P. 1
Principles of Smoke Management

Principles of Smoke Management

|Views: 352|Likes:
Published by pitong_manning

More info:

Published by: pitong_manning on Sep 10, 2011
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

07/31/2013

pdf

text

original

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 management, including pressurized stairwells, pressurized elevators, zoned smoke control, and smoke management 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 software for ar~alysis smoke management systems. of

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 niovement 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 University. 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 a registered professional engineer in the e 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 Protection 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 physics from Ursinus College. Dr. Mike has served as a research fire prevention engineer at the Building and Fire Research Laboratory, National Institute of Standards and Technology, as the fire protection engineer for Fairfax County, Virginia, 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 American Society of Civil Engineers. He is the chairman of the NFPA Technical Committee on Smoke Management 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
Editor

Barry Kurian

manager
Jayne Jackson
Pi-od~ictioi~ Assistant

Erin Howard
Assistant Editor

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
111

Chapter 7-Air Moving Equipment and Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 8 . 4 omputer Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 9-Hazard Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

119 129

Chapter 10-Stainvell Pressurization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 Chapter 1 l-Elevator Smoke Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 12-Zoned Smoke Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 13-Fundamental Concepts for Atria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 14-Atrium Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 15-Physical Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 171 181 199 217

Chapter 16-Computational Fluid Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225 Chapter 17-Commissioning and Routine Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235 Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243 247

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 ASMET Documentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix EAppendix FASET-C: A Room Fire Program for Personal Computers ................................. 321 329

Appendix GData and Computer Output for Stairwell Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337 Appendix HData and Computer Output for Zoned Smoke Control Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349 Appendix I-Inspection Procedures for Smoke Control Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355 Appendix JTest Procedures for Stairwell Pressurization Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361 Appendix KTest Procedures for Zoned Smoke Control Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365 Appendix L-Inspection Procedures for Atria Smoke Exhaust systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369 Appendix MTest 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 pressurization 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 mechanisms 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 analysis, 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 written 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 subcommittee 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 Institute 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 McGrattan 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 valuable 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 constructi\-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 concepts, 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 contributions to fire science and fire protection engineering.

CHAPTER 1

Introduction

S

moke is recognized as the major killer in fire situations. Smoke often migrates to building locations remote from the fire space, threatening life and damaging property. Stairwells and elevator shafts frequently become smoke-logged, thereby blockin,0 evacuation and inhibiting rescue and fire fighting. The MGM 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 occupants on upper floors were exposed to smoke for hours before rescue. The death toll was 85, and the majority of the deaths were on floors far above the fire.

The MGM Grand is not unique in this respect, as is 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 floor, but the majority of deaths were on upper floors (Figure 1.l).'

l , During the intensive activity of fire fighting and rescue, the locations of some of the bodies are not recorded. Thus Figure 1.1 is limited to the deaths for which the locations were known.

23 22 21 20 19 18 17

L

g
E

8 7

Retirement Center Fire Johnson City. T N Dec 24,1989

4
1

2 I 0 1 2 3

M M Grand Hotel Fire G Las Vegas, N V Nov 21,1980

Deaths

Note: Floors Renumbered for
2 I 0

1 2 3 4 5 6 7 8 9101112131415161718

Deaths

L.,-I

0 1 2 3 4 5 6 7 8

Deaths

Figure I .I Deaths byjloor for three fires where rhefire was locn~ed 017 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 grow and of how much smoke can be produced by a fire. This unawareness extends to many designers and other related professionals. Because such an awareness is necessary to the evaluation of design parameters for smoke management systems, the following example is provided. This example is fire test N-54, performed at the Health Care Test Facility at the National Institute of Standards and Technology Annex in Gaithersburg, Md. For technical details of this unsprinklered fire test, the reader is referred to a report by O'Neill et al. ( 1 980). The floor plan of the test facility is shown in Figure 1.2. In this test, various fabrics representing common clothing materials were hung on wire coat hangers and arranged loosely in a wooden wardrobe. A cardboard box containing crumpled newspaper was placed on the floor of the wardrobe. The test started when the crumpled newspaper was ignited by a match. Following ignition, the left-hand door of the wardrobe was closed tightly while the right-hand door was left partially open resulting in a 3 in. (76 mm) opening along the vertical edge of the door. At one second after ignition, no flame or smoke was visible. At 80 seconds, flames were visible flowing from the top of the wardrobe, a layer of smoke was covering the ceiling of the burn room, and smoke had flowed into the corridor forming a one-foot-thick layer just below the corridor ceiling. At 110 seconds, flames were flowing from the top two-thirds of the wardrobe opening, and the smoke flowing out of the burn room doorway had increased significantly. At 120 seconds after ignition, flames were flowing from the entire opening of the wardrobe door, and the layer of smoke in the corridor and lobby had descended to approximately 4 ft (1.2 m) below the ceiling. Such very rapid fire growth and accompanying smoke production represent a real possibility in .actual wardrobe fires and perhaps even closet fires. Many other fire scenarios are possible. For example, a latex or a polyurethane filled mattress ignited by an adjacent wastebasket fire would reach about the same stage of development in six minutes that wardrobe test N-54 reached in two minutes. Full-scale fire tests by Bennetts et al. (1997) and Lougheed et al. (2000, 2001) have shown that successfully sprinklered fires can continue to bum and produce enormous amounts of dense buoyant smoke after sprinkler activation. While it appears this smoke production is greatest for fires that are shielded from sprinkler spray, some unshielded fires still produced considerable amounts of buoyant smoke. The concept of smoke management has developed as a solution to the smoke migration problem.2 Smoke movement can be managed by use of one or more of the following mechanisms: compartmentation, dilution, airflow, pressurization, or buoyancy. These mechanisn~s are discussed in detail in Chapter 4. The use o f pressurization produced by mechanical fans is referred to as snloke control by NFPA 92A (NFPA 2000). By this definition, stairwell pressurization (Chapter 7), elevator pressurization (Chapter 8), and zoned smoke control (Chapter 9) are all types of smoke control systems. The primary emphasis of this book is on systems that cse pressurization produced by mechanical fans. The use of pressurization to control the flow of undesired airborne matter has been practiced for at least 50 years. For example, it has been used in buildings, such as experimental laboratories, where there is danger of
2. As discussed later in "Preliminary Design Considerations," smoke management is only one of many 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 migrating from one area to another; it has been used to control the entrance of contaminants where a dust-free environment is necessary; it has been used wheremdiation migration and contamination could occur; and it has been used in hospitals to prevent the migration of bacteria to sterile areas. However, the use of airflow and pressurization to control smoke flow from a building fire is a fairly recent adaptation.

