate compressibility
A CFD model developed specifically for fire appli- effects at high temperatures, it generally is not consid-
cations or a general purpose commercial CFD model ered appropriate for fire applications. However, for low-
can be used for smoke management applications. The temperature smoke at a distance away from a fire, the
FDS model is a product of the US. government and is in Boussinesq approximation can yield useful results.
CHAPTER 17

Commissioning and Routine Testing

ommissioning and routine testing are needed to Acceptable performance for a new system does not

C ensure that smoke management svstems will


function as intended during fire. situations.
Many o f the problenls encountered during acceptance
ensure that. years later, the system will perform accept-
ably during a fire. Components deteriorate with age and
can be inadvertently damaged during building modifica-
testing stem from nlisconceptions about the system's tions. For these reasons, annual testing of smoke man-
ability to control smoke and misunderstandings about agement systems is recommended to provide a level of
the intended function of a particular system. This chap- assurance that the system will function as desired in the
ter deals with determination of \\.hat type of measure- event of a fire. The methods orroutine testing should be
ments should be made and how to make them. Further, the same as those o f acccptance testing. Deficiencies
most smoke management s y s t e m should require adjust- encountered during routine testing should be corrected
ments of supply airflow rates or pressure relief vent as soon as possible. These corrections may include bal-
openings to accommodate the particular leakage charac- ancing to correct for changes in building leakage and
teristics of the buildings in which they are located. patching of gaps, holes, and cracks in barriers of smoke
These adjustments can be made in conjunction with the management systems.
acceptance test. Commissioning procedures for new
Inspection, testing, and balancing o r smoke man-
systems shouId include
agement systems can be conducted by the building
inspection of the system components, owner, the construction contractor, a testing and balanc-
: testing of the system operation. and ing contractor, a code official, or some other person.
balancing of the system to ensure perfornlance. Regardless of who performs the work, all measurements
made should be recorded for inspection. Typically, code
Testing and balancing of the system can be con- officials check for compliance with local codes, whereas
ducted together. Frequently, local authorities want to be building oivners and engineering and architectural firms
present at a formal cxeptance test of a smoke manage- also conduct inspections, checking for compliance with
ment system. Such a formal acceptance test should be the contract documents. Conlmissioning and routine
preceded by inspection, testing, and balancing. Before testing are simplified when conlpliance is checked or
acceptance testing, the owner, designer, builder, and measured against some standard. Contract documents
code officials should agree upon what constitutes can be prepared to reflect agreement between the owner,
acceptable perrormance. Acceptable performance designer, builder, and code official as to what constitutes
should be based on measurements of appropriate design acceptable performance. In the following discussion in
parameters. such as pressure differences, air velocities, this chapter and tlle referenced appendices, the phrase
and flow rates. If appropriate, the capabilities of the sys- "as specified" is used to mean as specified in accor-
tem to prevent smoke feedback into protected spaces dance with a standard or standards that have been
should be tested. agreed upon by the parties involved.
Chapter 17-Commissioning and Routine Testing

General information about testing and balancing of activation should be activated by putting an appropriate
HVAC systems is provided by SMACNA (1993) and initiating device into alarm.
ASHRAE (1999). Additional information about com-
missioning smoke management systems is available Pressurized Stairwells
from ASHRAE Guideline 5 (ASHRAE 1994). With all stairwell doors closed, pressure differences
across each stairwell door should be measured. Then
INSPECTION one door should be opened, and pressure difference
Inspection consists of checking smoke management measurements made at each closed stairwell door. This I, ,
system components, which include barriers, air-moving should be repeated until the number of doors opened
equipment, controls, and electric power supply. For equals the number of doors required by the code author-
pressurized stairwells, the barriers consist of the stair- ity to be opened.
well walls, ceiling, and doors. For zoned smoke control,
the barriers are the walls, floor, and ceiling separating Elevator Smoke Control
the zones. For elevator smoke control, the barriers The smoke control test depends on the type of ele-
would be of the elevator shaft and its lobbies. Walls, vator smoke control system. In general, the design pres-
partitions, floors, and ceilings should be checked for sure differences should be measured at the appropriate
obvious and unusual openings that could adversely locations for the particular design. If the intent oi' the
affect smoke control performance. Gaps around doors system is to pressurize enclosed.elevator lobbies, pres-
should be as specified. Automatic door closers that are sure differences across closed lobby doorsto the build-
part of the snloke control system should be of the type ing should be measured. If the intent of the system is to
specified. pressurize the elevator shaft to prevent smoke flow
The air-moving equipment to be checked includes through it, the pressure differences across the elevator
ducts, access openings in ducts, fans, fire dampers, and doors should be measured.
snloke dampers. The materials and construction of ducts
should be checked. Dampers should be the type speci- Atrium Smoke Management
fied and installed where and in the manner specified. Generally, designs for smoke management in large
Components of the control system should be checked to spaces will be based on providing specific exhaust
determine that they art: as specified. Any special electri- capacity from the upper region of the space and not
cal power requirements, such as standby power or dual exceeding an airflow at openings into the space. These
feeds, should be checked. General inspection procedures flows and velocities should be measured. Upper layer
are presented in appendix G, and these are only intended temperature of the space should be measured to ensure
as a guide for the development of specific procedures that considerations about smoke stratification in the
for individual smoke management systems. atrium are appropriate.

TESTING A N D BALANCING Caution about Smoke Bombs


S M O K E MANAGEMENT SYSTEMS A caution needs to be given concerning the use of
If standby power or other emergency poLver has smoke bombs. For zoned snloke control systems, a
been provided for the operation of the smoke manage- major problem with most snloke bomb tests of smoke
ment system, acceptance testing shall be conducted with control systems is that they are intended to test some
improvement of smoke conditions in the zone where the
emergency power and normal power. During one test
fire is located. This is based on the mistaken belief that
started under normal power conditions, the normal
smoke control is capable of producing a significant
power shall be shut off to determine the ability of the
improvement in tenable conditions within the zone
system and all associated systems to properly operate
where the fire is located.
under standby power or other emergency power.
Smoke bomb tests for zoned smoke control are
described here in general terms so that the reader car.
Zoned Smoke Control recognize this type of test and understand the problems
For zoned smoke control systems, one zolie should with them. The smoke control system is put in opera-
be put into the smoke control mode, and the pressure tion. In the zone that is being exhausted, a number of
differences at the boundaries of that zone should be smoke bombs are ignited. The smoke bombs produce all
measured. After snloke control operation in that zone of their smoke in a few minutes, and the zone rapidly
has becn deactivated, another zone should be tcsted in fills with smoke. Because the smoke control system is
the same manner. This should be repeated until all exhausting air and chemical smoke from this zone, the
smoke zones have been tested. Systems with automatic concentration of chemical smoke decreases with time. If
Principles of Smoke ~ a n a ~ e m e r h

at some specific time after ignition, a specific object


(such as an exit sign) is visible by a human observer at a
specific distance (such as 20 ft), the smoke control sys-
tem is declared a success.
The problems with this type of smoke bomb test are
numerous, and the unrealistic nature of these tests was
illustrated by the smoke control experiments at the Plaza
Hotel. The criterion for successfd operation is not
objective. Furthermore, the potential danger of exposing
the observer or other people to toxic chemical smoke
must be dealt with. The obscuration of smoke from a
building fire is very different from that of chemical
smoke. Most flaming fires produce a hot, dense, black
smoke, whde most smoke bombs produce a cool, white Figure 17.1 Setup for measzu.ing pressure difference
smoke. At present, no information is available relating across a door.
the smoke obscuration of chemical smoke to that of
smoke from building fires. These problems can be over-
come by modifications to the test method. However, this chemical smoke is described here. A number of smoke
would not yield a test relevant for a smoke control sys- bombs are placed in a metal container, and all bombs are
tem. Because a smoke control system is intended to simultaneously ignited. The container is located near an
maintain pressure differences at the boundaries of the exhaust inlet in the smoke zone being tested so that all
smoke zone, the system should be tested by measuring of the chemical smoke produced by the bombs is drawn
pressure differences. A very serious probleni with this directly into the exhaust airstream. If chemical smoke is
type of smoke bonib test is that it can give building detected in the supply air, its path should be determined,
occupants and fire service officials a false sense of secu- the path should be blocked, and then tlie smoke feed-
rity. The test can lead people to wrongly think that back test should be conducted again.
zoned smoke control is capable of achie\.ing a signifi- Smoke bombs or other tracers can be useful in
cant improvement in tenable conditions \vitliin tlie fire locating the leakage paths tliat sometimes defeat a
space. smoke control system. For example, if the construction
Testing the performance of zoned snioke control of a stainvell is unusually leaky, pressurization of that
systems with chemical smoke from smoke bonibs is not stairwell may not be possible with fans sized for con-
realistic for flaming fires in unsprinklered buildings. struction of average tightness. Chemical smoke gener-
Possibly the flow of unheated chernical snioke is similar ated within the stairwell will flow through tlie leakage
to that of smoke from a sprinklered fire or a smoldering paths and indicate their location so that they can be
fire. However, the gases produced by a large flaming caulked or sealed. General testing procedures are pre-
fire in a building are in the range of 1200°F to 1 800°F sented in appendices H and I. As with Appendix G,
(650°C to 1000°C). For chemical smoke to produce tlie these are intended as a guide for tlie development of
same buoyant pressure differences as these gases, the specific procedures for individual smoke control sys-
chemical smoke would have to be heated to tlie same tems.
temperatures. This is impractical because of the associ-
ated danger to life and property. DIFFERENTIAL PRESSURE
Smoke bonib tests of atriuni smoke management INSTRUMENTATION
are similar to those for zoned smoke control, except that
the smoke bombs are set off in tlie atrium with the The setup for measuring pressure difference across
atriuni system actwated. While atriuni smoke manage- a door is illustrated in Figure 17.1. The convention of
ment systems are intended to iniprove snioke conditions this sctup is that tlie instrument is on thc low-pressure
in tlie atrium, cliemical snioke from smoke bombs is so side of tlie door. Experience has shown that adherence to
different from hot snioke from a flaming tire tliat these a particular convention reduces conlision and, thus, the
tests are not realistic. As witli zoned smoke control, the potential for human error. A hose connected to the high-
concern witli this type of snioke bomb tcst is that it can pressure port of tlie instrument goes through a gap
give building occupants and firs scrvice otlicials a false underneath and is terminated with a tee on tlie liigli-
sense of security. pressure side of the door. The tec is used to minimize
Chemical smokc or a traccr gas (sucli as sulhr any pressure errors duc to air velocity. Alternatively, tlie
liexatlouridc) can bc used to test for snioke fecdback tube can end \vitIiout a tee, provided it is located so that
into supply air. Thc gcncral proccdurc for testing witli the dynamic pressure component is ncgligible. Rubbcr
Chapter 17-Commissioning and Routine Testing

or flexible plastic tube of 0.25 in. (6.4 mm) outside smoke control systems. The gage should have a stand so
diameter works well for most cases. A narrow gap may that it can be set on the floor or other flat surface. The
result in a pinched tube, invalidating any measurement. instrument has a zero adjustment that can correct for
Small diameter metal tubing can sometimes be used in minor deviations in surface level. Thus, an instrument
such cases, particularly through the gaps of some gas- level adjustment is unnecessary. A differential gage
keted doors. should be calibrated.
The differential pressure instrument should have a
sensitivity of at least 0.01 in. H20(2.5 Pa), and gener- Electronic Pressure Transducers
ally a range from 0 to 0.25 in. H20(0 to 62 Pa) is suffi- Most electronic differential pressure transducers are
cient. Occasionally an instrument with a range of 0 to of the diaphragm type. Changes in pressure across a dia-
0.50 in. H20 (0 to 124 Pa) is needed. phragm cause diaphragm displacement, which can be
measured by strain gages, piezoelectric elements, induc-
Inclined Liquid Manometer tance pickups, capacitance pickups, etc. These transduc-
An inclined manometer with a liquid reservoir is ers require electrical power and should be calibrated
illustrated in Figure 17.2. This device indicates pressure periodically. Many instruments are commercially avail-
by the height of a column of liquid. Before any measure- able with the necessary sensitivity and in appropriate
ments, the instrument must be ~djustedso that it is level. ranges. For many applications, a major advantage of
Generally, the scales of inclined manometers are com- these instruments is that they have analog voltage output
pensated for the liquid rise in the reservoir so that the suitable for monitoring by computer data acquisition
pressure difference can be read directly. The zero level systems. For field tests conducted with hand-held instru-
of these instruments can be adjusted by adding or ments, analog output seems to have little advantage. For
removing liquid from the reservoir or by changing the this reason and because of the expense of these instru-
position of the scale. Because the measurement princi- ments, they are not generally the instrument of choice
ple of these devices is so fundamental, it is believed that for smoke control testing.
com~nerciallyavailable inclined manometers are of suf-
ficient accuracy for smoke control testing \vithout inde- FLOW INDICATION AND MEASUREMENT
pendent calibration. During acceptance and routine testing, there are
many situations for which the knowledge of flow direc-
Differential Pressure Gages tion is desirable. Such cases abound during the initial
A gage without liquid has the advantage of conve- checkout of a smoke control system. A piece of paper
nience over the inclined manometer. Bourdon-tube placed in front of an air grille provides an immediate
gages are the most common type of pressure gages, but and simple indication of flow and flow direction. Air-
the friction of the mecha~~ical linkages of these instru- flow \\.ill cause a hanging strip of tissue paper to notice-
ments limits sensitivity. No Bourdon-tube gage is ably detlect diagonally at flow velocities as low as 15
known wit11 sullicient sensitivity for smoke control fpm (0.08 mls). Smoke flow from a punk stick or a ciga-
application. However, a magnetically coupled gage, as rette can also be used to detect such low airflows.
illustrated in Figure 17.3, is sufficiently sensitive, and This section discusses flow measurement appropri-
these gages have been used extensively for field tests of ate for smoke management applications, but more

Note: The absence of liquid makes his type of gage more


convenient than an inclined manometer.
High Low
--

Pressure

,Liquid Gage
Reservoir /Stand

1 Level
Adjustment 'zero Adjustment

Figu re 1 7.3 ~Magr?c~icallv


coupled d$ereritial p/-a-
swe gage.
Principles of Smoke Management

Figure 17.4 Flow hood being used to measure volzr-


mel~*icfloiv
of ceiling-tnounted supply.

detailed information on- this topic is available from


ASHRAE (1997~).For field measurement of. fan.
rL;;
H
rw1
Equal
Rectangular
Areas

Centers of Areas
Where Velocity is
Measured

Open Doorway
or Section of
Corridor

.. flows,
see AMCA (1 990b). Figure 17.5 Flow measurement t t a ~ e r s efor cot-ri-
dots and open doonvaj~s.
Volumetric Flow Rate
Airflow velocity through an open doonvay or the duct losses, the accuracy of flow hoods is believed to
across a section of a corridor is generally far from uni- be in the range of 10% to 15%.
form. Such flow is frequently characterized by the pres- When volun~etricflow is obtained from velocity
ence of large stationary vortices, especially flow measurements, a traverse should be made. Traversing
through open stairwell doorways. This makes accurate open doonvays or sections of corridor can be done in a
manner similar to that for rectangular ducts, as illus-
determination of volun~etricairflow difficult unless
trated in Figure 17.5. Velocity readings should be taken
extreme care is taken. Fortunately, airflow through large
in the center of equal areas over the cross section. For
openings is not the major principle of smoke control for flow in ducts, the cross section should be divided into 16
most building systems. It follows that for the majority of to 64 equals spaces. Because of the likely variations of
smoke control systems for buildings, flow measure- velocity in doonvays and corridors, these openings
ments in doorways and corridors are not necessary. should be divided into 30 to 64 equals spaces.
However, flow measurements of the supply and exhaust Flows through doorways in particular should be
of a smoke control system are often desired, and some- checked for stationary vortices by use of smoke from a
times information about the flows through doorways is punk stick or cigarette. If stationary vortices exist, care
also needed. should be taken that flows against the main flow direc-
tion should be assigned negati\ L values when calculat-
Flow can be measured directly by using a flow ing the average velocity. The volumetric flow rate for a
hood or determined indirectly from a set of velocity rectangular duct or other opening is calculated from the
measurements. Flow h o ~ d sare comn~erciallyavailable formula
instruments, which have a grid of static and dynamic
pressure taps from which the volunletric flow through
the hood is obtained and displayed directly on a meter. where
Figure 17.4 illustrates a flow hood being used to mea-
sure flow from a ceiling supply. The device can also be
p = volumetric flow rate, cfin (m3/s);
used to measure exhaust flows, and it can be oriented for H = height of opening, ft (m);
use with wall-mounted inlets and outlets. Provided that W = \vidth of opening, It (m);
the pressure loss through the hood is small compared to U = average velocity, rpm (rills).
Chapter 17- Commissioning and Routine Testing

Example 17.1 Volumetric Flow From Velocity Traverse


Calculation of the volumetric flow rate through a doorway 3 ft by 7 ft (0.91 m by 2.13 m) is desired, and the presence of a station-
ary vortex was observed with smoke. A traverse of 35 readings is like that shown in Figure 17.5, and the velocities are listed
below.

l l ~ b average
e velocity is 300 fpm (1.5 mls). Using Equation (17. I), the flow is 6300 cfm (3.0 m3/s).
I
Velocity Measurement /Pin Suppo"ng Note:
Pitot tubes, deflecting-vane anemometers, and ther- the Vane The aidlow causes
vane to deflect
mal anemometers are commonly used to measure air- // diagonally, and the
flow in building. These instruments are discussed in the
following sections.
these instruments
are low in mst and
Deflecting Vane Anemometer compact. they are
useful for soot checks
The deflecting vane anemometer consists of a vane and rough kstima~es.
hung from a pin such that air velocity ivill cause a diag- (a) Principle of operation of deflecting vane anemometer.
onal deflection of the vane, as illustrated in Figure 17.6.
Manufacturers rate the accuracy of these instruments at
5% for flows less than 100 fpm (0.5 rnls) and 10% for
greater flows. The ASHRAE Handbook identifies the
limitations of not being well suited for many airflow
readings and of needing periodic calibration. Because of
their low cost and compact size, these instruments are
popular for making spot checks and obtaining rough
estimates of \:elocity. However, it is not believed that
they are appropriate for acceptance or routine testing.

Pitot Tube
The stagnation pressure, is the pressure that
results when moving gas is brought to rest. An expres-
sion for this pressure can be obtained from Bernoulli's
equation,

i
/

(b) Deflecting vane anemometer in use.


Figure 17.6 Deflecting vatle anernometex
where
- stagnation pressure of the gas, in. HzO (mm
Ps,trs -
HzO): For an idealized frictionless fluid, the coefficient
/?~IOI = static pressure of the moving gas. in. H20 (mm C has a value of one, and the value differs for real tlu-
H2 0 ):
id:. Pitot tubes measure the stagnation pressure of a
moving gas, and some pitot tubes incorporate static
c,,, = correction factor (di~nensionless); pressure taps, as illustrated in Figure 17.7. Manufactur-
densit? of gas, lb/ft3(kdm3); ers of pitot-static tubes frequently supply information
P =
about the correction factor as a function of flow velocity
U = gas \.eloci~y,rpm ( d s ) ; or of Reynolds number. The velocity from Equation
K,,, = 1097 (4.427). (1 7.2) can be expressed as
Principles of Smoke Manageme*
. .

!
1 where
i
1
U = velocity, fpm (&S);
!
f = pressure difference from manometer, in. H20 (mm
Tzps Evenly
H2O); Spaced Around
Circunference
j 1 p = density of air, lb/ft3 (kg/m3);
Cp, = correction factor (dimensionless);

A pitot-static tube can be used to measure velocities


U Pstq
Hoses to Differential
Pressure Instrument

in the range of 400 to 2000 fpm (2 to 10 m/s) when con- Figure 17.7 Pitot-static tube.
nected to an inclined manometer. With an electronic dif-
ferential pressure transducer, a pitot tube can be used in
I the range of 200 to 3000 fpm (l to 15 d s ) . wire). For the constant-current type, a filum is subjected
I
to a constant electrical current and the temperature of
11 E x a m ~ l 17.2
e Velocitv from Pitot-StaticTube Reading
11The menometer connectki to a pitot-static tube reads 0.08 the filum depends upon the convective cooling of air
flowing past it. Thus, temperature is a measure of veloc-
in. H 2 0 2 . 0 3 mm H20), the air density is 0.075 lb/ft3 (1.2
ity. The constant-temperature type uses the same princi-
kg/m3), and the pitot tube correction factor is 1 .OS. ple in a different way. The electrical current through a
The velocity calculated from Equation (17.3) is l l l 0 f p n ~ filum is adjusted so that its temperature remains con-
(5.62 m/s). stant. For this instrument, current is a measurement of
velocity. Hand-held, battery-powered, temperature-
Thermal Anemometer compensated thermal anenlonleters are commercially
Thennal anemon~eters(also called hot-wire ane- available fbr air temperatures normally encountered in
nlometers and hot-film anemometers) are available in building heating and cooling systems. Such instruments
two types: constant-current and constant-temperature. have ranges of approximately 10 to 5000 fpm (0.05 to
Both types have a velocity probe with a filum (fine 25 m/s) with accuracies of about 5%.
Nomenclature

area of opening, leakage path, shaft, test depth ofsmoke layer below the smoke vent, ft
sample, atrium, or fire, ft2 (m2) (m)
wind exponent (dimensionless); ga'p equivalent diameter of flow path*
thickness"; dilution rate* diameter of fire, ft (m)
effective flow area, ft2 (m2) hydraulic diameter, in. (m)
distributed effective flo\v area per unit height, minimum smoke layer depth to prevent
ft (m) plugholing, ft (m)
vent area, ft2 (m2)
maximum specific flow, pers/min.fi (pers/s.m)
temperature factor, in. H20/ft(Pa/m);distance
diameter of visible axisymmetric plume, ft
from the opening to the balcony edge, ft (m);
or constant on N-gas model 127,000 for CO2 (m)
< 5% and -38,600 for CO2 > 5% E = effect of exposure (ppmmin)
flow or discharge coefticient (dimensionless); F = total door opening force, Ib (N)
gas or contaminant concentration; or specific f = friction factor of shaft or duct (dimensionless)
heat, Btu/lb°F (J/kg°C) F, = flow rate perdmin, (pers/s)
flow coefficient for elevator car FED = fractional effective dose (dimensionless)
(dimensionless)
F, = force to overcome the door closer and other
flow coefficient for exponential flow
friction, Ib (N),or Froude number
equation, ft3 min-I (in. H20)-" (m3 S-' Pa-");
(dimensionless)
constant pressure specific heat, Btu/lb°F (kJ/
specific flow, pers/min.ft (pers/sm)
kg°C)
pitot tube correction factor, (dimensionless) the flow factor, fpm ( m k j
conductivity factor, ft"' /s1I2(m''' / s " ~ ) acceleration of gravity
flow coefficient of the vent (dimensionless) height ofatrium, shaft, opening, ceiling above
the fire, upwind wall, balcony ceiling above
wind pressure coefficient (dinlensionless)
top of hel, fi (m)
depth ofsmoke layer below botto~nofexhaust
inlet, ft, (m); or distance from the doorknob to distanceabove neutral plane, ft (m); depth of
the knob side of the door. ft (m) smoke layer, ft (m)
density oroccupant tlon.. pers/ft' (pers/ni'), or conv&tive heat transfer coefticient, ~ t u / l i 's
dilli~sioncoellicient "F (w/m2 "C)
height o r fuel (m)
l.
: Units dcpc~itlon [tic spccilic equation height limit, li(m)
height of wind measurement, ft (m) RTI = response time index, ftlR S" (m1' sIn)
velocity factor (dimensionless) RTI, = virtual RTI, ft" S' (m'' S'')
thermal inertia ofa material (product ofk, p, and S = visibility, ft(m)
C), ~ t in.uh-' ~ft'' OF^ (kw2 m4 K-~s) Smi, = minimum separation between inlets, ft, (m)
friction factor of stairwell (dimensionless) T = temperature14;transmittance (dimensionless); I
length of gap*; length of shaft or duct, ft (m); or emc~ztiontime (minutes)
height of section of stairwell, ft (m) I = time*
I ;
lethal concentration, Ib ft-3 min (g m-3 min) lac1 = time of sprinkler actuation, S
niass of the sprinkler, Ib (kg) 1, = time for population to pass through constraint
mass flow rate, Ib/s (kgls) = transport time lag of ceiling jet, S
tQ
niass concentration of fuel burned, lb/@ (g/m3) = absolute centerline axisymrnetric plume
Tcp
mass of fuel burned or consumed, Ib (g) temperature, OR (K)
masitnum mass rate of exhaust without = absolute temperatureof the fire space, OR (K), or
plugholing, Ibis (kgls)
T/
temperature of gas in full scale facility, "F ("C)
inass concentration of particulate lb/ft3 (g/m3) = time in full scale facility S
f/
mass of particdates produced, Ib (g) T = temperature of the gases in the exhaust fan*
moment of the door closer and other friction, Ib 'g = growth time, s
ft (N m) = time to incapacitation du2 to thermal exposure,
mass of smoke, Ib (kg) min
fire location factor (diniensionless);,flow [m = modeled evacuation time for an egress route* or
esponent (din~ensionless) time in scale model, s
N-Gas model indicator (dimensionless) T,, = temperature in scale model, "F ("C)
number of exhaust inlets (dimensionless) T, = temperature of ambient or outside ai;
pressure ditTercnce (dimensionless) Tp = plume temperaturet
flo\~-rate (dimensionless) = transport time lag of plume, S
f
perimeter of duct or shaft, ft (m); or population T,. = absolute reference temperature, "R ("K)
pressure' U = velocity, fprn (mls)
absolute atniosplieric Uk = critical air velocity to prevent smoke backflow,
ambient pressilre* fpm
Prandtl number (diniensionless) U,,,,, = measured wind velocity. fpni (mls)
stagnation pressure of the gas, in. H20 (mm V = volumetric flow rate, cfm (m3/s)
c,,, = volunietric flow in scale model, ft3/s (m3/s)
static pressure of the moving gas, in. H20 (mm VC = volume of smoke in a space or test chamber, f$
H201 (m3
~vindpressure, in. H20 (Pa) = factor for CO2-induced hyperventilation
heat release densily. Btu/s ft2 (kw/mz) ''C = limiting average air velocity, fpni ( 4 s )
heat release of thc tire, Btds (kW) V, = volume of smoke, ft3 (m')
H R R at sprinkler actuation, kW (3tds) W = width ofdoor, corridor, opening, or plume. ft (m)
convective heat release rate, Btuk (kW) I = effective width of stair, in. (m); width of the
heat ge~ieratcdw i t h i n tlic control volume, Btu's opening from the fire rooni, ft (m); or spray
(kW) density, gpndft2 (nids)
radiant heat release of the fire. Bttds (kW) S = depth of gap in flow direction, in. (m), or
gas constant (J/kg K) distance of light travel or the path length. ft (m).
radius or horizontal distance from centerline of = particulate yield (dimensionless)
)'P
plun~e.fi (m) z = elevation. ft (m)

Reynolds number (di~nensionless) z, = mean flarne height, ft (m)

~cparationdistancc fiom the centcr of the fire to


a target. l1 (m) ''Units depend on the specilic equation.
Principles of Smoke Management

= height above balcony, ft (m) = time interval15


= maximum height at which plume is considered ATmin = minimum temperature rise of plume above
buoyant, ft (m) ambient, "F ("C)
= virtual origin correction of the axisyrnmetric E = turbulence dissipation rate; or roughness of the
plume, ft (m) inside surface of the duct, fi (m)
= extinction coefficient ft-I (m''); fire growth = wall heat transfer fraction (dimensionless)
coefficient, kw/s2 ( ~ t u l s ~or) ;plugholing
exponent (dimensionless) /1 = percentage obscuration (dimensionless) or
failure rate
= specific extinction coefficient, f t 2 ~ b(m21g)
= exhaust location factor (dimensionless)
p = absolute viscosity
= wind boundary layer height in the vicinity ofthe v = kinematic viscosity, fi2/s (m2/s)
building, .fi(m), or optical density per unit ll = dimensionless group of variables

- distance, ft-' (m-')


specific optical density (dimensionless)
= change in energy of the smoke layer, kJ
p

pfan
= density, lb/@ (kg/m3)
density of gases in exhaust fan, lb/@ (kg/m3)
=

p. = outside or ambient density, lb/fi3 (kg/m3)


= chemical heat of combustion, Btullb (kJkg)
T = time constant, s (S)
= mass loss of test sample, Ib (g)
= allo\vable fraction reduction in mass flow rate
= mass optical density, @/lb (m21g)
through fan
= boundary layer height in the vicinity of the wind
anemometer, ft (m) X, = convective fraction
Ap = pressure difl'erence, in. H 2 0 (Pa) z,. = radiatiue fraction
A = overall pressure difference from one side of a
building to another due to wind effect l 5 Units depend on the specific equation.
References

Abbott, M.B., and D.R. Basco. 1989. Co~iip~ttational ASHRAE. 1994. Guideline 5, Coni1nissio17ing sliioke
fluid dynamics: AI? introd~rction, for e~igi~leers. New managenlent systems. Atlanta: American Society of
York: Wiley. Heating, Refrigerating and Air-conditioning Engi-
Achakji, G.Y., and G.T. Tamura. 1988. Pressure drop neers.
characteristics of typical stairshafts in high-rise AS HUA E. 1997a. 1997 ASHRAE handbook-F~rnda-
buildings. ASHRAE fi-ansactiom (I): 1223- 1236. mentals, Chapter 15, Airflow around buildings.
Atlanta: American Society of Heating, Refrigerat- Atlanta: American Society of Heating, Refrigerat-
ing and Air-conditioning Engineers, Inc. ing and Air-conditioning En,'o~neers.
Ahonen, A., M. Kokkala, and H. Weckman. 1984. Burn- ASHRAE. 1997b. 1997 ASHRAE handbook-F~r17da-
ing characteristics of potential ignition sources of n~entals,Chapters 25 and 26. Atlanta: American
room fires, Research Report 285. Technical Society of Heating, Refrigerating and Air-Condi-
Research Center of Finland. tioning Engineers, Inc.
Alpert, R.L. 1972. Calculation of response time of ceil- AS HRAE. 1 9 9 7 ~ .1997 ASHRAE handbooX--F~o7da-
ing-mounted fire detectors. Fire Technology 8(3): mentals, Chapter 14, Measurement and instruments.
181-195. Atlanta: American Society of Heating, Refrigerat-
ing and Air-conditioning Engineers, Inc.
AMCA. 1987. Air Sj~tetns,AMCA Pub. 200. Arlington
Heights, Ill.: Air Movement and Control Associa- ASHRAE. 1999. 1999 ASHRAE handbook-Applica-
lions, Chapter 36, Testing, adjusting and balancing.
tion.
Atlanta: American Society of Heating, Refrigerat-
AMCA. 1990a. Fans and Sjate~iis,AMCA Pub. 20 1-90. ing and Air-conditioning Engineers, Inc.
Arlington Heights, 111.: Air Moiwnent and Control
AS HRAE. 2000a. ASHRAE Handbook-*ems and
Association.
Applications. Atlanta: American Society of Heat-
AMCA. 1990b. Field pe1fofor7nance of ~ n e a s ~ o n , r eof~ i t ing, R~frigeratingand Air-conditioning Engineers.
fat7 sjate~ns, AMCA Pub. 703-90. Arlington
AS H U E . 2000b. ANSIIASHRA E Standa~d149-2000,
Heights, Ill.: Air Movement and Control Associa-
Laboratory Methods of Testing Falls Used to
tion. Exhaust Smoke in Smoke Manageme~u Sjs/e/ns.
Anderson, D.A., J.C. Tarmehill, and R.H. Pletchec Atlanta: American Society of Heating, Refrigerat-
1984. Con7pr/atio17al flitid nreci1m7ic.s and /lea/ ing and Air-conditioning Engineers.
t/-ansfer.. New York: Hemisphere. ASH RAE. 2000c. ASHRAE H a n d b o o k - E ~ L ~ ~ ~ I I I E I I ~ ,
Aris, R. 1962. Vecto~*.s,ten so^.^, a d //re l)o.sic e q m / i o ~ n Chapter 18-Fans. Atlanta: American Society of
New York: Dover.
off7rcid 1nec/7atiic~. Heating, ~ e f r k p a t i n gand Air-conditioning Engi-
Arpaci. V.S., and A. Agarwal. 1999. Scaling laivs of tur- neers.
bulent ceiling fires. C o n ~ I ~ r ~ . s mu1
/ i o ~ iF1c111re 1 16: ASME. 1987. A~iiericans / a n d a ~ sqf+
d codeJor- eleva-
84-93. t o n . escaluto~s,drc~nb~t~ni/enls
trlrd /no\.i17g 11nlkS.
References

AI 7. I. New York: American Society of Mechanical Blockley, W.V. 1973. Biology Data Book. Bethesda,
Engineers. . Md.: Federation of American Societies of Experi-
Aynsley, R.M. 1989. The estimation of wind pressures mental Biology.
at ventilation inlets and outlets on buildings. BOCA. 1999. National Building Code. Country Club
ASHRAE Transactions 95(2): 707-72 1. Hills, Ill.: Building Oficials and Code Administra-
Babrauskas, V. 1990. The cone calorimeter Heat tor International, Inc.
Release in Fires, Chapter 3. V. Babrauskas and S.J. Borisenko, A.I., and I.E. Tarapov. 1968. Vector and ten-
Grayson eds. Elsevier, New York. sor analysis with applications, transulated by R. A. , ,
Babrauskas, V., et al. 1991. Toxic measurement for fire Silverman. New York: Dover.
hazard analysis, NIST Special Publications 827. Bryan, J.L. 2002. Behavioral response to fire and
Gaithersburg, Md.: National Institute of Standards smoke. SFPE Handbook of Fire Protection Engi-
and Technology. neering. Quincy, Mass.: National Fire Protection
Babrauskas, V. 2002. Burning rates. SFPE Handbook of Association.
Fire Protection Engineering, Chapter 3-1. Buckingham, E. 1915. Model experiments and the form
Bethesda, Md.: Society of Fire Protection Engi- of empirical equations. Transactions ASME 37:
neers. 263-296.
Barrett, R.E., and D.W. Locklin. 1969. A computer tech- Bukowski, R. W., et al. 1989. ~echnicalReference Guide
nique for predicting smoke movement in tall build- for- HAZARD I Fire Hazard Assessment Method.
ings. Symposiuni on'movement of smoke on escape NIST Handbook 146, Vol. Il. Gaithersburg, Md.:
routes in buildings, pp. 78-87. Watford, Herts, National lnstitute of Standards and Technology.
U.K.: Watford College of Technology.
Bukowslc~,R.W., and R.C. Spetzler. 1992. Analysis of
Baum, H.R., K.B. McGrattan, and R.G. Rehm. 1996. the Happyland Social Club fire with HAZARD I.
Large eddy simulation of smoke movement in three Joltrnal of Fire Protection Engineering 4(4): 1 17-
dimensions. Interflam 1996, 7th International Inter- 17 1
1Jl.
flam Conference, March 26-28, 1996, Cambridge,
Burden, R.L., J.D. Faires, and A.C. Reynolds. 1981.
England.
N1m7erical analpis, 2d ed. Boston, Mass.: Prindle,
Baum, H.R., K.B. McGrattan, and R.G. Rehm. 1997.
Weber & Schmidt.
Three dimensional simulations of fire plume
dynamics. Fire Safety Science Proceedings, 5th Butcher, E.G., P.J. Fardell, and P.J. Jackman. 1969. Pre-
International Symposium, March 3-7, 1997, Mel- diction of the behavior of smoke in a building using
bourne, Australia. a computer. Syn~posiumon movement of smoke in
escape routes in buildings, pp. 70-75. \Vatford,
Bazjanac, V. 1977. Simulation of elevator performance
Herts, England: Watford College of Technology.
in high-rise buildings under conditions of erner-
gency. Human response to tall buildings, ed. by Butcher, E.G., T.H. Cottle, and T.A. Baily. 1971. Smoke
D.J. Conway. Stroudsburg, Pa.: Dowden, Hutchin- tests in the pressurized stairs and lobbies of a 26-
son & Ross, pp. 3 16-328. story ofice building. Building Setvice Engineet- 39:
Bennetts, I.D., et al. 1997. Simulated shopping centre 206-2 10.
fire tests. Fire code Reform Centre, Broken Hill Butcher, E.G., et al. 1976. Smoke control by pressuriza-
Proprietary Company Ltd., Melbourne, Australia. tion. Fire Engineers Journal 36(103): 16-19.
Berl, W.G., and B.M. Halpin. 1980. Hunian fatalities Carhanan, B., H.A. Luther, and J.O. Wilkes. 1969.
from unwanted fires. Johns Hopkins APL Technical Applied n~tmericalmethods. New York: John Wiley
Digest l(2): 129-134. & Sons.
Best, R., and D.P. Demers. 1982. Investigation report on Cetegan, B.M., E.E. Zukoski, and T. Kubota. 1982.
the MGM Grand Hotel fire, Las Vegas, Nevada, Entrainment and flame geometry of fire plumes,
November 2 1, 1980. Quincy, Mass.: National Fire Ph.D. thesis of Cetegan, California Institute of
Protection Association. Technology, Pasadena.
Beyler, C. 1986. Fire plumes and ceiling jets. FireSafey Chow, W.K., and A.C.W. Lo. 1995. Scale modelling
J., l l , JulyISeptember, p. 53-75. studies on atrium smoke movement and the smoke
Block, J.A. 197 1. A theoretical and experimental study filling process. Journnl o f Fire Prolecrion Et@-
of nonpropagating free-burning fires, 13th Sympo- neering 7(2): 56-64.
sium (International) on Combustion, 1970 August Chow, W.K., and W.M. Siu. 1993. Visualization of
23-29, Salt Lake City, Utah, Combustion Institute, smoke movement in scale models of atriums. JOLW-
pp. 97 1-978. nal of Appliecl Science 3(2): 93- 1 1 1.
Principles of Smoke Management' .- .-.-.