INTENT
The primary intent of this book is to provide practical state-of-the-art infoimation to engineers who have been charged with design of smoke management systems. The book is also intended to provide information for the review of designs and development of codes and standards,. This chapter contains general background informati6n; Chapter 2 deals with fire development and the heat :release rate of fires. Chapter 3 discusses the nature of:s.moke, including toxicity, heat exposure, and visibility through smoke. Chapter 4 ciiscusses people movement during fire evacuation. Chapter 5 is devoted to smoke movement in buildings, and the individual driving forces of smoke movement are discussed in detail. Chapter 6 contains a fimdamental discussion of topics that are essential for the design of systems to manage smoke movement. It discusses the mechanisms of compartmentation, dilution, airflow, pressurization, and buoyancy, which are used by themselves or in combination to manage smoke conditions in fire situations. Background information is provided about ducts, fans, fire dampers, smoke dampers, and fan-powered ventilation systems in Chapter 7. Chapter 8 is a description of the computer programs that are used for the analysis of smoke management systems. Chapters 9 through 14 address hazard analysis, stairwell pressurization, elevator smoke control, zoned smoke control, and atrium smoke management. For applications for which these conventional methods are inappropriate, the methods of scale modeling and computational'fluid dynamics (CFD) can be used (Chapters 15 and 16). Chapter 17 addresses the important topic of commissioning and routine testing. It may be noted that pressurized corridors have 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 . concern with pressurized corridors is that if a fire room door is blocked open, the corridor pressurization system can force smoke into other rooms off the corridor. For this reason, pressurized corridors are not generally recommended except for applications where practical
'
~ ~

methods are employed to minimize the possibility of doors being propped open. While advances in tenability analysis have made engineering analysis of smoke shafts feasible, these systems are not included in this book. The idea of smoke shafts is that smoke flows up the shaft due to bgoyancy where the smoke flows away from the building, but the authors have concerns about the fundamental effectiveness of smoke shafts. Further, there seems to be little interest in smoke shafts. The stair systems known as "smokeproof' towers are misnomers, in that there is nothing about them that ensures no smoke migration into stairs. Originally, these towers were separate from the building and were connected to it only by walkways open to the outside. Some versions of these towers used relatively small openings in exterior vestibule walls in place of the separate walkways. In the absence of an engineering analysis of these systems, it can only be stated that the benefits of these systems are questionable. For these reasons, separated stair towers are not included in this book, and it is recommended that the term "smokeproof' towers not be wed.

EQUATIONS AND UNITS OF MEASUREMENT
Considering that this book is primarily intended for design, it seems most appropriate that units should be specified for every equation. However, the topic of smoke management is relatively new, and there is no test to refer to for the derivation of many of the equations used. Further, it was desired that the text be in both Inch-Pound (IP) units and the International System (S[) units. It would be unacceptably cunlbersome to present derivations using both commonly used English units and S1 units. The equations used for derivations are dinlensionally homogeneous, and they can be used with the S1 system, the slug pound system, and the pound mass poundal system (Appendix A). These dimensionally homogeneous equations are easily identified because no units are specified for them in the text. Howe\;er, all of the equations t h a ~ reader is IikeIy to use the for design analysis are given in both English and S1 units. These equations are easily identified because the appropriate units for the equation are specifically indicated in the text.