CIBSE. 1995. Relationships for smoke control calcula- DeCicco, P.R. 1973. Smoke and fire control in high-rise
tions, Technical Memoranda TM19: 1995. London: office buildings-Part I: Full-scale tests for estab-
Chartered Institute of Building Services Engineers. lishing standards. Symposium on experience and
Cooper, L.Y. 198 1. Estimating safe available egress applications on smoke and fire control at the
time fiom fires. National Bureau of Standards ASHRAE Annual Meeting, Louisville, Ky., pp. 9-
(U.S.), NBSIR 80-2 172. 15. Atlanta: American Society of Heating, Refriger-
Cooper, L.Y. i982. A mathematical model for estimat- ating and Air-conditioning Engineers, Inc.
ing available safe egress time in fires. Fire Mater. Dias, C. 1978. Stairwell pressurization in a high-rise
6(3/4). commercial building. ASHRAE Journal 20(7): 24-
Cooper, L.Y. 1985. ASET-A computer program for 26. Atlanta: American Society of Heating, Refriger-
calculating available safe egress time. Fire Safe& ating and Air-conditioning Engineers, Inc.
Journal 9: 29-45. Dols, W.S., K.R. Denton, and GN. Walton. 2000. CON-
Cooper, L.Y., et al. 1981. An experimental study of TAMW user manual. Gaithersburg, Md.: National
upper hot layer stratification in full scale multiroom Institute of Standards and Technology.
fire scenarios. Paper 8 I-HT-9, American Society of Drysdale, D. 1985. An introduction to five dynamics.
Mechanical En,'olneers. New York: Wiley.
Cooper, L.Y., and G.P. Forney. 1990. The consolidated Dyrbye, C., and S.O. Hansen. 1997. Wind loads on
compartment fire model (CCFM) computer code structures. New York: Wiley.
application CCFM. VENTS, Part I: Physical basis. Evans, D.D. 1993. Sprinkler fire suppression algorithm
Cooper, L.Y., and D.W. Stroup. 1982. Calculating safe for HAZARD. Gaithersburg, Md.: National Insti-
egress time (ASEV-A cornputer program and tute of Standards and Technology, NISTIR 5254.
user's guide. National Bureau of Standards (U.S.), Evans, D.D., and D.W. Stroup. 1986. Methods to calcu-
NBSIR 82-2578. late the response time of heat and smoke detectors
CRC. 1985. Handbook of Chernislty and Pl~ysics,66th installed below large unobstructed ceilings. Fire
ed. Boca Raton, Fla.: CRC Press, Boca Raton. Technology 22(1): 54-65.
Cresci, R.J. 1973. Smoke and fire control in high-rise Evans, D.D., D.W. Stroup, and P. Martin. 1986. Evaluat-
ofice buildings-Part 11: Analysis of stair pressur- ing thernlal fire detection systems (S1 Units),
ization systems. Symposium on experience and NBSSP 71. Gaithersburg, Md.: National Bureau of
applications on smoke and fire control at the Standards.
ASHRAE Annual Meeting, Louisville, Ky., pp. 16- Evans, D.H. 2001. Two Las Vegas hotcl and casinos
23. Atlanta: An~ericanSociety of Heating, Refriger- demonstrate new ways of thinking about smoke
ating and Air-conditioning Engineers, Inc. control. HPAC Engineering 73(1).
Davis, W.D. 1999. The zone model JET: A model for Evers, E., and A. Waterhouse. 1978. A computer model
the prediction of detector activation and gas tem- for analyzing smoke movement in buildings. Bore-
perature in the presence of a smoke layer, NISTIR hamwood, Herts, U.K.: Building Research Est.
6324. Gaithersburg, Md.: National lnstitute of Stan- Fang, J.B. 1980. Static pressures produced by room
dards and Technolo~y. fires. National Bureau Standards, NBSIR 80-1984.
Davis, W.D., and L.Y. Cooper. 1989. Estimating the Feng, J.B., and J.N. Breese. 1980. Fire development in
environment and the sprinkler links in compartment residential basement rooms. National Bureau of
fires with draft curtains and fusible-link-actuated Standards, NBSIR 80-2 120.
ceiling vents, Part 11: User guide for the computer Forney, G.P. 1989. Personal communications at the Cen-
code LAVENT, NISTIR 89-4122. Gaithersburg, ter for Fire Research. Gaithersburg, Md.: National
Md.: National lnstitute of Standards and Technol- Institute of Standards and Technology.
ogy. Forney G.P., W.D. Davis, and J.H. Klote. 1992. Simulat-
Davis, W.D., G.P. Forney, and J.H. Klote. 1991. Field ing the effect of beamed ceilings on smoke flow,
modeling of room fires. Gaithersburg, Md.: Part I. Comparison of Numerical and Experimental
National lnstitute of Standards and Technology, Results. Gaithersburg, Md.: National lnstitute of
NISTlR 4673. Standards and Technology, NISTIR 4994.
Davis, W.D., K.A. Notarianni, and K.B. McGrattan. Forney, G.P., and W.D. Davis. 1992. Analyzing strate-
1996. Comparison of fire modcl predictions with gies for elimination of flame blow-down occurring
experiments conducted in a hangar with a 15 meter in the Navy's 19F4 fire fighting trainer. Gaithers-
ceiling. Gaithersburg, Md.: Nabonal Institute of burg, Md.: National lnstitute of Standards and
Standards and Technology, NlSTlR 5927. Technology, NlSTlR 4825.
References

Forney, GP., and K.B. McGrattan. 2000. user's guide Harlow, EH., and P.I. Nakayama. 1968. Transport of
for smokeiiew version 1.0- A tool for visualizing turbulence energy decay rate. Los Alamos, N.M.:
fire dynamics simulation data. Gaithersburg, Md.: Los Alamos Scientific Laboratory of the University
National Institute of Standards and Technology, of California.
NISTIR 65 13. Hartzell, GE., A.F. Grand, and W.G Switzer. 1990. Tox-
Forney, G.P., R.W. Bukowski, and W.D. Davis. 1993. icity of smoke containing hydrogen chloride, fire
Field modeling: Effects offlai beamed ceilings on and poiymers-Hazards identification and preven-
detector and spr-inkier- response. Quincy, Mass.: tion, Nelson, G.I., ed., ASC Symposium Series 425,
National Fire Protection Research Foundation. American Chemical Society, Washington, D.C., pp.
Frantzich, H. 1996. Study of movement on stairs during 12-20.
evacuation using video analysing techniques. Lund Hay, GE. 1953. Vector and tensor analusis. New York:
Institute of Technology. Dover.
Friedman, R. 1992. An international survcy of computer Heskestad, G. 1972. Similarity relations for the initial
models for fire and smoke. Journal qf Fire Protec- convective flow generated by fire. Paper 72-WN
tion Engineering 4(3): 8 1-92. HT-1 7, American Society of Mechanical Engi-
Fruin, J.J. 1987. Pedestrian planning and design, neers.
revised ed. Mobile, Ala.: Elevator World Educa- Heskestad, G. 1975. Physical modeling of fire. J. Fire &?
tional Services Division. Flatnnzability 6: 253-272.
Gann, R.G 200 l . Toxic hazard of building products and Heskestad, G. 1983. Virtual origins of fire plumes. Fire
furnishings. Proceedings of the Internatio17al Fire Safety Jozrrnal 5(2): 109- 114.
Sajety Conference, March 11-14, 2001, San Fran- Heskestad, G. 1984. Engineering relations for fire
cisco, Fire Retardant Chemicals Assoc., Lancaster, plumes. Fire Safe02 Journal 7(1): 25-32.
Pa., pp. 85-9 1. Heskestad, G. 1986. Fire plume air entrainment accord-
Gross, D. 1962. Experiments on the Burning of Cross ing to two competing assumptions, Twenty-first
Piles of Wood, Journal of Research of NBS, Vol. Symposium (International) on Combustion, Com-
66C, NO. 2, pp. 99- 105. bustion Institute, Pittsburgh, Pa., pp. 111-120.
Gross, D., and W.L. Haberman. 1988. Analysis and pre- Heskestad, G. 1989. Inflow of air required at wall and
diction of air leakage through door assemblies. Fire ceiling apertures to prevent escape of fire smoke.
Safety Science, Proceedings of the 2nd Interna- FMRC J. I. OQ4E4. RU. Nonvood, Mass: Factory
tional Symposium, Tokyo, Japan, pp. 169-1 78. Mutual Research Corporation.
Gwynne, S., and E.R. Galea. 1999. A review of the Heskestad, G. 2002. Fire plumes. SFPE handbook offit-e
methodologies and critical appraisal of computer protection engineering. Boston, Mass.: Society of
models used in the sin~ulationof evacuation from Fire Protection Engineers.
the built- environment. Bethesda, Md.: Society of Heskestad, G., and M.A. Delichatsios 1977. Environ-
Fire Protection En,'omeers. ments of fire detectors-Phase 1: Effect of fire size,
Gwynne. S., et al. 1999. Escape as a social response, ceiling height and materials. Volumes I and 11-Mea-
Research report. Bethesda, Md.: Society of Fire surenients (NBS-GCR-77-86; NBS-GCR-77-95).
Protection Engineers. Gaithersburg, Md.: National Bureau ofstandards.
Haber, F. 1924. Funf Vortrange aus den jaren 1920- Hirsch, C. 1988. Nzrmerical computation ofinternal and
1923, Verlag v011Julius Spanger, Germany. external flows, Vol. I: Fztnda~nentalsof ntrmerical
Hadjisophocleous, G.V., G.D. Lougheed, and S. Cao. disct-etization. New York: Wiley.
19,99. Numerical study of the effectiveness of Hirsch, C. 1990. Numerical computation ofinternal a17d
atrium smoke exhaust systems. .4SHRAE Tramac- external flo~rs. Vol. 2: Con7p1rtationalmethods for
1io17.7105(1): 699-71 5. inviscid a17d visco~rsjlo~vs.New York: Wiley.
Hagglund, B., R. Jansson, and K. Nireus. 1985. Smoke Hoffmann, K.A. 1989. Conzplrtational fluid ajwan~ics
filling experiments in a 6 x 6 ~ 6meter enclosure. for engineers. Austin, Tex.: Engineering Education
Stockl.~olm:National Defense Research Institute of System.
Sweden. Hottel, H.C. 1961. Fire modeling. The use of models in
Handbook of' HVAC Desigtl. 1990. Nils Grimm and fire research, Publication 786, National Academy
Robert Rosaler, eds. New York: McGraw. of Sciences, Washington DC, pp. 32-47.
Harland, W.A., and W.D. Woolley. 1979. Fire fatality Huggett, C. 19SO. Estimation of rate of heat release by
study-University of Glasgow. Building Research means of oxygen consumption measurements. Fzre
Establishment, Borehamwood Information Paper IP and Materials 4(2): 6 1-65.
18/79. ICBO. 1997. Uniform building code. Whittier, Calif.:
Principles of Smoke Management, l
i
ICC. 2000. International building code. Society of Heating, Refrigerating and Air-Condi-
Idelchik, I.E. 1986. Handbook of hydraulic resistance, tioning Engineers, Inc.
2d edition. New York: Hemisphere Publishing. Klote, J.H. 1982. A computer program for analysis ,of
Incropera, F.P., and D.P. DeWitt. 1985. Fundamentals of smoke control systems. Gaithersburg, Md.:
heat and mass transfer, 2nd ed., Wiley, New York. National Bureau of Standards, NBSIR 82-2512.
! Jin, T. 1974. Visibility through fire smoke, in main Klote, J.H. 1983. Elevators as a means of Fire Escape.
reports on production, movement and control in ASHRAE Transactions 89( I B): 362-378. -

buildings, Japanese Association of Fire Science and Klote, J.H. 1984. Smoke control for elevators. A S m E
Engineering, pp. 100-153. Journal 26(4): 23-33.
Jin, T. 1975. Visibility thorough fire smoke. Report of Klote, J.H. 1986. Smoke control at Veterans Adminis-
the Fire Research Institute of Japan 5(42): 12- 18. tration hospitals. National Bureau of Standards
Jin, T. 1985. Irritating effects of fire smoke on visibility. (U.S.), NBSIR 85-3297.
Fire Science and Technology 5(1): 79-90. Klote, J.H. 1987. A computer model of smoke move-
Jones, W.W. 1983. A review of compartment fire mod- ment by air conditioning systems (SMACS),
els. Gaithersburg, Md.: National Bureau of Stan- NBSIR 87-3657. Gaithersburg, Md.: National
dards, NBSIR 83-2684. Bureau of Standards.
Jones, W.W., and X. Bodart. 1986. Buoyancy driven Klote, J.H. 1988. An analysis of the influence of piston
flow as the forcing function of smoke transport effect on elevator smoke control, NBSIR 88-375 1.
models. Gaithersburg, Md.: National Bureau of Gaithersburg, Md.: National Bureau of Standards.
Standards. NBSIR 86-3329. Klote, J.H. 1989. Considerations of stack effect in build-
Jones, W.W., et al. 2000. A technical reference for ing fires. Gaithersburg, Md.: National lnstitute of
CFAST: An engineering tools for estimating fire Standards and Technology, NlSTlR 89-4035.
growth and smoke transport, NIST TN 143 1. Gaith- Klote, J.H. 1990. Fire experiments of zoned smoke con-
ersburg, Md.: National Institute of Standards and trol at the Plaza Hotel in Washington, D.C.
Technology. ASHRAE E-ansactions 96(2): 397-4 16.
Jorgensen, R. 1983. Fan engineering. Buffalo, N.Y.: Klote, J.H. 1995. Design of smoke control systems for
Buffalo Forge Co. elevator fire evacuation including wind effects. 2nd
Joyeux, D. 1997. Natural fires in closed car parks: Car Symposium on Elevators, Fire, and Accessibility,
fire tests, INC - 961294d - DJ/NB. Baltimore, Md. April 19-21. New York: American
Juillerant, E.E. 1964. Jacksonville Hotel disaster. NFPA Society of Mechanical En,'Olneers.
Quarterly 57(4): 309-3 19. Klote, J.H. 1999a. An engineering approach to tenabil-
Kandola, B.S. 1986a. A wind tunnel building model for ity systems for atrium smoke management.
the investigation of smoke movement problems. ASHRAE i5-atisactions 105(1): 716-729.
Fire Safety Journal 1 O(3): 203-209. Klote, J.H. 1999b. CFD sirnulations of the effects of
Kandola, B.S. 1986b. Comparison of wind tunnel pres- HVAC-induced flows on smoke detector response.
sure measurements and smoke movement computer ASHRAE i5-ansactions 105(1): 395-409. Atlanta:
predictions inside a five-storey model building. American Society of Heating, Refrigerating and
Fire Safety Journal lO(3): 229-238. Air-conditioning Engineers, Inc.
Kandola, B.S. 1986c. The effects of simulated pressure Klote, J.H., D.M. and Alvord. 1992. Routine for analy-
and outside wind on the internal pressure distribu- sis of the people movement time for elevator evacu-
tion in a five-storey building. Fire Safep Journal ation. Gaithersburg, Md.: National Institute of
1 O(3): 2 11-227. Standards and Technology, NlSTlR 4730.
Karlsson, B., and J.G. Quintiere. 2000. Encloswe fire Klote, J.H., and X. Bodart. 1985. Validation of network
dynatnics. New York: CRC Press. models for smoke control analysis. ASHRAE Tram-
Keating, J. 1982. The myth of panic. Fire Jo~rmal76(3): actiota 91(2b): 1134- 1 145. Atlanta: American
57-61, 147. Society of Heating, Refrigerating and Air-Condi-
Keating, J.P., and E.F. Loftus. 1977. Vocal alarm sys- tioning Engineers, Inc.
tems for high-rise buildings-A case study. Mass Klote, J.H., and J.W. Fothergill. 1983. Design of smoke
Enietgeticies 2: 25-34. control systems for buildings. Atlanta: American
Kendik, E. 1985. Technology Report 85-4. Bethesda, Society of Heating, Refrigerating and Air-condi-
Md.: Society of Fire Protection En,'wneers. tioning Engineers, Inc.
Klote, J.H. 1980. Stairwell pressurization. ASHRAE Klote, J.H., H.E. Nelson, S. Deal, and B.M. Levm.
Tratlsactiot7s 86(1): 604-623. Atlanta: American 1992. Staging areas for persons with mobility limi-
References

tations. NISTIR 4770. National Institute of Stan- gases and low oxygen atmospheres produced by
dards and Technology. fires. Fundomental and Applied Toxicology 9: 236-
Klote, J.H., and GT. Tamura. -1986a. Elevator piston 250.
effect and the smoke problem. Fire Safefy Journal Levin, B.C., et al. 1988. Toxicological effects of diffkr-
1l(3): 227-233. ent time exposures to fire gases: Carbon monoxide
Klote, J.H., and GT. Tamura. 1986b. Smoke control and or hydrogen cyanide or to carbon monoxide com-
fire evacuation by elevators. ASHRAE Transactions bined with hydrogen cyanide or carbon dioxide.
92(1A): 23 1-245. Society Of Plastics Industry, Polyurethanes 88, 31St .

Klote, J.H., and GT. Tamura. 1987. Experiments of pis- Annual Technicalhlarketing Conference, Philadel-
ton effect on elevator smoke control. ASHRAE phia, PA, October 18-2 1, pp. 240-248.
Transactioi7s 93(2a): 221 7-2228. Atlanta: American Levin, BC., et al. 1989. Synergistic effects of nitrogen
Society of Heating, Refrigerating and Air-Condi- dioxide and carbon dioxide following acute expo-
tioning Engineers, Inc. sures in rats, NISTIR 89-4105. Gaithersburg, Md.:
National Institute of Standards and Technology. .
Klote, J.H., et al. 1992. Feasibility and design consider-
ations of emergency evacuation by elevators. Levin, BiC., et al. 1995. Further development of the N-
National Institute of Standards and Technology, gas mathematical model: An approach for predict-
NISTIR 4870. ing the toxic potency of complex combustion mix-
Koplon, N.A. 1973a. Report of the Henry Grady fire tures, fire and polymers 11: Materials and tests for
tests. Atlanta: City ofAtlanta Building Department. hazard prevention. ACS Symposium Series. No.
Koplon, N.A. 1973b. A partial report of the Henry 599. August 2 1-26, 1994, Washington, D.C., Amer-
Grady fire tests (Atlanta, July 1972). Symposium ican Chemical Society, Washington, DC., pp. 293-
on Experience and Applications on Smoke and Fire 311.
Control at the ASHRAE Annual Meeting, Louis- Liu, H. 199 1. Wind engineering-A handbook for struc-
ville, Ky., pp. 3-8. Atlanta: American Society of tural engineers. Englewood, N.J.: Prentice Hall.
Heating, Refrigerating and Air-conditioning Engi- Lougheed, G.D. et al. 1999. Large-scale physical model
neers, Inc. studies for an. atrium smoke exhaust system.
Kreith, F. 1965. Prit~ciples of heat fran.fer, 2d ed. ,4SHRA E Transactions 105(1): 676-698.
Scranton, Pa.: International Textbook Co. Lougheed, G.D., and G.V. Hadjisophocleous. 1997.
Kumar, S. 1983. Mathematical modelling of natural Investigation of atrium smoke exhaust effective-
convection in fire-A state of the art review of the ness. ASHRAE Transactions 1 03(2): 1- 1 5.
field modelling of variable density turbulent flow. Lougheed, GD., and GV. Hadjisophocleous. 2000. The
Fire and ,liaterials 7(1): 1-24. smoke hazard from a fire in high spaces. ASHRAE
Launder, B.E.. and D.B. Spalding. 1974. The numerical Transactions 107(1).
computation of turbulent flows. Computer Methods Lougheed, GD., C. McCartney, and B.C. Taber. 2000.
in Applied ~~eclianics and Engineering 3269-289. Smoke movement for sprinklered fires. ASHRAE
Law, M. 1982 Air-supported structures: Fire and smoke Transactions 106(1): 605-619.
hazards. Fire Prevention, Vol. 48, pp. 24-28. Lougheed, G.D., C. McCartney, and B.C. Taber. 2001.
Law, M. 1986. A note on smoke plun~esfrom fires in Sprinklered mercantile fires. ASHRAE Transactions
multi-level shopping malls. Fire Safe& Journal 10: 107(1).
197. MacDonald, A.J. 1975. Wind loading on buildings. New
Lawson, J.R., W.D. Walton, and W.H. Twilley. 1984. York: Wiley.
Fire performance of furnishings as measured in the Madrzykowski, D. 1998. Office building fire research
NBS furniture calorimeter, Part I. Gaithersburg, program: An engineering based approach to fire
Md.: National Bureau of Standards. safety design, fire and material, 1998 International
Lees, F.P. 1980. Loss prevention in the process indus- Conference, Interscience Communications, Lon-
tries, Volume 2, Buttenvorths, London. don, UK, pp. 23-33.
Levin, B.C. 1996. New research avenues in toxicity: 7- Madrzykowski, D., and R.L. Vettori. 1992. A sprinkler
gas N-gas model, toxicant suppressants, and genetic fire suppression algorithm for the GSA engineering
toxicology. Toxicology. No. 1 15, pp. 89- 106. fire assessment system. Gaithersburg, Md.:
Levin, B.C. 7000. Personal communication between National Institute of Standards and Technology,
b i n and klote about the N-gas model and LCSo NISTlR 4833.
values, March 2000. Markatos, N.C. 1986. The mathematical modelling of
Levin, B.C., et al. 1987. Effects of Exposure to Single of turbulent flows. Applied Mathematical Modeling
Multiple Conibinations of the predominant toxic 1 O(3): 190-220.
. .
Principles of Smoke Management,

McCaffrey, B.J. 1983. Momentum implications for Moodie, K., et al. 1988. Fire at King's Cross Under-
buoyant diffusion flames. Combustion and Flame ground station, 18th November 1987-Part 11:
52(2): 149-167. Scale model fire growth tests, Incident Report,
McGrattan, K.B., H.R. Baum, and R.G Rehm. 1996. Health and Safety Executive, Derbyshire, U.K.
Large eddy simulations of smoke movement in Morgan, H.P. 1979. Smoke control methods in enclosed
1 three dimensions, NISTIR 6030. Gaithersburg, shopping complexes of one or more stores: A
Md.: National Institute of Standards and Technol- design summaqr,Building Reseerch Establishment.
1 . ogy. Morgan, H.P., and GO. Hansell. 1987. Atrium build-
McGrattan, K.B., H.R. Baum, and R.G Rehm. 1999. ings: Calculating smoke flows in atria for smoke-
Large eddy simulations of smoke movement. control design. Fire Safefy Journal 12(1987): 9-35.
ASHRAE Pansactions 105(1): 426-436. Atlanta: Morgan, H.P. et al. 1999. Design methodologies for
American Society of Heating, Refrigerating and smoke and heat exhaust ventilation. London: CRC
Air-conditioning Engineers, Inc. Ltd.
McGrattan, K.B., and GP. Forney. 2000. Fire dynamics Morita, M., and M. Hirota. 1989. Numerical analysis
simulator-User's manual, N[ST[R 6469. Gaithers- and experiments of convective heat flow in fire
burg, Md.: National Institute of Standards and compartments. Fire Science and Technology 9(1):
Technology. 11-24.
McGrattan, K.B. et al. 2000. Fire dynamics simulator- Morton, B.R., G. Taylor, and J.S. Turner. 1956. Turbu-
Technical reference guide, N[ST[R 6467. Gaithers- lent gravitational convection from maintained and
burg, Md.: National Institute of Standards and instantaneous sources. Proceedings of tize Royal
Technology. Sociel)~of London 234: 1-23.
McGuire, J.H., and G.T. Tamura. 1975. Simple analysis Mowrer, F.W. 1990. Lag times associated with fire
of smoke-flow problems in high buildings. Fire detection and suppression. Fire Technology 26(3):
Technology 1 1 (I): 15-22. 244-265.
Mowrer, F.W., and R.B. Williamson. 1987. Estimating
Milke, J.A., and F.W. Mowrer. 1994. Computer aided
room temperatures from fires along walls and in
desig~ifor smoke management in atria and covered
corners. Fire Techt~olog?- 23(2): 133- 145.
malls. ASHRAE Tt-amac/iotu lOO(2): 448-456.
Mudan, K.S., and P..4. Croce. 1995. Fire hazard calcula-
Miller, R.W., and L.S. Han. 1971. Pressure losses for
tions for large open hydrocarbon fires. SFPE Hand-
laminar floupin the entrance region of ducts of rect-
book of Fire Ptorectiotl E11gir7eet-i11,o, 2nd edition.
angular and equilateral triangular cross sections.
Quincy, Mass.: Sational Fire Protection Associa-
Journal ofApplied Mechanics 38: 1083-1087.
tion.
Mills, F.A. 2001. Case study of a fire engineering Mulholland, G. 2002. Smoke production and properties.
approach to a large, unsprinklered, naturally ve~lti- SFPE HaridDook o f Fits Ptorection Engineering, 2d
lated atrium building. ASHRAE T,.misac/ions ed. Quincy, Mass.: National Fire Protection Associ-
107(1). Atlanta: American Society of Heating, ation.
Refrigerating and Air-conditioning Engineers, Inc. Mulholland, G., et al. 1981. Smoke filling in an enclo-
Mitler, H.E. 1984. Zone modeling of forced ventilation sure, Paper 81 -HT-8. New York: American Society
fires. Cotm!~itst.Sci. Tech. 39: 83-106. of Mechanical Engineers.
Mitler, H.E. 1996. input data for fire modeling, 13th Murosaki, Y., H. Hayashi, and T. Nishigaki. 1986.
Meeting of the UJNR Panel on Fire Research and Effects of passage width on choice of egress route
Safety, March 13-20, 1996, NlSTlR 6030, pp. 187- at a T-junction in a building. 1st International Sym-
199. Gaithersburg, Md.: National Institute of Stan- posium on Fire Safety Science. NenTYork: Hemi-
dards and Technology. sphere, 593-600.
Mitler, H.E. 2000. Notes on burning of several Suel Nam, S., and R.G. Bill. 1993. Numerical simulation of
packages, Personal com~iii~nicatio~~ between Henri thermal plumes. Fits Sqfery Joitrna/ 2 l(3): 23 1 -
Mitler and John H. Klote on April 27, 2000. 256.
Mitler, H.E., and H.W. Emmons. 198 l . Documentation NBFU. 1939. Sniokr hazards of air-conditioning sys-
for CFC V, the Fifth Harvard Computer Code. tems. NFPA Quarret-i~~ 33(2): 1 13-1 22.
L
Home Fire Projcct Tech. Rcp. #45, Harvard Univer- Nelson, H.E. 1987. An engineering analysis of the early
sity. stages of fire dr.t.elopment-The fire at the Dupont
Mitler, H.E., and J.A. Rockeu. 1986. How accuratc is Plaza Hotel and Casino-December 3 1 , 1986.
mathematical lire niodeling? Gaitlicrsburg. Md: Gaithersburg. Xld.: National lnstitute of Standards
National Bureau of Slandards. NBSIR 86-3459. and Techno1og~-.SlSTlR S7-3560.
References

Nelson and MacLennan. 2002. Emergency movement. Pauls, J; 2002. Mcvement of people. SFPE Handbook of
SFPE Handbook of Fire Protection Engineering. Fire Protection Engineering. Quincy, Mass.:
Quincy, Mass.: National Fire Protection Associa- National Fire Protection Association.
tion.. P a u l ~ ,J.L., and E. Juillet. 1989. Recent technical and
Newman, J.S. 1988. Princioles of fire detection. Fire social developments influencing the life safety of
Technology 24(2): 116-127. people with disabilities. Presentation at Pacific Rim
NFPA. 1976. Life safety code--Code for safety to life Conference of Building Officials, Honolulu,
from fire in buildings and stiuctures, NFPA 101. Hawaii, April 9-13.
Quincy, Mass.: National Fire Protection Associa- Peacock, R.D., et al. 1991. Software user's guide for
tion. HAZARD I, Version l. l. NIST Handbook 146, Vol.
NFPA. 1977. Life Safity Code. NFPA 101. Quincy, I. Gaithersburg, Md.: National Institute of Stan-
Mass.: National Fire Protection Association. dards and Technology.
NFPA. 1988. Reconmended practice for smoke control Peacock, R.D., et al. 1993. CFAST, the consdlidated
systems, NFPA 92A. Quincy: Mass.: National Fire model of fire growth and smoke transport. Gaithers-
Protection Association. burg, Md.: National Institute of Standards and
NFPA. 1995. Fire safety concept tree, NFPA 550. Technolegy, NIST Technical Note 1299.
Quincy, Mass.: National Fire Protection Associa- Peacock, R.D., et al.. 1997. A user's guide for FAST:
tion. engineering tools for estimating fire growth and
NFPA. 1997. Fire Protection Haildbook. Quincy, Mass.: smoke transport. Gaithersburg, Md.: National Insti-
National Fire Protection Association. tute of Standards and Technology, NIST Special
NFPA. 1999. National fire alarm code, NFPA 72. Publication 92 1 .
Quincy, Mass.: National Fire Protection Associa- Peacock, R.D., et al. 2000. A user's guide for FAST:
tion. engineering tools for estimating fire growth and
NFPA. 2000a. Recommended practice- for smoke con- smoke transport, NIST Special Publication 92 1.
trol systems, NFPA 92A. Quincy, Mass.: National Gaithersburg, Md.: National Institute of Standards
Fire Protection Association. and Technology.
NFPA. 2000b. Guide for smoke I I I O I U ~ ~ I IS V~ SI I~~ I I ~inS
Peacock, R.D., P.A. Reneke, R.W. Bukowski, and V.
malls, atl-in, and large areas. NFPA 92B. Quincy,
Babrauskas. 1999. Defining flashover for fire haz-
Mass.: National Fire Protection Association.
ard calculations. Firs Safety Journal. Vol. 32, pp.
NFPA. 2000c. Llfe safefy code, NFPA 101. Quincy,
33 1-345.
Mass.: National Fire Protection Association.
Pitts, W.M. 1994. The global equivalence ratio concept
Nowler, S.P. 1987. Enclosure environment characteriza-
and production of carbon monoxide in enclosure
tion testing for the baseline validatiorl of cornputer
fires, NIST Monograph 197. Gaithersburg, Md.:
fire sinlulation codes. Sandia National Laborato-
National Institute of Standards and Technology.
ries.
O'Neill, J.G., W.D. Hayes, and R.H. Zile. 1980. Full Prasad, B. 1995. Testing fire dampers at varying air
flows and temperatures. ASHRAE fin~aactions
scale tests with automatic sprinklers in a patient
I 0 l (2).
room phase 11, NBSIR 80-2097. Gaithersburg, Md.:
National Institute of Standards and Technology. Predtechenskii, V.M., and A.I. Milinskii. 1978. Plannins
Parker, W.J. 1982. Calculation of heat release rate by for foot traffic flow in buildings. New Delhi: Arner-
oxygen consunlption for various applications. ind Publishing Co.
Gaithersburg, Md.: National Bureau of Standards, Press, W.H., et al. 1986. Monerical recipes. New York
NBSIR 82-2427-1. Cambridge Univ. Press, pp. 258.
Pauls, J. 1977. Management and movement of building Proulx, G. 1995. Evacuarion times and movement times
occupants in emergncies, DBR Paper NO. 788. in apartment buildings. Fire Safety JOUI-nal 24: 229-
Ottawa, Canada: National Research Council. 246.
Pauls, J. 1980. Building evacuation: Research findings Proulx, G., and R. Fahy. 1997. The time delay to start
and recomniendations. Fires and Hu~nanBehuv- evacuation: Review of tive case studies. 5th Inter-
iour, D. Canter, ed. New York: John Wiley. national Symposiun~ on Fire Safety Science.
Pauls, J.L. 1988. Review of standards & codes plus rec- IAFSS, 783-794.
ommendations for accessible means of egress, Proulx, G., and^. Sime. 1991. To prevent panic in an
Funded by U S . Architectural & Transportatio~i underground enlcrgency. W l ~ ynor tell people the
Barriers Compliance Board. Dept. of Education. timth. 3rd International Symposium on Fire Safety
Washington, D.C. Scicncc. Ncw York: Elsevier. 843-853.
Principles of Smoke Management

Purser, D.A. 2002. Toxicity assessment of combustion Schetz, J.A. 1993. Boundary layer analjsis. Englewood,
products, Chapter 1-13. SFPE Handbook of Fire Cliffs, N.J.: Prentice Hall.
Protection Engineering, 2d ed. Quincy, Mass.: Schlichting, H. 1960. Boundary layer theory, 4th ed,., J .
National Fire Protection Association. Kestin, Translato . New York: McGraw.
Quarantelli, E.L. 1979a. Panic behavior in fire situa- Seader, J., and I. Einhom. 1976. Some physical, chemi-
tions: Findings and a model from the English lan- cal, toxicological, and physiological aspects of fire
gliagi rescarch literature. Columbus, Ohio: Ohio smokes, NSF Report, Utah University.
State University. S FP E. 2002. Handbook of fire protection engineering.
Quarentelli, E.L. 1979b. Fire papers from the panel ses- Boston, Mass.: Society of Fire Protection Engi-
sion on panic, Second International Seminar on neers.
Human Behavior in Fire Emergencies, October 29 - Shaw, C.T., and GT. Tamura. 1977. The calculation of
November 1, 1979, in Proceedings of Seminar, B. air infiltration rates caused by wind and stack action
M. Levin and R. L. Paulsen (ed.), NBSIR 80-2970. for tall buildings. ASHME Transactions 83(2):
Washington D.C.: National Bureau of Standards. 145- 158. Atlanta: American Society of Heating,
Quintiere, J.G. 1989a. Fundamentals of enclosure fire Refrigerating andr Air-conditioning Engineers, Inc.
"zone" models. Journal of Fire Protection Engi- Shaw, C.Y., J.T. Reardon, and M.S. Cheung. 1993.
neering l(3): 99- 119. Changes in air leakage levels of six canadian office
Quintiere, J.G. 1989b. Scaling applications in fire buildings. ASHRAE Journal 35(2): 34-36. Atlanta:
research. Fire Safefy Journal 1 5(1): 3-29. American Society of Heating, Refrigerating and
Quintiere, J.G. 1998. PI-inciples of fire behavior. New Air-conditioning Engineers, Inc.
York: Delmar Publishers. Sherman, F.S. 1990. Escousflow. New York: McGraw.
Shields, et al. 1996. Escape of disabled people from fire.
Quintiere, J.G., and M.E. Dillon. 1997. Scale model
reconstruction of fire i n an atrium. 2nd International A measurement and classification of capability for
assessing escape risk, BRE Report 301, Boreham-
Symposium on Scale Modeling, June 23-27, Uni-
wood, Building Research Establishment.
versity of Kentucky, Lexington, Kentucky, pp. 73-
86. Siegel, R., and J.R. Howell. 1992. Thermal radiation
heat transfer, 3rd ed., Hemisphere, Washington,
Quintiere, J.G., B.J. McCaffrey, and T. Kashiwagi. 1978.
D.C.
ConlBustion Science and Tecl~nology18: 1- 19.
Simcox, S., N.S. Wilkes, and LP. Jones. 1988. Fire at
Ramachandran, G. 199 1. Informative fire warning sys-
King's Cross Underground Station, 18th November
tems. Fire Teclmology 27(1): 66-8 l .
1987: Numerical simulation of the buoyant flow
Read, R.E.H., and M.P. Shipp. 1979. An investigation of and heat transfer. Hanvell Laboratory, UK.
fire door closer forces, BRE Report. Borehamwood, Simcox, S., N.S. Wilkes, and I.P. Jones. 1989. Computer
Herts, U.K.: Fire Research Station. simulation of the flows of hot gases from the fires at
Rehm, R.G., and H.R. Baum. 1978. The equations of King's Cross Underground Station. Institution of
motion for thermally driven, buoyant flows. Jour- Mechanical Engineers (IMechE) King's Cross
nal of Resea~.cllof NBS 83: 297-308. Underground Fire: Fire Dynamics and the Organi-
Rilling, J. 1980. Mechanism and conditions of smoke zation of Safety, London, UK, pp. 19-25.
control through a door opening. Champs Sur Simiu, E., and R.H. Scanlan. 1996. Wind effects on
Marne, France: Centre Scientifique et Technique du structures: Fundamentals and application to design,
Batiment (CSTB). 3rd ed. New York: Wiley.
Rockett, J.A., et al. 1987. Comparison of NBSIHarvard Sinclair, R. 200 I . CFD simulation in atrium smoke man-
V1 simulations and full scale, multi-room fire test agement system design. ASHRA E Transactions
data, NBSlR 87-3567. Gaithersburg, Md.: National 107(1). Atlanta: American Society of Heating,
Bureau of Standards. Refrigerating and Air-conditioning Engineers, Inc.
Sander, D.M. 1974. FORTRAN IV program to calculate SMACNA. 1987. HVAC sysfems applications. Vienna,
air infiltration in buildings, National Research Va.: Sheet Metal and Air Conditioning Contractors'
CounciI Canada, DBR Computer Program No. 37. Association, Inc.
Sander, D.M., and G.T. Tamura. 1973. FORTRAN IV SMACNA. 1990. ~ ~ ~ ~ - s ~ ducf s f edesign,
r n s 3d ed.
program to stimulate air movement in multi-story Chantilly, Va.: Sheet Metal and Air Conditioning
buildings, National Research Council Canada, DBR ~ontractor~"~ssociation, Inc.
Conlputer Program No. 35. SMACNA. 1993. HVAC Systems Esfing, Adjlrsring and
SBBCI. 1 999. Standard building code, Birmingham, Balancing, 2d ed. Vienna, Va.: Sheet Metal and Air
Ala. Conditioning Contractors' Association, Inc.
- ~eferinces

Steckler, K.D., H.R. Baum, and J.G Quintiere. 1986. Transactions 96(2): 373-383. Atlanta: American
Salt water modeling of fire induced flows in multi- Society of Heating, Refiigerating and Air-Condi-
compartment enclosures. 2 1st Symposium (Interna- tioning Engineers, Inc.
tional) on Combustion, Combustion Institute, pp. Tamura, GT. 1990~.Fire tower'tests of stair pressuriia-
143-149. tion systems with mechanical venting of the fire
Steckler, K., J.G. Quintiere, and J.H. Klote. 1990. The floor. ASHRA E Transactions 96(2): 3 84392.
Johnson City Fire, (Letter Report); Center for Fire Atlanta: American Society of Heating, Refrigerat-
Research. Gaithersburg, Md.: National Institute of ing and Air-conditioning Engineers, Inc.
., ,
Standards and Technology. Tamura, GT. 1991. Determination of critical air veloci-
Stewart, et al. 1973. Experimental human response to ties to prevent smoke backflow at a stair door open-
high concentrations of carbon monoxide. Archifec- ing on the fire floor. ASHRAE Transacfions 97(2):
frrral Etzvironmenfal Health 26(1): 1-7. 627-633. Atlanta: American Society of Heating,
Stoll, A.M., and M.A. Chianta. 1969. Method and rating Refrigerating and Air-conditioning Engineers, Inc.
system for evaluation of thermal protection. Aet-o- Tamura, G.T. 1994. Fire tower tests on vestibule pressur-
space Medicine 40: 1232-1238, November. ization for protection of stairshafts. ASHRAE Trans-
Strakosch, G.R. 1983. Verficaltransporfafion:Elevafors actions lOO(2). Atlanta: American Society of
and escnlafot-S,2d ed. New York: Wiley & Sons. Heating, Refrigerating and Air-conditioning Engi-
Streeter, V.L., and E.B. Wylie. 1979. Fluid Mechalzics, neers, Inc.
7th Ed. New York: McGraw. Tamura,G.T., and J.H. Klote. 1987a. Experimental Fire
Stioup, W.D. 1999. Scotch pine christmas tree fire tests, Tower Studies on Elevator Pressurization Systems
Report of Test FR 4010. Gaithersburg, Md.: for Smoke Control. ASHRAE Transactions 93(2):
National Institute of Standards and Technology. 2235-2257.
Stroup, W.D., et al. 2000. Large fire research facility Tamura, G.T., and J.H. Klote. 1987b. Experimental fire
(Building 205) exhaust hood heat release rate mea- tower studies on mechanical pressurization to con-
surement system. Gaithersburg, Md.: National trol smoke movement caused by fire pressures. Pro-
Institute of Standards and Technology, NISTIR ceedings of the 2nd Infernational Sytnposilrnl 017
6509. Fire Safe01 Scietice, ToXyo, Japan.
Sumka, E.H. 19% Presently, elevators are not safe in Tarnura, G.T., and J.H. Klote. 1988. Experimental fire
fire emergencies. Elevafol- World 36(11): 34-40. tower studies on adverse pressures caused by stack
S~~ppIemen~ 10 /he nafinnal building code o f Cannda. and wind action: Studies on smoke movement and
1985. Chapter 3, Measures for fire safety in high control, ASTM International Symposiuni on Char-
buildings. Associate Committee on the National acterization and Toxicity of Smoke, December 5,
Building Code, National Research Council of Can- 1988, Phoenix, Ariz.
ada, Ottawa, NRCC 23178. Tamura, G.T., and C.Y Shaw. 1973. Basis for the design
Tamanini, F. 1976. The application of water sprays to of smoke shafts. Fire Tecl7nol0,qy 9(3): 209-222.
the extinguishnient of crib fires. Combusfion Sci- Tamura, G.T., and C.Y. Shaw. 1976a. Studies on exterior
ence atid Technolog~Vol. 14, pp. 17-23. wall air tightness and air infiltration of tall build-
Tamura, G.T. 1978a. Exterior wall venting for smoke ings. ASHRAE Transactions 82(1): 122-134.
control in tall office buildings. ASHRAE J O Z I I - I ~ ~ Atlanta: American Society of Heating, Refrigerat-
20(8): 43-48. Atlanta: American Society of Heat- ing and Air-conditioning Engineers, Inc.
ing, Refrigerating and Air-Conditioning Engineers, Tamura, G.T., and C.Y. Shaw. 1976b. Air leakage data
Inc. for the design of elevator and stair shaft pressuriza-
Tamura, G.T. 197%. Experimental studies on exterior tion systems. ASHRAE T?ansactions 82(2): 179-
wall venting for smoke control in tall buildings. 190. Atlanta: American Society of Heating, Refrig-
ASHRAE 7bansactions SI(2): 204-2 15. Atlanta: erating and Air-conditioning Engineers, Inc.
American Society of Heating, Refrigerating and Tamura, G.T., and C.Y. Shaw. 1978. Experimenta! stud-
Air-conditioning Engineers, Inc. ies of mechanical venting for smoke control in tall
Ta~nura,G.T. 1990a. Field tests of stairshaft pressuriza- office buildings. ASHRAE Tramactions 86(1): 54-
tion systems with overpressure- relief. ASHRAE 7 1. Atlanta: American Society of Heating, Refriger-
Trn1l.soc/io17s96(1): 95 1-958. Atlanta: American ating and Air-conditioning Engineers, Inc.
-
Socicty of Healing, Refrigerating and Air-Condi- Tamura, G.T., and A.G. Wilson. 1966. Pressure differ-
~ioni~igEngineers, Inc. cnces for a nine-story building as a result of chim-
Tamura, G.T. 1990b. Fire tower tests of stair pressuriza- ney effcct and ventilation system operation.
tion systems \vitli overprcssure relief. ASHRAE ASHRAE Tt.tn~snc/ions72(1): 180-189. Atlanta:
Principles of Smoke Management

i American Society of Heating, Refrigerating and Walton, GN. 1989. AIRNET-A computer program for
Air-conditioning Engineers, Inc. building airflow network modeling. Gaithersburg,
Tamura, GT., and A.G Wilson. 1967a. Building pres- Md.: National Institute of Standards and Technol-
sures caused by chimney action and mechanical ogy.
ventilation. ASHRAE Transactions 73(2). Walton, GN. 1997. CONTAM96 user manual, NISTIR
Timura, GT., and A.G Wilson. 1967b. Pressure differ- 6056. Gaithersburg, Md.: National Institute of Stan-
ences caused by chimney effect in'three high build- dards and Technology.
l ings. ASHRA E Transactions 73(2). Walton, W.D. 1985. ASET-B: A room fire program for
I '
Tanaka, T. 1983. A model of multiroom fire spread. personal computers. Gaithersburg, Md.: National
Gaithersburg, Md.: National Bureau of Standards, Bureau of Standards, NBSIR 85-3144-1.
NBSIR 83-27 18. Walton, W.D. 1988. Suppression of wood crib fires with
Tewarson, A. 2002. Generation of Heat and chemical sprinkler sprays: Test results. Gaithersburg, Md.:
compounds in fires, Chapter 3-4. SFPE Handbook National Bureau of Standards, NBSIR 88-3696.
of Fire Protection Engineering, 2d ed. Quincy, Walton, W.D., and P.H. Thornas. 1995. Estimating tem-
Mass.: National Fire Protection Association. peratures in compartment fires. SFPE Handbook of
Fire Protection Engineer-irlg, Chapter 3-6.
Thomas, P.H. 1970. Movement of smoke in horizontal
Bethesda, Md.: Society of Fire Protection Engi-
corridors against an air flow. Institute of Fire E@-
neers.
neers Quarterly 30(77): 45-53.
Waters, R.A. 1989. Stansted terminal building and early
Thomas, P.H., et al. 1963; Investigation into the flow of atrium studies. low-nal of Fire P)-otcction Engi-
hot gases in roof venting, Tech. Paper No. 7. Bore-
rieeririg 1(2): 63-76.
ham Woods, Herts, UK: Fire Research Station. White, F.M. 1974. fiscozrs ,fluid floiv. New York:
Tsujirnoto, M., T. Takenouchi, and S. Uehara. 1990. A McGran..
scaling law of smoke movement in atrium. I l th Wood, Peter G. 1971. The behaviour of people in fires.
Joint Panel Meeting of the UJNR Panel on Fire Fire Research Note 953. Borehamwood, U.K.: Fire
Research and Safety. Gaithersburg, Md.: National Research Station.
Institute of Standards and Technology. Wrap, C.P., and G.K. Yuill. 1993. An evaluation of algo-
UL. 1999a. Standard forfire darupers, UL 555. North- rithms for analyzing smoke control systems.
brook, Ill.: Underwriters Laboratories, Inc. ASHRAE Transactions 99( 1 ): 160- 1 74.
UL. 1999b. Standard for smoke dampers, UL 555s. Yii, E.H. 1998. Exploratory salt \vater experiments of
Northbrook, Ill.: Underwriters Laboratories, Inc. balcony spill plume using laser induced fluorescent
Underwriters Laboratories. 1990. Standard for air leak- technique. Fire Engineering Research Report 9517,
age tests of door assemblies-UL 1784, North- University of Canterbury, Neu. Zealand.
brook, 111. Yoshida, H. et al. 1979. A FORTRAN IV pro,oram to
Wakamatsu, T. 1977. Calculation methods for predicting calculate smoke concentrations in a multi-story
smoke movement in building tires and desiging building. Ottawa, Canada: National Research
smoke control systems, Fire Standards and Safety, Council.Best, R., and D.P. Demers. 1982. Investiga-
ASTM STP-6 14, A. F. Robertson, ed., pp. 16s-193. tion report on the MGM Grand Hotel fire-Las
Philadelphia, Pa.: American Society for Testing and Vegas. Nevada, November 21, 1980. Quincy,
Materials. Mass.: National Fire Protection Association.
Appendix A

Units of Measurement and Physical Data

hysical quantities such as length, weight, and become base units and the other is a derived unit. Theo-
time are expressed in temls of standard units of retically, any three can be selected as base units. How-
measurement. In this book, both English units ever, the only two combinations to be used extensively
and international system (SI) units are used. are:
Newton's second law of motion states that the force,
F, on a body of fixed mass, m , is proportional to the Base Units Derived Unit
product of the mass and the acceleration, a: mass, length, and time force
force, length, and time mass

There are three c0111111011 English systems with Because force is a derived unit in the S1 system,
regard to mass and force: the pound mass and pound that convention is used in the following discussion for
force system, the slug and pound system, and the pound the inch pound (I-P) system. For convenience, the unit
mass and poundal system. Introduction of the propor- of mass in the I-P system \\.ill be taken to be the slug. A
tionality constant 11% into the abo1.e relation yields slug can be thought of as a mass that has a weight of
32.174 pounds at sea level. In the I-P system, the unit of
force is the pound, Ib, which is the force required to
accelerate a mass of one slug at a rate of one foot per
second squared. In the S1 system, the unit of force is the
Table A-I lists the units for these systeins and the newton, N, which is the force required to accelerate a
S1 system along with the values of g, for each. Gener- mass of one kilogram at a rate of one meter per second
ally, a pound is thought of as a unit of force. However, in squared.
some engineering applications, the pound also has been The base units and derived unit discussed above
used as a unit of mass. One pound mass (Ibm) is the relate force and mass, but many more units are needed
mass of a body that weighs one pound (Ib) at sca level. for engineering calculations. The base units and derived
One slug equals 32.174 Ibni, and one poundal is a force units needed for smoke control applications are listed in
of 0.03108 pounds. For the systems listed in Table A-l Tables A-2 and A-3. In the S1 system, prefixes are used
for which the value of ,q. is one, Ne\vton's second law to form decimal multiples and submultiples of the S1
can be written as units. The S1 prefixes are listed in Table A-4.
Unit systems with g, of 1 can be referred to as
F = I110 .
I~omogeneousunit systems (Table A-l), and in this text
This formulation of Neuton's law simplifies homogeneous unit systems are referred to as being
derived equations and calculations. I t is accomplished either the S1 system or the slug and pound system with
by defining one of the four units (len_rth,mass, time, and the base units and derived units as listed in Tables A-?
- force) in terms o f ~ l i other
e three. Thus, three of h e units and A-3.
Appendix A- Units of Measurement and Physical Data

Table A-l:
Units Relating Force and Mass in Various Systems
Pound Mass and Slug a n d Pound Mass a n d
Pound Force Pound Poundal International
Quantity System System System System (SI)
length foot (ft) foot (ft) foot (ft) meter (m)
time second (S) second (S) second (S) second.(s)
mass pound mass (Ibm) slug . pound mass (Ibm) kilogram (kg)
force pound force (Ibt) pound (Ib) poundal Newton (N)
32.174 Ibm ft 1 slug ft 1 Ibm ft I kgm
gc 2 2
I lbf s lbf s poundal s N s2

Table A-2:
Base Units
S1 System English System
Quantity Unit Symbol Unit Symbol
len,oth meter m foot ft
mass kilogram kg slug slu~
time second S second S

thennodynamic (absolute) kelvin K O R O R

temperature

Table A-3:
Derived Units
S1 System English System
Quantity Unit Symbol Formula Unit Synlbol Formula
force newton N kg I pound Ib slug ft/s2
pressure pascal Pa ~lm' lblft2
energy, work or heat joule J Nm Ib ft
power, enersy release rate watt W Jls Ib ftls
mass flow rate kgls slugls

Table A-4:
SI Prefixes
Prefis Sxmbol Multiplication Factor
giga G I o9 = I 000 000 000
niega
kilo

nano 11 I = 0.000 000 00 1


I. Thc prelis ccnri is ro hc atoidcd wlwc pssiiic.
Principles of Smoke Management

Most of the conversion factors listed in Tables A-5


and A-6 have been rounded off to four significant fig-
ures. This level of accuracy may be excessive for most
smoke control calculations, but users can easily round
down as desired. Table A-7 lists constants for accelera-
tion of gravity, gas constant of air, and standard atmo-
spheric pressure.
Absolute temperature is mezsured using the Kelvin
scale in the S1 system and the Rankine scale in the I-P
system. In addition, temperature is frequently measured
in the Celsius or the Fahrenheit scale. Because Celsius
and Fahrenheit scales are so commonly used by design
engineers, these scales are used exclusively in the dis- where
cussions in the text and figures. However, caution
should be exercised to ensure that absolute temperatures TF = temperature, OF,
are used in calculations where necessary. The following
TC = temperature, OC,
equations can be used to convert between temperature
scales:
TR = temperature, OR,

TK = temperature, K.

Tables A-8 and A-9 list density, specific heat, vis-


cosity, and thermal conductivity of air. For further infor-
mation concerning the S1 system, the reader is referred
to Guideline for- the Use of the International System of
Units, 1995 Edition, NlST Special Publication 8 11,
National Institute of Standards and Technology, Gaith-
ersburg, Md.
-Appendix A - Units of Measurement a n d Physical Data

Table A-5:
Factors for Conversion to SI Units
Multiply BY To Obtain
atmosphere, standard (atm) 101325 pascal (Pa)
atmosphere, standard (atm) 101.325 kilopascal (kPa)
British thermal unit (Btu) 1055 joule (J)
British thermal unit (Btu) 1 .055 kilojoule (U)
British thermal unit per hour (Btulh) 0.293 watt (W)
British thermal unit per pound (Btdlb) 2330 joule per kilogram (Jkg)
British thennal unit per pound degree Fahrenheit [Btu/(lb°F)] 4187 joule per kilogram kelvin [J/(kg+K)]
British thermal unit per second (Btuls) watt (W)
British thermal unit per second (Btu/s) kilowatt (kW)
British thermal unit foot per hour square foot degree watt per meter kelvin [W/(m.K)]
Fahrenheit [Btu ft/(h f t Z O ~ ) ]
British thermal unit inch per hour square foot degree watt per meter kelvin [Wl(m-K)]
Fahrenheit [Btu in./(h f t Z O ~ ) ]
calorie (cal) joule (J)
centimeter of mercury (cm Hg) pascal (Pa)
centimeter of mercury (cni Hg) kilopascal (kPa)
centimeter o r water (cm H@) pascal (Pa)
centipoise (cP) pascal second (Pa-S)
centistokes (cSt) meter squared per second (m2/*)
3
cubic foot (ti ) - cubic meter (m3)
cubic foot (ft3) liter (L)
cubic foot per ~ninutc(li-'Irnin or cfin) cubic meter per sccond (rn3/s)
cubic I'oot per ~ninute(li3in1in or clin) liter per second (Lls)
cubic tbot per second (ct3/s) cubic meter per second (m'/s)
cubic inch (in.3) cubic meter (m3)
cubic inch per minutc (in.'imin) cubic meter per second (m3/s)
cubic yard (yd3) cubic meter (m3)
cubic yard pcr minute (yd'/n~in) cubic meter per second (1n3/s)
dyne (dyn) newton (N)
dyne centinleter ( d y n a n ) netvton meter (N.ni)
dyne per square centillleter (dyn/cm2) pascal (Pa)
erg (erg) joule (J)
erg per second (ergls) watt (W)
foot (li) meter (m)
Toot of mcrcury. conven~ional(It tig) pascal (Pa)
foot of mercury. conventional (It Hg) kilopascal (kPa)
root ofwater ( f 1H20) pascal (Pa)
h o t ol'water (li 1i20) kilopascal (kPa)
foot pcr hour (lilh) nieter per second (mls)
h o t per minutc (lilmi~ior Ipnl) meter per sccond (m!s)
h o t per sccond (lils) metcr pcr sccond (mls)
Principles of s m o k e M a n a g e m e n t

Table A-5: (Continued)


Factors for Conversion to SI Units

foot pound-force (ft-lbf) 1.356 joule (J)


foot pound-force per hour (ft.lbt%) watt (W)
foot pound-force per minute (ft-lbflmin) watt (W)
foot pound-force per second (ftdbfls) watt (W)
gallon (Imperial) (gal) cubic meter (m3)
gallon (Imperial) (gal) liter (L)
gallon (US.) (gal) cubic meter (m3)
gallon (US.) (gal) liter (L)
gallon (US.) per minute (gpm) (gall niin) cubic meter per second (m3/s)
gallon (US.) per minute (gpm) (gallmin) liter per second (Lls)
horsepower (550 ft.lbf/s) watt (W)
horsepower (boiler) watt (W)
horsepower (electric) watt (W)
hour (hr or h) second (S)
inch (in.) nleter (m)
inch (in.) centimeter (cm)
inch of mercury (in. Hg) pascal (Pa)
inch of mercury (in. Hg) kilopascal (kPa)
inch o f water (in. H 2 0 ) pascal (Pa)
inch per second (in.1~) meter per second (mls)
kilogram-force (kg0 newton (N)
kilogram-force meter (kgfm) newton meter ( N m )
kilogram-force per square centimeter (kgf/cm2) pascal (Pa)

kilogram-force per square centimeter (kgf/cm2) kilopascal (kPa)

kilogram-force per square meter (kgf/m2) pascal (Pa)


kilometer per hour (kmlh) meter per second (mls)
kilowatt hour (kWh) joule (J)
kilowatt hour (kWh) ~negajoule(MJ)
kilowatt-hour (kWh) joule (J)
knot (nautical mile per hour) meter per second (mls)
liter (L) cubic meter (m3)
mil (0.00 1 in.) meter (m)
mil (0.001 in.) millimeter (mm)
mile (mi) meter (m)
mile (mi) kilometer (km)
mile per hour ( m i h or mph) meter per second (mls)
mile per hour (milh o r mph) kilome~erper hour (kmlh)
mile per minute (milmin) n~eterper second (mls)
mile per second (mils) meler-persecond (mls)
mile, nautical nleler (m)
millimeter of mercury (mmHg) pascal-(Pa)
niillimeter of water (1111iiH~O) pascal (Pa)
minute (min) second ( S )
ounce (avoirdupois) (oz) 0.02835 kilogram (kg)

263
Appendix A-Units of Measurement and Physical Data

Table AS: (Continued)


Factors for Conversion to SI Units
Multiply BY To Obtain
ounce (avoirdupois) (02) 28.35 gram (g)
ounce (troy or apothecary) (02) kilogram (kg)
ounce (boy or apothecary) (02) gram (g)
poise (P) . pascal second (Pas)
pound (avoirdupois) (Ib) kilogram (kg)
pound (troy or apothecary) (Ib) kilogram (kg)
pound force (Ibf) newton (N)
pound per cubic foot (lblft3) kilogram per cubic meter (kg/m3)
pound per hour (lblh) kilogram per second (kg/s)
pound per minute (Iblmin) kilogram per second (kgk.)
pound per second (Ibls) kilogram per second (kgls)
poundal newton (N)
pound-force (Ibf) newton (N)
pounbforce foot (Ibf.ft) newton meter (N-m)
pound-force inch (Ibfh.) newton meter ( N m )
pound-force per square foot (lbf/ft2) pascal (Pa)
pound-force per square inch (psi) ( l b ~ i n . ~ ) pascal ( W
pound-force per square inch (psi) (1bflin.l) kilopascal (kPa)
revolution per minute (rpm) (rimin) radian per second (radls)
slug (slug) kilogram (kg)
slug per cubic foot (slug/ft3) kilogram per cubic meter (kg/m3)
slug per foot second [slug/(ft.s)] pascal second (Pas)
square foot (ft2) square meter (m2)
square foot per second (ft2/s) meter squared per second (m2/s)
square inch (in2) square meter (m2)
square inch (in2) square centimeter (cm')
square yard (yd2) square meter (m2)
standard cubic feet per minute (scfm)' standard cubic meter per second (sm3/s)
standard cubic feet per minute (scfm)' standard lltre per second (sL/s)
standard cubic feet per minute (scfm)' kilogram per second (kgls)
stokes (St) meter squared per second (m2/s)
ton of refrigeration watt (W)
ton, long (2240 Ib) kilogram (kg)
ton, metric (t) kilogram (kg)
ton, short (2000 Ib) kilogram (kg)
ton-force (2000 1bf) newton (N)
ton-force (2000 Ibf) kilonewton (kN)
watt hour (W.h) joule (J)
yard (yd) 0.9 144 meter (m)
I. scfni i s a form o f mass f l o u rate used for air movement. and for this text. ir is at 70°F (21°C) and one armospherr.
Principles of S m o k e Management
. -

Table A-6: .,.

Factors for Conversion to the English Units


Multiply BY To Obtain
atmosphere, standard (atm) 2116.2 pound-force per square foot (lbflft2)
atmosphere, standard (atm) pound-force per square inch (psi) (~bflin.~)
atmosphere, standard (atm) inch of water (in. H20)
British thermal unit (Btu) foot pound-force (ft-lbf)
British thermal unit per hour (BtuAi) foot pound-force per second (ft4bfls)
British thermal unit per hour (Btu/h) British thermal unit per second (Btu/s)
British thermal unit per minute (Btulmin) foot pound-force per second (ft-lbf/s)
British thermal unit per minute (Btulmin) British thermal unit per second (Btu/s)
British thermal unit per second (Btds) British thermal unit per minute (Btdmin)
British thermal unit per second (Btu!s) British thermal unit per hour (Btdh)
British thermal unit per second (Btuk) foot pound-force per second (ft.lbf/s)
Btdlbm foot pound-force per slug (ft4bfIslug)
centimeter of mercuq (cm Hg) pound-force per square foot (1bflft2)
centimeter of mercup (cm Hg) pound-force per square inch (psi) ( ~ b f / i n . ~ )
centimeter of mercun (cm Hg) inch of water (in. H20)
centimeter of water, (cm H20) pound-force per square foot (lbf/Ft2)
centimeter of water, (cm HzO) pound-force per square inch (psi) (lb~in.')
centimeter of water, (cm H,O) inch of water (in H20)
cubic foot (ft3) cubic inch ( h 3 )
cubic foot (ft3) gallon (US.) (gal)
cubic foot per minute (sfin) (ft3/lnin) cubic foot per second (ft3/s)
cubic foot per second (ft3/s) cubic foot per minute (cfm) (ft3/min)
cubic inch ( h 3 ) cubic foot (ft3)
cubic inch per minute (in.3/n~in) cubic foot per minute (cfrn) (ft3/min)
cubic inch per minute (in.'lmin) cubic foot per second (Ft3/s)
cubic yard (yd') cubic foot (ft3)
cubic yard 0.d') cubic inch (in.))
cubic yard (yd3) gallon (U.S.) (gal)
cubic yard per n~inute(vd3/n~in) cubic foot per minute (cfm) (ft3/min)
cubic yard per minute (yd3/min) cubic foot per second (ft3/s)
foot of mercury, con\.entional (ft Hg) pound-force per square foot (lbf/ft2)
foot of mercury, conventional (ft Hg) pound-force per square inch (psi) (lbf/in2)
foot of mercury, conventional (ft H
:
) inch of water (in. H20)
foot of water (ft HIOl pound-force per square foot (lbf/ft2)
foot of water (ft H,O) pound-force per square inch (psi) (lbf/in2)
foot of water (ft H 2 0 ) inch of water (in. H20)
foot per hour (ft/h) foot per second (ft/s)
foot per hour (ftlh) foot per niinute.(ft/min)
foot per minute (ftlmin) foot per second (Ws)
foot per second (fi/s) foot per minute (Wmin)
foot ~ound-force(li.lht'l British thermal unit (Btu)
Appendix A -Units of Measurement and Physical Data

Table A-6: (Continued)


Factors for Conversion to the English Units
Multiply BY To Obtain
gallon,(Imperial) (gal) 0.1605 cubic foot (ft3)
gallon (Imperial) (gal) cubic inch ( h 3 )
gallon (Imperial) (gal) gallon (US.) (gal)
gallon (U.S.) (gal) cubic foot (8')
gallon (US.) (gal) cubic inch (im3)
gallon (US.) (gal) gallon (Imperial) (gal)
gallon (U.S.) per minute (gpm) (gal/ min) cubic foot per minute (cfm) (ft!/min)
gallon (U.S.) per minute (gpm) (gal/ min) cubic foot per second (ft3/s)
horsepower foot pound-force per second (ft.lbf/s)
hour (h) second (S)
inch (in.) foot (ft)
inch (in) yard (yd)
inch o f mercury (in. Hg) pound-force per-square foot (lbf/ft2)'
inch o f mercury (in. Hg) pound-force per square inch (psi) ( l b ~ i n . ~ )
inch o f mercury (in. Hg) inch of water (in H 2 0 )
inch of water (in. H 2 0 ) pound-force per square foot (lbf/ft2)
inch of water (in. HzO) pound-force per square inch (psi) (lbf/im2)
inch per second (ink) foot per second ( f ~ k )
inch per second (ink) foot per minute (fvmin)
kilogram (kg) pound (Ib)
kilometer per Iiour (kmlh) foot per second (~L's)
kilowatt ( k W ) foot pound-force per second (ft-lbf/s)
kilowatt (kW) British thermal unit per second (Btuls)
kilowatt hours (kW h) rt ~ b f
knor (nautical milc pcr hour) foot per second ((L'S)
knot (nautical mile per hour) milc per hour ( m i h )
liter per sccond (Lls) cubic foot per second (ft3/s)
liter per second (Lls) cubic foot per minute (rt3/rnin)
meter (m) foot (ft)
mil inch (in.)
mile (mi) foot (11)
mile (mi) yard (yd)
mile (nii) mile. nautical
milc per I~our(nii/h) foot per second ( ~ L ' s )
milc per hour (milli) foot per minute (f~'min)
milc per hour (milh) knot (nautical milc per hour)
mile, nautical mile (mi)
millirncter of mercury (mniHs) pound-forcc per square root (lbf/ft2)
niilli~ncterof nlcrcury (nitiiHg) pound-forcc pcr squarc inch (psi) (lbf~ir,.~)
millimctcr of mercury (mmHg) ofwatcr.(iti H2O)
i~~cli
millirncter of water (mmI1,O) pound-forcc per squarc l h t (lbflft')
niillinietcr of\vatcr (1ii1n14~0) pound-li~rccper squarc inch (psi) (lbflin.')
millimctcr ofwatcr (mrnl~l,O) inch oI'w;~tcr(in H?O)
Principles of Smoke Management

Table AB: (Continued)


Factors for Conversion to the English Units
Multiply BY To Obtain
minute (min) 60 second (S)
ounce (avoirdupois) (02) 0.9115 ounce (troy or apothecary) (oz)
ounce (avoirdupois) (02) 0.0625 pound (Ib)
ounce (troy or apothecary) (02) 1 .097 ounce (avoirdupois) (oz)
pascal (Pa) 0.0209 pound-force per square foot (lb0ft2)
pound (Ib) 16 ounce (avoirdupois) (02)
pound (Ib) 0.03 108 slug (slug)
pound per cubic inch ( ~ b l i n . ~ ) 1728 pound per cubic foot (lblft3)
pound per cubic yard (lblyd3) 0.037037 pound per cubic foot (lblft3)
poundal 0.03 109 pound-force (IbQ
pound-force per square foot (lbf/ft2) 0.006944 pound-force per square inch (psi) ( l b f h 2 )
pound-force per square foot (lbf/ft2) 0.1922 inch of water (in. H20)

pound-force per square inch (psi) (1bf/im2) 144 pound-force per square foot (1bf/ft2)
pound-force per square inch (psi) (lbf/im2) 27.68 inch of water (in H20)
slug (slug) 32.174 pound (Ib)
2
square foot (ft ) 0.1111 square yard (yd2)
square foot (ft2) 0.006944 square inch ( h 2 )
square inch (in2) 7.7 168-04 square yard (yd2)
square inch (in ) 2
144 square foot (ft2)
square yard (yd2) 1296 square inch ( h 2 )
square yard (yd2) 9 square foot (ft2)
standard cubic feet per minute (scfm)' 3 . 8 9 10-5
~ slug per second (slugls)

standard cubic feet per minute (scfm)' 0.00 125 pound per second (Ibls)
ton (refrigeration) 12,000.00 British thermal unit per hour (Btuh)
ton (refrigeration) 2594 foot-pound per second (ft Ibfls)
ton, long (2240 1b) 2240 pound (Ib)
ton, metric (t) 2205 pound (Ib)
ton, short 2000 pound (Ib)
watt (W) 0.7376 foot pound-force per second (ft.lbf/s)
watt (W) 9.4788-04 British thermal unit per second (Btuls)
watt (W) 0.7376 foot-pound per second (ft Ibfls)
yard (yd) 3 foot (ft)
I. scfm IS a form o f mass flo\\. rate used for air movemenr and, for this text. ir IS ar 70°F (2I0C) and one atmosphere
Appendix A - Units of Measurement and Physical Data

Table A-7:
Ccnstants
Acceleration of gravity at sea level, g 9.80665 mls2

Gas c21lsianio f air, R 287.0 Jlkg-K


53.34 ft 1bWlbm OR
1716. ft IbWslug OR
0.06858 Btullbm O R

Standard atmospheric pressure, P,,, 101325 Pa


14.696 psi

Table A-8:
Properties of Air in English units1
T P G P v k
(OF) (lbmlft') (Btullbm "F) (Ibn~lftS) (ft2/s) (Btulhr ft OF)
0 0.086 0.239 1.110~10‘~ 0. l 30x 1 0-j 0.0133
Principles of Smoke Management

Table A-9:
Properties of Air in SI units1

(K) (kglm3) (Jlkg 'C) (Wm (m21s) (Wlrn 'C)


200 1.7684 1.0061~10~ 1.3289x 10" 7.5 14x10-~ 0.0 1809

I. Note: Noration listed ar bottom ofTable A-S.

Table A-10:
Thermal Properties of Materials in IP Units
Thermal
Density specific
kjeat Conductivity
P C k kpC
Material lblf? Btullb F Btulh ft OF t3trr2 ft4 OF-' h-'
Aluminum (pure) 1 69 0.2 1 119 4300
Steel (0.5% Carbon) 190 0.1 1 31 1700
Concrete l50 0.18 0.92 25
Brick 162 0.19 0.46 14
Glass, Plate 169 0.19 0.44 14
BrickIConcrete Block 119 0.20 0.42 10
Gypsum Board 59.9 0.26 0.10 1.5
Plywood 33.7 0.60 0.07 1.4
Fiber Insulation Board 15.0 0.30 0.3 1 1.4
Chipboard 19.9 0.30 O.OP7 I .3
Aerated Concrete 31.2 0.23 0.15 1.1
Plasterboard 59.3 0.20 0.092 1. I
Calcium Silicate Board (Marinits XL)' 13.7 0.27 0.064-0.08 I 0.74-0.95

Alurnina Silicate Block ( ~ a o \ v o o l ) '


Glass Fiber Insulation 5.75 0.19 0.02 1 0.0 15
Expanded Polystyrenc 1.25 0.36 0.020 0.0088
I. Trade namcs imply no ~ r d u r s c ~ i i ~h!.
. t iIIIC
l ;lulhors or tlic ~ U ~ I I J I I C ~ S
Appendix A-Units of Measurement and Physical Data

Table A-ll:
Thermal Properties of Materials in SI Units
Thermal
Density specific Conductivity
P C k 103 PC
Material (kglm3 (kJ/kg K) (kWIm K) k w 2 m4 K%
Aluminum (pure) 2710 0.895 206 500
Steel (0.5% Carbon)
Concrete
Brick
Glass, Plate
BrickIConcrete Block
Gypsum Board
Plywood
Fiber Insulation Board
Chipboard
Aerated Concrete
Plasterboard
Calcium Silicate Board (Marinite XL )'
Alumina Silicate Block (~aowool)'
Glass Fiber Insulation 60 0.8 0.037 0.0018
Expanded Polystyrene 20 1.5 0.034 0.0010
I. Trade names imply no endorsement by the authors or the publishers.
Appendix B