HISTORY OF SMOKE VENTING
Smoke venting has been used extensively to manage smoke flow during theater fires. The acceptance of such venting resulted from several major theater fires, including those at the Brooklyn theater, which killed 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 m

m

............ :::::.--p ..... ........... ............ ............ m............ m

............. ....... m

iiilial:l

. m . Fire Area 3 on Floor l 0
n m

I
n

m

3 Smoke Shafl 4 ElwatorlSiau Lobby Supply

Figure 1.3 Typical floor plan of the office building at 30 Church Street.
in 188 1; the Theater Royal, which killed 186 in 1887; and the Iroquois theater, which killed 571 in 1903. All of these fires started on the theater stage and resulted in major loss of life in the audience. The Palace theater fire in Edinburgh in 1911 was an exception. In this fire, smoke venting through the stage roof was credited for helping to prevent any loss of life. The buoyancy of the hot smoke forced the smoke flow through the vent openings, and this venting is called natural venting or gravity venting. Over the past few decades, fan-powered smoke exhaust has become the standard for almost all atria in North America. In other areas, such as Europe, Australia, and New Zealand, both natural venting systems and fan-powered exhaust systems have become common for atria. Modem atria smoke management designs are based on engineering analysis developed over the last few decades. These analytical methods are primarily based on research in smoke plumes andzone fire modeling. Information about these analytical methods is provided in Chapters 13 and 14.

Figure 1.4 Typicalfloor plan of 117e NRCC exper-hentalfire tower.

HISTORY OF PRESSURIZATION SMOKE CONTROL
The idea of smoke protection by pressurization systems is .to restrict the movement of smoke from a building fire. To study the effectiveness of pressurization smoke control, the Brooklyn Polytechnic Institute conducted a series of fire experiments at a 22-story office building at 30 Church street in New York City (DeCicco 1973). This building was scheduled for demolition. Materials representative of fuels that would be in an office were burned on floors 7 and 10, as shown in Figure 1.3. This project demonstrated that pressurization could provide "smoke free" exits during large unsprin-

klered fires. The term "smoke free" is used to mean essentially free of smoke, with the possibility of such insignificant amounts of combustion products that tenability is maintained. Other full-scale fire tests also demonstrated that pressurization could provide "smoke free" exits during large unsprinklered fires (Koplon 1973a, 1973b; Butcher et al. 1976). Cresci (1973) describes visualization experiments using a model of the stair shaft at the Church Street building, where stationary vortices \\.ere observed at open doonvays. These vortices are the reason that the flow coefficient through an open stainvell door is about half of what it \i.ould be otherwise. This significant effect on airflow is discussed in Chapter 6. The Research Tower near Ottawa (Figure 1.4) was used for a joint National Institute of Standards and Technology (NIST) and National Research Council Canada (NRCC) study of elevator smoke control. Again, i t was demonstrated that pressurization could control smoke from large unsprinklered fires (Tamura and Klote 1987a, 1987b, 1988; Klote and Tamura 1986a, 1986b, 1987). In the spring of 1989, NIST conducted a series of experiments of zoned smoke control at the Plaza Hotel in Washington D C , as shown in Figure 1.5 (Klote 1990). A zoned smoke control system is a system that uses pressurization to restrict smoke migration to the zone of fire origin. Once again, it was demonstrated that pressurization could control smoke from large unsprinklered fires. An analysis based on first principles of engineering was made of the pressure differences produced by the smoke control'system during the fires at the Plaza Hotel. As is done with zone fire modeling, the pressures n.ithin rooms were considered hydrostatic. The general trends of calculated values were in agreement with the msasurements (Figure l h ) , and this indicates a levc.1 of

Principles of Smoke Management

applicability of zone fire modeling for analysis of pressurization smoke control systems. OBJECTIVES O F SMOKE MANAGEMENT Some objectives of a smoke management system are to reduce deaths and injuries from smoke, reduce property loss from smoke damage, and to aid firefighters. Many designers feel that life safety is the primary objective of smoke management; however, systems have been built with the primary objective of protecting property--especially high-value equipment. Regardless of the objective, the methods of design analysis presented in this book are applicable. Theoretically, a smoke management system can be designed to provide a safe escape route, a safe refuge area, or both. However, a pressurization (smoke control) system can meet its objectives even if a small amount of smoke infiltrates protected areas. For this book, pressurization systems are designed on the basis that no smoke infiltration will occur. Hazard analysis (Chapter 9) can be used for the design of systems that maintain tenability even when people come into contact with some smoke. PERFORMANCE-BASED DESIGN In recent years, performance-based codes have become a topic of considerable attention. Traditional codes prescribe requirements, while performance-based codes require a level of performance. A perforrnancebased design is developed to meet the level of performance 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

barriers, such as partitions and closed doors. Atrium smoke exhausts often are designed to keep smoke from descending below a specific level. Further, various types of smoke management systems can be designed to maintain tenable conditions within specific spaces. PRELIMINARY DESIGN CONSIDERATIONS Smoke management should be viewed as only one part of the overall building fire protection systems. Two basic approaches to fire protection are to prevent fire ignition and to manage fire impact. Figure 1.7 shows a simplified decision tree for fire protection. The building occupants and managers have the piimary role in preventing fire ignition. The building design team may incorporate features into the building to assist the occupants and managers in this effort. Because it is impossible to prevent fire ignition completely, managing fire impact has assumed a significant role in fire protection design. Compartmentation, suppression, control of construction materials, exit systems, and smoke management are examples. The NFPA Fire Protection Handbook (NFPA 1997), SFPE Handbook of Fire Protection Engineering (SFPE 2002), and NFPA 550 (NFPA 1995) contain detailed information about fire safety.