Bibliography

Ammosov, F.A., L.I. Karpov, and V.I. Dubovik. 1977. Billings, L. 198 1. Mt. Sinai Medical Center: A new con-
Investigation of hydraulic drag coefficients in stair- cept in smoke control. ASHRAE Journal 23(2): 34-
wells. Fire Pt-eve~tion,Information Bulletin No. I I, 37.
Moscow, USSR, pp. 63-66. Bryner, N.P., E.L. Johnsson, and W.M. Pitts. 1994. Car-
Appleton, I.C. 1976. A model of smoke movement in bon monoxide production in compartment fires-
buildings. CIB Symposium on Co17t1.01of Smoke Reduced-scale enclosure test facility. National
Movetnent it1 Bzrilding Fires, Vol. I . Fire Research Institute of Standards and Technology, NISTIR
Station, Garston Watford, U.K., pp. 127-138. 5568.
Atkinson, B. 1992. Fire safety engineering: The devel- Buchman, A.H. 1998. Modelling post-flashover fires
opment of an Australian standard hot smoke test for with FASTLite. Journal of Fire protectiotl Engi-
large compartments, 2nd CIB Workshop at the Fire neering 9(3): 1- 1 1.
Research Station, Jan 30-3 1, 1992, Borehamwood, Bukowski, R.W. 1978. Smoke measurements in large
U.K. and small scale fire testing. National Bureau of
Standards, NBSIR 78- 1502.
Atkinson, B., and T. Atkinson, T. 1992. Case study
NFPA 92B and the extra dimension. Proceedings of Bukowski, R.W. 1991. Fire models, the future is now!
the I11tert~afiot7al
Fire Safety Engineering Coffer- NFPA Jownal2(85): 60-69.
e w e , 18-20 October 1992, Sydney Australia. Bukowski, R.\V. 1986. Quantitative determination of
smoke tosicity hazard-Practical approach for cur-
Atkinson, B., and R. Marchant. 1992. Case study-The
rent use. Proceedings of the First International
Deeds Road experiments. Proceedings of the Infer-
Symposium on Fire Safety Science, October 7-11,
nafional Fire Safefy Engineering Conference, 18-20
1985, C. Grant and P. Pagni, eds., pp. 1089-1 100.
October 1992, Sydney Australia.
New York: Hemisphere Publications.
Barrett, R.E., and D.W. Locklin. 1969. A computer tech- Butcher, E.G., and A. C. Parnell. 1979. Smoke control in
nique for predicting smoke movement in tall build- fire safety design. London: E. & F. N. Spon.
ings. $n7posi~m on kIo1~eti1etifof Smoke on Escape
Butcher, E.G., T.H. Cottle, and T.A. Baily. 197 1. Smoke
Roufes in Buildings, Watford College of Technol-
tests in the pressurized stairs and lobbies of a 26-
ogy, Watford, Herts, U.K., pp. 78-87.
story ofice building. Building Service Engineer 39:
BenjaminIClarke Assoc. 1984. Operation San Fran- 206-210.
cisco-SmokeISprinkler Test, Operation Life Butcher, E.G., P.J. Fardell, and P.J. Jackman. 1969. Pre-
Safety. diction of the behavior of smoke in a building using
Beyler, C. 1993. Revicw of design of smoke manage- a computer. Symposi~tm017 movemenf of smoke in
ment systems. Jo~r~nc~l
($Fire PtafecfionEngit1eer- escape mires in buildings, pp. 70-75. Watford,
itlg 5(1): 33-34. Herts, England: Watford College of Technology.
Appendix B - Bibliography

Chow, W.K. 1990. Field tests on a staircase pressuriza- American Society of Mechanical Engineers, pp. 76--
tion system installed in a commercial building. 79.
ASHRAE Transactions 96(2). Degenkolb, J.G 1975. Firesafety for atrium type build-
Chow, WK. 1993. Simulation of.the atrium fire environ- ings. Building Standard 44(2): 16-18.
ment in Hong Kong. ASHRAE Transactions 99(2). Degenkolb, J.G. 1983. Atrium. Building Standat-&
chow, W.K. 1995. Simulation of car park fires using 52(1): 7-14.
zone models. Journal of Fire Protection Engineers Dias, C. 1978. Stairwell pressurization in a high-rise
7(2): 65-74. commercial building. ASHRAE Journal 20: 7: 24-
Chow, W.K. 1999. Atrium smoke filling process in 26.
shopping malls of Hong Kong. Journal of Fire Pro- Ditzel, P. 1982. Theater of death-December 30, 1903:
tection Engineering 9(4): 8-30. Chicago's Iroquois Theater becomes a stage for an
Chow, W.K., and L.W. Lam. 1993. Evaluation of a stair- inferno that.kills 602. Firehouse, pp. 52-58.
case pressurization system. ASHRAE Transacrio~a Drysdale, D. 1985. At7 ill/mduction IO fire dyna~nics.
99(2). New York: Wiley.
Chow, W.K., and K.W. Lau. 1995. Field tests on atrium Earp, D. 1998. Natural and mechanical smoke control
smoke control systems. ASHRAE Transactions systems. Seminar Fire Safety-Smoke Control:
lOl(1). Standards and Practice, CIBSE-ASHRAE Group and
Chow, W.K., and C.W. Lo. 1995. Scale modeling studies Colt International. ..
on atrium smoke movement and the smoke filling Edwards, J. 1987. Preventing fire risk in the theatre-A
process. Jorwnal of Fire ProfecfionEngitieers 7(2): lengthy process ofevolution. Fire 77(980): 30.
55-64. Edwards, J. 1987. Theatre fires-The importance of
smoke ventilation. Fire Prevetition (198): 37-38.
Clark, F.R.S. 1963. Visibility in smoke. 3rclSynposizrtr7
Egilsrud, P.E. 1984. Pre\:ention and control of highway
0 1 1 Cotnbus~ibili~~, atid Plastics, Toronto Oct 24-25,
tunnel fires. U.S. Department of Transportation,
1983, National Research Council of Canada.
Report FHWA/RD-83!032.
Clark, F.R.S. 196s. Strategies for improving visibility in
Evans, D.H. 1999. Guide to the 1997 UBC smoke-con-
fires. Canadian Brrilcfir~gDigesr, CBD 246, Janu-
trol provisions. International Conference of Build-
ary.
ing Officials, Whittier. CA.
Clark, J.A. and J.W. Harris. 1989. Stairwell pressuriza-
Evers, E. and A. Waterhouse. 1978. A computer model
tion in a cold climate. ASHRAE Ti-amactions95(l). for analyzing smoke movement in buildings. Bore-
Clarke, F.B. 1997. Physiological effects of smoke: man- hamwood, Herts, U. K.: Building Research Est.
aging escape. ASHRAE Jorrrnal39(3):47-56. Gablin, K.A. 1997. The memorial tunnel fire ventilation
Collins, B.L., M.S. Dahir, and D. Madrzykowski. 1992, test program. ASHRAE Jo~irnal39(2): 26-30.
Visibility of exit signs in clear and smoky condi- Galbreath, M., and G.W. Shorter. 1965. Air pressures
tions. Jout-nal o f /lhrminating Engineering Socie!~:. contribute to a smoke-free corridor, National
Cooper, L.Y. 1988. Ceiling jet-driven wall flows in Research Council of Canada Fire Research Yote
compartment fires. Co117busf.Sci. aiid Tech. 62: No. 3 1966, Reprinted from Fire Fighting in Can-
285-296. ada.
Cooper, L.Y., G.P. Fomey, and J.H. Klote. 1990. Hadjisophocleous, G.V., and M. Cacambouras. 1993.
FANRES-Calculate the flow through a fadduct Computer modeling of comparhnent fires. Joro-nal
forced ventilation system. CCFM Computer Code of Fire Protecfio17Et7,oineet-s 39-52.
Application CCFM. VENTS-Part Ill: Catalog of Hagglund, B., R. Jansson. and K. Nireus. 1985. Smoke
algorithms and subprogranis, Cooper L.Y. and For- filling experiments in a 6 x 6 ~ 6meter enclosure.
ney, G.P., editors, National Institute of Standards Stockholni, Sweden: National Defense Research
and Technology, NISTIR 4344. institute of Sweden.
Cuzzillo, B.R., and P.J. Pagni. 1998. Thermal breakage Hansell, GO., and H.P. Morzan. 1988. Smoke control in
of double-pane glazing by fire. Jour~nlo f f i r e Pro- atrium buildings using depressurization, PD 66'88,
recfiou Etigi17eering9( l ): l - l l . Borehamwood. Fire Research Station.
DeCicco, P.R., and R.J. Cresci. 1975.Smoke and tire Hansell, G.O., and H.P. Morgan. 1994. Design
control in large atrium spaces. ASHRAE fimunc- approaches for smoke control in atrium buildings,
t i o n ~8 l(2): 3 19-335. BR 258. Borchamwood. Hem, U.K.: Fire Research-
Degenkolb, J. 1991. Elevator usage during a building Station.
tire. ASME sy~liposiulnon tire and elevators. Balti- Hasemi, Y. 1985. Analysis of failures of automatic
niorc. IM~.,February 19-20. 1991. New York: sprinklers in actual fires. UJNR Panel on Fire
Principles of Smoke Management

Research and Safety, Joint Meeting May 13-21, Hobson, PJ., and L.J. Stewart. 1973. Pressurization of
1985. Tsukuba, Japan, pp. 794-807. escape routes in buildings, heating and ventilating
He, Y. V. and Beck. 1996. Estimation of neutral plane in research association. Berkshire, U.K.: Brachnell.
high rise buildings. Journal of Fire Sciences 14: Joshi, A.A., and P.J. Pagni. 1991. User's guide to
235-248. Heating and Ventilating Research kssoci- BREAKI, The Berkeley algorithm for br~aking
ation. window glass in a compartment fire. Gaithersburg,
Heselden, A.J.M. 1970. Smoke travel in shopping malls Md.: National Institute of Standards and Technol-
experiments in co-operation with Glasgow fire bri- ogy, NIST-GCR-91-596.
gade-Part 2, F. R. Note No. 854. Borehamwood, Kennedy, W.D. 1997. The influence of the memorial
Herts, U.K.: Fire Research Station. tunnel fire tests on transit tunnel fire emergency
ventilation analysis, Draft of unpublished paper.
Heselden, A.J.M. 1976. Studies of smoke movement
Klem, T.J. 1987. Ninety-seven die in arson fire at
and control at the fire research station. CIB Sympo-
dupont plaza hotel. Fire Journal 81(3): 77, 79, 83,
sium on cont1-01 of smoke movement in building
104, 105.
fires, Vol. I, Fire Research Station, Garston Wat-
Klote, J.H. 1980. Stairwell pressurization. ASHRAE
ford, U. K., pp. 185- 198.
Transactions 86(1): 604-623. Atlanta: American
Heselden, A.J.M. 1978. studies of fire and smoke Society of Heating, Refrigerating and Air-Condi-
behavior relevant to tunnels. BRECP, Boreham- tioning Engineers, Inc.
wood, Herts, U.K.: Building Research Establish- Klote, J.H. 1982. A Computer program for analysis of
ment, smoke control systems. National Bureau of Stan-
Heselden, A.J.M., and R. Baldwin. 1976. The move- dards, NBSIR 82-2512.
ment and control of smoke and escape routes in Klote, J.H. 1982. Smoke movement through a sus-
buildings, BRECP, Building Research Establish- pended ceiling system. National Bureau of Stan-
ment, Borehamwood, Hertfordshire, U.K. dards, NBSIR 8 1-2444.
Heselden, A.J.M., and P.L. Hinkley. 1970. Smoke travel Klote, J.H. 1983. Designing effective zoned smoke con-
in shopping malls experiments in co-operation with trol systems. Building Design and Constructior7
glasgow fire brigade-Part I, F. R. Note No. 832, 24( 1 l ): 90-93.
Fire Research Station, Borehamwood, Herts, U.K. Klote, J.H. 1984. Field tests of the smoke control system
Heskestad, G. 1972. Detem~ination of gas venting at the San Diego VA Hospital. National Bureau of
geometry and capacity of air pollution control sys- Standards, NBSIR 84-2948.
tem at factory mutual research center, FMRC Serial Klote, J.H. 1984. Smoke control for elevators. ASHRAE
No. 2058 1, Factory Mutual Research, Nonvood, Journal 26(4): 23-33. Atlanta: American Society of
Mass. Heating, Refrigerating and Air-conditioning Engi-
Heskestad, G., and J.P. Hill. 1987. Propagation of fire neers, Inc.
smoke in a corridor. P,-oceeditigs of the 1987 Klote, J.H. 1985. Field tests of the smoke control system
ASME-JSME Thermal Engineering Joint Cotfer- at the Bay Pines VA Hospital. ASHRAE Transac-
ence, 1 March 22-27, 1987, Honolulu, H.I. Arneri- tions 91 (I B): 802-8 19. Atlanta: American Society
can Society of Mechanical Engineers, Martro, P. J. of Heating, Refrigerating and Air-conditioning
and Tanasawa, I., Editors, pp. 37 1-3 79. Engineers, Inc.
Klote, J.H. 1985. Computer modeling for smoke control
Hinkley, P.L. 1970. A preliminary note on the move-
design. Fire Safeg Jout.nal9: 181- 188.
ment of smoke in an enclosed shopping mall, Fire
Klote, J.H. 1985. Smoke control in VA Hospitals.
Research Note 806. Borehamwood, Herts, U.K.:
ASHRAE Journal 27(4): 42-45.
Fire Research Station.
Klote, J.H. 1988. Inspecting and testing air moving sys-
Hinkley, P.L. 1970. The flow of hot gases along an tems for fire safety, heatinglpipinglair conditioning,
enclosed shopping mall: A tentative theory, F.R. April.
Note No. 807. Borehamwood, Herts, U.K.:Fire Klote, J.H. 1989. Considerations of stack effect in build-
Research Station. ing fires. Gaithersburg, Md.: National Institute of -
Hinkley, P.L. 197 1. Some notes on the control of smoke Standards and Technology, NISTIR 89-4035.
in enclosed shopping centres, Fire Research Note Klote, J.H. 1991. A general routine for analysis of stack
875. Borehamwood, U.K.: Fire Research Station. effect. Gaithersburg, Md.: National Institute of
Hinkley, P.L. 1975. Work by thc firc research station on Standards and Technology, NISTIR 4588.
- the control of smoke in covcrcd shopping centres. Klote, J.H. 1993. Design of smoke control systems for
Borehamwood. U.K.: Fire Rescarch Station. areas of refuge. ASHRAE Transactions 99(2): 793-
Appendix B- Bibliography

807. Atlanta; American Society of Heating, Refj-ig- can Society of Heating, Reffigerating and Air-Con-
erating and Air-conditioning Engineers. ditioning Engineers, Inc.
Klote, J.H. 1994. Fire and smoke control: An historical Marshall, N.R 1985. The behavior of hot gases flowing
perspective. ASHRAE Journal 36(7): 46-50. within a staircase. Fire Safety Journal 9(3): 245-
Klote, J.H. 1994. Method of predicting smoke move- ,255.
ment in atria with application to smoke manage- Marshall, N.R 1986. Air entrainment into smoke and
ment. Gaithersbig, Md.: National Institute of hot gases in open shafts. Fire Safe07 Journal lO(1):
Standards and Technology, NISTIR 5516. 37-46.
Klote, J.H. 1998. North American approaches to smoke Matsushita, T., and I.H. Klote. 1992. Smoke movement
management, Seminar Fire Safety-Smoke Con- in a corridor-Hybrid model, simple model and
trol: Standards and Practice, CIBSE-ASHRAE comparison with experiments. Gaithersburg, Md.:
Group and Colt International. National Institute of Standards and Technology,
Klote, J.H., and X. Bodart. 1985. Validation of network NISTIR 4982.
models for smoke control analysis. ASHRAE Pans- Mawhinney, J.R., end G.T. Tamura. 1994. Effect of auto-
actions 9 l(2B): 1134-1 145. Atlanta: American niatic sprinkler protection on smoke control sys-
Society of Heating, Refrigerating and Air-Condi- tems. ASHRAE Transactions lOO(1).
tioning Engineers, Inc. McGuire, J.H. 1967. Control of smoke in buildings. Fire
Klote, J.H., and L.Y. Cooper. 1989. Model of a simple Technology 3(4): 281-290.
fan-resistance ventilation systenl and its application McGuire, J.H. 1967. Smoke movement in buildings.
to fire modeling. Gaithersburg, Md.: National Insti- Fire Techr?ology3(3): 122-133.
tute of Standards and Technology, NISTIR 89- McGuire, J.H., and GT. Tamura. 1971. Smoke control
4141. in high-rise buildings, CBC 134. National Research
Klote, J.H., and G.P. Fomey. 1993. Zone fire modeling Council, Ottawa, Canada.
with natural building flows and a zero order shaft Milke, J.A., and J.H. Klote. 1998. Smoke management
model. Gaithersburg, Md.; National Institute of in large space buildings. Melbourne, Australia:
Standards and Technology, NISTIR 525 1. Broken Hill Proprietary Company Limited.
Klote, S.H., and R.IH. Zile. 1981. Smoke movement and Morgan, H.P. 1977. The flow of buoyant fire gases
smoke control on merchant ships. Gaithersburg, beneath corridor ceilings: A theory, FR Note 1076.
Md.: National Bureau of Standards, NBSIR S1- Borehamwood, Herts, U.K.: Fire Research Station.
2433. Mogan, H.P. 1979. Smoke control methods in enclosed
Klote, J.H., D.M. Alvord, B.M. Levin, and N.E. Groner. shopping complexes of one or more stores: 4
1992. Feasibility and design considerations of design summary. Building Research Establishment.
emergency evacuation by elevators. Gaithersburg, Morgan, H.P. 1991. Smoke control in shopping malls
Md.: National Institute of Standards and Technol- and atria, fire & safety in buildings. Symposium in
ogy, NlSTlR 4870. Hong Kong, Hong Kong Polytechnic.
Klote, J.H., and R.H. Zile. l98 1. Smoke movement and Morgan, H.P. 1993. Fire safety-Smoke control: devel-
smoke control on merchant ships. NBSIR 81-2433. opments in European Standards, Seminar fir5
Gailhel-sburs, Md.: National Bureau of Standards. safety--Smoke control: Standards and practice.
Klote, J.H., H.E. Nelson, S. Deal, and B.M. Levin. ClBSE-ASHRAEGroup and Colt International.
1992. Staging areas for persolis with mobility liiiii- Morgan, H.P. 1998. Sprinklers and fire safety design.
tations. Gaithersburg, Md.: National Institute of Fire Safety Engirleer-irlg 5(1): 16 - 20.
Standards and Technology, NISTIR 4770. Morgan, H.P., and J.P. Gardner. 1990. Dcsign principles
Lamming, S.D., and J. Salmon. 1994. Wind data for for smoke ventilation in enclosed shopping centres,
design of smoke control systems, Final Report 816- BR 186. Borehamwood, Herts, U.K.: Fire Research
TRP. Atlanta: American Society of Heating, Refrig- Station.
erating and Air-Conditioni~gEngineers, Inc. Morgan, H.P., and (3.0.Hansell. 1987. Atrium build-
Lougheed, G.D. 1997. Expected size of shielded fires in ings: Calculating smoke flows in arria for smoke-
sprin klcred office buildings. ASHRA E T,-nn.suctiot~s control design. Fire Safely Jozo-nal 12(1987): 9-35.
103j l ). -Atlanta: American Society of Heating. Morgan, H.P., and N.R.~arshall. 1975. Smoke hazards
Refrigerating and Air-conditioning Engineers, Inc. in covered, ,multi-level shopping malls: An experi-
Lougheed, G.D., and D.W. Carpenter. 1995. Probability mentally-based theory for smoke production. Bore-
of occurrence and espectcd size of shielded fires in hamwood, U.K.: Building Research Establishment.
sprinklcrcd buildings: Phase 2 full scale tire tests. Morgan, H.P., and N.11. Marshall. 1978. Smoke hazards
Final l1spo1-l01' ASIIRAE RP-S38. Atlanta: Amcri- in covercd, mulli-lcvcl shopping malls: A method
Principles of Smoke Management

of extracting smoke from each level separately. sium on Control of Smoke Movement in Building
Borehamwood, U.K.: Building Research Establish- Fires, Vol. I, Fire Research Station, Garston Wat-
ment. ford, U.K., pp. 99- 126.
Morgan, H.P., and GO. Hansell. 1987. Atrium build- Steckler, K.D. 1989. Fire induced flows in corridors-A
ings: Calculahg smoke flows in atria for smoke review of efforts to model key features. Gaithers-
control design. Fire Safety Journal 12: 9-35. burg, Md.: National Institute of Standards and
Morgan, H.P., and N.R. Marshall. 1975. Smoke hazards Technology, NIST3R 89-4050.
in covered, multi-level shopping malls: An experi- Tamura, G.T. 1969. Computer analysis of smoke move-
mentally based theory for smoke production. CP ment in tall buildings. ASHRAE Transactions 75(2):
48/75. Borehamwood, Hertfordshire, U.K.: Build- 8 1-93. Atlanta: American Society of Heating,
ing Research Establishment, Fire Research Station. Refrigerating and Air-conditioning Engineers, Inc.
Morgan, H.P., and N.R. Marshall. 1978. Smoke hazards Tamura, G.T. 1972. Pressure difference for a nine-story
in covered, multi-level shopping malls: A method building as a result of chimney effect and ventila-
of extracting smoke from each level separately. tion system operation. ASHRAE Transactiom
Borehamwood, Hertfordshire, U.K.: Building 72(1): 180- 189. Atlanta: American Society of Heat-
Research Establishment. ing, Refrigerating and Air-Conditioning Engineers.
Morgan, H.P., N.K. Marshall, and B.M. Gladstone. Inc. .
1976. Smoke hazards in covered multi-level shop- Tamura, G.T. 1974. Experimental studies on pressurized
ping malls: Smoke studies using a model 2-story escape routes. ASHRAE Tratuactiom 80(2): 224-
mall. Borehamwood, U.K.: Building Research 237. Atlanta: American Society of Heating, Refrig-
Establishment. erating and Air-conditioning Engineers, Inc.
Morgan, H.P., N.R. Marshall, and B.M. Gladstone. Tamura, G.T. 1978. Experimental studies on exterior
1976. Smoke hazards in covered, multi-level shop- wall venting for smoke control in tall buildings.
ping malls: Smoke studies using a model 2-story ASHRAE Transactiotis 84(2): 204-2 15. Atlanta:
mall. CP 45/76. Boreham\vood, Hertfordshire, American Society of Heating, Refrigerating and
U.K.: Building Research Establishment. Air-conditioning Engineers, Inc.
Moulen, A.W., and S.G. Grubits. 1975. Stairwell pres- Tamura, G.T. 1978. Exterior wall venting for smoke
surization in a 26-story building. North Ryde, control in tall office buildings. ASHRAE Jotit-tin/
N.S.W. Australia: Experimental Building Station. 20(8): 43-48. Atlanla: American Society of Heat-
Moulen, A.W., and S.G. Grubits. 1975. Stainvell Pres- ing, Refrigerating and Air-conditioning Engineers.
surization in a 22-story building. North Ryde, Inc.
N.S.W. Australia: Experimental Building Station. Tamura, G.T. 1980. The performance of a vestibule pres-
Quintiere, J.G. 1989. Fundamentals of enclosure fire surization system for the protection of escape routes
"zone" n~odels.Journa/ of Fire Protectioti Engi- of a 17-story hotel. ASHRAE Transactiotis 86(1):
neet-itig l(3): 99- 1 19. 593-603. Atlanta: American Society of Heating.
Quiter, J.R. 1996. An application of performance based Refrigerating and Air-conditioning Engineers, Inc.
concepts at the stratosphere tower Las Vargas, Tamura, G.T. 1982. A smoke control system for high-
Nevada. Presented at NFPA Research Foundation rise office buildings. ASHRAE Jourtial 24(5): 29-
Syn~posiun~, San Francisco, Calif.. June 26, 1996. 32. Atlanta: American Society of Heating, Refriger-
Said, M.N.A. 1988. A review of smoke control models. ating and Air-Conditioning Engineers, Inc.
ASHRAE Journal 30(4): 36-40. Tamura, G.T., and A.G. Wilson. 1967. Pressure differ-
Sander, D.M. 1974. FORTRAN IV program to calculate ences caused by chimney effect in three high build-
air infiltration in buildings. National Research ings. ASHRAE Transac~iom73(2): 111. 1-111. 10.
Council Canada, DBR Computer Program No. 37. Atlanta: American Society of Heating, Refrigerat-
Sander, D.M., and G.T. Tamura. 1973. FORTRAN IV ing and Air-conditioning Engineers, Inc.
program to stimulate air movement in multi-story Wakamatsu, T. 1968. Calculation of smoke movement
buildings. National Research Council Cauada, DBR in buildings-1st Report, Building Research Insti-
Computer Program No. 35. tute, Japan.
Schmidt, W.A., and J.H. Klote. 1952. In case of fire- Wakamatsu, T. 197 1 . Calculation of smoke movement
Use the stainvclls, Elevators aren't safe. Spec$i~ing in buildings-2nd Report, Building Research Insti-
Engitieer 47(5). tule, Japan.
Shannon, J.M.A. 1976. Compulcr analysis of [he mo\/e- Wakamatsu, T. 1976. Unsleady-state calculation of
ment and conttol of smoke in buildings with s~nokemovement in an actually tired building. CIB
mccha~iical and natural vcntilat~on CIB Sympo- Syn~posiumon Control of Smoke Movement in
Appendix B-Bibliography

Building Fires, 1: 81-98. Garston Watford, U.K.: Pa.: American Society for Testing and Materials,
Fire Research Station. pp. 168-193.
Wakamatsu, T. 1977. Calculation methods for predicting Yoshida, H., C.Y. Shaw, and GT. Tamura. 1979. A
smoke movement in building fires and designing FORTRAN IV program to calculate smoke concen-
smoke control systems, fire standards and safety, trations in a multi-story building. Ottawa, Canada:
ASTM STP-614, A. F. Robertsan, ed. Philadelphia, National Research Council.
Appendix C

Calculation of Elevator Evacuation Time

NOMENCLATURE Subscripts
2 T = end of leveling car motion (also end of travel)
a = acceleration, Ws (nl/s2)
J = number of elevators 1 = end of constant acceleration motion
m = number of round trips 2 = end of transitional acceleration motion
N = number of people entering or leaving the elevator 3 = end of constant velocity motion
Nd,,: = number of people entering or leaving the elevator 4 = end of transitional deceleration motion
during the dwell time 5 = end of constant deceleration motion
S = distance, ft (m)
During building fires, elevators are almost always
ST = total floor to floor travel distance for trip, fi (m) taken out of service, and vertical evacuation is by the
t = time, s (S) use of stairs. Elevators have been used for emergency
I, = elevator evacuation start-up time, s (S) evacuation in a few unique situations (subway stations
several stories underground, luxury apartments, and
~d = time for elevator doors to open and close, s (S)
towers). The Life Safety Code (NFPA 1977) recognizes
tdL, = dwell time for elevator doors, s (S)
elevators as a second means of egress for towers.
t, = e\.acuation time, s (S) This appendix presents a detailed method of analy-
t,, = time for leveling of elevator car, s (S) sis of people movement by elevators during emergency
t, = time for A' people to enter elevator car, s (S) building evacuation, based on principles of elevator
t,, = time for one person to enter elevator car, s (S) engineering (Strakosch 1983). Bazjanac (1977) and
I, = travel time from ele\.ator lobby to outside or to Pauls (1977) have developed methods of calculation of
other safe location, s (S) evacuation time by elevator, but the method presented
= time for elevator car to make a round trip, s (S) here incorporates more detail about elevator motion and
t,.
elevator loading and unloading. The ELVAC computer
t, = standing time, s (S)
program by Klote and Alvord (1992) is based on the
t,, = time for A1people to leave elevator car. s (S) analysis of this appendix, and an example ELVAC anal-
!,, = time for one person to leave elevator car. s (S) ysis is included in this appendix.
V = \.eloci t): ftk (mk) The sequence of elevator operation for emergency
V,,, = normal operating velocity. Ws ( d s ) evacuation is complicated and has many possible varia-
a = basic transfer inefficiency tions. The following general sequence is presented to
provide a framework for the method of analysis pre-
p = total transfer inefficiency. ,U = a + E + ;/
sented in this paper. Upon activation of emergency
E = door transfer inefXcicncy evacuation, elevators in normal service will go to a dis-
y = other transfer inefticiency charge floor where any passengers on the elevators will
11 = trip inefficiency exit. This discharge floor may either lead to the outside
Appendix C-Calculation of Elevator Evacuation Time

or lead to an area of relative safety where people~may elevators being out-of-service depends on a rlumber of
stay during the fire. The elevators .will make a number factors, including the age of the elevators and the quality
of round trips to transfer occupants from other floors to of maintenance. Because the out of service condition
the discharge floor. During evacuation, the elevators can significantly increase elevator evacuation time, any
may be under a special emergency evacuation mode of analysis of elevator evacuation should take this condi-
automatic control or under manual control. tion into account.
The evacuation time addressed in this paper is an
idealized time for people movement that does not START-UP TIME
account, for the complex human behavior that often The elevator evacuation start-up time is the time
occurs during emergencies. It is believed that the analy- from activation to the start of the round trips that evacu-
sis of this paper is about as accurate as that for evacua- ate people. For automatic elevator operation during
tion by stairs. evacuation, a simple approach is to start elevator evacu-
ation after aH of the elevators have been moved to the
EVACUATION TIME discharge floor. For this approach, the start up time, I,
Analysis of people movement during elevator evac- consists of the time for elevators to go to the discharge
uation must take into account the number and arrange- floor plus the time for the passengers to leave the eleva-
ment of elevators in a building. Genkrally, elevators are tors. This can be expressed as
located in groups of up to eight elevators. Elevators in a
group
- . arc located near each other and are controlled
together to efficiently move people. Arrange~nentsof where rr is the travel time for the elevator car to go from
elevator groups are discussed later. The method of anal- the farthest floor to the discharge floor. l,, is the time for
ysis and the computer program of this paper are for the passengers to leave the elevator, td is the time for the doors
calculation of the evacuation time for one group of ele- to open and close once, and p is the total transfer ineffi-
vators. For buildings with multiple groups of elevators, ciency. These terms are discussed in detail later.
the approach presented in this paper can be applied sep-
An alternative to the simple approach discussed
arately to each group of elevators.
above consists of starting the evacuation operation indi-
Ideally, the time to evacuate a number of people vidually for each elevator when it reaches the discharge
using one group of'elevators consists of the sum of all floor. This alternative could result in slightly reduced
the round trip times divided by the number of elevators evacuation time. However, this alternative is not dis-
plus the time needed to start up the elevator evacuation cussed further here because of its limited benefit and
and the travel time from the elevator lobby to the out- added complexity.
s ~ d (or
c to another safe location). Accounting for ineffi-
For manual elevator operation, the tinie for ele\.ator
ciencies of elevator operation, this e\acuation lime can
operators to be alerted and then get to the elevators must
be expressed as
be included in the estimate of start-up time. This addi-
tional time may be considerably greater than that calcu-
lated from Equation ( C l ) .

ELEVATOR R O U N D TRIP TIME


where t,? is the timc for round trip j. 111 is the number of The round trip starts at the discharge floor and con-
round trips, J is the number of elevators. 11 is the trip ineffi- sists of the following sequence: elevator doors close. car
ciency, t,, is elevator evacuation start-up time, and 1 , is the travels to another floor, elevator doors open, passengers
travel ti~ilefrom the elevator lobby to the outside or to enter the car, doors close, car travels to discharge floor,
another safe location. The round trip time depends on the doors open, and passengers leave the car. The round trip
travel timc of the elevator and on the nulnbcr ol'pcople car- time, t,, can be written as
ried by the elevator, as discussed latcr. The travcl timc from
the elevator lobby to a safe location can be evaluated by
conventional~n~ethods of people niwen~cnt(Chapter 4).
The trip inefticiency accounts tbr trips to empty floors and where t, is the standing tinie and f7-isthe travel time for one
trips to pick only a few stragglers. The elevator evacuation way of the round trip. This equation is based on the eleva-
start-up time is discussed in the next section. tor only stopping at one floor to pick up passengers. It is
The nunibcr of elevators, .J, used in Equation ( C l ) expected that most elevators will fill up on one lloor and
may bc less than the number ofclcvators in the group to procccd to the discliary floor. Whnt coristitutcs a full ele-
account h r out-of-service elevators. Thc probability of' vator is disci~sscdlater. If an clcvator stops to pick up pas-
Principles of Smoke Management

sengers at more than one floor during a round trip, chosen to be 0.05 for hospital elevators. Generally, for
Equation (C3) can be modified accordingly. However, the o f i c e buildings, y is taken as zero.
trip hiefficiency accounts for such multiple stops.
The time, fd,for the doors to open and close depends
on the width and type of the doors, as listed in Table
STANDING TIME
C-l. The kinetic energy of closing doors is limited by
The standing time is the sum of the time to open elevator safety codes and is usually not more than 0.29
and close the elevator doors twice, the time for people to J (7 ft poundal'6). This is why doors from different man-
enter the elevator, and the time for people to leave the ufacturers take about the same time to open and close.
elevator. Considering transfer inefficiencies, the stand- Types of elevator doors are shown in Figure C3. Door
ing time for a round trip can be expressed as operating time is important because of the many times
that doors open and close during an evacuation. Further,
an elevator cannot leave a floor bzfore the doors are
where p = a + E t y. closed and locked, and passengers cannot leave an ele-
vator until the doors are fully opened or nearly fully
The basic transfer inefficiency, 4 allows for round-
ing off of probable stops, door operating time, door opened. Generally, elevator doors do not open until the
starting and stopping time, and the unpredictability of car has stopped and is level with the floor. However,
people. Typically, a value of 0.10 is used for thebasic some center opening doors start opening while the car
transfer inefficiency for commonly accepted arrange- is leveling, and the times listed in Table C- l should be
ments of elevator groups, as illustrated in Figure C I . For reduced by one second for these preopening doors.
each of these arrangements, the configuration of the ele- The time, fi, for pcople to enter an elevator depends
vator lobby is such that passengers can recognize which on the number, N,of people entering and on the door
elevator has arrived and get on the elevator without
operation. As previously stated, it is expected that most
excessive delay. Further, these lobbies have sufficient
elevators will fill up on one floor and proceed to the dis-
space so that people exiting one elevator will have a
charge floor. However, elevators will be less than full
minimal impact on the flow of people leaving another
elevator. when there are not enough people waiting in the lobby
to fill an elevator or elevators. Thus, the analysis must
Arrangements of elevator groups other than those
include partially filled elevators. Strakosch (1983) has
commonly accepted can be less efficient and require an
observed elevator loadings in which passengers do not
increased value of the basic transfer inefficiency. These
unusual arrangements include cars separated, too many board an elevator and choose to wait for the next one.
cars in a line, angular arrangement, and cornered These observed values are based on 0.22 m2 (2.3 ft') of
arrangement (Figure C2). Separation of elevators results floor space in the elevator car per person. It should be
in increased boarding time for people waiting by one noted that the ASME A17.1 (1987) elevator standard
elevator who have to walk to another when it arrives. If allows a maximum loading at 0.14 m2 (I .S ft2) per per-
the separation is too large, some passengers choose to son, but this high density is not achieved in normal prac-
let elevators go by without boarding. Use of too many tice. The observed values of Strakosch are suggested as
elevators in a line has similar inefficiencies. With the the number of persons in a full elevator car, and these
angular arrangement [Figure C ~ C ]cars , at the narrow loadings are listed in Tables C-2 and C-3.
end tend to be too close together while cars at the wide
When elevator doors open, the doors remain open
end tend to be too far apart. In the cornered arrangement
(Figure C2d), passengers entering or leaving corner cars for a least fixed time, referred to as the dwell-time, rd,,-
tend to interfere with each other. The time that the door is open can be extended beyond
The door inefficiency, E. is used to adjust for any the dwell-time by blocking of the light beam across the
increase in transfer time over that of a 1200 mm (48 in.) door opening or by pushing the door safety edge. The
wide center opening door. Values of E are listed in Table time, ri, for N people to enter an elevator car can be
C-1. The inefficiency, y, is used to account for any other
expressed as
inefficiencies in people transfer into or out of elevators,
such as increased movement times within an elevator I6.The poundal is the unit of force in the pound
car due to an unusual elevator car shape or limited phys- mass-poundal system of units, and one poundal equals
ical capability of passengers. For example, y often is 0.03 1 1 pounds force.
~ ~ ~ eC -
n calculation
d k of Elevator ~vacuationT~A

B l A U
B = 1 S A , but not less thzn 1.8 m (6ft)
1.5Ai; B r 2A
(a) Two Car Group (c) Three Car Group
(b) TWO Car Group

B = 1.5A, but not less than 2.4 m (8ft)


(e) Four Car Group
(d) Three Car Group

-r- - l
2- - - _--_J

Both Ends of
I Open or B
L
B
A
I
-I
Closed
,J----- - Lobby Open

1.5As B c 2 A B = 2A
(f) Four Car Group (g) Six Car Group

-r
B
- 1
, Open or
Closed
Both Ends of
Lobby Open
L A _1

1.75As B 52A B = 2A
(h) Six Car Group (i),Eight Car Group
Figure C1 Conzmor71~~
accepted elevafor- a~ar7gemer71s.
Principles of Smoke Management

ti = { tdw forNI2.
tdH,+ tio(N- Ndw) for N > 2
(C51
t,, for N I 2
{ldw+ tuo(N- Ndl) for N > 2. (c61

where Nd, is the number of people entering the elevator


For the computer program of this paper, the dwell-
during the dwell time, and tio is the average time for one
person to enter the elevator. The number of people entering time is taken to be 4 seconds, the average time for one
the elevator during the dwell time is the tern (td,JIio), Passenger to enter an elevator is taken to be 1 second,
rounded down to the nearest integer. The time for Npeople and the average time for one Passenger to leave an ele-
to leave an elevator can be expressed in a similar manner. vator is taken to be 0.6 seconds.