0 '

S

l0

l5

20

25

3 ;

i7me (minutes)

(a) Pressure Difference Near Ceiling

0

00

5

10 15 20 lime (minutes)

25

30

(b) ~ressure Dierence Near Floor

Figure 1.5 Secot7djloor-plnt~oJthe Plnzn Ho~el.

Figure 1.6 Co117par-isoti tneaszrt-ed and calczrlated o/ ~ I ~ S I I d f - t~ n c e s ji-ot~l Plaza Hotel I i s 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 e. Sin~plifiedfir.eprotection 17 decision tree.
FLEXIBILITY A N D RESILIENCY
To help ensure smoke management system performance, the approaches of flexibility and resiliency can be employed. The concept of flexibility consists of using design features that allow for easy adjustment of a smoke management system in order to achieve acceptable performance. A resilient system is one that resists serious adverse effects due to pressure fluctuations. During the design of a new building, the leakage paths throughout the building can only be estimated. Therefore, the smoke management design calculations constitute only an approximate representation of the pressures and airflows that will occur as a result of the smoke management system in the actual building. The introduction of flexibility into a smoke management system allows for variations in leakage from the originally estimated values. Because it is difficult to measure leakage paths in existing buildings, the concept of flexibility is also useful for retrofit of smoke management in existing buildings. In many systems, flexibility can be achieved by the use of fans with sheaves3 to allow several flow rates, a variable flow fan for the same purpose, or by dampers that can be manually adjusted to obtain desired pressure differences. Pressure fluctuations often occur during a fire when doors are opened and closed and when windows are opened, closed, or broken. To resist such fluctuations, resiliency can, be incorporated in a system by use o f
3. A sheave is tlic whcel with a groovcd rim, sonieti~ncs callcd a bclt whecl. By exchanging a sheave for onc of anothcr dinmetcr. thc rotational spced of the fan and its flow ratc are changed.

Many factors will affect the design of a slnoke management system. Before the actual mechanical design of the system can proceed, the potential constraints on the system should be determined and the design criteria established. This section introduces some considerations peculiar to smoke management system design, some of which are merely listed below, since detailed discussion is beyond the scope of this book. However. published works on some of these subjects are cited in the bibliography in Appendix B. Occupancy type and characteristics Evacuation plan Refuge areas Distribution of occupant density Human life support requirements Form of detection and alarm Fire service response-to-alarm cliaracteristics Fire suppression system characteristics Type of heating, ventilating, and air-conditioning (HVAC) system Energy ~na~iagement system Building security provisions Controls Status of doors during potential fire condition Potential lirc threats characInternal compartmentation and arcliitectr~ral teristics Bu~ldmg leakage paths Exterior temperatures Wind vcloc~ty

Principles of Smoke Management

automatic control to reduce the pressure fluctuations. For example, in pressurized stairwells, automatic control can be used in the supply fan bypass system to reduce the effect of opening and closing stairwell doors. An alternative is.to keep the exterior stairwell door open during pressurization. This eliminates what is probably the major source of fluctuations; that is, the opening and closing of the exterior stairwell door. The concepts of flexibility and resiliency are discussed further where they apply to specific smoke management applications. SAFETY FACT0RS
/

W '

/

9,

Smoke management is still a relatively new field, and it should come as no surprise that there is no CO sensus concerning safety factors, which are commonly used in many branches of engineering to provide a level of assurance of system performance. Further, the topic of safe@ factors has attracted little attention in smoke control design. Safety factors for sizing fans of pressurization systems are very different from those intended to maintain a tenable environment in an atrium or other application based on a hazard analysis. If a pressurization fan is undersized, it will not maintain acceptable pressure differences. This should be apparent and corrected during commissioning. Ideally, an analysis of a system intended to maintain a tenable environment would be based on detailed and accurate capabilities of simulating smoke transport, physiological effects of fire-related exposures, human response to fire, and evacuation analysis. However, this technology is not so advanced, and these calculations are of necessity based on a number of conservative assumptions with conservative design parameters. It can be argued that such conservative calculations may result in conservative designs even in the absence of any safety factors. The specifics of the design and the meth06s of analysis would be expected to have a significant impact on any approach to safety factors. ~ e & u s e the absence of any accepted approaches of to safety factors, this topic is not included in the methods of analysis of this book. FIRE SUPPRESSION SYSTEMS Automatic suppression systems are an integral part of many fire protection designs, and the efficacy of such systems in controlling building fires is well documented. However, it is important to recognize that while the functions of fire suppression and smoke management are both desirable fire safety features, they should not be readily substituted for each other. One of the best ways to deal with the smoke problem is to stop smoke production. To the extent that a suppression system