Table C-l:
Door Operating Time and Transfer Inefficiency

~ i r n e to
' Open Door Transfer
Width and Close Inefficiency
Door Type mm (in.) td (S) E
Single-Slide 900 (36) . 6.6 0.10
Two-Speed
center-opening 2
Single-Slide
Two-Speed
center-opening 2
Two-Speed
center-opening2
Two-Speed
center-opening 2
Two-Speed
center-opening 2
Two-Speed: Center-

opening 2
I . Time 10 open and close doors includes 0.5 second for car to stafl.
2. When preopening can be used, the time to open and close these doors can be reduced by I second.

(a) Cars Separated (b) Too Many Cars in a Line

(c) Angular Arrangement (d) Cornered Arrangement


Figure C2 C't~tis~ial
elevator a1~rn7gen1enfs
t-esulfingin inejjicienf people tno\~ernet7f.
'
Appendix C- calculation of Elevator Evacuation Time

-
-
2 L
(d) Center Opening Doon
(a) Single-Sliding Doa

I
(b) Two-Speed Sliding Doom

00
0
I L
(cl Two-Speed. Center-Opening O m
(e) Venical Bipaning Doom

Figure C3 Types of elevator doors.

Table C-2:
Car Size and Observed Loading in SI Units
Car Inside (mm) Observed
2
Capacity kg (Ib) Wide Deep Area (m ) Loading' (people)
1200 (2640) 2100 1300 2.73 10
1400 (3080) 2100 l450 3.05 12
1600 (3520) 2100 1650 3.47 16
1600 (alt.) 2350 1450 3.4 1 16
1800 (3960) 2100 1800 3.78 18
1800 (alt.) 2350 1650 3.88 18
2000 (4400) 2350 1800 4.23 20
2250 (4950) 2350 1950 4.58 22
2700 (5940) 2350 2150 5.05 25
I. Sec footnote on Table C-3.

Table C-3:
Car Size and Observed Loading in English Units
Car Inside (in.) Observed
Capacity (Ib) Wide Deep Area (ft2) ~ o a d i n (people)
~'
2000 68 5l 24.1 8
2500
3000
3500
3500 (alt.)
4000
4000 (alt.)
4500
5000

I. This loading is givcn by Strakosch (1953) as (hat for which passenprs will not board an elevator and
choose to wait for rhc nuxl one.
Principles of Smoke Management

TRAVEL TIME The distance traveled by the end of transitional


Elevator motion is depicted in Figure C4 for most acceleration is
trips. Motion starts with constant acceleration, followed
by transitional acceleration and constant velocity
motion. Constant acceleration ends when the elevator
reaches a predetermined velocity, which is typically
about 60% of the normal operating velocity (V, = 0.6 The one-way travel time is
V,,,). For office buildings, the normal operating velocity
is generally from 1 to 9 m/s (200 to 1800 fpm), and
acceleration is from 0.6 to 2.4 m/s2 (2 to 8 ft/s2). Decel-
eration has the same magnitude as the acceleration, and The leveling time must be added to the above time
the total acceleration time equals the total deceleration
.. to get the total travel time for a one-way trip.
time (l2 = 1, - I ~ ) The
. method of analysis that follows
takes ad\-antageof this symmetry.
Analysis of ele\-ator motion that reaches the normal
operating velocity is presented nest. For short trips, ele- Usually elevators do not stop exactly at the desired
vators do not always reach the nonnal operating veloc- floor at the end deceleration, so the elevator must be
ity, and methods of analysis for these short trips are moved slowly up or down to get it nearly level with the
presented later. floor. Unless there are better data, a leveling time, th, of
0.5 seconds is suggested.
Motion Reaching Normal Operating Velocity
The time to complete constant acceleration motion Motion Reaching Transitional Acceleration
(going to point I on Figure C4) is If the trip is too short for the elevator to reach the
normal operating velocity, but it reaches transitional
acceleration, the velocity is represented by Figure C5a.
The time, 1 1 , and distance, S I , traveled during constant
The distance traveled during constant acceleration acceleration are given by Equations (C7) and (CS). The
is velocity at the end of transitional acceleration is

The time at the end of transitional acceleration is


Transitional accsleration is approximated by con-
sidering rhe product of velocity and acceleration to be a
constanr. The time to reach the end of transitional accel-
eration (point 2 of Figure C4) is
The one-way travel time is

Transitional Transittonal
/ Leveltng
Accelerabon Decelerat~on\ \
Constant I
Accelerat~on

0 t, t2 t, t.4 t 5 t6
Time
Figure C4 Id o c i of
~ ~elevarot- reaching normal operating ve1ocif.v. V,,.
Appendix C-Calculation of Elevator Evacuation Time

Transitional Transitional
Acceleration Deceleration Leveling

Constdnt Constant

-l
Acceleration Deceleration

-
0 fl f2 4 5 fr 0 t1 5 tr
Time Time

( a ) Car Reaching Transitional Acceleration (b) Car Not Reaching


Transitional Acceleration
Figure C 5 V e l o c i of
~ elevators not reaching normal operating velociv

Motion Not Reaching 89.1 s (the same as calculated in Example Cl). In order
Transitional Acceleration to move 90 people from floor 21, the elevator trips are
considered to consist of five trips with a full car (16 peo-
When the trip does not go beyond constant acceler-
ation, the motion is illustrated in Figure C5b. The one- ple) plus one trip of a partially filled (10 people) car.
way travel time is The time for the partially filled round trip is 78.6 s (not
shown in Table C-5). Thus, the total trip time to move
90 people from floor 21 is 5(89. l) + 78.6 = 524.1 S. This
time is listed under the heading "Time per Floor" for
floor 2 1 in the table.
COMPUTER EVACUATION ANALYSIS On floor 10 of this example, 3% of 90 people are
The computer program ELVAC, written in Quick evacuated-this is rounded up to three people. Because
BASIC, calculates evacuation time for one group of ele- this is done by one trip, the round trip time of 45.8 s
vators. For buildings with multiple groups of elevators, listed in Table C-5 is for moving three people rather than
the program can be used a number of times to calculate the full car load of sixteen. The total round trip time of
the evacuation time for each group. 5395.6 s is the sum of all the round trips to move people
Discussion of Table C-5 provides insight into the from all the floors. The evacuation time of 1258.3 S
computer program. The round-trip time for floor 21 is using five elevators was calculated from Equation (Cl).
Principles of Smoke Management

Table C-4: -

Parameters for Example C3


Number of stories 21
Number ~f elevator cars 5
Number of people per floor 90
Percent of people evacuating by elevators from floors 2 to 10 3
Percent of people evacuating by elevators from floors 11 to 21 100
Height between floors 3.2 m (10.5 ft)
Operating velocity of'elevator car, V, 3.0 m/s (590 fpm)
Car acceleration, a 1.20 m/s2 (3.94 ft/s2)
Other transfer inefficiency, y 0
Trip inefficiency, g 0.10
Car full load 16 people

Table C-5:
Elevator Trip and Evacuation Time Calculated by ELVAC Computer Program

One-way
Number of
Elevation Trip Round Trip People on Percent
Round Time per
Floor (m) (ft Time (S) Time (S) Floor Elevator Evacuation Trips ~loor-(S)
21 64.0 210.0 24.4 89.1 90 100 6 524.1

Total round trip time (S) = 5395.6


Start-up time (S) -
- 41.3
-
Time to get outside after leaving the elevator (S) 30.0
Evacuation time using five elevators (S) = 1258.3
'
Appendix C-Calculation of Elevator Evacuation Time

Example C1 Round Trip in IP Units


A 3500 Ib elevator in an office building makes a round trip fiom the ground floor to pick up a h11 ioad of passengers fiom the

I 21st floor and return them to the ground floor. The operating velocity is 600 fpm with an acceleration of 4 ft/s2, and the elevator
door is 48 in. wide, center-opening. The distance between floors is 10.5 fl, and the total travel distance, SF is 210 ft.
From Table C-3, the number of people in the full elevator is approximated at 16. From Table C-l, td is 5.3 S, and g is 0. The ele-
vator shape is not unusual and the passenger capability is normal, so yis 0. The total transfer inefficiency is
p = a + E + y = 0.10 + 0 + 0 = 0.10.

I From Equation (C5), the time for 16 people to enter the elevator is ti = N = 16

From Equation (C6), the time for 16 people to leave the elevator is I, =
S.

4 + 0.6(N- 6) = 4 + 0.6(16 - 6) = 10 S.

From Equation (C4). the standing time is


Is = (ti+tU+2td)(l + p ) = ( 1 6 7 10+2(5.3))(1 +O.l) = 40.26s.

lI The normal operating velocity is V,,, = 600-- ft I m i n = 10ft/s.


min 60 s

Consider VI is 60% of V,,,, then V, = 0.6VnI = 0.6(10) = 6 ft/s.

From Equation (C7), the time at the end of constant acceleration is f, = Vl/a = 6/4 = 1.5 s

Equation (C8), the distance traveled during constant acceleration is S1 =


v:
- --
2a 2(4)
= 4.5 f t .

From Equation (C9), the time at the end of transitional acceleration is r2 = i 1 + L= 1 .S + ( 1 0 ) ~ - ( 6 ) ~ = 2 , 8 3 ~ .


2a VI 2(4)(6)

1 From Equation (CIO), the distance traveled by the end of transitional acceleration is

11 The one-way travel time is calculated from Equation (C1 I):

lI The total travel time is calculated from Equation (C12): I T = t5 + I,, = 23.6 + 0.5 = 24.1 s

The round trip time is calcu!ated from Equation (C2): I, = 21, + 1, = 2(%. 1) + 40.3 = 88.5 s
Principles of Smoke Management

7 , C2 -
Examnle Round Trin Time in S1 Units -- -.. ~. lI
11 A 1600 kg elevator in an ofice building
r-----

11
-~ ~ ~ - - - -

a mund trip fmm the ground floor to pick up a fdl load of passengers horn h e

Il 2 1st floor and return them to the ground floor. The operating velocity is 3 d s with an acceleration of 1.2 d s 2 , and the elevator
door is 1200 mm wide, center-opening. The distance between floors is 3.2 m, and the total travel distance, SF is 64 m.

From Table C-2, the number of people in the full elevator is approximated at 16. From Table C-I, !;is 5.3 S,and E is Q. >e ele-
vator shape is not unusual and the passenger capability is normal, so y is 0. The total transfer ineficiency is
Il
p = a + ~ + y = O . 1 0 + 0 += 0 0.10.'

1 From Equation (CS), the time for 16 people to enter the elevator is ti = N = 16 S .
II
From Equation (C6), the time for 16 people to leave the elevator is

tU=4+0.6(N-6)=4+0.6(16-6) = 10s.

I From Equation (C4), the standing time is


1, = ( t i + t,, + 2td)(1 + p ) = (16 + 10 + 2(5.3))(1 + 0.1) = 40.26 S.

11l
Consider V1 is 60% of V,,,, then V, = 0.6Vn, = 0.6(3) = 1.8 d s .

From Equation (C7), the time at the end of constant acceleration is t l = Vi/a = 1.8/1.2 = 1.5 S.
II
I From Equation (Cgj, the distance traveled during constant acceleration is
Il
1 From Equation (C9), the time at the end of transitional acceleration is

I From Equation (CIO), the distance traveled by the end of transitional acceleration is

(1 The one-way travel time is calculated from Equation (Cl l): II

I The total travel time is calculated from Equation (C12): t T= t5 + th = 23.9 + 0.5 = 24.4

The round trip time is calculated from Equation (C2): t, = 21, + t, = 2(24.4) + 40.3 = 89.1 s
S

ll -- Example,C3 Elevator Evacuation Time


Estimate the time needed for elevator evacuation of all the people from the upper I I floors of a 21-story building to the outside
of the building. Additionally, 3% of the people on the other~floorsare included in the elevator evacuation. The rest of the people
on the lower floors will use the stairs. Each floor is occupied by 90 people. A group of six 1600 kg (3500 Ib) elevators are used -
for the evacuation, and the ele\.ator doors are 1200 mm (48 in) wide, center opening. One of the six cars is considered out-of-
service; thus, only five of the cars are used in the analysis. Other parameters of this example are listed in Table C-4.
.
Table C-5 lists trip times and the evacuation time calculated by ELVAC. The evacuation time using five elevators is calculated
at 1258 s or about 2 1 minutes.
Appendix D

Application of CONTAMW

INTRODUCTIOP 5. Attachment 2 (Floor Layouts)--provides floor layouts


with zones and airflow paths identified in this example
CONTAMW is a computer program developed by
application.
George Walton at the National Institute of Standards and
Technology. The program is a multizone indoor air qual- 6. Attachment 3 (Simulation Results)-provides the
ity and ventilation analysis program that is useful in a tabular and schematic results of the simulation for this
variety of applications. For smoke management pur- example application. An example of the shaft report
poses, the program can be used to help calculate room- generated by the program is also included.
to-room airflows and pressure differences induced by
mechanical and natural forces. SECTION 1: DESCRIPTION OF EXAMPLE
The CONTAMW computer program is described in
What is the purpose of this appendix? this appendix by way of the following example.
Building Description. The building has 12 stories
This appendix describes the use of CONTAMW by and a roof. The overall dimensions of the building are
way of an example application. The data for this exam-
246 ft (75 m) by 328 ft (100 m) by 157 ft (48 m) (in
ple application are selected as an example of illustrating
height). The building has two stairwells that have the
the input and output form of CONTAMW and are not
dimensions 7.9 ft (2.4 m) by 33 ft (10 m). Each stair
intended to represent recommended values for this pro-
door is 2.9 ft (0.91 m) in width and 7 ft (2.13 m) in
gram.
height. The center of the doorknob is 3 in. (0.076 m)
from the edge of the door, and gaps around the closed
How is this appendix organized? doors are 0.125 in. (0.00305 m) (vertical gaps) and 0.25
This appendix is organized as follows: in. (0.00636 m) (horizontal gaps). The interior doors of
both stairs are open on the 1st floor. All other stair doors
Section I (Description of Example)-provides a brief are closed. Wall and floor construction in this building is
narrative of the example application. "loose."
Section 2 (Data Input Process+provides a detailed Smoke Management Description. The building
description of how to input data into the CONTAMW has zoned smoke control and two pressurized stairwells.
program using the input data from the example appli- The 6th floor is the smoke zone and is exhausted at a
cation. rate of 30,000 scfm (14,158 SUS).The 5th and 7th floors
are pressurized at this same flow rate. The stairwells are
Section 3 (Data Output Process+provides a descrip- each pressurized by a fan on the roof. The minimum and
tion of how to run simulations in CONTAMW and maximum allowable pressure differences for the stair-
how to read the output data. wellare0.05 in. W.C. (12.5 Pa)and 0.36in. W.C. (90 Pa).
Attachment I (Input Databprovides input data used InteriorIExterior Environmental Description.
in this example application. The building temperature is 72°F (22°C). The stairwell
Appendix D - Application of CONTAMW

Figure D1 Simple schematic offirstf200~

tempirature is 76°F (24OC) on the 1st floor and refer to the CONTAMW Help Program or the
increases linearly to 90°F (32OC) on the 12th floor. The CONTAM96 User's Manual.
outside temperature is 85°F (20°C), representing a sum- Note: Since this is a schematic diagram, the draw-
mer condition. The wind speed is 10 mph (4.5 d s ) at 30 ing does not need to be to scale. However, geometric
ft (9.1 m) above ground level. There are two wind coef- relationships of relevant building features will facilitate
ficients-0.8 and -0.8-that are used to simulate the review of model outputs. In addition, the user should
pressures on windward and leeward walls, respectively. refrain from creating additional building levels before
The terrain around the building is considered "urban."
identifying all zones and airflow paths on the floor level
Other quantitative and qualitative input data are drawn initially. This will allow all building elements to
provided in Attachment 1.
line up from floor to floor. The creation of new floor
SECTION 2: DATA INPUT PROCESS levels is discussed further in Step 7. When drauing the
roof level of a building, all walls and zones should be
The data input process for the example application deleted from the drawing. In general, the only elements
is broken into the following seven steps: that show on the roof level are airflow elements, such as
1. Drawing Building Components leaks, stair openings, fans, etc. Further discussion on
2. Identifying Zones roof drawings is discussed further under Steps 4 and 8.
3. Describing Zones Example Application: Four exterior walls, two
stairs, and one shaft (representing an elevator shaft) are
4. Identifying Airflow Paths
presented. Figure D1 illustrates the program screen for
5. Describing Airflow Paths drawing the example schematic.
6. Describing Weather Data
7. Creating and Defining New Levels Step 2: Identifying Zones
Each of these steps is discussed in detail in this sec- Zones indicate a volume of air with uniform tem-
tion. The purpose of each step is presented, followed by perature and contaminant concentrations. The user can
its use in the example application. Examples of the pro- identify a zone as any area of interest (i.e., stair, room,
gram's screens are provided to assist the user in under- smoke management zone, etc.). Once walls have been
standir.2 the narrative. Notes are identified in areas drawn, "normal" zone icons are placed within the wall
where the user should be cautious during the data input
boundaries of the building. In addition to "normal"
process.
zones within the building interior, the model represents,
by default, an "ambient" zone for the exterior environ-
Step 1: Drawing Building Components
ment surrounding the building.
Use the SketchPad to draw the desired structural
components of the building, including exterior walls. Note: Every physical division in the building, as defined
interior walls, stairs, ducts, shafts, etc. For further assis- by the schematic, must be identified with a single and
tance on drawing building components. the user should uniquely defined zone icon.
Principles of Smoke Management

'Normal" zone (has

I zone
(~nddned ICo16.~or14 Ilwel<l,:l of1
11
A,

Figure D2 Normal and ambient zones on thejirstjloor.

EXAMPLE APPLICATION: "Normal" zones are


identified as follows:
1. Stair l (Stl)
2. Stair 2 (St2) ...... --- -......... -. ............... -,-p I

3. Building interior (Rml) (ignore compartmentation Tempedre: 24.1-


within the building interior) pressure: 10. PO _r]
G Variable
4. Elevator shaft c Constant
In addition, the computer identified the "ambient zone" by
default. Figure D2 illustrates the proram screen used to
volume: 196. r n Il
identify zones.

Step 3: Describing Zones


Each zone must be identified by numeric parame-
ters. The zone data include the name of the zone asso-
).l i
a n I
d ~ il
ciated with the zone icon, the zone temperature, Figure D3 zone properties for stair I on rhe first
pressure (constant, variable, or none), volume/floor jlooy.
area, and initial contaminant concentration if applicable.

Note: The zone name must be unique for this level and is zone properties. Contaminant data were not used in this
limited to four characters. In addition, the user inputs either application.
the floor area or volume information. The program will
automatically calculate the field that is not entered by the Step 4: Identifying Airflow Paths
user.
An airflow path indicates some building feature by
Example Application: Variable pressure is used which air can move from one zone to another. The user
where pressures are determined by the model for each should identify any flow element on an exterior wall,
zone based on an analysis of conservation of mass. Fig- interior wall, door, or floor (e.g., leakage paths. other
ure D3 illustrates the program screen used to describe openings, fans or shafts).
Appendix D- Application of CONTAMW

I -- ,-- ZJ
p - - - - t o l S. ~ p IS
w L ~ V O I (1 >. 1 0f1 n

Figure D4 Selecting an aitflowpath onjrstjlooc

Note: Floor leaks must be identified at 0 meter elevation age characteristics. The height of the flow element
for the floor level in question. The propm describes these should be entered as the midheight elevation of the ele-
leaks as airflow between that level and the level below. In ment.
addition, airflow paths must be identified on each exterior
wall ifthe user is taking into account wind effects. Path ele- Note: Once a new element is defined, the identified ele-
ments can be identified for each wall or one path element ment is stored in the user-defined library. The user can edit
can be identified and the areas involved described in paral- an existing element at any time and the changes will apply
lel. When adding a fan on the roof for stair pressurization, to all airflow paths identified as that particular element.
the user should make sure that the airflow element is joined Example Application: The filter and schedule tab
up such that it is within the boundary of the stair enclosure. is not used in this example application. New user-
"Large" openings and "small" openings can be used inter- defined elements were added for the horizontal and ver-
changeably at the user's discretion to help facilitate review. tical airflow paths in this example. Figure D5 illustrates
Example Application: Horizontal and vertical air- the program screen used to describe a new airflow ele-
flow paths are identified. Horizontal airflo\v paths ment. Airflow path properties are broken into three
include "large" openings, such as stair doors, exterior intermediate steps: defining airflow element characteris-
doors, and elevator doors. They also include "small" tics, identifying flow paths, and determining wind pres-
openings, such as leaks in exterior walls, interior walls, sures (where applicable).
and closed doors. Vertical airflow paths include "large"
openings, such as stair and elevator enclosures. They Itzternzediate Step I : Defining Airflow
also include "small" openings such as leaks between Eleineizt Characteris tics .
levels. Since the 1st floor does not have a level below it,
there is no floor leak identified from that floor. Figure Airflow elements describe the mathematical rela-
D4 illustrates the program screen used to identify an air- tionship between the flow through an airflow path and
flow path. the pressure drop across the path. CONTAMW includes
choices of several types of flow elements and mathe-
matical models relating the pressure difference, area,
Step 5: Describing Airflow Paths
and mass flow. A mathematical model must be selected
Once placed on tlie SketchPad, the user can define for each new airflow path identified.
airflow characteristics and move, copy, and delete them.
The user should refer to the CONTAMW Help Program Note: Fields in the model input screens are either user-
or CONTAM96 User's Manual for additional guidance defined or default values. Refer to Attachment 1 for details
on describing airflow paths. When describing airflo\v on the values used for this example application.
element characteristics, flow paths are specified as
Example Application: The following models are
either new elements or as an existing element in the
used:
CONTAMW library. Input data for each airflow path
include information on the zones that the paths connect, 1. One-way flow using powerlaw model, orifice area
their height, and olhcr quantitative information on leak- data, for all closed and open stair doors.
Principles of Smoke Management

Figure D5 New zuer-defined elernenfs idenfifiedas EXTWALLI.

__r W_D e s a r p t u n C Tes(DataCLpom!J


-
C , C I
__ - _ _ _ _- _
T ~ F l o Models
w - - -
r One Openmg r T w Opening
........... - ..... - ... -.........

Figure D6 Airjlow elenlenf models in C O N T A M

2. One-way flow using powerlaw model, leakage area


data, for all leaks in walls and floors.
3. One-way flow using powerlaw model, stairwell, for all
vertical stair openings behveen levels.
4. One-way flow using powerlaw model, shaft, for verti-
cal elevator shaft openings behveen levels.
Hydraulic Diorneter. 10 9 8 4 3 7 8 n
R e y n o l d s Number: 130
5. Fan and forced flow model, constant mass flow, for all Description:
mechanical fans (exhaust and pressurization fans). Cross-seC18onal a r e a is hcdf of the d o o r

Figure D6 illustrates the different airflow element


models that are available in !he CONTAMW program.
1 C L a r g e oponing 0 I
Figures D7 through D1 1 illustrate the inpilt data . . . . . . . . . . -1
required for the models used in the example application.
I 0K I Cancel

Intermediate Step 2: Ident$j~zg Flow Paths Figure D7 Pon.erlaw model, orifice area dafa.
Once numeric parameters for flow characteristics
have been added to the model fields, flow path proper- flow elements is most accurate at midheight of the open-
ties (flow path tab shown in Figure 0 5 ) must be defined. ing. The default value in the program is the mid-height
When defining flow path properties, the elevation of of the room.
Appendix D - Application of CONTAMW

..
Figure D8 Powerlaw model, leakage area. Figure D 11 Powerlaw model, stainoefi.

Perimeter. (121.d
Roughness: 1
0.1
7-
Descaipuon:
elt~calfloor leak In shafl loose mnstruaon. 3m X 3m Design (mm) flow rate: i
- - -

rate
~ ~ I O W that provides desired d e s ~ g npressure i i l
1 G Small openmg o
l
C Large opening 0
l
-- .I

-,.. 1
01
.1 5 -
1cvlasl

Figure D 10 Powerlaw model, shoff. Figure D 11 Fan andforced-fiu, models.


Principles of Smoke Management

Note: For stairs and the elevator shaft, the elevation of the sure option is used for the leaks on ,the roof and all air-
horizontal flow path is the midheight of the staidelevator flow paths inside the building interior. Figure D13
doors and the elevation of the vertical flow path is zero. In illustrates the screen used to describe wind pressure.,
addition, the user must identify the positive flow d i i t i o n For variable wind pressures, three inputs are
of an airflow path f ~allr fans (or other element type where required: a wind pressure modifier, a wall azimuth
a flow rate is designated). For all other elements, the pro- angle, and a wind pressure profile. The wind pressure
pm arbitrari!y se!ects the positive flow direction. modifier is determined using the equation
Example Application: In this example, the posi-
tive flow direction from ambient to the building interior
is defined for the fans on the roof and the 5th and 7th
where Ch is the wind pressure modifier, A, and a depend
floors (in this manner, these fans supply rather than
exhaust air to the floor). The positive flow direction on the terrain around the building (ASHRAE 1989, p.
from the building interior to ambient is defined for the 14.3), and H is the height of the roof or wall.
fan on the 6th floor. Elevation of flows was at midheight The wall azimuth angle is defined as the direction
of level for all walls, mid-height of doors for all doors, the wall faces with north being 0 degrees, east 90
or at floor level for all vertical flows. Figure D12 illus- degrees, south 180 degrees, and west 270 degrees. A
trates the program screen used to define flow path eleva- default azimuth angle is provided based on the orienta-
tion and flow direction. tion of the wall on the SketchPad with the top of the
SketchPad. being north. The wind pressure profile is
Intermediate Step 3: based on wind coefficients and their respective wind
azimuth angles.
Determining W i n d Pressure
Example Application: The wind pressure modifier
Wind pressure characteristics are included only for was calculated using the building height (48 meters) and
elements for which a flow exists between "normal" and urban terrain factors (A, = 0.35 and a = 0.40). Figure
"ambient" zones. Three wind pressure options for open- D14 illustrates the weather and wind parameters screen
ings exist in the CONTAMW program: no wind pres- from which weather and wind characteristics are pre-
sure, constant pressure, and variable pressure sented. In addition, the default azimuth angle of 0
(dependent on wind speed and direction). degrees is used in this example application. The wind
coefficients used are 0.8 for the windward wall and -0.8
Note: Data entry corresponding to the wind pressure
for all other walls. As mentioned, the windward wall
option selected is required.
azimuth angle is 0 degrees (north). Figure D15 illus-
Example Application: Variable wind pressure is trates the wind pressure display based on these wind
chosen for all exterior airflow paths. The no wind pres- coefficients.

- ..
1 Falter dnd schedule
+-U .. -L.,.

l
- .
Flow Element
. l'.
h o w Path Wqnd &br&r%
,...-, -.
Path Number 6- '< '
>
7 .

.
R e l a w Elevation: -1

Figure D12 .+!irflowpath elevation and direction for- a door:


-
Appendix D Application ofCONTAMW -

FigureD13 Describing wind pressure information fir

Note: If the building-site pressure is unknown, the locati.on


tab allows for input of the building site altitude, which
CONTAMW will use to determine a default barometric
pressure.
Example Application: Transient weather is not
addressed in this example. Steady-state weather data
have been used. Default values are used for all parame-

Local Terrain Constmt v Il


ters with the exception of ambient temperature, which is
20°C. Figure D16 illustrates the location specific
Velocity Profile Eqonent 1 3 5 II weather parameters used.

Wnd Speed Modifier. 10 4797


ll Step 7: Creating and Defining New Levels
The CONTAMW program is organized by levels,
and each level is represented by a plan view drawing. A
default level is created so that the user can begin work-
- I 1
ing on a drawing right away without having to create a
Figure D14 Wind chat-acferisficsfor urban seffing. new level. Each level should be given a name, an eleva-
tion of the level above ground, and the height of the
level from floor to ceiling. Whenever a new level is cre-
ated-whether it is a blank level or a copy of another
Step 6: Describing Weather Data level-XONTAMW will give it a default name that will
consist of a number enclosed within the "<" and '5"
Weather parameters (i.e., ambient temperature, characters. The user can modify the default names as
baroixctric pressure, wind speed, and wind direction) required for the particular building.
are also included in the model. The weather parameters Note: The user must be careful when copying levels to
can be either steady-state or transient. The weather and ensure that the connections between building levels are
wind parameter fields define steady-state weather and presented in a manner that makes sense for the user's pur-
wind data. CONTAMW uses default values for tempera- poses. As previously mentioned, it is recommended that
ture and pressure (with no wind): 20°C and 1 atmo- the first level be completed first with all "typical" zones
sphere (approximately 101 kPa), respectively.The wind and airflow paths, so that the elements line up from floor to
speed field is used to address the reference velocity used floor. Other elements can be added or deleted as deemed
for the wind. necessary. In addition, vertical leakage paths (i.e., stairs,
- Principles of Smoke Management

Figure D15 Ifindpress~ti-eprofile for defined wind coej7ccients.

shafts, floor leaks) are included when new levels are cre-
ated since leaks via the floor are indicated oh the levels l
WenIher Wind 1 Locaaon Wind Pressure Display /
above. The user must be careful to make sure that all air-
These values are used tor deterrnln~ng
mnd pressure
flow paths specific to each level are accounted for. jdtsplayfen(ures onlyi

.bmbrent~e&erahlre
All level names must be unique. In addition, the
value for the "distance to the level above" is used by Atlsoluie h s s u r e T I P , 11
CONTAMW to calculate zone volumes based on the Wind Speed y l m p h d
floor area of each zone. Wind DseQion. 7 degrees

Example Application: Thirteen levels are included


in the example application (including 12 floor levels and
a roof level). The levels are copied from the 1st floor
Figure D16 Bztilding site related weather- data.
drawing, and the appropriate elements are added or
deleted for each subsequent floor. For example, fans
have been added on the 5th, 6th, and 7th floors, as well Once the results are available, the user can view
as the roof, and floor leaks have been adde6 on the 2nd results in two primary formats: schematic form and tab-
through roof levels. The elements referring to the exte- ular form.
rior doors have been deleted from the 2nd through roof
levels. Attachment 2 provides the layout of each level Schematic Results. The CONTAMW program dis-
plays color-coded bars indicating the relative airflow
(including the zones and airflow paths present).
rates and pressure drops associated nith each airflow
path on the current level of the Sketchpad. Airtlow rates
SECTION 3: DATA OUTPUT PROCESS are shown in blue and pressure dift-crences in red.
Downward positioned bars indicate that airflows/pres-
CONTAMW provides the user with several types of
sures are going from that level to the level below.
simulation results. The results available after a simula- Upward positioned bars indicate that airf~ows/pressures
tion depend on the simulation method and output are going from that level to the level above. Schematic
parameter settings. The simulation is run based on set results for the example application arc provided in
simulation parameters. For this example application, the Attachment 3.
simulation parameters used were default values pro-
vided by CONTAMW. Figure D17 illustrates the tab Note: When a airflow path icon is highlighted, the respec-
that is used to run the simulation. tive airflow and pressure results (in addition to level and
Appendix D - Application of CONTAMW

aifflow element name) will appear in the lower left-hand direction of airflow on every level of the shaft where
corner of the prograin screen.
- -
there are'airflow path icons in the same location on the
Sketchpad. As with the comma-delimited results, the
Tabular Results. CONTAMW can export results to shaft report can be saved as a text file. Tabular results
comma-separated format file that can later be imported (comma-delimited and shaft report format). for the
to a spreadsheet program for further analysis. The user example application are provided in Attachment 3, Fig-
can plot aifflow, contaminant, exposure, and pressure ure D18 illustrates the program screen used to export
results. In addition to comma-delimited format, CON- comma-delimited results and generate a shaft report.
TAMW generates a shaft report-a special reporting
feature f i r shafts. The shaft report generated by the pro- NOTE: A simulation must be run first in order to export
gram will display pressure drop, airflow rates, and results/generateshaftreport

Figure D17 Rmnitzg the CONTAMWsimlclation.