a

slows down the burning rate, it reduces the smoke problem. From fires that are suppressed rather than extinguished, smoke is produced. This smoke can move through the building due to various driving forces discussed in Chapter 5. OII the other hand, well-designed smoke management systems can maintain tolerable conditions along critical egress routes but will have little effect on the fire. In addition to the fact that the systems perform different functions, it is important that the designer consider the interaction between smoke management and fire suppression. For example, in the case of a h l l y sprinklered building, the pressure difference needed to control smoke movement is probably less than in an unsprinklered building, due to the likelihood that the maximum fire size will be significantly smaller than in an unsprinklered building. A pressurization (smoke control) system can adversely affect performance of a gaseous agent (such as halon, CO2, or NZ)suppression system when the systems are located in a common space. In the event that both systems are activated concurrently, the smoke exhaust system may exhaust the suppressant gas from the room, replacing it with outside air. Because gas suppression systems commonly provide a single application of the agent, the potential arises for renewed growth of the fire. A general guideline would be that the gaseous agent suppression system should take precedence over the smoke control system. An extremely desirable feature in such spaces would be the ability to purge the residual smoke and the suppressant gas after the fire is completely extinguished and to replace them with fresh air. This ability to replace the atmosphere in these spaces in the post-fire period is very important from a life-safety viewpoint, since some gas suppressants are asphyxiants at normal design concentrations. ENERGY CONSERVATION The smoke management system must be designed to override the local controls in a variable air volume HVAC system so that the air supply necessary to pressurize nonfire spaces is supplied. Also, if there is an energy management system or a 24-hour clock system, the designer must ensure that the smoke management system will take precedence over the local control system so that the necessary air is supplied or exhausted according to the design approach. It is a good general rule that smoke management should take precedence over energy conservation features in both new designs and retrofits.

Chapter l - Introduction

SYSTEM ACTIVATION System activation is probably the major area of disagreement in the field of smoke control. Primarily, this disagreement is about automatic activation versus manual activation. In the early days of smoke control, there was general agreement that activation of "pressure sandwich" systems should be automatic upon alarm from smoke detectors. Automatic activation by smoke detectors located in building spaces has the clear advantage of fast response. Some building designers and fire service officials began to realize that smoke detectors could go into alarm on a floor far away fiom the fire. Thus, automatic activation by smoke detectors could result in pressurization of the zone in which the fire occurred. This would result in the opposite of the desired operation; that is, smoke would be forced into other zones. As a result, a vocal minority of officials feel that smoke control should only be activated manually by fire fighters after they are sure of the fire location. However, many involved professionals are concerned that such manual activation could be so late in the fire development that significant hazard to life and damage to property would result. Such delayed activation can suddenly transport a body of smoke that is highly charged with unbumed hydrocarbons, carbon monoxide, and other toxic gases and depleted of oxygen to remote locations. This can result in a wave-like movement of toxic gases or flame to remote areas. The most recent view on the subject is that zoned smoke control should be automatically activated by an alarm from either heat detectors or sprinkler water flow. This can only be accomplished if the detector or sprinkler zones are compatible with the smoke control zones. Using heat detector or sprinkler flow signals for activation increases the likelihood of proper identification of the fire zone. For smoldering fires, this approach would result in a significantly longer response time, and smoke detectors would probably be better suited for applications where smoldering fires are of particular concern. However, for flaming fires, it is believed that the response time with this approach would be short enough so that significant benefit would be realized by the operation of the smoke control system. It is hoped that advances in smoke detector technology and application will significantly improve the ability of these detectors to positively identify the fire zone. Throughout all of this controversy, there has been complete agreement that zoned smoke control should not be activated by alarms from manual stations (pull boxes). The reason can be illustrated by the scenario ofa man who, while observing a fire on an upper floor of a building, decides that the first thing he should do is to

get out of the building. On the way down the stairs, he thinks of his responsibility to the other occupants. He stops on a lower floor long enough to actuate a manual station. If that alarm activated the smoke control system, the wrong zone would be identified as the fire zone. Because of the long response time and the maintenance problem of clogging with airborne particles, it is generally agreed that smoke detectors located in HVAC ducts should not be the primary means of smoke control system activation. A means of activation of higher rellability and quicker response time is needed. However, an alarm from a duct-located detector can be used in addition to such a primary means of activation. A signal fiom only this secondary means might be unusual, but it should be able to activate the smoke control system. Most stairwell pressurization systems operate in the same manner regardless of where the fire is located. Therefore, it generally is agreed that most stairwell pressurization systems can be activated by the alarm of any fire alarm-initiating device located within the building. A possible exception to this is large buildings with horizontal separations, such that smoke is not expected to have an impact on some stairwells remote from the fire. It is recommended that zoned smoke control systems be equipped with a remote control center from which the smoke control system can be manually overridden. This center should be easily identifiable and accessible to the fire department. Such a remote control center allows fire fighters to change the mode of smoke control system operation in addition to system shutdown. Activation of smoke management systems for atria and other large spaces is addressed in Chapter 10. RELIABILITY O F S M O K E MANAGEMENT The intent of this section is to provide insight into the need for acceptance testing and routine testing and the relative importance of system simplicity: The following should not be thought of as an exhaustive treatment of smoke management reliability. Due to the difficulty of obtaining data about the reliability of components of smoke management systems, the simple calculations that follow are only very rough estimates. However, it is believed that the insight gained justifies this treatment despite these limitations. Further, the same reliability concerns that apply to smoke management systems apply to all life safety systems, and the following discussion may be of general interest beycnd smoke management. The discussion is limited to series systems, which are systems that operate only if all the components operate, as is true of many smoke management system designs. Redundancies (such as backup power) are not 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 System Fans 3 No. of Other Components Reliability1 of New System Before Commissioning

Mean Lifez of Commissioned
System (months) 11 6

System 1 2

3 4 5
1.

0 3 5 5

0 3 9 1 8 54

0.97 0.83 05 .6 0 31 . 0.03

46 1 4 8 3

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, the.comof ponents. .-:

Usually, discussions of reliability progress from this point with the assumption that all components operate initially and that failures occur with time after system installation. For this assumption to be appropriate, a program of acceptance testing and defect correction is necessary. Such commissioning must include an installation check of all components, tests of system performance during all modes of operation, repair of defects, and retesting until all defects are corrected. Current construction practices are such that system commissioning is not always this exhaustive. For this reason, attention is first given to reliability of systems without commissioning followed by a discussion of reliability of systems for which all components operate after commissioning.

missioned system is if that component is part of an HVAC system. In hot or cold weather, building occupants 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 system fan will approach unity. Based on field observations, it is believed that other components will have a lower reliability. The following reliabilities were chosen for example calculations for new systems that have not been commissioned: Fans of a forced air HVAC system 0.99 Other components 0.94 These values were arbitrarily selected, but the relative values between them are based on the discussion above. Table I. 1 lists calculated reliabilities of such systems made up of many components. It can be observed from this table that the more components a system has, the less likely the system is to operate before it has been commissioned. The most reliable new system would be one that only uses the HVAC system fans. A large complicated system consisting of many components (Table 1.1, system 5) has very little chance of operating before commissioning. The trend of lower reliability for complicated systems agrees with observations of the author during nunixous field tests of systems of various degrees of complexity. Probably the most important point to be made from this discussion is the need for commissioning of new systems.

RELIABILITY BEFORE COMMISSIONING
For newly installed components, the reliability can be thought of as the likelihood that the component will both be installed properly and be in good working condition when it is delivered to the construction site. There are an enormous number of errors that can occur during manufacture, transportation, storage, and installation that can cause a component to fail to operate. Problcms such as motors wired for the wrong voltage, motors not connected to power, dampers failing to close, fans running backward, holes in walls, and automatic doors failing to close have been observed in newly built smoke management systems. Based on experience \\lit11 tield testing of smoke management systems, it is estimated that the reliability of components i n noncommissioned systems is 0.90 or highcr. An imporlant consideration regarding the reliabilily of a component in a noncorn-

MEAN LIFE OF COMMISSIONED SYSTEMS
For this discussion, all system components are considered to operate-at the end of the commissioning process. A commonly used relation for the reliability of components is the exponential distribution,
R; = exp(-),,r)

.

(1.2)

Chapter l - Introduction

I II

Circuit Breakers DistrobutionTransfomen

"

I

I 1 /

Mechanical

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 life, L, of a system is

selected for example calculations, but their relative values are based on the above discussion: Fans of a forced air I-[VAC system 1 o - per hr ~ Other components
Io-' per hr

Some typical ranges of failure rates of some coniponents and systems are shown in Figure 1.8. It can be seen that failure rates vary over large ranges and that failure rates vary considerably with equipment type. It seems that the failure rate of HVAC system fans would be lower than those of other components. If these fans fail, building occupalits desiring heating o r cooling tend to put pressure on maintenance personnel to get fans repaired quickly. Smoke management systems are only needed for a short time over the life of a building. Thus, when an HVAC system fan is called uron for smoke management duty, it seems that it will be more likely to operate than other components. To account for this, the effective failure rate of HVAC system fans can be thought of a s being much smaller than other components. The following failure rates were arbitrarily

Table 1.1 shows mean lives of systems composed of various numbers of components. It can be observed that systems composed of a few components have long mean lives, while those made up of very many components have short lives. This tends to support the view that simple systems are more reliable, and this view is supported by obsenations in the field. However, it should be cautioned that systems should not be overly simple; that is, they should have the features needed to achieve desired performance at likely conditions during a fire. Further, the above simple analysis did not include the beneficial effects of redundancies. However, it is safe to conclude that unnecessary system complexities should be avoided. The mean lives listed in Table 1 .l also indicate that routine testing and repair of smoke management systems is needed so that the systems will probably be in good working order when they are needed. A similar statement can be made concerning all life safety systems.

CHAPTER 2

Fire and Heat Release

ky 1
robably the most important aspect of a building fire is the heat release rate (HRR). The temperature and amount of gases produced by a fire are directly related to the HRR, and predictive computer models use the HRR as input. When talk about the size of a fire or how big a fire is, they almost always are referring to the HRR. Other indicators of fire size are the fire area and fire perimeter, but neither of these is commonly used to depict how big a fire is in the predictive models that have gained a high level of acceptance in recent years. For these reasons, the term jr-e size is used in this book to mean HRR. The intent of this chapter is to provide basic information about fire size and development that should be helpful concerning evaluation and deterniination of design fires. A design fire is the challenge that a smoke management system is designed to withstand. Because the presence of sprinklers often plays a role in the determination of a design fire, sprinklers are also included. The design fire can be a steady fire or an unsteady one. While the steady fire is not physically realistic, it can result in very conservative designs and it can simplify design analysis.