! ........
Zone h b t ! JenllD0:OO:OO
=l .
Lwel cl>: l ol 13
--....-F -. .I

Figure D 18 E.vpo~rir~g
~zslrlfs.
Principles of Smoke Management
Powerlaw Model: Leakage Area
Element Airflow properties Flow Path Wind Pressure
N:lme 1,eskage Discharge Flow Preswre Relative Multiplier Positive Flow Limits Wind Pressure Wind Pressure Wall Azimuth
c : Cocfficicnt Exponent I)rop Elcv:~tio~~ Direction Option Modifier Angle
c\tbuw:llll le~ult tlcl>~ult dclault 2 tlrli~ul[ cdculated by Nonc Variable 0.47972 1 901270
0 1 4 ,,,2
Droeraln I
, i II I I I Y
II I I

c t s r , l , c l ~ l t ilcf;~ul~ (~C~~~LII~ j 2 dclhult calculated by


program
None Variable 0.479721 011 80

estwall l default default default 2 default calculated by None Variable 0.47972 1 011 80
0, n12
Droeram
B U

estwall2 default default default 2 default calculated by None Variitble 0.479721 901270
0,1051n2
program
1loorlc;lk 1,275 ,,,2 tlcSault default dcftlul[ 0 default calculated by None None .None None
I I I I I II I l . -
Drocraln l II 4

Illcakslilil 0,00153 ,,,2 dePdult defidtrlt default 0 dcfault calculated by None None None None
I .program ,
Illc;~ks~rl 0 , 0 0 4 ~ 8 tlcli~ult dclhult dd;lult 0 tlcliult calc~~latcd by None None None None
program
/ intivallrhft 0,0042 m 2 default default default 2 default calculated by
program
' None None None None

intwallstr 0,014 m2 defadt default default 2 default calculated by None None ' None None
I program 1 II
Powcrlaw Model: Stairwell
Airlluw I'ropcrtics Wind I'rcssitre 1
Dist;~ncc Cross-Section:~l Density of' Stair llclative Multiplier Positivc Flow Limits Wind Pressure Wind Presst4r-e Wall Azimuth
Between arc;^ People Trei~tls Elev:~tio~~ Dircctiou Option Modifier
Lcwls m

4 111 24 1n2 Dcljult Closed 0 delh~lt calculated by None None None


program

Powcrl:~rvModel: Shaft
Element Airflow Properties 1:low 1';1tl1 Wind I'ressure
Name Distance Cross-Sectional Perimeter Roughness Relative Multiplier Positive Flow Limits Wind Pressure Wind Pressure Wall Azimuth
Iletwccn Arca Elevation Direction Option Modifier Angle
Levels
flleakslift2 4111 9 m2 12 m default 0 default calculated by None None. None None
program
lI JOOJaql q8no~qiJOOU puosas aql uaam~aqyea1 JOOU ps!uaA I

II 01 P'Z L 1000'0

1I P 1 SL 1 Sfooo'o I z\le~\lxslI
Appendix D - Application of CONTAMW

ATTACHMENT 2
FLOOR LAYOUTS

extslrwall2
,
slairdwrl
6-
/
-- . +-+p-- 1
,intwallslr
i
extwalll 'extdwr .
i
I

FIRST FLOOR

edstrwall2

flleakslr2 -+
I
- extwalll

p -- .+st,.
e~stwaIl1,--.-\.
-b

' stairdoor2
I

j Elev
l

Rml

L;
61.'
'
flwrleak
elevdoor

extwalll

7.
Q-.-

SECOND THROUGH FOURTH FLOORS


a 11 cl
EIGtlTH TIIROUGH T\YELTH FLOORS
Principles of Smoke Management -

Elev
I

FIFTH AND SEVENTH FLOORS

I
floorleak
, St2 :
elevdoor openstdoor .
A.. . . *i' extstwa!l

si :.

extwalll fan5

SIXTH FLOOR
Appendix D - Application of CONTAMW

flleakshftl
@ --6

ROOF
Principles of Smoke Management

ATTACHMENT 3
SIMULATION RESULTS FOR EXAMPLE APPLICATION

FIRST FLOOR

SECOND FLOOR
Appendix D - Application of CONTAMW

THIRD FLOOR

FOURTH FLOOR
Principles of Smoke Management

FIFTH FLOOR

SIXTH FLOOR
SEVENTH FLOOR

EIGHTH FLOOR
Principles of Smoke Management

NINTH FLOOR

TENTH FLOOR

. . . . . . . . -- - . . . - . . -.. ' -["' '.. " - '.. '. '


Z W l ) Rrnl /<l[D. TO. 23T.Vol: 31)O[Y)m' , Jml/B!lWW i L-
1 ~ 1 1 00d 13
~
Appendix D - Application of CONTAMW

ELEVENTH FLOOR

TWELFTH FLOOR
Principles of Smoke Management

ROOF
Appendix D - Application of CONTAMW

TABULAR PRESSURE AND


AIRFLOW DATA OUTPUT
project: CONTAM project Sun D& 03 10:22:54 2000

description:

simulation date : JanI


simulation time : 00:00:00
ambient temperature : 20.0 "C
barometric pressure : 101325.0 Pa
wind speed : 10.0 mph
wind direction : 0.0 deg

level: < l > elevation: 0.0 m

zone P T path From Flow 1 Flow2


Stl 53.5 24.0 flleakstr2 St 1/<2> 7119.15
extstnvall2 Amb t -48.70
stairdoor 1 Ambt -268.35
intwallstr Rml/<l> -101.72
extstnvall l Ambt -226.93
openstdoor Rml/<l> -6473.46

Elev flleakshft2 Elev/<Z>


inkallshft Rml/<l>
elevdoor Rml/<l>

Rml floorleak Rm 1/<2>


estwalll Ambt
estdoor Ambt
inhvallstr Stl/<l>
openstdoor Stl/<l>
inhvallshft Elev/< l >
esnvall2 Ambt
esnvall2 Ambt
elevdoor Elev/< l >
openstdoor St2/<1>
inhvallstr SW< 1>
extwall l Ambt
estdoor Ambt

flleakstr2 St2/<2>
openstdoor Rml/<l>
intwallstr Rml/<l>
extstnvall I Ambt
stairdoor l Ambt
eststn\~all2 Ambt

level: <2> ele\,ation: 4.0 m

zone P T path from Flow l Flou.2


St l 7.2 24.7 tlleakstr2 St 1/<3> 7718.22
eststnvall2 Ambt -49.28
tlleakstr2 Stl/<l> -7119.15
int\\,allstr Rm 1/<2> -1 38.00

312
Principles of Smoke Management

Ambt
Rml/G>

flleakshfu
intwallshft
flleakshfu
elevdoor
f
4..
I
$
Rml -19.6 20.0 floorleak Rm1/<3>
1
I exhvall I Ambt
inhvallstr Stl/G>
stairdoor2 St 1/<2>
intwallshft Elev/G>
exhvall2 Ambt
exhvall2 Ambt
elevdoor Elev/G>
floorleak Rml/<l>
stairdoor2 St2/<2>
inhvallstr St2/<2>
exhvall1 Ambt

flleakstr2 st2/<3>
stairdoor2 Rm1/<2>
inhvallstr Rm 1/<2>
extst&alll Ambt
flleakstr2 St2/<1>
extsmvall2 Ambt

level: <3> elevation: 8.0 m

zone P T path from Flow I Flow2


Stl -38.9 25.5 flleakstr2 St 1 /<4> 8335.10
extstrwall2 Ambt -49.96
flleakstr2 St 1/G> -77 18.22
inhvallstr Rm 1 /<3> - 145.27
extstrwalll Ambt -231.78
stairdoor2 Rm 1/<3> - 189.89
Elev -73.1 20.0 flleakshft2
inhvallshft
flleakshft2
elevdoor

floorleak Rm 1/<4>
exhvall1 Ambt
inhvallstr St l / - + -
stairdoor2 St1/<3>
inhvallshft Elev/:3>
exhvall2 Ambt
eshvall2 Ambt
elevdoor Elev/<3>
floorleak Rm 1/<2>
stairdoor2 St2/<3>
inhvallstr St2/<3>
extwall l Ambt
Appendix D - Application of CONTAMW

St2 -39.0 25.5 flleakstr2 StU<4>


stairdoor2 Rrn1/-=3>
intwallstr N1/<3>
extstrwalll Ambt
flleakstr2 sue>
extstnvall2 Ambt

level: <4> elevation: 12.0 m

zone P path from


St l -84.8 flleakstr2 St 1/ G >
extstnvall2 Ambt
flleakstr2 St 1/<3>
inhvallstr Rm l /<4>
extstnvall 1 Ambt
stairdoor2 Rrn 1 /<4>

Elev flleakshft2 Elev/<5>


intwallshft Rm 1/<4>
flleakshft2 Elev/<3>
elevdoor Rm 1/<4>

Rm l floorleak Rrn 1 / G >


exhi.all1 Ambt
intwallstr St 1/<4>
stairdoor2 St 1/<4>
intwallshft Elev/<4>
exhvall2 Arnbt
exhvall2 Arnbt
elevdoor Elev/<4>
floorlrak Rm 1/<3>
stairdoor2 St2/<4>
intwallstr St2/<4>
extuall I Arnbt

flleakstr2 St2/<5>
stairdoor2 Rm 1/<4>
intwallstr Rn1 1/<4>
extstn5,alll Arnbt
flleakstr2 St3/<3>
extstn\.all2 Ambt

level: <5> elevation: 16.0 m

zone P T path from Flow 1 Flow2


St l - 1 30.5 26.9 flleakjtr2 St 1/<6> 9177.54
cxtstn5-all2 Ambt -51.61
flleakstr2 St 1 /<4> -8945.69
intwallstr Rrn 1/<5> -100.10
extstnvall l Ambt -238.22
stairdoor2 Rrn 1/<5> -141.91

Elcv tlleakshft2 Elev/<6> -1249.5 1


intwallshli Rm 1 /<S> 34.99
Illeakshfi2 Elev/<4> 891.96
Principles of Smoke Management

elevdoor Rm1/<5>

floorleak Rm l/<6>
extwalll Ambt
intwallstr St 1 / 4 >
stairdoor2 st
fan3 Ambt
intwallshfi Elev/<S>
extwall2 Ambt
extwall2 Ambt
elevdoor Elev/<S>
floorleak Rm l/<&
stairdoor2 St2/<5>
intwallstr St2/<5>
extwall I Arnbt

flleakstr2 St2/<6>
stairdoor2 Rrn 1 /<S>
intwallstr Rm1/<S>
extstnvall l Arnbt
flleakstr2 St2/<4>
extstnvall2 Ambt

level: <6> elevation: 20.0 m

zone P T path from Flow l Flow2


Stl -176.1 27.6 flleakstr2 St l/<7> 10253.70
extstnvall2 An1bt -52.59
flleakstr2 St 1 /<S> -9477.54
intwallstr Rm 1/<6> -220.1 1
extstnvall I Ambt -242.02
stairdoor2 Rm 1/<6> -26 1.43

Elev -2 14.7 20.0 flleakshft2 Elev/<7>


intwallshft Rrn 1/<6>
flleakshft2 Elev/<S>
elevdoor Rni l /<G>

Rml -231.8 20.0 tloorleak Rm 1 /<7>


extwall I Ambt
intwallstr St 1/<6>
stairdoor2 St 1/<6>
fan4 Arnbt
intwallshft Elev/<6>
extwall2 Ambt
extwall2 Ambt
elevdoor Elev/<6>
tloorleak Rm I /<5>
stairdoor2 St2/<6>
intwallstr St2/<6>
eshvall l An1bt

flleakstr2 St2!<7>
stairdoor2 Rni l /<G>
inlwallstr Rni I /<G>

3 l5
Appendix D- Application i f CONTAMW

extstrwalll Ambt
flleakstr2 St2/<5>
extstrwall;? Ambt

level: a>elevation: 24.0 m


zone P T path from
Stl -221.5 28.4 flleakstr2 St l/<8>
extstrwall2 Ambt
flleakstr2 St l/<6>
intwallstr Rm
extstrwalll Ambt
stairdoor2 Rm l/<7>

Elev flleakshfu Elev/<8>


intwallshfl Rm 1/ < P
flleakshfu Elev/<6>
elevdoor Rrn1/<7>
floorleak Rm l/<8>
extwall l Arnbt
intwallstr St1/<7>
stairdoor2 St1/<7>
fan3 Ambt
intwallshft Elevl<7>
extwall2 h b t
extwall2 Ambt
elevdoor Elev/<7>
floorleak Rm l/<6>
stairdoor2 StU<7>
intwallstr stU<7>
extwalll Ambt

flleakstr2 StU<8>
stairdoor2 Rm 1/<l'>
intwallstr Rm1/<7>
extstrwalll Ambt
flleakstr2 StU<6>
extstrwall2 Ambt

level: <8> elevation: 28.0 m

zone P T path from Flow I Flow2


St l -266.7 29.1 flleakstr2 St 1/<9> 11573.08
extstrwall2 Ambt -54.85
flleakstr2 St1/<7> -10851.84
intwallstr Rm 1/<8> -186.12
extstrwall I Ambt -256.91
stairdood Rm l/<8> -229.35

Elev flleakshft2 Elev/<9>


intwallsh ft Rm 1/<8>
flleakshft2 Elev/<7>
elevdoor Rm 1/<8>

316
Principles of Smoke Management

Rml -309.6 20.0 floorleak Rm1/<g>


exhvall 1 Ambt
inhvallstr St1/<8>
stairdoor2 St 1/<8>
intwallshft Elev/<8>
extwall2 Ambt
ext;;.all2 h b t
elevdoor Elev/<8>
floorleak RmlM>
stairdoor2 St2/<8>
inhvallstr St2k8>
exhvall I Ambt

flleakstr2 St2/<9>
stairdoor2 Rni 1/<8>
inhvallstr Rm 1/43>
extstrwalll Ambt
flleakstr2 St2/<7>
extstrwall2 Ambt

level: <9> elevation: 32.0 m

zone P T path from Flow1


St l -31 1.6 29.8 flleakstr2 Stl/<lO> 12356.22
extst1kall2 hlbt -56.14
fl leakstr2 St 1 !<8> -11573.08
intwallstr Rnl1/<9> -2 14.84
extstnvall I Arnbt -255.97
stairdoor2 Rm 1 /<9> -256.19

Elev -356.3 20.0 flleakshft2


inhi-allshft
flleakshft2
elevdoor

Rrn l -365.4 20.0 floorleak


exhvalll
inhvallstr
stairdoor2
inh\.allshft
eshvall2
exhvall2
elevdoor
floorleak
stairdoor2
inhvallstr
eshvall I

flleakstr2 St?.'<lO>
stairdoor2 Rm 1 /<9>
inh\-allstr Rrn 1 /<9>
eststnval l l Anbt
flleakstr2 St7/<S>
eststnvall2 Arnbt
Appendix D - Application of C ~ A M W

level: <10> elevation: 36.0 m

zone path fiom Flowl Flow2


Stl flleakstr2 Stl/<l l> 13 179.90
extstrwall2 Ambt -57.54
flleakstr2 St 1/<9> - 12356.22
intwallstr Rml/<lO> -232.47
extstnvall I Ambt -261.49
stairdoor2 Rm1/<10> -272.18

Elev flleakshft2 Elev/<lI>


intwallshft Rml/<lO>
flleakshft2 Elev/<9>
elevdoor Rm1/<10>

Rm l floorleak Rml/<l l>


extwa l l l Ambt
intw?!lstr Stl/<lO>
stairdoor2 Stl/<lO>
intwallshft Elev/<l O>
extwall2 Ambt
extwall2 Ambt
elevdoor Elev/< l O>
floorleak Rm 1 /<9>
stairdoor2 St2/< 10>
intwallstr St2/<10>
exhvall l Ambt

St2 flleakstr2 St2/< I I >


stairdoor2 Rtnl/<lO>
inhvaktr R1111/< 1 0>
extst~walll Ambt
flleakstr2 St2/<9>
extstnvall2 Ambt

level: <l l> elevation: 40.0 111

zone path from Flow I Flow2


St l flleakstr2 St1/<12> 14035.98
extstnvall2 Ambt -59.05
fl leakstr2 Stl/<lO> - 13179.90
intwallstr Rm1/<1 l> -245.65
extstnvall l Ambt -267.47
stairdoor2 R1111/<1l> -283.9 1

Elev flleakshft2 Elev/< 12>


intwallr;hli R~iil/<II>
tlleakshfi2 Elev/< l O>
elevdoor R~nl/<ll>

R111I floorleak Rm1/<12>


extwall l Ambt
intwallstr Stl/<l l>
stairdool" Stl/<l l>
intwallslili Elev/< l l >

31s
Principles of Smoke Management

ex,twall2 Ambt
extwall2 Ambt
elevdoor Elev/<l l >
floorleak Rml/<lO>
stairdoor2 St2K 1 l>
intwallstr St2/<1l>
_extwalll Arnbt

flleakstr2 SW< 12>


stairdoor2 Rml/<ll>
intwallstr Rml/<l l>
extstrwalll Ambt
flleakstr2 St2k1o>
extstrwall2 Ambt

level: < l 2> elevation: 44.0 m

zone P T path f?om dP Flow lFlow2


St l -445.1 32.0 flleakstrl Ambt -76.0 -77.4 1
fan l Ambt -76.0 15000.00
extstrwall2 Ambt -70.4 -60.68
flleakstr2 Stl/<ll> -1.2 -14035.98
intwallstr Rm1/<12> -72.5 -257.60
extstrwall 1 Ambt -79.6 -273.92
stairdoor2 Rm1/<12> -72.0 -294.4 1

flleakshft l Ambt
intwallshft Rm1/<12>
flleakshft2 Elev/< l l >
elevdoor Rm1/<12>

floorleak Ambt
extwall1 Ambt
intwallstr St1/<12>
stairdoor2 St1/<12>
intwallshft Elev/<l2>
extwall2 Ambt
extwall2 Ambt
elevdoor Elev/< l 2>
floorleak Rml/<l l>
stairdoor2 St2k 12>
intwallstr St2k12>
extwall l Ambt

fan2 Ambt
flleakstr l Ambt
btairdoor2 Rm1/<12>
intwallstr Rm1/<12>
extstrwall l Arnbt
flleakstr2 St2/<1 l>
eststrwall2 Ambt

level: < 13> elevation: 48.0 nl

zone P T path
Appendix D - Application of CONTAMW

Note:
flows in scfm
pressures in Pa
temperatures in "C
* indicates limit exceeded
EXAMPLE OF SHAFT REPORT FOR STAIR 1
project: CONTAM project
shaft report

levellzone [pal [scfmI [pal [scfm] zone

+----------v---------------- +
< 1 2>Rm I 72.5 < 257.60 Stl 72.5> 257.60 Rml
I ------ ------ l
<l I>/Rml 67.2 < 245.65 Stl 67.2 > 245.65 Rml
1 ------ ------ I
< I O>/Rm I 61.7 < 232.47 Stl 61.7 > 232.47 Rml

Rml

Rml
- - p - - -

I
Stl 24.3 > 127.32 Rml
------ I
Stl 56.3 > 220.11 Rml
I ------ --
16.7 < 100.10 Stl Rml
- p - - - -

I
Stl 28.0 > 140.33 Rml
------ I
St l 29.5 > 145.27 Rml
-p----

l
Stl 27.2 > 138.00 Rml
-p----

l
St l 17.0 > 101.72 Rm l
Appendix E

ASMET Documentation

l
1 NOMENCLATURE TCP = absolute centerline plume temperature at eleva-
tion z, K
A = cross-sectional area of the atrium, m2
t, = growth time, s
a = fire growth coefficient, kw/s2
Tp = average plume temperature at elevation z, 'C
Cl = 0.071
C2 = 0.026 V = volumetric smoke flow at elevation z, m3/s
z = height above top of fuel, m
ZJ = mean flame height, ni
I, = virtual origin of the plume, m
CS = 9.1
j = convective fraction of heat release
C7 = 0.235
p = density of air or plume gases, kg/m3
C8 = 0.0018
C9 = 0.166 p, = density of ambient air, kg/m3

Clo = 1.11 pp = density of plume gases at elevation z, kg/m3


Note: The variables above are given in S1 units
only, because internal calculations in ASMET are in SI.

= specific heat of plume gases, 1.005 kJ/kg-K PART 1: ASMET DESCRIPTION


Cp
= diameter of fire, m Below are the equations used in each section of
DJ
ASMET, except for ASET-C, which is discussed in
g = acceleration of gravity, 9.807 rn/s2 Appendix F.
H = ceiling height above the fire, m
Steady Filling Equation (Solve for z)
n = mass flow in plume at height z, kg/s
P = absolute pressure, Pa
Q = heat release rate of the fire, kW
QC = convective heat release rate of fire, kW
R = gas constant, 287 J kg.K
Steady Filling Equation (Solve for t)
t = time, S
T = absolute temperature, K
Ta = ambient temperature. OC
-Appendix E- ASMET Documentation

Unsteady Filling Equation (Solve for z) The density of air and plume gases:

Plume Centerline Temperature


Unsteady Filling Equation (Solve for t)
Plume cecterline temperature:

Simple Plume Equations


Mass flow of plume: The virtual origin of the plume and the mean flame
height by the equations of the previous section, "Plume
1/3 5/3
iil = ClQC z +C8Qc with Virtual Origin Correction."

Mean flame height: Convective portion of the heat release rate:


- - C Q2/5
:l- 9 c

Average plume temperature: The convective fraction, E,, is generally taken as 0.7
for design. However, when burning a known fuel (as in
acceptance testing), it may be desired to use the specific
value for the fuel.

The volumetric flow of a plume: PART 2: ASMET USERS GUIDE


ASMET is a collection of tools that can be used for
analysis of atria smoke management systems. This pro-
gram is for a personal computer with a DOS operating
system, and the program was \vritten in C. When
The density of air and plume gases: ASMET is in the active directory, the program is acti-
vated by typing "ASMET" follo\ved by pressing the
<Enter> key. When the program starts, the main menu
appears on the screen as shown in Table E-l.
Plume with Virtual Origin Correction
The equations used for each routine are listed in
Mass flow of plume: Appendix C, except for ASET-C, which is described in
I /;
Appendix E. Theequations of Appendix C are also
til = C l QC (Z- z ~ ) ' / ~1 [+ c ~ Q ~ / ' ( z - z0 ,-5/3] (E10) addressed in the body of the text.
The first time the program is run, it starts in S1
This equation can be rearranged to simplify calculation: units, and the user can change units by pressing E for
I ; 5/3 English units or I for S[ units. The program stores a unit
tii = c l Q C (I-z0) +C8Qc (El 1) indicator in file UNITS so that it M-ill start up with the
unit selection from the last time the program was run.
Virtual origin of the plume: The other menu items are selected by pressing the key
715 that is in bold type (or yellow on a color monitor).
z0 = C;@ - 1.02 Of (E121 The first menu item is selected'by pressing S, and
the screen for this menu is shown in Table E-2. There
Mean flame height: are two ways to enter data from this menu. The first is
by pressing the key that is in bold for that menu item.
zf = c,Q'/'- 1.020, (E131 The second is by moving the indicator at the right of the
menu item with the up and down arrows. This indicator
Average plume temperature: is next to the first menu item (ceilinz height above fire)
in Table E-2. Once an item has been selected, the num-
QC
TI' = T , + -
111 c, (E14)
ber for that item is entered followed by <Enter>. Table
E-3 shows the screen after data has been entered.
The data displayed on the screen can be sent to the
The volumetric flow of a plume: printer by pressing P. and pressing D returns the user to
the main menu. To send results to a file, press f and
f . = c4 !! (E15) enter the lile name. Use of the other items in the main
p, menu is similar to that discussed above.
Principles of Smoke Management

-Table E-l:
Main Menu Screen of ASMET.
ASMET: Atria Smoke Management Engineering Tools

Menu
Steady Fiiiing Equation (Solve for z)
Steady Filling Equation (Solve fort)
Unsteady Filling Equation (Solve for z)
Unsteady Filling Equation (Solve for t)
Simple Plume Equation
Plume with Virtual Origin Correction
Plume Centerline Temperature
ASET-C (C language version o f ASET-B)
Input units (S1 or English): S1
Exit

Table E-2:
Screen for Steady Filling Equation (Solve for z)
Steady smoke filling
Height of smoke layer during atrium filling from a steady fire
ceilins height above fire H (m):
cross-sectional area of atrium A (mA2):
heat release rate of tire Q (kW):
time t (S):
Print results (to LPTI)
Print results to file disabled

Table E-3:
Screen for Steady Filling Equation After Data are Entered
Steady smoke filling
IHei=ohtof smoke layer during atrium filling from a steady fire
ceiling height above fire H (m): 80.00
cross-sectional area of atrium A (mA2): 20000.00
he31 release rate of fire Q (kW): 10000.00
time t (S): 1200.00 +
X l Prinr results (to L m l )
Print results 10 file disabled
Done (rcrum to main mcnu)
Hr.i:hr of smoke layer ahovc lire, z, is 17.6 m or 57.8 A
Appendix E- ASMET Documentation

EXAMPLE OUTPUT (S1 UNITS)


Steady Filling Equation (Solve for z)

Height of smoke layer during atrium filling from a steady fire

ceiling height above fire H (m): 30.00


cross-sectional area of atrium A (m2): 5000.00
heat release rate of fire Q (kW): 5000.00
time t (S): 300.00

Height of smoke layer above fire, z is 17.4 m or 57.2 ft


-------------------------------------------------------------------------
Steady Filling Equation (Solve fort)

Atrium filling time for steady fire

ceiling height above fire H (m): 40.00


cross-sectional area of atrium A ( I ) : 10000.00
heat release rate of fire Q (kW): 5000.00
height of smoke layer above fire z (m): 8.00

Filling time is 1290 seconds or 2 1.5 min.


-------------------------------------------------------------------------
unsteady Filling Equation (Solve for z)

Atrium tilling time for unsteady fire

ceiling height above tire H (m): 30.00


cross-sectional area of atrium A( I ) : 8000.00
fire growth constant (Menu) a (kw/sZ): 0.04659
time t (S): 800.00

At 800 seconds, the tire is 300 l0 kW or 28445 Btds.

Height ofsmoke laycr above tire, z, is 10.7 m or 35.0 ti


-------------------------------------------------------------------------
Unsteady Filli~lgEquation (Solve fort)

Atrium tilling time for unsteady tire

ceiling height above fire H (m): 50.00


cross-sectional arca ofatrium A (m Z): 12000.00
fire growth constant (Menu) a ( k ~ l s ' ) : 0.04659
height of smoke layer ab0i.e tire z ( m ) 10.00

Filling time is 1237 seconds or 20.4 min.


At this time, the fire is 7 1754 k W or 68014 Btuts.
.........................................................................
Simple plume equation
Mass flow and temperature rise of an plume
U illlout correction for virtual origin

Elwation r (m): 50.00


Heat release rate ol'lirc Q (k\V): 25000.00
Ambicnt tcmpcl-atcw Ta (C): 7 1 .00
Principles of Smoke Management

At elevation z, the plume has:


Mass flow of 1282.4 kg/s 2827.2 Ib/s
Volumetric flow of 1117.2 m3/s 2367016 c h -
Average temperature of 35°C 94°F
Mean flame height of 8.3 m 27.1 ft
.............................................................
Plume with V i a l Origin Correction -

Mass flow rate and average plume temperature

Elevation z (m): 50.00


Heat release rate of fire Q (kW): 25000.00
I fire diameter Df (m): 4.00
Ambient temperature Ta ("C): 21.00

At elevation z, the plume has:


Mass flow of 1254.7 kg/s 2766.1 Ib/s
Volumetric flow of 1094.2 mA3/s 23 1 8 122 cfm
Average temperature of 35°C 95°F
Virtual origin at 0.7 m 2.3 fi
Mean flame height of 9.4 m 30.9 fi

Plume Centerline Temperature


Calculate centerline plume temperature
. .
Elevation z (m): 50.00
Heat release rate of fire Q (kW): 25000.00
fire diameter Df (m): 4.00
Convective fraction of heat release (0.6 to 1): 0.70
Ambient temperature Ta ("C): 21.00

At elevation z, the plume has:


Centerline temperature 46°C 115°F
Virtual origin at 0.7 m 2.3 fi
Mean flame height of 9.4 m 30.9 fi
EXAMPLE OUTPUT (ENGLISH UNITS)
Height of smoke layer during atrium filling From a steady fire

ceiling height above fire H (R): 98.40


cross-sectional area of atrium A (g): 53800.00
heat release rate of fire Q (Btu%): 4740.00
time t (S): 200.00

Height of smoke layer above fire, z, is 17.4 m or 57.2 fi


.........................................................................
Steady Filling Equation (Solve for t)
Atrium filling time for steady fire

ceiling height above fire H (R): 131.00


cross-sectional area of atrium A (g): lO7OOO.OO
heat release rate of fire Q (Btuls): 4740.00
height of smoke layer above tire z (fi): 26.20
Appendix E- ASMET Documentation

Unsteady Filling Equation (Solve for z)


Atrium filling time for unsteady fire

ceiling height above fire H (R): 98.40


cross-sectional area of amum A (g): 86100.00
fire growth constant (Menu) a ( ~ t u l s ~ ) : 0.04444
time t (S): 800.00

At 800 seconds, the fire is 30006 kW or 28442 Btuk


Height of smoke layer above fue, z, is 10.7 m or 35.0 R
-------------------------------------------------------------------------
Unsteady Filling Equation (Solve for t)
Atrium filling time for unsteady fire

ceiling height above fire H (ft): 164.00


cross-sectional area of atrium A (P): 129000.00
fire growth constant (Menu) a (~tuls'): 0.0444
height of smoke layer above fire z (ft) 32.80

Filling time is 1236 seconds or 20.6 min.


At this time, the fire is 71650 kW or 67914 Btuk

Simple plume equation


Mass flow and temperature rise of a plume
without correction for virtual origin

Elevation z (ft): 164.00


Heat release rate of fire Q (Btds): 23700.00
Ambient temperature Ta (F): 70.00

At elevation z, the plume has:


Mass flow of 12s1.9 kg/s 2826.2 Ib/s
Volumetric flow of I 117.2 m3/s
2367054 cfm
Average temperature of 35°C 94°F
Mean flame height of 8.3 m 27.1 ft
-------------------------------------------------------------------------
Plume with Virtual Origin Correction
Mass flow rate and average plume temperature

Elevation z (ft): 164.00


Heat release rate of fire Q (Btds): 23700.00
fire diameter D f ( ft): 13.10
Ambient temperature Ta (F): 70.00

At elevation z, the plume has:


Mass flow of 1253.9 kg/s 7764.4 Ib/s
Volu~netricflow of 1093.9 m3/s 2317599 cfm
Average temperature of 3j3C 95°F
Virtual origin at 0.7 m 2.3 ft
Mean flame height of 9.4 m 30.9 fi
Principles of Smoke Management

Plume Centerline Temperature


Calculate centerline plume temperature

Elevation z (R): 164.00


Heat release rate of fire Q (Btuls): 23700.00
Fire diameter Df (R): 13.10
Convective fraction of heat release . (0.6 to l): 0.70
Ambient temperature Ta (F): 70.00

At elevation z, the plume has:


Centerline temperature 46OC
Virtual origin at 0.7 m
Mean flame height of 9.4 m
Appendix F

ASET-C: A Room Fire Program


for Personal Computers
INTRODUCTION version of ASET-B with improved interactive input and
a few added features. The interactive input was made to
Cooper (1981) of the Center for Fire Research,
be consistent with the other ASMET routines. The
National Bureau of Standards, introduced ASET, a
added features consist of allowing fire data input from a
mathematical model for estimating available safe egress
file and the use of a t-squared fire. Most of the material
time in fires. Cooper and Stroup (1982) published a
in this appendix is adapted from Walton's (1985) paper
computer program to perform the calculations in the
on ASET-B and, in many places, the adaptation con-
mathematical model; thus, the computer program also
sisted only of changing ASET-B to ASET-C.
became known as ASET. ASET was not specifically
written for the personal computer environment because
at the time it was being developed, personal computers DESCRIPTION OF THE MODEL
were just emerging as a tool for use in the engineering The mathematical model that is the basis for ASET,
office. ASET-B, and ASET-C has been presented in detail by
Since the introduction of ASET, the use of personal Cooper (1981, 1982) and will be only summarized here.
computers has become widespread and there has been It is based on a single room or enclosure with all doors,
significant interest in running ASET on personal com- windows, or vents closed except for a small leak at floor
puters. In response to this interest, Walton (1985) intro- level. This leak prevents the pressure from increasing in
duced ASET-B, a program for personal computers the room. A fire starts at some point below the ceiling
based on the original ASET mathematics1 model. The B and releases cnergy and produc:~ of combustion. The
was used to indicate basic, brief, BASIC, and beta. rate at which energy and products of combustion are
ASET is a 1500-line FORTRAN program that has released may change with time. The hot products of
many features. ASET-B is a 100-line BASIC program combustion form a plume, which, due to buoyancy, rises
that was developed to be as simple and fast as possible. toward the ceiling. As the plume rises, it draws in cool
The most significant change in ASET-B is the use of a air from the room, which decreases the plume's tempera-
different mathematical procedure to solve the primary ture and increases its volume flow rate. When the plume
equations. ASET-B employs an equation solver that is at reaches the ceiling, it spreads out and forms a hot gas
least five times faster than that used in ASET, while layer, which descends with time as the plume's gases
retaining mathematical agreement to within a fraction of continue to flow into it. There is a relatively sharp inter-
a percent. ASET-B is an interactive program requiring a face between the hot upper layer and the air in the lower.
minimunl of input. These features make ASET-B easy to part of the room, which, in this model, is considered to
learn and apply. In many con\ ersations with practicing be at ambient, temperature. The only interchange
fire protection engineers, the author has found that between the air in the lower part of the room and the hot
ASET-B has become very popular. upper layer is through the plume. ASET could therefore
This appendix describes the ASET-C routine, which be described as a two-layer or zone model. The basic
is p a ~ ofthe
t ASMET program. ASET-C is a C language fire phenomena are shown schernatically in Figure F 1.
Appendix F- ASET-C: A Room Fire Program for ~ e r s o n acomputers
i

the smoke layer, 20, and height of the base of the fire, F,
are the dimensior~alvalues for these variables in feet
divided by a characteristic length CL, which is also in
feet. Here, as in the ASET program, CL is simply taken
Air at Approximately ---c
as one foot. Thus. the dimensionless lengths Z, ZO, and
Ambient Temperature F are the same as their physical lengths in feet. The
dimensionless time, T, is the actual time divided by a
characteristic time, CT, of one second. The dimension-
Leak at Floor Level
l "-- I less time, T, is therefore numerically equal to the actual
time in seconds. Since engineering units are used in
1

Figure F1 Schematic offire phenonzena.


ASET, this convention has been continued here for con-
sistency. Conversion to S1 units is provided in the com-
puter pro,oram.
The two unknowns in ASET-C are the height of the The d~fferentialequations for the dimensionless
hot layer interface above the fire, Z, and the average height of the layer above the fire, 2,and average tem-
temperature of the upper layer, P. It should be noted that perature of the layer, P, are given below.
the notation used here to describe the model is consis-
tent with the variable names u;ed in the computer pro- -Cl . Q T - C 2 . Q T ' / ' 2 5 1 3 for 0 < Z < zo
gram. The unknowns, Z and P, are often referred to as
the (dimensionless) height and temperature of the
smoke layer since, consistent with the model formula- 0 for Z = -F
tion, smoke can only be found i n the plume and the hot
P [ C I . P T - ( P - 1 ) c 2- Q T 1 / 3 t 5 / 3 1 / ( z 0 + Z ) for o < Z < zo
upper layer. The known quantities are the.area and
height of the room, A and H, the height of the base of the
fire above the floor, F, and the acceleration due to grav-
ity, G. In addition, the ambient temperature, PA, density,
DA, and specific heat, CP, of air must be kno\vn. The 2 I /3
C2 = (0.21 . C T / A ) [ ( I - L R ) . QO. G.CL / ( D A - C P - P A ) ]
final known quantities are the rate at which heat is
released by the fire as a function of time, QT, the frac- In order to solve the equations for Z and P, the ini-
tion of the total heat release, which is given off as radia- tial conditions must be known. One set of initial condi-
tion, LR, and the fraction of total heat release rate, which tions, which were derived in Cooper (1981, 1982) and
is lost to the contents and surrounding surfaces of the will be used here, assume that the fire starts with a small
room, LC. heat release rate, Q0, at time T = 0. Under such condi-
tions, the initial conditions are
The unknown height and temperature are deter-
mined by using conservation of mass and energy in con-
junction with equations describing the plume. Since the
height and temperature of the smoke layer will vary
with time, T, their solutions are obtained by solving two Although dPldT is indeterminate in the above equa-
differential equations. In developing the original equa- tion at T = 0, its actual value has been found in Cooper
tions for ASET (Cooper 198l , 1982), two dimensionless (1981, 1981) to be
groups of problem parameters, C1 and C2, were intro-
duced. Also introduced were dimensionless forms of the . + (Cl + C2 - ZO"')
n_P - ~ 2DQO
variables: time, height, and temperature of the smoke d T - C2 6.zo~/~
layer, initial height of the smoke layer, height of the base
of the fire, and the rate of heat release. These variables where DQO = dQT/dTat time T = 0.
are made dimensionless by dividing them bv a charac-
teristic quantity with the same dimensions or units. SOLUTION OF THE EQUATIONS
Thus, the dimensionless temperature, P, is the actual In general, the differential equations for-Z and P
temperature of the smoke layer. PF (converted to R), cannot be solved explicitly; that is, an algebraic expres-
divided by the ambient temperature, PA (R). Similarly, sion cannot be written that describes Zand P at any time
the din~ensionlessrate of heat release, QT, is the actual 7. As a result, the equations must be solved numerically.
rate of heat release, QA (kW), divided by the initial rate ASET sol\-esthe difl'erential equations using a variation
of heat release, Q0 (kW). Finally, the dimensionless of the fourth-order Runge-Kutta method with variable
variables, height of the smoke layer, Z,initial height of time step. While this mcthod has a high degree of accu-
Principles of Smoke Management
3
j
1 racy, it has been determined that the improved Euler's time step. If the difference is greater than 0.001, the pre-
method has sufficient accuracy for this problem. The dicted values become the corrected values and the deriv-
2
improved Euler's method is a simple predictor-corrector atives at time T + DT are recalculated. New corrected
type and is described in most books on numerical meth- values are then calculated. In ASET-C, this procedure is
ods (Carhanan, Luther, and Wilkes 1969). The improved repeated for a maximum of thirty times. If the differ-
Euler's method used in ASET-C requires substantially ences are still greater than 0.001, a warning is printed,
fewer calculations than the method used in ASET, and the program proceeds to the next time step.
resulting in ASET-C running much faster than ASET. The evaluation of the derivatives of Z and P
The improved Euler's method as applied in ASET-C requires the dimensionless heat release rate, QT, be
is basically a technique for stepping the solution for- known for all times, I: For heat release rates that are not
ward in time. Given the values of Zand P at a particular constant with time, ASET-C requires the heat release be
time, T, the method is used to determine the values of Z specified for each one-second time interval. To simplify
and P at time T + DT, where DT is a small time incre- this procedure, ASET-C uses point specified heat
ment. This process is started at time T = 0 and continued release rates with linear interpolation. Heat release rates
until Z and P are known at all times of interest. In the can be specified at as many as 100 different times. Lin-
case of ASET-C, an increment of one second has been ear interpolation is then performed to determine the heat
found to yield results that agree well with ASET for release rate at each time step.
problems of practical interest.
In ASET-C, ZI, and PI are used to indicate the val- R U N N I N G THE P R O G R A M
ues of Z and P at time i7 For the first step, these are the
General Instructions
initial values at time T = 0. 22 and P2 are used to indi-
cate the values of Z and P to be calculated at time T + ASET-C is written as an interactive program; that
Di7 To determine 22 and P2 it is observed that the dif- is, the program prompts the user with questions. As pre-
ferential equations for Z and P represent the time rate of viously stated, ASET-C is part of the ASMET package
change of these quantities. The time rate of change mul- of routines for atrium analysis, and a description of this
tiplied by the time step yields the change that occurs package is provided in Appendix E. The mechanics of
over the time step. This would be an exact result if the input for ASET-C, are consistent with the other routine
equations were linear or the time steps were infinitely in this package. To use ASET-C, data niust be entered
small. Since the equations are nonlinear, and it is for the items discussed below.
impractical to make the time step infinitely small, an
Program Inputs
approximation must be used. In the improved Euler's
method, 22 and P2 are first predicted using the deriva- Heat Loss Fraction. The first input is tlie heat loss
tives evaluated at time I: Using 22 and P2, the deriva- fraction. This quantity is the instantaneous faction of the
tives are then evaluated at time T + DT. Corrected values heat release rate of the fire that is lost to the bounding
of Z2 and P2 are then calculated using the average of the surfaces of the room and its contents. Cooper (1 98 1.
derivatives evaluated at times T and T + DT. 22 and P2 1982) has provided guidelines for selecting this parame-
are predicted by ter, which is called Lambda C (?), or ALMAC in ASET.
He has detirmined that the approximate range is 0.G -
2 2 = 21 + DZI - DT , 0.9. The lower value corresponds to high aspect ratio
P2 = PI + DRI - DT . spaces (ratio of ceiling span to room height) with
smooth ceilings and fires positioned far away from the
where DZ1 = dZldT and DP1 = dPldT are evaluated using walls. The intermediate to high values corresponds to
Z = ZI and P = PI. The derivatives at time T + DT, D22 =
low aspect ratio spaces, rooms with irregular surfaces.
dZlfl: and D M = dPld7; are then evaluated using Z = 22
or rooms in which the fire is within one ceiling height of
and P = P2. Corrected values for Z2 and F2 are calculated the wall. The temperature of the upper layer is a func-
using the average derivatives tion of the heat loss fraction and the heat release rate of
Z2C = Z l + [ ( D Z I + D Z 2 ) / 2 ] . D T , the fire. The greater the heat loss fraction, the lower the
P2C = P1 + [ ( D P I + D P 2 ) / 2 ] . DT .
temperature in the xpper layer. The heat loss fraction for -
a room mith insulated walls will be lower than the frac-
The predicted values of Z and P are then compared tion for the same.room with uninsulated walls.
to the corrected values. In ASET-C, if the absolute value Both ASET and ASET-C treat tlie heat loss parame-
of the difference between the predicted and corrected ter as a constant. That is, the heat lost from the room is a
values is less than 0.001, the solution is considered to constant fraction of tlie heat release rate of the fire. As
have converged and the program proceeds to the next the heat release rate of the tire changes. tlie quantity of
-Appendix F-ASET-C: A Room F i e Program for Personal Computers

heat lost will also change, but in direct proportion to the Heat release rates entered as less than 0.1 kilowatt
fire. Therefore, the room will not cool down even will be converted to that value. The program will auto-
though the heat release rate of the fire goes to zero. matically assume a starting value 0.1 kilowatt at time
Height of the Base of the Fire. The second input is zero. A heat release rate at time zero does not have to be
the height of the base of the fire above the floor in feet. entered unless a greater initial heat release rate is
For fuel items of relatively uniform surface height, such required. When all of the desired times and heat release
as beds, this is simply the height of the surface. For rates have been entered, a -9,-9 followed by a return is
three dimensional h e 1 items, such as sofas, an average entered to terminate the data entry and begin the calcu-
height weighted to reflect the distribution of surfaces lations. Actually, any negative time followed by a heat
should be used. The rate of growth of the upper layer is release rate will result in the same action.
strongly dependent on the difference between the height Optional Upper Limit on Fire. Fire growth may
of the base of the fire and the height of the smoke layer be approximated by the t-squared curve for some time.
interface. Because of the action of a suppression system, limita-
Room Ceiling Height and Floor Area. The third tions of fuel, or limitations of combustion air, t-squared
and fourth inputs are the room ceiling height in feet and fire growth eventually must stop. The optional upper
the floor area in square feet. According to Cooper limit on fire growth allows the user to specify a heat
(1981, 1982), the calculations may not be valid when release rate at which the fire curve reaches steady burn-
applied to room length-to-width aspect ratios greater mg.
than 10: 1 or with a ratio of height to minimum horizon- Send Results to Printer or to File. To sent results
tal dimension exceeding one. The equations are based to the printer, press P. To send results to a file, press t
on the assumption that the upper layer is well mixed and and enter the file name.
at a uniform temperature. Therefore, the results for a Run Simulation. To run ASET-C, press R. If heat
square room and a rectangular room of equal height and release rate by point entry has been selected from the
area will be the same. Menu, the data points will be requested after the run
Output Interval. The fifth input is the output inter- starts.
val. This is the time step for results that are sent to the Program Outputs. The output of the ASET-B pro-
screen or printed. The output interval of ASET-B was gram is a summary of the input data and a table of the
set at five seconds, and this is the default interval for conditions in the room as a function of time. The first
ASET-C. colunln in the table is the simulation time in seconds.
Maximum Time. The sixth input is the niaximuni The second and third columns are the temperature in the
time for the simulation in seconds. The results of the upper layer in degrees Celsius and Fahrenheit. The
calculations will be printed at five-second intervals until fourth and fifth columns are the height above the floor
the maximum time or until the end of the heat release of the interface between the upper and lower layers. The
data. sixth and seventh columns are the heat release rate of
Fire Growth Constant. The seventh input is the the fire in kilowatts and Btu per second. The output has
description of heat release rate of the fire. A fire gro\vth the same number of significant digits as does ASET-B,
constant can be entered to define a t-squared fire, or the which allows users to verify that this program produces
Menu can be activated that allows selection of a fire the same results as ASET-B for the same input.
growth constant for typical fires (slow, medium, fast, or
ultra-fast). From the menu, the user also can choose to LIMITATIONS OF ASET
enter data as sets of points, as was done with ASET-B. The use of ASET-C or any design aid requires the
When the user selects data points, the computer waits design engineer to make the final evaluation as to the
for the run command to request the data. However, the appropriateness of the design. The ASET-C programs
following is a discussion of input by data points. are based on certain engineering approximations of the
As described earlier, the program can accommodate fire environment and should be used to supplement
up to 100 pairs of times and comesponding heat release rather than replace sound engineering judgment. The
rates. The program performs a linear interpolation program results should be treated as approximate and
between the specified points to determine the heat the user is encourayed to become familiar with how
release rates at the required times during the calcula- changes in the input variables affect the program results.
tions. The data are entered by typing the time in sec- The temperature of the upper layer and the height of the
onds, follo\ved by a comma, followed by the heat interface respond differently to changes in the input
release rate i n kilowatts. A return or enter is then typed data. Appropriate factors of safety should be applied to
to proceed to the nest linc. either the input data or ths program results.
1 Principles of Smoke Management