P

(

quences of a fire after ignition but not with the causes of Ignition. Growth: After ignition, fire growth is determined by the material burning, with little Or no influence from the compa*ment. This stage is characterized by an bundance ofin afor the fire. Figure 2.2 showschair and fire 2 starting air corner of an upholstered an office growing until it spreads to other objects. As the fire grows, the temperature in the room rises. A fire with sufficient combustion air is called a fuel confrolled fire, and such a fire is also referred to as burning infr-ee air. Flashover: In engineering, most processes of interest consist of gradual changes, but flashover is an exception. Flashover is a sudden change from an apparent steady fire confined to a relatively small space to a fire that involves a much larger space, such as the entire room. For the office fire of Figure 2.2 (c), materials throughout the room are subject to thermal radiation from the Flames and the smoke layer under the ceiling. When this radiation is sufficiently high, some of these materials ignite. This is followed by other materials

i
I
l

8

I

STAGES O F FIRE DEVELOPMENT Fires in rooms or other compartments are often described in terms of the stages of fire development, shown in Figure 2.1. These stages are useful in discussing 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 fire begins. Smoke management deals with the conse-

I

I
I

Il II II
II

I

Post Flashover
I
I

I

l f--

Gr~wth

l1

I

Dewy

Time

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 Yield for Room ~ i r e s *
CO yield**

7

Flaming fires in "free air" Fully involved fire (in a room without cellulosic materials on ceiling or upper portion of walls)***

0.04

Measure Temperature, Flow Rate, & Gas Concentrations.

0.2

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. In a very large room, such as an open office floor plan, only a portion of the room may flashover. The smoke layer temperature at which flashover occurs is generally in the range of 930°F to 1300°F (500°C to 700°C). The criteria for flashover is sometimes taken to be a smoke layer temperature of 1100°F (600°C) or a radiant heat flux of 1.8 ~ t u l f st (20 kw/m2) at the floor ~ of the fire r o a n (Peacock et al. 1999). Fully Developed Fire: This stage of fire development has the highest temperatures. For small and medium rooms, the HRR of a fully developed fire depends on the amount of air that reaches the fire. Such a fully developed fire is ventilation cotztt-olled. In a ventilation controlled fire, more volatile gases are produced by the burning materials than can be bumed in the room with the oxygen available, and the fire can be characterized by flames consisting of burning volatile gases extending from open doonvays of the fire room. For very large rooms, as in an open office floor plan, the fire may not ever become ventilation controlled. Fully developed fires are characterized by inefficient combustion and high production of CO (Table 2.1). Decay: As the fuel is consumed, the HRR of the fire and the temperature of the room drop. The fire may change from ventilation controlled to fuel controlled. Strictly speaking, the term post-flashover fire includes both fully developed and decay stages, but the term is often used to mean a fully developed fire. measurements would be meaningless. Various schemes . to keep the load cell from heating up were devised, but they all interfered to some extent with the measurements. The situation was even worse when pieces of burning material would fall from the load cell. To further exacerbate the difficulties with such HRR determinations, many items burned are composites of several different materials, each with its own heat of combustion. For example, a desk might be made of wood, fiberboard, sheet plastic and molded plastic doors, and drawer fronts. Not only do these materials have different heats of combustion, but they burn at different times during the course of a fire. For these reasons, an HRR estimated from measured mass losses is often unreliable.

Oxygen Consumption Calorimetry
In the 1980s, fire research laboratories around the world worked to develop a method of calorimetry that was not subject to the problems of the old method discussed above. The new method is based on the osygen used up in the fire and is called oxygetz conszrt?zptioncaloritnetty (and sometimes oxygetz depletion calorinzetry). While oxygen consumption calorimeters often have load cells, the measurements from these cells are for sepante information and not for calculation of the HRR. The key to this technology is that the heat released per unit oxygen consumed is almost a constant for most materials. Huggctt (1980) found that this heat release constant is 5,630 Btu per Ib of oxygen consunled (13.1 MJ per kg of oxygen consumed). For most materials involved in building fires, this constant has an uncertainty of about 6%. Figure 2.3 shows a calorimeter where furniture is burned under a hood connected to an exhaust, such that all the smoke is drawn into the exhaust. From measurements of the mass flow of exhaust and the O2 content of

MEASUREMENT OF HEAT RELEASE RATE
In the early days of fire research, determination of the HRR during a fire was very crude. Typically, materials were burned on a load cell (scale), and the HRR was estimated from the mass loss and the heat of combustion of the material. If the load cell became too hot, tlie mass

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 calculated. From this, the HRR can be calculated. Because some of the O2 is not completely consumed, gas measurements also include CO and CO2 Parker (1982) presents 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 not always as intended. Some of the smoke may not be captured by the hood, or burning materials may fall off the fire and away from the calorimeter. With such unintended operation, uncertainties in excess of 20% could result. For a well-calibrated calorimeter operated as intended, the uncertainty of measured HRR may be in the neighborhood of 10%. For more information about the uncertainty of ovygen consumption calorimeters, see Stroqp et al. (2000). Open air calorimeters (Figure 2.3) are sometimes called furniture calorimeters because they are often used for furniture. However, they can be used for any fuel package provided that ( l ) all of the smoke from the fire is collected, and (2) the heat released does not damage the calorimeter including the pollution control equipment. Typically, these calorimeters are located indoors to protect the fire from the wind. The hoods are usually l0 to 20 ft (3 to G m) square, but the size is only constrained by the practicalities of construction. Other types of Oz consumption calorimeters are the room calorimeter and the cone calorimeter. The room calorimeter (Figure 2.4) is used when the effects of the walls and ceiling on the HRR are thousht to be signifi-

Time (S)

Figure 2.5 Three kiosk fires iIIzcstrate iypical repeatabiIiry of burni~~g materials (data Ji-onl MifIer-[I 9961).
cant. The cone calorimeter is a "bench scale" laboratory instrument developed at NIST (Babrauskas 1990).