Some of the limitations of the program have been VERIFICATION O F ASET


presented in conjunction with the input data require-
ments. There are, however, some additional limitations. Results of the ASET program have been compared
The mathematical procedure used in ASET-C is very to data from a limited number of actual fire experiments
3;j harrly; that is, the procedure will normally converge and (Cooper 1981, 1982). These comparisons can be
produce results. There are combinations of input data extended to the ASET-B and ASET-C programs since
for which the program will either fail to converge or halt they produce results that are within a few percent of
due to an illegal mathematical operation. If the proce- those produced by ASET. The fire experiments consid-
'I
dure for solving the equations fails to converge, a warn- ered a mockup of a hospital room-corridor building
ing will be printed and the solution will continue. The
space. Comparisons were found to be generally favor-
results following this message may be in error and
able. This does not necessarily mean that the compari-
.
,. should be treated as such. The failure to converge is usu-
ally a result of a heat release value that changes too rap- son will be favorable ia all cases. Clearly, additional
idly. In most cases, this problem can be corrected by studies are required in this area and that work is ongo-
minor smoothing of the input heat release curve. ing.
Appendix F- ASET-C: A Room Fire Program for Personal Computers

FIRE HEIGHT =
-
SAMPLE RUN (ENGLISH UNITS)
HEAT LOSS FRACTION 0.80
1.OOft
ROOM HEIGHT = 9.00ft
ROOM AREA = 225.00sq ft

Fire curve input manually


TIME (sec), HEAT RELEASE RATE (kW):
TIME (sec), HEAT RELEASE RATE (kW):
TIME (sec), HEAT RELEASE RATE (kW):
TIME (sec), HEAT RELEASE RATE (kW):

TIME TEMP TEMP LAYER LAY ER FIRE FIRE


sec C F ft kW Btuls
0.0 21.3 70.3 9.0 0.1 0.1
5.0 23.4 74.2 8.7 10.1 9.6
10.0 24.9 76.7 8.3 20.1 19.0
15.0 26.3 79.3 7.8 30.0 28.5
20.0 27.7 81.8 7.3 40.0 37.9
25.0 29.2 84.6 6.9 50.0 47.4
30.0 30.8 87.5 6.5 60.0 56.9
35.0 32.6 90.6 6.0 70.0 66.4
40.0 34.4 93.9. 5.7 80.0 75.9
45.0 36.4 97.5 5.3 90.0 85.4
50.0 38.6 101.4 5.0 100.0 94.8
55.0 40.9 105.6 4.7 1 10.0 104.3
60.0 43.3 110.0 4.4 120.0 113.8
65.0 46.0 114.7 1. I 130.0 123.3
70.0 48.8 119.8 3.9 140.0 132.8
75.0 51.8 125.2 j.6 150.0 142.3
80.0 55.0 130.9 3.4 160.0 151.8
85.0 55.3 137.0 3.2 170.0 161.2
90.0 61.9 143.5 3.0 180.0 170.7
95.0 65.8 150.4 2.8 190.0 180.2
100.0 69.8 157.6 2.6 200.0 189.7
105.0 74.2 165.5 2.5 218.8 207.5
110.0 79.0 174.2 2.3 237.5 225.3
1 15.0 84.3 183.7 2.1 256.2 243.1
120.0 90.0 194.0 2.0 275.0 260.5
125.0 96.2 205.1 I .S 293.8 278.6
130.0 102.9 217.2 1.7 3 12.5 296.4
135.0 110.1 230.1 I .5 33 1.2 314.2
140.0 1 17.7 243.9 1.3 350.0 332.0
145.0 125.9 258.7 1.2 368.8 349.8
150.0 134.6 274.3 1 .o 387.5 367.5
155.0 143.7 290.7 0.8 406.2 385.3
160.0 153.3 307.9 0.6 425.0 403.1
165.0 163.3 325.9 0.4 443.8 420.9
170.0 173.7 344.6 0.2 462.5 438.7
175.0 184.5 364.2 0.0 48 l 2 456.5
180.0 195.9 384.7 0.0 500.0 474.2
Principles of smoke ~ a n a ~ e m e h t

SAMPLE RUN (S1 UNITS)


HEAT LOSS FRACTION = 0.80
FIRE HEIGHT = 0.00 m
ROOM HEIGHT = 3.00 m
ROOM AREA = 20.00 sqm
fire growth constant (KWlsA2):0.046890

TIME TEMP TEMP LAYER FIRE FIRE


sec C F m kW Btds
0.0 21.2 70.2 3.0 0. I 0.1
5.0 21.5 70.6 2.9 1.2 1.1
10.0 21.9 71.5 2.8 4.4
15.0 22.6 72.6 2.7 10.0
20.0 23.3 74.0 2.5 17.8
25.0 24.3 75.7 2.3 27.8
30.0 25.4 77.8 2.2 40.0
35.0 26.8 80.2 2.0 54.5
40.0 28.4 83.1 1.8 71.2
45.0 30.3 86.6 1.6 90.1
50.0 32.5 90.5 1.5 111.2
55.0 35.1 95.2 1.3 134.5
60.0 38.1 100.5 1.2 160.1
65.0 4 1.5 106.6 1.1 187.9
70.0 45.3 113.6 I .o 217.9
75.0 49.8 121.6 0.9 250.2
80.0 54.8 130.6 0.8 284.6
85.0 60.5 140.8 0.7 321.3
90.0 66.9 152.4 0.6 360.2
95.0 74.1 165.3 0.5 401.4
100.0 82.1 179.9 0.4 444.8
105.0 9 1.2 196.1 0.3 490.3
110.0 101.3 214.3 0.2 538.1
115.0 112.5 234.4 0. I 588.2
120.0 124.8 256.7 0.0 640.4
125.0 138.5 281.3 0.0 694.9
130.0 153.8 308.8 0.0 751.6
135.0 171.0 339.8 0.0 810.6
140.0 190.3 373.5 0.0 871.7
145.0 211.9 413.4 0.0 935.1
150.0 236.2 457.1 0.0 1000.7
155.0 263.5 506.3 0.0 1068.5
160.0 293.3 56 1.7 0.0 1138.6
165.0 329.0 613.1 0.0 1210.8
170.0 368.1 694.6 0.0 1285.3
175.0 412.5 774.4 0.0 1362.1
180.0 362.7 864.9 0.0 1441.0
Appendix G

Data and Computer Output


for Stairwell Example
his appendix lists the data and CONTAM output for Example 10.4. Figure G1 is the CONTAM representation
of the building. The design parameters are listed in Table G-l and the flow areas are listed in tat.!^ G-3. The
CONTAM runs are summarized in Table G-3, and the CONTAM simulated pressure differences from the stair
to the building are listed in Table G-4. Table G-5 is a listing of the entire CONTAM flow output for run 1. For a dis-
cussion of the results of this example, see Example 10.4 in Chapter 10.

(a) Ground F!oor

(b) Typical Floor Notes.


l . Values of flow areasare Iksted in Table E2.
Symbols 2. This door is either opened or closed
Single-door Leakage area of dosed single door (Single-door or Open-door)
Double-door Leakage area of closed doubles door
Own-door Flow area of opened single door
Elev-door Leakage area of closed elevator door
Ext-wall Leakage area of canstruclion cracks and gaps in eflerior walls
Elev-wall Leakage area of construction cracks and gaps in elevator shafi walls
SW-wall Leakage area of mstruction cracks and gaps in slairwe!l walls
81dg-floor Leakage area of construction cracks and gaps in budding floor
El vent Vent to the outside at the IOD of the elevator shafl
Sj[r_floor Effectwe area to a m u n t fo; fncl~onlosses m slamell
€1-floor Etfecl ve area to a m u n t for fnct~onlosses in e1eva:or snaft
SWI Stairwell 1
SW2 Stawell2
FL Open plan space on the flool
EL Elevafor shafl

Figure G 1 CONTAM t-e~~~-~setitatiot~~fot-


Exatnple 10.4: (a) gro~o~d~floot.
atid (1,) ~picalfloot:

337
Appendix G - Data and ComputerOutput forStairwell Example

Table G-1:
Design Parameters for Example 10.4
Design number of open doors from stairwell to building 4
Number of stories 15
Height between stories 12.0 ft (3.66 m)
Outside winter design temperature 14OF(-1 O°C)
Outside summer design temperature 93OF (34OC)
Building design temperature 73OF (23OC)
Minimum design pressure difference 0.05 in. H20 (12.4 Pa)
Maximum design pressure difference 0.30 in. H20 (87 Pa)

Table 6-2:
Flow re as' for Example 10.4
CONTAM Tight Building Loose Building
Path Name f? mz ft2 m2
Doors:
Single - Closed Single-door
Double -Closed Double-door
Single - Opened Open-door
Elevator - Closed Elev-door
Walls (per floor):
Exterior Ext-wall
Elevator Elev-wall
Stairwell to Building SW-wall
Stairwell to Outside SW-wall
Building Floor Bldg-floor
Elevator Shaft Vent Elev vent
Effective Areas of shafts2
Stairwell SW-floor 41 3.8 41 3.8
Elevator Elev-floor 1290 120 1290 120
I. A flow cocfficient, C, of 0.65 is used for all flow areas cscepr for open stairwell doors for which C = 0.35.
2. Effective area of a shaft is the area that results in a pressure drop equal to the friction losses of flow in the shaft. See Examples 6.9 and 6.10

Table G-3:
Summary of CONTAM Runs for Example 10.4
Building Stair Supply ~ i r '
Run File Season Leakage Stair Doors open2 cfm rn3/s
1 EX-10-4D Summer Loose G, 2 , 3 , 4 , 5 20,500 9.67
2 EX-10-4A Summer Loose G 20,500 9.67
3 EX-10-4C Winter Loose G, 12, i3, 14, 15 20,500 9.67
4 EX-10-4B Winter Loose G 20,500 9.67
5 EX-10-4E Summer Tight G, 2 , 3 , 4 , 5 13,900 6.56
6 EX-10-4F Winter Tight G 13,900 6.56
The flow orsupply pressurization air was obtained by running thc computer program several times for runs I and 6 toobrain pressure differences
1.
that are 31 leas1 the mininiuni d c s i g value of 0.05 in. H+ (12.4 Pa).
2. G indicates thc exterior ground lloor stainvell door.
Principles of Smoke Management

Table G-4:
Pressure Differences Across interior1 Stairwell Door for Example 10.4
Run 2 Run 3 Run 4 Run S Run 6
in. H 2 0 Pa in. H 2 0 Pa in. H 2 0 Pa in. H 2 0 Pa in. H 2 0 Pa
N.% .NA NA NA NA NA NA NA NA NA
0.171 42.5 0.204 50.8 0.110 27.4
0.162 40.3 0.214 53.2 0.1 10 27.4
0.159 39.6 0.228 56.7 0.110 27.4
0.152 37.8 0.239 59.5 0.110 27.4
0.149 37.1 0.248 61.7 0.1 10 27.4
0.147 36.6 0.253 61.9 0.110 27.4
0.145 36.1 0.256 63.7 0.109 27.1
0.144 35.8 0.257 63.9 0.109 27.1
0.143 35.6 0.259 64.4 0.109 27.1
0.141 35.1 0.262 65.2 0.109 27.1
0.139 34.6 0.267 66.4 0.108 26.9
0 . 1 34.1 0.271 67.4 0.108 26.9
0.134 33.3 0.275 68.4 0.108 26.9
0.133 33.1 0.276 68.7 0.107 26.6
I . All interior stair\rc.ll door is one br.t\;-.cn 11ic s~sinv,
2. NA indicates "no[ applicahlc" hcc2ci: therc. is 110interior stairwell door on rl~cground floor.
A p p e n d i x G - D a t a and C o m p u t e r O u t p u t for Stairwell Example

Table GS. CONTAM Flow and Pressure O u t p u t for Example 10.4 Run 1

project: EX-10-4D
description: E x a m p l e 1 0 . 4 Surrrmer - 4 SW-Doors Opened

simulation date: Janl


s i m u l a t i o n t i m e : 00:00:00
ambient temperature: 93.0.F
barometric pressure: 2 9 . 5 i n . Hg
wind s p e e d : 0 . 0 mph
wind d i r e c t i o n : 0.C d e g

level: G e l e v a t i o n : 0.0 f t

zone P T . path Flow


EL 0.103 73.4 Elev-f l s o r 579.50
Elev w i l l -32.15
~levrdoor - 273.68
Elev-docr -27.3. 68

926.11
32.15
273.68
273.68
-167.61
-167.61
80.68
80.68
-1038.41
- 293.32

73.4 SW-£ l o o r 11643.36


Open-docr -11451.74
SW-wa l l - 1 l C . 94
SW-wall - 80.68

73.4 SW-f l o o r 11643.36


Open-doo: -1lrsi.74
SW-wall -80.68
SW-wa l l - 110.94

level: 2 e l e v a t i o n : 12.0 f t

zone P T path Flow


EL - 0.069 73.4 Elev-floor - 271.45
Elev-wall 47.20
Elev-flocr - 579.50
Elev-door 401.87
Elev-door 401.87

FL -0.OC8 73.4 Bldg-floor - 155.15


Elev-wall - 47.20
Elev-door - 401.87
Eiev-door -401.87
Open-docr 1712.12
0per.-door 1712.12
Blcig-floor -926.11
SW-wa l l 16.66
S5i-w a l l 16.66
Ext-wall - 1525.34
Principles of Smoke Management

SW1 -0.005 73.4 SW-floor


Open-door
supply
SW-wall
SW-wall
SW-floor

SW2 -0.005 73.4 ~w-floo; SW2/3


Open-door FL/2
supply Supply
SW-wall FL/2
SW-wall Ambt
SW-£ loor SW2/G

level: 3 elevation: 24.0 ft

zone P T path from Flow


EL -0.241 73.4 Elev-floor EL/4 -1104.74
Elev-wall FL/3 46.22
Elev-floor EL/2 271.45
Elev-door FL/3 393.53
Elev-door FL/3 393.53

FL/4
EL/3
EL/3
EL/3
SW1/3
SW2/3
FL/2
3W1/3
SW2/3
Ambt

SW1 -0.181 73.4 SW-floor 3W1/4


Open-door FL/3
SW-wall Ambt
SW-wa11 FL/3
SW-floor sh'1/2

SW2 -0.181 73.4 SW-floor 3W2/4


Open-door FL/3
SW-wall FL/3
SW-wall Ambt
SW-floor SW2/2

level: 4 elevation: 36.0 ft

zone P T path r rom Flow


EL. -0.412 73.4 Elev-floor EL/5 -1917.62
Elev-wall FL/4 45.09
Elev-floor EL/3 1104.74
Elev-door FL/4 383.89
Elev-door FL/4 383.89
Appendix G-Data and Computer Output for Stairwell Example

SW1/4
SW2/4
EL/3
SW1/4
SW2 / 4
Ambt

SW1 -0.355 73.4 SW-floor SW1/5


Open-door EL/4
SW-wa l i Ambt
SW-wall FL/4
SW-floor SW1/3

SW2 -0.355. 73.4 SW-floor SW2/5


Open-doo r FL/4
SW-wall FL/4
SW-wall Ambt
SW-floor SW2/3

level: 5 elevation: 48.0 ft

zone P T path - Flow


EL -0.584 73.4 Elev-floor -2711.67
Elev-wall 44.05
Elev-f loor 1917.62
.Elev-door 375.00
Elev-door 375.00

-721.49
-44. 05
-375.00
-375.00
1361.43
1361.45
156.d.7
13.24
13.24
-1390.31

SW1 -0.529 73.4 SW-floor SW1/6 -3039.04


. . OpeR-do cr
SW-wa 11
FL/5
Ambt
-1361.45
-102.81
SW-wa1 1 FL/5 -13.24
SW-£ loo= SW1/4 4516.54

w 2 -0.529 73.4 SW-floor SW2/6 -3039.04


Open-door FL/5 -1361.45
SW-wall FL/5 -13.24
SW-wall Ambt -102.81
SW-floor SW2/4 4516.54

level: 6 elevation: 60.0ft

zone P T path from dP Flow


L -0.756 73.4 Elev-floor EL/7 0.000 -2598.ii
Elev wall FL/6 -0. G01 -6.21
~ l e v ~ f l o c r EL/5 0.000 2711.67
Elev-docr FL/6 -0.001 -53.35
Elev-door FL/6 -0.001 -53.35
Principles of Smoke Management

FL/7
EL/6
EL/6
EL/6
SW1/6
SW2/6
FL/5
SW1/6
SW2/6
Ambt

SW1 -0.701 73.4 SW-floor SW1/7


Single-door FL/6
SW-wall Ambt
SW-wall FL/6.
SW-floor SW1/5

SW2 -0.701 73.4 SW-floor SW2/7


Single-door FL/6
SW-wall FL/6
SW-wall Ambt
SW-f loor SW2/5

level: 7 elevation: 72.0 ft

zone P T path Flow


EL -0.927 73.4 Elev-floor -2302.12
Elev-wall -16.45
Elev-floor 2598.71
Elev door -140.07
~1ev:door -140.07

FL/8
EL/:
EL/:
EL/?
SW1/7
SW2/7
FL/6
SW1/7
SW2/7
Ambt

SW1 -0.874 73.4 SW-floor SWl/%


Single-door FL/7
SW-wall Ambt
SW-wall FL/7
SW-floor SW1/5

SW2/8
FL/7
FL/7
Ambt
SW2/6

level: 8 elevation: 84.0 ft

zone P T path Flow


EL -1.099 73.4 Elev-f loor -1994.48
Elev-wall -17.07
Appendix G-Data and Computer Output for Stairwell Example

Elev-f loor
Elev-door
Elev-door

FL -1.107 73.4 Bldg-floor -FL/9


Elev-wall EL/8
Elev-door EL/8
Elev-dqor EL/8
Single-door SW1/8
Single-door SW2/8
Bldg-f loo r FL/7
SW-wall SW1/8
SW-wall SW2/8
Ext-wall Ambt

SW2 -1.046 73.4 SW-floor


Single-door
SW-wall
SW-wall
SW-floor

level: 9 elevation: 96.0 ft

zone P T path from dP Flow


EL - 1.271 73.4 Elev-floor EL/10 0.000 -1688.91
Elev-wall FL/9 -0.008 -16.95
Elev-f loor EL/8 0.000 1994.48
Elev-door FL/9 -0.008 -144.31
Elev-door €L/9 -0.008 -144.31

€L/10
EL/9
EL/9
EL/9
SW1/9
SW2/9
FL/8
SW1/9
SW2/9
Ambt

SW1 -1.218 73.4 SW-floor SW1/10


Single-door FL/9
SW-wall Amb t
SW-wall FL/9
SW-floor SW1/8

SW2 -1.213 73.4 SW-floor


Single-dozr
SW-wall
SW-wall
SW-floor

Levei: 1 0 elsvation: 108.0 f z


Principles of Smoke Management !,

zone P T path from Flow


EL -1.442 73.4 Elev-floor EL/ 11 -1399.17
Elev-wall FL/lO -16.07
Elev-floor EL/9 1688.91
Elev-door FL/10 -136.83
Elev-door IL/lO -136.83

"/l1 111.30
ZL/ 10 16.07
EL/10 136.83
ZL/lO 136.83
SXl/lO 151.44
S:i2/10 151.44
- 3/9 -87.32
s;.:1/10 69.41
sx2/10 69.41
.=nbt -755.43

SXl/ll -1462.62
'L/10 -151.44
.'?mbt -88.81
fL/10 -69.41
%1/9 1772.29

SW2 -1.390 73.4 SW-floor -C;.;2/11 -1462.62


Single-door 3/10 -151.44
SW wall 3/10 -69.41
SWIW~ll .=Xbt
S>;2/ 9
-88.81
SW-floor 1772.29

level: i l elevation: 120.0 ft

zone P T patn dP Flow


EL -1.614 73.4 Elev-floor 0.000 -1137.18
Elev-wall -0.006 -14.53
Elev-floor 0.000 1399.17
Elev-door -0.006 -123.73
Elev-door -0.006 -123.73

FL - 1.620 73.4 Bldg-floor


Elev-wall
Elev-door
Elev-door
Singie-door
Single-door
91dg-floor
S%' wall
SWIW~ ll
Ext-wa 11

SW1

SW2 1 6 2 73.4 SW-f loor


Singie-door
SW-wal l
Appendix G - Data and Computer Output for Stairwell Example

SW-wall Ambt
SW-floor SW2/10

level: 12 elevation: 132.0 ft

zone P T path from Flow


EL -1.786 73.4 Elev-floor EL/13 -911.05
- Elev-walf FL/12 -12.54
Elev-floor EL/11 1137.18
Elev-door FL/12 -106.80
Elev-door FL/12 -106.80

FL/13
EL/12
EL/12
EL/12
SW1/12
SW2/12
FL/11
SW1/12
SW2/12
Ambt

SW1 -1.734 73.4 SW-floor SW1/13


Si~.gle.-door FL/12
SW-wa ll Ambt
SW-wall FL/12
SW-f loor SW1/11

SW2 -1.734 73.4 SW-floor SW2/13


Single-door FL/12
SW-wall FL/12
SW-wall Ambt
SW-floor SW2/11

level: 13 elevation: 1'44.0 ft

zone P T path from Flow


EL -1.958 73.4 Elev-floor EL/14 -726.25
Elev-wall FL/13 -10.25
Elev-f loor EL/12 911.05
Elev-door FL/13 -87.24
Elev-door FL/13 -87.24

FL/14
EL/13
EL/13
EL/13
SW1/13
SW2/13
FL/12
SW1/13
SW2/13
Ambt

SW1/14
FL/13
Ambt
FL/13
SW1/12
. . Principles of Smoke Management

SW2 -1.906 73.4 SW-floor SW2/14 0.000 -568.54


Single-door =/l3 -0.055 -145.53
SW-wall FL/13 -0.055 -66.70
SW.-wall Ambt -0.079 -79.85
SW-floor SW2/12 0.000 860.62

level: 14 - elevation: 156.0 ift

zone T path from Flow


EL 73.4 Elev-f loor EL/15 -584.77
Elev-wall FW14 -7.85
Elev-floor EL/13 726.25
Elev-door FL/14 -66.85
Elev-door FL/14 -66.85

FL FL/lS 78.72
EL/14 7.85
EL/14 66.85
EL/14 66.85
SW1/14 143.91
SW2/14 143.91
FL/13 . .. -106.76
SW1/14 65.96
SW2/14 65.96
Ambt -533.25

SW1 SW1/15 -281.99


FL/ 14 -143.91
Amb t -76.68
FL/14 -65.96
SW1/13 568.54

SW2 SW2/15 -281.99


FL/i4 -143.91
FL/14 -65.96
Ambt -76.68
SW2/13 568.54

level: 15 elevation: 168.0 ft

zone P from Flow


EL -2.301 EL/16 -473.52
FL/lS -6.17
EL/14 584.77
FL/15 -52.54
FL/15 -52.54

EL/15 6.17
EL/l5 52.54
EL/15 52.54
SW1/15 143.04
SW2/15 143.04
FL/14 -78.72
SW1/15 65.56
SW2/15 65. 56
Ambt -449.73

FL/15 -143.04
Ambc -73.39
Appendix G-Data and Computer Output forStairwell Example

SW wall FL/15
~~Ifloor SW1/14

SW2 -2.249 73.4 Single door FL/15


SW wali FL/15
~ ~ 3 a l l Ambt
SW-floor SW2/14

level: 16 elevation: 180.0 ft

zone P T path from dP Flow


EL .-2.473 73.4 Elev-Vent Ambt -0.008 -473.52
Elev-floor EL/15 0.000 473.52

systems:
name air flows:
recirc outside
Exhust 0.00 0.00
supply 0.00 40999.97

Note:
flows in scfm
pressures in i ' n . ~ 2 0
temperatures in F
* indicates limit exceeded
Appendix H

Data and Computer Output for


Zoned Smoke Control Example
his appendix lists the data and CONTAM output for Example 12.5. The example is an eight-story building
with zoned smoke control and two pressurized stainvells. With the exception of the number of stories, the
design parameters and flow areas o f this example are the same as Example 10.4 (Appendix G), and Figure G I
is applicable.
The CONTAM runs are summarized in Table H-l, and the CONTAM simulated pressure differences from the
stair to the building are listed in Table H-2. Table H-3 is a listing o f the entire CONTAM flow output for run l .
For a discussion of the results of example, see Example 12.5 in Chapter 12.
Table H-l:
Summary of CONTAM Runs for Example 12.5
Fire Floor Adjacent Floor
Building Eshaust SUPP~Y Staircwll Supply
Run File Season Leakage ~loorl cfn' m3/s cfm ni3/s cfni n19s
1 EX-12-5A Summer Loose G 2800 1.32 2800 1.32 l600 0.761
2 EX-12-5C Summer Loose 2 2800 1.37 2800 1.32 l600 0.761
3 EX-12-5B Summer Loose 7 2800 1.37 2800 1.37 l600 0.761
4 EX-12-5E Winter Loose G 2800 1.37 2800 1.32 l600 0.761
5 EX-12-5D Winter ~oose --I 2SOO 1.32 2800 1.32 1600 0.764
6 EX-12-SF Winter Loose 7 2800 1.32 2800 1.32 l600 0.764
I. G indicates the exterior ground lloor stairwell door.

Table H-2:
Pressure Differences Calculated by CONTAM for Example 12.5

Stairwell to Fire Floor Floor Below to Fire ~ l o o r ' Floor Above to Fire Floor
Run in. HzO Pa in. HzO Pa in. HzO Pa

2 0.065 16.2 0.060 11.9 0.066 16.4


3 0.053 13.2 0.053 13.2 0.072 17.9
4 0.063 l 5.7 NA NA . 0.069 17.7
5 0.054 13.1 0.087 3 1.6 0.051 13.4
6 0.103 75.6 0.087 21.6 0.091 23.1
1. N A indicates "not spplic;~hlc."
Appendix H- Data and Computer Output for Zoned Smoke Control Example

Table H3. CONTAM Flow and Pressure Output for Example i 2 . 5 , ~ u n1

project: EX-12-5A
description: Example 12.5 Summer - Loose Building - Fire on Floor G

simulation date: Janl


simulation time: 00:00:00
ambient temperature: 93.b F
barometric pressure: 29.9 in. Hg
wind speed: 0.0 mph
wind direction: 0.0 deg

Levei: G elevation: 0.0 ft

zone D T path from dP Flow1


EL 0.044 73.4 Elev-floor EL/2 0.000 855.97
Elev-wall FL/G -0.061 -47.48
Elev-door FL/G -0.061 -404.24
Elev-door FL/G -0.061 -404.24

€L -0.018 73.4 Bldg-floor FL/2


Elev-wall EL/G
Elev-door EL/G
Elev-door EL/G
Single-door Ambt
Single-door Ambt
return Exhust
SW-wa ll SWl/G
SW-wall SW2/G
Ext-wall Ambt
Double-door Ambt

S?; l 0.069 73.4 SW-floor SW1/2 0.000 317.03


Single-door Ambt -0.067 -160.52
SW-wall Ambt -0.066 -72.87
SW-wall FL/G -C.086 -83.64

Sii2 0.069 73.4 SW-floor SW2/2 0.000 317.03


Single-door Ambt -0.067 -160.52
SW-wall FL/G -0.086 -83.64
SW-wall Ambt -0.066 -72.87

level: 2 elevation: 12.0 ft

zcne P T path f rom dP Flow1


EL -0.128 73.4 Elev-floor EL/3 0.000 322.91
Elev-wall FL/2 0.024 29.57
Elev-floor EL/G 0.000 -855.97
Elev-door FL/2 0.024 251.74
Elev-door FL/2 0.024 251.74

€'L -0.104 73.4 Bldg-floor FL/3 -0.G24 -479.55


Elev-wall EL/2 -0.024 -29.57
Elev-door EL/2 -0.024 -251.74
Elev-door EL/2 -0.024 -251.74
Single-door SW1/2 0.001 19.87
supply Supply n/a 2800.00
Principles of Smoke Management

SW2/2
FL/G
SW1/2
SW2/2
Ambt

SW1 -0.103 73.4 SW-floor


Single-door
supply
SW-wall
SW-wall
SW-floor

SW2 -0.103 73.4 %-floor SW2/3


Single-door FL/2
supply Supply
SW-wall FL/2
SW-wall Ambt
SW-floor SW2/G

level: 3 elevation: 24.0 ft

zone P T path from


EL -0.300 73.4 Elev-floor EL/4
Elev-wall FL/3
Elev-floor EL/2
Elev-door FL/3
Elev-door FL/3

FLi4
EL/3
EL/3
EL/3
SW1/3
SW2/3
FL/2
SW1/3
SW2/3
Ambt

SW1 -0.275 73.4 SW-floor SW1/4


. Single-door FL/3
SW-wall Ambt
SW-wall FL/3
SW-floor SW1/2

SW2 -0.275 73.4 SW-floor SW2/4


Single-docr FL/3
SW-wall FL/3
SW-wall Ambt
SW-floor SW2/2

level: 4 eleva~ion:36.0 ft

zone P T path from


EL -0.47i 73.4. Elev-floor EL/5
Elev wail FL/4
Appendix H-Data and Computer Output for Zoned Smoke Control Example

EL/3
FL/4
FL/4

FL/5
EL/4
EL/4
EL/4
SW1/4
SW2 / 4
FL/3
SW1/4
SW2 / 4
Ambt

-0.447 73.4 SW-floor SW1/5


Single-door FL/4
SW-wall Ambt
SW-wall FL/4
SW-floor SW1/3

SW2/5
FL/4
FL/4
Ambt
SW2/3

level: 5 elevation: 48.0 ft

zone P T path from


EL -0.643 73.4 Elev-f loor EL/6
Elev-wall FL/5
Elev-f loor EL/4
Elev-door FL/5
Elev-door FL/5

FL -0.64 4 73.4 Bldg-f loor FL/6


Elev-wall EL/5
Elev-door EL/5
Elev door EL/5
single-door SW1/5
Single-door SW2/5
Bldg-f loor FL/4
S W-wa ll SW1/5
SW-wall SW2/5
Ext-wall Ambt

SW1 -0.618 73.4 SW-floor SW1/6


Single-door FL/5
S W-wa ll Fmbt
SW-wall FL/5
SW-floor SW1/4

SW2 SW2/6
FL/5
FL/5
Ambt
Principlesof Smoke Management

SW-floor

level: 6 elevation: 60.0 ft

zone P T path from Flowl


EL -0.8 15 73.4 Elev-f loor EL/7 668.01
Elev-wal l FL/6 -4.68
~lev-f loor EL/5 -583.62
Elev-door FL/6 -39.86
Elev-door FL/6 -39.86

FL -0.8 15 73.4 Bldg-f loor FL/7


Elev-wa 11 EL/6
Elev-door EL/6
Elev-door EL/6
Single-door SW1/6
Single-door SW2/6
Bldg-floor FL/5
SW-wall SW1/6
SW-wall SW2/6
Ext-wall Ambt

SW1 -0.790 73.4 SW-floor SW1/7


Single-docr FL/6
SW-wall Ambt
SW-wall FL/6
SW-floor SW1/5

SW2 -0.790 73.4 SW-floor SW2/7


Single-door FL/6
SW-wall FL/6
SW-wall Ambt
SW-floor SW2/5

level: 7 elevation: 72.0 ft

zone P T path from Flowl


EL -0.986 73.4 Elev-f loor EL/8 609.22
Elev-wall FL/7 3.26
Elev-floor EL/6 -668.01
Elev-door 'FL/7 27.77
FL/7 27.77

FL/8
EL/7
EL/7
EL/7
SW1/7
SW2/7
FL/6
SW1/7 .
SW2i7
Ambt

SW1/8
FL/7
Ambt
Appendix H- Data and Computer Output for Zoned Smoke Control Example

SW-wall FL/7 -0.024 -44.35


SW-£ loor SW1/6 0.000 368.28

S W ~ -0.962 73.4 SW-floor SW2 /8 0.000 -178.95


Single-door FL/7 -0.024 -96.76
SW-wall FL/7 -0.024 -44.35
SW-wall Ambt -0.029 - -48.22
SW-f loor SW2/6 0.000 368.28

level: 8 elevation: 84.0 ft

zone P T path from Flowl


EL -1.158 73.4 Elev-floor EL/9 456.28
Elev-wall FL/8 8.48
Elev-f loor EL/7 -609.22
Elev-door FL/8 72-23
Elev-door FL/8 72.23

EL/8
EL/8
EL/8
SW1/8
SW2/8
FL/7
SW1/8
SW2/8
Ambt

FL/8
Ambt
FL/8
SW1/7

SW2 - 1.134 73.4 Single-door


SW-wall
SW-wall
SW-f loor

level: 9 elevation: 96.0 ft

zone P T path from Flowl


EL -1.330 73.4 Elev-Vent Ambt 456.28
Elev-floor EL/8 -456.28

systems:
name air flows:
recirc outside
Exhust 0.00 0.00
Supply 0.00 6000.00

Note:
flows in scfm
pressures in in.H20
temperatures in F
* indicates limit exceeded
Appendix I

Inspection Procedures for


Smoke Control Svstems
SCOPE Check duct installation. Duct installation,
including the hangers, must not reduce the
T h e inspection procedures described in this appen-
fire resistance rating of structural members
dix apply to smoke control systems that are dedicated
and of assemblies. Frequently, structural
only to controlling smoke in building fires o r that make
members and asse~iiblieshave fire protec-
use o f air-moving equipment with another function,
tive coverings, such as drywall construction
such as heating and air conditioning. These procedures
or a sprayed-on layer. Check that ducts are
are, o f a general nature, intended as a guide for tlie
installed in such a manner that these protec-
development of specific procedures for individual
tive coverings are not damaged. Check that
smoke control systems. These procedures address tlie
clearance from ducts to conibustible con-
major components of smoke control systems but, by
struction is as specified. In addition, check
their general nature, cannot address all possible conipo-
that where ducts pass through walls, floors,
nents. In this appendix, the phrase "as specified" is used
or partitions, the openings in construction
to mean as specified in accordance with a contract docu-
around tlie ducts are as specified.
ments, a code, or some other standard o r standards that
have been agreed upon by the owner, designer, builder, Clieck that installation and materials of duct
code official, and other involved parties. connectors and flexible duct connectors are
as specitied. CAUTION: Become 11le cllar-
BARRIERS acteristics of duct co1it7ectors atid j1e.rible
drtc~co~itiecforsare diffe~wir,orie sliorrld
a. Clieck walls, partitions, floors, and ceilings not be srrbs~i~rrtedfor 111eotliet:
of barriers of smoke control systems for
Check duct coverings and linings to verify
obvious and unusual openings that could
that their fire safety requirements are as
adversely affect smoke control perfor-
specified. Check that duct coverings do not
mance.
conceal any service opening.
b. Check tliat gaps around doors do not exceed
the limits specified. If gasketing is required, Check direct access and inspection provi-
check that it is as specitied. sions. Service openings and telescoping or
removable duct sections are used for direct
c. Check that automatic door closers in barri-
access and inspection. Check tliat a service
ers of smoke control systems are as speci-
opening or a telescoping or removable duct
fied.
section is provided in ducts as specified
adjacent to fire dampers, smoke dampers.
AIR-MOVING EQUIPIMENT
and smoke detectors. Check that these
a. Check ducts to veriQ that materials ofduct access openings are identified wit11 letters as
material and construction are as specified. specified. Check that service openings are
. -
Appendix I - Inspection Procedures for Smoke Control Systems

provided in horizontal ducts and plenums with normal air flow to ensure that they are
where specified. not held open by the airstream. Remember
to reinstall all hsible links that have been
f Check air filters to verify that they have the
removed during inspection.
classification specified.
g. Check that the location, fire protection rat- CONTROLS
ing, and installation of fire, ceiling, and
smoke dampers are as specified. Generally, a. Check manual controls. Check that devices
fire, ceiling, and smoke dampers should be for manual activation and deactivation of , .,
installed in accordance with the conditions the smoke control system are of materials
of their listing and the manufacturer's instal- and installation as specified.
lation instructions that are supplied with the b. Check automatic controls. Check that
damper. Further, check installation by devices for automatic activation and deacti-
removing hsible link (where applicable) vation and control of the smoke control sys-
and operate damper to verify that it fUUy tem are of materials and installation as
closes. It is desirable to operate dampers specified.
Principles of Smoke Management

Table 1-1:
inspection Checklist-Barriers of Pressurized Stairwells
Project:

Inspection agent: Date:

YES REMARKS
General:
1 All materials in plenums appropriate
2 Air filters appropriate
3 Fan inlets protected by scree~is
4 Heating equipment installation appropriate
5 Cooling equipment installation appropriate
6 Manual controls installed
7 Automatic controls installed
S
Ductwork:
I Duct material appropriate
2 Duct installation appropriate
3 Duct connectors appropriate
4 Duct coverings appropriate
5 Duct linings appropriate

Duct access and inspection provisions:


1 Access at all required locations
2 Access properly identified

Dampers:
I Fire dampers located where required
2 Fire dampers of appropriate rating
3 Fire dampers installed appropriately
4 Ceiling dampers located where required
5 Ceiling dampers of appropriate rating
G Ceiling dampers installed appropriately
7 Smoke dampers located where rcquired
8 Smoke dampers of appropriate rating
9 Smoke dampers installed appropriately
10 Combination fire and sniuke dampers
located where required
I I Ccmbination fire and snioke dampers
of appropriate rating
12 Combination tire and smoke dampers
installed appropriately
Comments:
A p p e n d i x I - Inspection Procedures f o r smoke Control Systems

Table 1-2:
Inspection Checklist-Barriers of Elevator Smoke Control Systems
Project:

Inspection agent: Date:

DESCRIPTION YES REMARKS


General:
l All materials in plenums appropriate
2 Air filters appropriate
3 Fan inlets protected by screens
4 Heating equipment installation appropriate
5 Cooling equipment installation appropriate
6 Manual controls installed
7 Automatic controls installed

Ductwork:
1 Duct material appropriate
2 Duct installation appropriate
3 Duct connectors appropriate
4 Duct coverings appropriate
5 Duct linings appropriate

Duct access and inspection provisions:


I Access at all required locations
2 Access properly identified

Dampers:
I Fire dampers located where required
2 Fire dampers of appropriate rating
3 Fire dampers installed appropriately
4 Ceiling dampers located where required
5 Ceiling dampers of appropriate rating
6 Ceiling dampers installed appropriately
7 Smoke dampers located where required
8 Smoke dampers of appropriate rating
9 Smoke dampers installed appropriately
10 Combination fire and smoke dampers
located where required
I I Combination fire and smoke dampers
of appropriate rating
12 Combination fire and smoke dampers
installed appropriately
Comments:
PrincipIes of Smoke Management

Table 1-3:
Inspection Checklist-Barriers of Zoned Smoke Control Systems
Project:

Inspection agent: Date:

DESCRIPTION YES NO REMARKS


General:
l All materials in plenums appropriate
2 Air filters appropriate
3 Fan inlets protected by screens
4 Heating equipment installation hppropriate
5 Cooling equipment installation appropriate
6 Manual controls installed
7 Automatic controls installed

Ductwork:
1 Duct material appropriate
2 Duct installation appropriate
3 Duct connectors appropriate
4 Duct coverings appropriate
5 Duct linings appropriate

Duct access and inspection provisions:


I Access at all required locations
2 Access properly identified

Dampers:
I Fire dampers located where required
2 Fire dampers of appropriate rating
3 Fire dampers installed appropriately
4 Ceiling dampers located where required
S Ceiling dampers of appropriate rating
6 Ceiling dampers installed appropriately
7 Smoke dampers located where required
8 Smoke dampers of appropriate rating
9 Smoke dampers installed appropriately
10 Combination fire and smoke dampers
located where required
I I Combination fire and smoke dampers
of appropriate rating
12 Combination fire and smoke dampers
installed appropriately
Comments:
Appendix I - Inspection Procedures for Smoke Control Systems

Table 14:
Inspection Check List-Fire Safety Controls in HVAC Systems
Project:

Inspection agent: Date:-

DESCRIPTION
Manual shutdown:
1 Appropriate fans stopped
2 Appropriate smoke dampers
fully and tightly closed

Automatic shutdown by return detector:


1 Appropriate fans stopped
2 Appropriate smoke dampers
fully and tightly closed

Automatic shutdown by supply detector:


1 Appropriate fans stopped
2 Appropriate smoke dampers
fully and tightly closed

Automatic shutdolvn by detector system:


1 Appropriate fans stopped
2 Appropriate smoke dampers
fully and tightly closed

Comments:
Appendix J

Test Procedures for


Stairwell Pressurization .Svstems
SCOPE STAIRWELL PRESSURIZATION TEST
The test procedures described in this appendix Activate the stairwell pressurization systems by a putting a
apply to systems for stainvell pressurization. ' detector in alarm as required by the contract documents.
Test each pressurized stairwell by conducting the following
EMERGENCY POWER steps.
If standby power or other emergency power has a. With all stairwell doors closed (except for
been provided for the operation of the stainvell pressur- the exterior ground floor door if it is
ization control system, acceptance testing shall be con- required to be opened upon system activa-
ducted with emergency power and normal power. tion), measure and record pressure differ-
During one test started under normal power conditions, ences across each closed stainvell door.
the normal power shall be shut off to determine the abil- b. Open tlie exterior ground floor stairwell
ity of the stairwell pressurization systems and all associ- door (except if the exterior ground floor
ated systems to properly operate under standby power or door is required to be opened upon system
other emergency power. activation), and measure and record pres-
sure differences across each closed stairwell
NORMAL OPERATION TEST door. For stainvells without a ground floor
exterior door, another highly severe open
With all building HVAC systems in normal opera-
door condition must be tested. This can be
tion, any zoned smoke control systems shut off, and the
an exterior door not at the ground floor or a
stairwell doors closed, measure and record the pressure
large flow path to the outside created by
differences across each stainvell door. The sign conven-
opening the stainvell door and other doors,
tion for all pressure difference readings in the stairwell
including an exterior building door.
tests is: a pressure dityerence resulting from a flow from
the stairwell is positive, and a pressure difference result- c. Open an additional stainvell door, and mea-
ing from a flow to the stainvell is negative. sure and record pressure differences across
Evaluate these pressure differences to detennine each closed stainvell door. Repeat this step,
that they are appropriate for the balanced HVAC system. ope!ling anotiier door each time, until the
Generally, this would be about 0.01 inches of water required number of d o o k is opened. The
gage, but pressure differences as large as 0.03 inches required number of doors is that nuniber
water gage are not a causc for concern. However, higher that must be opened during testing as stipu-
pressure differences may occur for special systems such lated i l l tlie applicable codes or contract
as those intended to control airbornc pollutants. Addi- docu~nents.
tionally, greater pressure dillkrcnccs can be caused by d. With the required number of doors opened,
stack effect (as explained in Chapter 5). clicck flow direction tliroi~gh open door-
Appendix 1-Test Procedures for stairwell Pressurizatiori Systems

ways using a 6 ft strip of tissue paper pressure differences are too low after these
secured at the top of the door h m e . actions, excessive air leakage paths in the
e. Check that the measured pressure difference construction should be filled, caulked, or
is within the acceptable range, as defined in sealed as appropriate. (Often it is very diffi-
the contract documents. If the pressure dif- cult to locate leakage paths in buildings.
ference is not in the acceptable range, dou-
Chemical smoke from smoke bombs can be
ble check that the states of fans, dampers,
used to find these leakage paths. The stair-
and doors is as required. if any of these were
not as required, they should be fixed and the well is filled with chemical smoke and pres-
zone retested. After this, if the pressure dif- surized, while the low-pressure side of the
ference is not acceptable, the flow rate of air stairwell barriers is examined for smoke
to the stairwell in question should be mea- leakage that indicates the location of a leak-
sured and adjusted as appropriate. If the age path.) Then the zone should be retested.
Principles of Smoke Management

Table J-1:
Test Work Sheet-Pressurized Stairwell
Project
Stairwell No.
Test Agent:
Pressure Difference - Flow Direction
Doors in Pressurized Stairwell (inches of water gage) From Stair To Stair

Comments:
Appendix K

Test Procedures for


Zoned Smoke Control Systems
SCOPE gage are not a cause for concern. However, higher pres-
sure differences may occur for special systems, such as
The test procedures described in this appendix
those intended to control airborne pollutants. Addition-
apply to zoned smoke control systems that are either
ally, greater pressure differences can be caused by stack
dedicated systems or part of systems for heating, venti-
effect (as explained in Chapter 5).
lating, and air conditioning (HVAC).
SMOKE MODE TEST
EMERGENCY POWER
Each smoke zone is to be individually tested by per-
If standby power or other emergency power has forming the following sequence.
been provided for the operation of the zoned smoke con-
Activate smoke control system operation in
trol system, acceptance testing shall be conducted with
the zone. This should be accomplished by
emergency power and normal power. During one test
putting one of the detectors into alarm that
stated under normal power conditions, the normal power
are intended to activate the snioke control
shall be shut off to determine the ability of the zoned
system in that zone.
smoke control systems and all associated systems to
Check that the operation of fans is as
properly operate under standby power or other emer-
required by the contract documents.
gency power.
Check that the position of smoke dampers is
SMOKE CONTROL DIAGRAM as required by the contract documents.
Also, check that any smoke dampers
Identify the exact location of each smoke control required to be closed are fully and tightly
zone. If it is not part of the building plans, make a closed.
smoke control zone diagram of the building. This dia- Check to verify that all doors required by
gram should include the locations of all zone boundaries the contract documents -:o be closed during
and of all doors in those boundaries. smoke control system operation are fully
closed and that they operate freely, allowing
NORMAL OPERATION TEST use during evacuation without becoming
With all building HVAC systems in normal opera- jammed in their door frames. This should
tion, the zoned smoke control system shut off, and the include doors in the boundary of the smoke
smoke barrier doors closed, measure and record the zone being tested.
pressure differences across each smoke barrier door. Measure and record pressure difTerences
Evaluate these pressure differences to determine that across all the closed doors in the boundary
they are appropriate for the balanced HVAC system. of the smoke zone being tested. Pressure
Generally, this would be about 0.01 inches water gage, differences resulting from air flowing to the
but pressure differences as large as 0.03 inches \vater snioke zone being tested are to be recorded
Appendix K-Test Procedures for Zoned Smoke Control Systems

as positive values, and pressure differences age path. Exterior walls, interior partitions,
resulting fiom air flowing fiom the -smoke floors, and ceilings, including areas above
zone being tested are to be recorded as neg- suspended ceilings, must not be overlooked
ative values. when hunting for excessive leakage areas.)
Check that the measured pressure difference Then the zone should be retested.
f.
is within the acceptable range, as defined in g. Test for smoke feedback into supply air.
the contract documents. If the pressure dif- Place six smoke bombs (three-minute dun-
ference is not in the acceptable range, dou- tion size) in a metal container, simulta-
ble check that the state of fans, dampers, neously ignite all bombs, and locate
and doors is as required. If any of these are container near exhaust inlet in smoke zone
not as required, they should be fixed and the being tested so that all of the chemical
zone retested. After this, if the pressure dif- smoke produced by the bombs is drawn
ference is not acceptable, the flow rates of directly into the exhaust airstream. Check
air to and from the smoke zones in question that air supplied to other zones of the build-
should be measured and adjusted as appro- ing has no trace of chemical smoke. If
priate. If the pressure differences are too low chemical smoke is detected in this supply
after these actions, excessive air leakage air, its path should be determined, the path
paths in the construction should be filled, should be blocked, and then the smoke feed-
caulked, or sealed as appropriate. (It is often back test should be conducted again. (The
very difficult to locate leakage paths in two most likely causes of smoke feedback
buildings. Chemical smoke from smoke
are a leaky or party opened return air
bombs can be used to find these leakage
damper and an outside air inlet located in
paths. The high-pressure sides of smoke
the vicinity of the exhaust air outlet.)
barriers are exposed to heavy concentrations
of chemical smoke, while the low-pressure h. Make sure that this zone has been returned
side of the barrier is examined for smoke to its normal setting before continuing to
leakage that indicates the location of a leak- test other zones.
Principles of Smoke Management

Tz51e K-l:
Test Work Sheet-Zoned Smoke Control System in Normal Operation
Project:
Test Agent: Date:
Pressure Difference Flow Direction
Doors of Smoke Control Zone (inches of water gage) From Zone To Zone
1

Comments:
Appendix K-Test Procdures for Zoned Smoke Control Systems

Table K-2: ~.
-

Test Work Sheet-Zoned Smoke Coritrol System in Smoke Control Mode


Project:
Test Agent:
- Date:
Yes No
Fans operating appropriately -
Smoke dampers in required position
Pass feedback test

Pressure Difference Flow Direction


Doors of Smoke Control Zone (inches of water gage) From Zone To Zone

Comments:
Appendix L

Inspection Procedures for


Atria Smoke Exhaust Systems
SCOPE as specified. CAUTION: Because the char-
acteristics oJ duct connectors and flexible
The inspection procedures described in this appen-
duct connectors are d~fferent,one should
dix apply to atrium smoke exhaust systems. These pro-
not be substitutedJor the otheu.
cedures are of a general nature, intended as a guide for
the development of specific procedures for individual d. Check duct coverings and linings to verify
smoke control systems. These procedures address the that their fire safety requirements are as
major components of smoke control systems but, by specified. Check that duct coverings do not
their general nature, cannot address all possible compo- conceal any service opening.
nents. In this appendix, the phrase "as specified" is used e. Check direct access and inspection provi-
to mean as specified in accordance with a contract of sions. Service openings and telescoping or
documents, a code, or some other standard or standards removable duct sections are used for direct
access and inspection. Check that a service
that have been agreed upon by the owner, designer,
opening or a telescoping or removable duct
builder, code offkial, and other involved parties.
section is provided in ducts, as specified
adjacent to fire dampers, smoke dampers,
AIR-MOVING EQUIPMENT and smoke detectors. Check that these
a. Check ducts to verify that materials of duct access openings are identified with letters as
material and construction are as specified. specified. Check that service openings are
b. Check duct installation. Duct installation, provided in horizontal ducts and plenums
including the hangers, must not reduce the where specified.
fire resistance rating of structural members f. Check air filters to verify that they have the
and of assemblies. Frequently, structural classification specified.
members and assemblies have fire protec- g. Check that the location, fire protection rat-
tive coverings. such as drywall construction ing, and installation of fire, ceiling, and
or a sprayed-on layer. Check that ducts are smoke dampers are as specified. Generally,
installed in such a manner that these protec- fire, ceiling, and smoke dampers should be
tive coverings are not damaged. Check that installed in accordznce with the conditions
of their listing and the manufacturer's instal-
clearance from ducts to combustible con-
lation instructions that are supplied with the
struction is as specified. In addition, check
damper. Further check installation by
that where ducts pass through walls. floors, removing the fusible link (where applicable)
or partitions, the openings in construction and operate damper to verify that it fully
around the ducts are as specified. closes. It is desirable to operate dampers
c. Check that installation and materials of duct with normal airflow to ensure that they are
connectors and tkxible duct connectors are not held open by the airstream. Remember
Appendix L-Inspection Procedures for Atria Smoke Exhaust Systems

to reinstall all hsible links that have bken the smoke control system are of materials
removed during inspection. and installation as specified.
b. Check automatic controls. Check that
CONTROLS devices for automatic activation and deacti-
vation and control of the smoke control sys-
a. Check manual controls. Check that devices tem are of materials and installation as
for manual activation and deactivation of specified.
Test Procedures for
Atria Smoke Exhaust Svstems
SCOPE to properly operate under standby power or other emer-
gency power.
The test procedures described in this appendix
apply to systems for atrium smoke exhaust systems. EXHAUST OPERATION TEST

EMERGENCY POMrER With all building HVAC systems in normal opera-


tion and any pressurized stainvells, zoned smoke control
If standby power or other emergency power has systems, and other smoke management systems shut off,
been provided for the operation of the atrium smoke activate the atrium smoke exhaust system by a signal
exhaust system, acceptance testing shall be conducted from a smoke detector or initiating device. After activa-
with emergency power and normal power. During one tion, determine that the smoke exhaust fans are operat-
test started under normal power conditions, the normal ing as intended. The volun~etricflow of the smoke
power shall be shut off to determine the ability of the exhaust fans should be measured before the eshaust
atrium smoke exhaust system and all associated systems operation test.
Principles of Smoke Management

Index Carbon dioxide (CO2) 27,36,37, 252


Carbon monoxide (CO) 8, 27, 34, 36, 38, 252, 254,
256, 271
A Carboxyhemoglobin (COHb) 38
Acceleration of gravity 66, 90, 93, 121, 122, 183, 184,
Chimney effect (see stack effect)
191, 196,197,207,218,219,222,243,261,268,321
Church Street fire tests 4
Activation 8, 148, 154, 168, 205, 206, 208, 236, 249,
Clear height 202, 203
277,278,356,36 1,370,371
Colebrook equation 10 1
Air
Commissioning 3, 7, 9, 105, 146, 152, 161, 167, 175,
density 67,74,78,79,81,82,92,93, 143, 153, 158,
235,236,247
191, 195-196,207,241
Communicating spaces 197,210,224
gas constant 67, 190,261,268
Compartmentation 2, 3, 5, 6, 32, 87, 129, 172, 180,
properties 97,220,268,269
199, 291
specific heat 269- 270
Computational fluid dynamics (CFD) 3, 197, 247, 250
Airborne matter 2,63
Confined Flow 190
Airflow 2-4, 6, 70, 71, 74, 78, 79. 87-95, 97, 109, 112,
Conservation of energy 123, 125,219
113, 115, 117-119, 120-122, 142, 148, 150, 154, 158,
Conservation of mass 84, 121, 123, 125, 219, 229,
169, 172, 173, 175, 179, 181, 197, 206, 207, 210, 213,
291, 330
226, 235, 236, 238-240, 247, 257, 289-293, 295-301,
Conservation of momentum 219,228,230
3 12,369
CONTAM 119-122, 130, 132, 137, 139, 154, 155, 161,
Anemometer 78,240,241,245
165, 180, 206, 257, 289, 290, 292, 293, 295-298, 312,
ASCOS 119, 120, 122
320,337,338,349
ASET 120, 122, 123, 126, 199,202,203,249,257,321,
323,329-333 Contaminant 3,87, 88, 120-123,243,290,291, 298
ASMET 120, l26,32 l-323,329,33 1 Control volun~e123, 125, 158, 188, 189, 225, 244
Atria 4, 8, 120, 131, 181-185, 189, 192, 195, 196, 199, Convective fraction 24, 182, 184, 202, 204, 205, 223,
201, 206, 207, 2 10, 215, 2 17, 221, 223, 225, 253, 254, 245,321,322,325,327
274,275,322,323 Critical air velocity 89, 90, 244
Atrium
mechanical exhaust 199
D
Dampers6,9, 111, 113, 114, 117, 139, 169, 175, 178,
natural venting 4, 190, 199, 203,207
236, 357-359, 362,366, 369
smoke filling 129, 199,201,205,221,248,272,323
balancing 1 17
Attenuation coefficient (see extinction coefficient)
barometric 148, 149, 168
AZONE 120, 123, 195, 199? 200, 202, 203, 205, 206,
bypass 149
211-215
chatter 149, 169
control 1 17
B
Barriers 5,87.88,2 10,235,236,254,355,357-359,362, curtain 1 17
366 fire 3, 117. 149,254,257, 355, 357-359
Base fuel package 23- 26 leakage classification 1 18, 178
Benioulli's equation 93, 240 multi-blade 117, 1 18
Boundary conditions 229, 232, 233 return 178, 179, 366
Boundary layer 56, 75, 78. 226.230, 245, 255 smoke 3, 79, 87, 1 17, 1 18, 178, 179,257, 355-360,
Buoyancy 2-4, 66, 71, 73, 74. 79. SO. 87, SS, 92, 107. 365,368,369
129, 150, 175, 176, 179, 181, 189-191, 195, 207, 217, Darcy-Weisbach equation 101
220,221,251,329 Decision tree 5, 6
DETACT-QS 2 l , 120, 126
C DETACT-T2 2 1, 120
Calorimeter 13, 14, 25, 252 Detectors 19, 127, 169, 247, 250, 365
cone 14,248 Diameter
open air 14, 15 fire 182, 184,325-327
oxygen consumption 13. I ? hydraulic 92, 93, 95, 96, 101, 104, 243, 299
room 14 plume 183
Index

Differential pressure (see Pressure difference) airfoil blade 115


l
Differential pressure instruments 237 axial 115
Dilution 2,3,45-47,87,88, 130, 172, 177,243 backward flow 115, 1 17 I
Dimensional Analysis 2 17 backward rotation 115, 117
Dimensionless groups 2 19,330 centrifugal 114, 115, 117, 141, 146, 148, 154
Door-opening force 105-107, 145 exhaust 111,113,120,169,172,175-177,205,206,
Duct 3, 8, 79, 101-104, 111, 112, 114, 117, 118, 140, 244,245,371
149, 169, 173, 178, 179, 236, 239, 243-245, 253, 255, forward curved 115
272,290,355- 359,369 propeller 115, 141, 142
return 113, 179
Duct, access 357- 359
roof-mounted 11 1, 140, 141
supply 7, 1 13, 154, 179,225
E temperature 176, 177
Economizer 1 13
tubeaxial 115, 1 17
Egress 7, 27, 5 1-53, 56, 57, 59, 60, 88, 107, 120, 126,
vaneaxial 115, 117
161,244,249,253,254,277,329
variable flow 6, 169
Elevator 1- 4, 49, 63, 80-82, 89, 97-99, 133, 139, 142,
FAST21, 120, 126, 130, 1)2, 137,254,271
143, 155, 157-159, 161, 165-169, 171, 172, 236, 247,
Fire
248, 250-252, 256, 272-274, 277-279, 281- 287, 291- building 3,4,5,7, l l , 13,29,36,49,63,7 1, 79, 88,
293,295,301,338,358 90, 107, 131, 139, 157, 166, 167, 177,237,
car motion 68,69, 158, 277 251,252,257,271-273,275- 277.355
evacuation 119, 120. 157, 158, 161, 165, 166, 167, design 11,21, 129, 180, 188, 199,203, 205. 207
277,278,285,287 fighters 8,81, 139, 149, 166, 167
piston effect 66,79, 129, 160,252 flaming 8, 13,29, 32,33,34,237
ELVAC 119,120,277,284,285,287 fully developed 13, 18, 37, 133, 188
Energy conservation 7, 11l , 149, 157 growth coefficient 22.245, 221
English units (I-P units) 3, 259, 265, 268, 282, 322, growth time 206
325, 334 research tower 7 1, 96
Evacuation 1, 3, 6, 7, 27, 29, 37, 48-52, 56-63, 87, 119, 'scenario 2 , 21,4 1 , 50, 129, 249
120, 130, 133, 140, 141, 146, 148, 157-159, 161, 165- size 7, 11,21, 192, 250
168, 175, 199- 202,205, 207, 244, 250- 252, 254, 274, smoldering 8, 29, 237
277-279,284,285,287,365 spread 87, 172, 185,257
component-by-component 57,59,60 sprinklered 2, 7, 19, 107, 180, 188, 232, 237, 252
constrained flow 57 steady 1 1, 21, 192, 200-203, 205, 207, 2 11, 215,
density 6,52,53, 55, 56, 58-61 323-325
empir~calcorrelations 5 I suppression 5, 6, 7, l 1. 19, 25, 9 1, 129, 199, 249,
252,253, 257.332
hydraulic a~alogy5 1. 56
test 2,4,23,28,44, 73.21 7,248,251,252,255,256,
velocity 5 1,52,53, 55,58-62
271,273, 274
Evacuation 53 t-squared 18, 21-23, 192, 201, 202, 214, 215, 329,
Exhaust fan 176 332
Exhaust inlets, number 193, 194,2 10,213, 244 unsteady 21, 192,201,203,205,207,21 l , 324,326
Exhaust inlets, separation 175, 194 ventilation controlled 13, 18, 188
Expansion 66,74,89, 129. 172, 175 Fire Dynamics Simulator (FDS) 23 1, 253
Exponential flow equation 94, 96, 97, 243 Flame height 19, 182-186, 204, 205, 244,321,322,325-
Extinction (attenuation) coefficient 28, 29,3 1, 32, 245 327
Flashover 1 1, 13, 254
F Flexibility 6, 7 l
Fan 2-4, 6, 9, 10, 66, 79, 87-89, 92, 109, 111, 112, 114, Flow area
121, 129, 139, 140, 146, 138, 149, 152, 158, 161, 167- effective 63-66, 70, 79-51, 143-145, 150, 152. 161,
169, 171, 175, 177, 236, 737, 239, 247, 251, 272, 274, 172-174,243
289-295, 297, 301, 315, 316, 319, 357-360, 362, 365, parallel paths 63-65
366,368 serles paths 63-65
Principles of Smoke Management,

Flow coefficient 4, 63-66, 70,78,79, 81, 82, 93, 94, 96, Manual stations 8
97, 103, 133, 145, 152, 159, 163, 164, 174, 191, 196, Mass optical density 28,30,32,34,46,47, 131, 133,245
243,338 Metric units (see International system units)
Fractional effective dose (FED) 36,46, 131,243 MGM Grand fire 1,71, 157,248,257
Fractional incapacitating dose (Fm) 40,41, 44, 131 Modeling
Friction losses 66, 93, 143, 150, 161, 163, 164,338 detector activation 120, 126, 192,226, 227
Fuel package 14, 19,23-26, 129,253 Froude 21 7,221,222,224
network 104, 180
G pressure 221,222
Gas law (see Ideal gas law) salt water 256
Governing equations 2 19,220,225,230,231 saltwater 221
turbulence 229- 232
H zonefire4,5, 180, 181,211,274
Haber's Law 36
Hazard analysis 3, 5, 7, 61,87, 119, 122, 129, 130, 131, K
133, 168,207,248 Navier-Stokes (NS) equation 94
Heat exposure 3,27,44,45,47, 130, 131, 133,207 Neutral plane 63,67, 70,71, 73,74, 82- 85,243,273
Heat release density 22, 184,244 Newton Raphson method 102
Heat release rate (HRR) Newton's second law 2 18
automobile 17 N-Gas model 39,40,42,243,244,252
Christmas tree 14, 15 Nomenclature 277,321
cribs 17
furniture 13-15, 21 0
kiosk 14 Objectives, smoke management 5
pallets 17, 22 Open doors 74,87,97, 105, 140, 141, 154,338
peak 14, 15, 17,25 Optical density 28,29,3 1,32,34,46,47, 13 1, 133, 245
sprinklered fires 19 Orifice equation 70,93, 94, 96,97, 100
Heat Transfer Scaling 223 Oxygen (02)8, 13,37,38,39,41,42, 74, 79, 89,90-92,
Height limit 145-147,243 175,250,252,254
HVAC 6-10, 79, 88, 1 1 1-113, 115, 117, 123, 129, 139,
172, 175, 176, 178, 179, 226, 236, 250, 25 1, 255, 360, P
361,365 Panic 49,50,25 1,254,255
Hydrogen bromide (HBr) 34,36 Percentage obscuration 28-30,245
Hydrogen chloride (HCl) 34,36,250 Perfect gas law (see Ideal gas law)
Hydrogen cyanide (HCN) 34,36,252 Physical (scale) modeling 130, 197, 217,219, 22 1
Pirot tube 240,241
I P!ugholing 120, 181, 193-195, 210,211,213,244, 245
Ideal gas law 67, 143 Plume
Ignition 2, 5, 11, 19, 22-25, 34,45, 124, 129, 223, 237, average temperature 188, 189,208,325,326
247 axisymmetric 181-186, 188, 189, 199, 204, 21 1,
Inspection 88, 235, 236,355-360, 369, 370 244,245
lnternational system (SI) units 3, 259, 261 balcony 186,187, 197,204,257
centerline temperature 126, 182-184,322,323,325,
J 327
JET 2 1, 120, 122, 126, 127,249 corner 185, 186, 188
Johnson City Retirement Center fire I maximum height 189,245
wall 185, 186, 188
; L window 188,204
LAVENT 2 l , 120, 122, 126 Poiseuille Flow 94
, Leakage area (see flow area) Post-flashover fire 13,27 1
Power law 75
M
Manometer 238,24 1 Prandtl number 2 17, 2 18, 220
Index

Pressure difference Stack effect 66, 70, 71, 73, 79-84, 107, 108, 129, 142,
average 142, 145, 162 167,179,251,273,361,365
critical 158, 159 Stack effect, normal 66,67,175
design 107, 109, 162, 168, 172, 175,338 Stairwell pressurization
Pressure sandwich 8, 17 1 analysis 147
Pressurization 2-5,7,8,87&9,97-99, 105, 113-115, 119, compartmentation 141
120, 122, 129, 139, 140-146, 148-150, 152-155, 157- multipleinjection 140, 141, 146, 148, 150
159, 161, 165-169, 173-176, 180, 210, 226, 237, 248, pressure profile 142, 145, 146
249,251,256,272,273,275,292,293,338,361 single injection 140
Pull box (see Manual station) vestibules 141
Purging 87, 88, 149, 168, 172 with open doors 146, 148, 150
Stairwell, pressure losses 104
R Standard atmospheric pressure 67, 73, 93, 120, 190,
Radiant fraction 24 261, 268
Reliability 8, 9, 62, 166 Stratification 207,208,236,249
Remote control center 8 Symmetry 104,105, 146, 152,229,283
Resiliency 6, 7
Response time index (RTI) 20,21, 127,244 T
Reynolds averaging 229 Temperature, conversion 261
Reynolds number 78, 92-96, 101, 217, 220-224, 240, Thermal inertia 223, 224, 244
244,299 Thermal radiation (radiant heat flux) l l, 23, 24, 27, 45,
Roosevelt Hotel fire 1 47,48,125, 130, l31,255
Roughness 101-103,245,300 Thornas's equation 89,90,92
Time lag
S ceiling jet 191, 192, 205
Safety factors 7, 146, 152, 161, 167 plume 192, 205
Scaling relations 217, 221-223 Toxicity 3, 27, 34,36-39, 42,43,45, 47, 109, 13 1. 133,
Shopping malls 92, l81,252,272-275 135,250,252,255,256,271
Similitude 217,219,22 1 Tracer gas 237
Smoke Transient fuel 21,9 1
backflow 88-92,181,244,256 Transmittance 27-29,244
bombs (see Smoke, chemical)
chemical 236,237,362,366 U
dampers (see Damper, smoke) Units of measurement 259
definition 27
detectors 8,208,209, 224, 249, 355, 369 v
exhaust 4, 5, 7, 87, 114, 123, 129, 149, 169, 175, Vector 227,228,247,248,250
194, 195, 203-206, 210. 214, 250, 252, Vestibules 141
369,371 Virtual origin 182-185, 245,250, 321-327
filling 129, 18 1, 199, 200, 201, 205, 248, 272,323 Viscosity, dynamic (absolute) 93,94,217-2 19,227,228,
horizontal flow 126, 195, 196,295, 301 23 1,245,268
layer interface 122, 190, 195,204, 332 Viscosity, kinematic 92,95,96, 218,245
shaft 3,87, 142, 149,150, 169, 172, 175,256 Visibility 3,27,29,3 1,32,34,45-48, 130, 13 l, 133. 134,
venting 3,4, 129, 149, 169, 172, 190,207,210 136,244,251,272
Specific heat, constant pressure 125,2!8,228,265 Volumetric flow 64, 74, 92- 94, 96, 103, 1 12, 144, 150,
Specific heat, constant volume 125, 2 18, 21 9 152, 174, 176, 177, 190, 194, 204, 205, 213, 222: 223,
Specific heat, ratio 20, 46, 89, 125, 177, 189, 204, 2 12, 239,240,244,322,325,326,37 1
219,243,261,269,270,321,330
Sprinkler W
activation 2 Weather data 109, 290, 296, 297
time constant 19,20 Wind 6, 66-69, 74, 75, 78-80, 104, 107-109, 120. 129,
Stack action (see Stack effect) 141-143, 148, 172, 207, 226, 243-245, 248, 249, 251,
Principles of Smoke Management

252,255,256,274,290,292,295,296,299-301,3 12 Zoned smoke control 2, 3,4, 8, 74, 89, 120, 139, 142,
Wind data 78, 109,274,296 149, 171, 172, 175, 178-180, 236, 237,273, 349, 359,
361,365,367,368,371
z
Zero floor leakage idealization 70, 7 1, 142

You might also like