HRR OF SOME OBJECTS
When duplicate objects are burned, there are deviations in HRR as illustrated with the three kiosk fires of Figure 2.5. These kiosks are for selling T shirts. The deviations of HRR are due to a number of factors, including (I) minor variations in arrangement of the Tshirts, (2) variations in composition of T-shirts, (3) variations in the dimensions of the kiosk, (4) variations in materials of the kiosk, and (5) variations in the air currents near the kiosk. However, the shapes and peak HRRs of kiosk curves are similar. Figures 2.6 to 2.19 show HRRs of other objects. The peak HRR of Scotch pine Christmas trees burned by Stroup et al. (1 999) were in the range of' 1800 to 5000 Btu1 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 were in the range ofabout 40 to 620 Btds (42 to 650 kW). All of these Christmas tree fires had rapid growth stages followed by decay as the tree was burned up. Data for a burning dresser (Figure 2.7) and bunk bed (Figure 2.8) were obtained by Mitler (2000). Like the Christmas tree fires, the dresser had rapid growth

Time (S)

Figure 2.6 Scotch pine Chrislrnas tree (adapledfi-orn S~roup al. [ l 9991). et

stages followed by decay. Many other objects b u m 4 under an open air calorimeter will show the same type of rapid growth followed by decay as the material burns UP. Lawson et al. (1984) burned an assortment of furnilure (Figures 2.9 to 2.16). In general, all these curves are of the s a n e generd shzpe as the proceeding HRR 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 differently 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. Madrzykowski and Vittori (1992) burned workstations. These workstations are simulated offke workspaces, including a chair, shelves or a desk, paper, personal computer, and dividers separating the worksta-

O 'V

300

600

960 l ~ O O 1&0 Time (S)

0 1/00

Figure 2.7

Wooder? dresser- fda/n ji-on? hfitler[2000]).

Figure 2.9 Innerspring tnat~ress filled wilh polyurelhane foam (dala fvom Lawson et al. [ I 9841).

5000

5000 4C30 3000

-$

..-.

4000

m 2000
I
I L

-

3000

z
5
0 3 6 0 9 0 1 Time (S) 0 1&0 0 l&

1000 '0 300 600 900 1200 Time (S)

2000 I 1OGO 0 1500 1800

Figure 2.8 HEN/1-eleasc rn/cjur- b~rnk (dalnfiani bed Miller- [2000]).

Figure 2.10 M C I ~ fi-ame chair wilh polyurethane I 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)

10 ; 0

10 ;0

0 1fO 8O

Figure 2.11 Upholstered chair with polyurelhanefoa~n padding and weighing 25 lb (11.5 kg) (datafroni Lawson et al. [1984]).

Figure 2.14 Metal wardrobe w'th cotton andpolyesrer garments (data from Lawson cl al. [1984]).

0 0

300

600

900
lime
(S)

1200 1500 1800
Time (S)

Figure 2.1 2 Upholsfered chair ~ Y t polytrerhn~~e h foam padding and weighing 62 IB (28.3 kg) (dnia.f,.otn Lauson et al. [l 9841).

Figure 2.15 Wardrobe of 0.5 in. (12.7 I ~ I ) p!~-~t~ood wirh cotton nnd po!~wtetgarnienrs (dnro ,/ram Lnwson et al. [l 9841).

8000

-Unfinished
Fire Retardant
1

.8000 6000

.
A

6000 -

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

In

Paint:

1 Coat 2 Coats

- 4000
2000

S 2
K I

K

,

.
1

S ;

8.

.
lime (S)

.foam Figure 2.13 Sofa wit/? po~v~tretl~at~e padding (datafi-0171 Laws017 et al. [19S4]).

Figure 2.16 War-drobeof 0.125 in. (3.2 ~mnj p11.1t~ood ~dth cotton a ~ i d pol~:este~garnze~its(darn ,from Lnwson et al. [1984]).

Principles of Smoke Managemerit

20W

2000 1500 l000

m , 1000
I

.

; 1500 ;

F E
I

500

500

OO

6W

1200 1800 2400 lime (S)

3000 3600

0

Time (min)

Figure 2.17 Two-divider workstation with conventional desk and credenza (data from Madrzykowski and Vettori [1992]).

Figure 2.19 Automobiles (data from Joyeux [1997]).

lime (S)

Figure 2.18 Three-divider wo~kstationwith an open work top and shelf (data from Madrzykm~ski Vettori [l 9921). a17d tion from other spaces. The two-divider workstation (Figure 2.17) has a peak HRR of 1700 Btuis (1800 kW) at 140 S. The three-divider workstation (Figure 2.18) has a peak HRR of 6400 Btuis (6800 kW) at 550 S. A major reason for the higher HRR of the three-sided workstation is probably the increased radiation feedback from the additional divider and the shelves. For further information about the HRRs of workstations, readers are referred to Madrzykowski (1998). Figure 2.19 shows HRR data of automobiles measured 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 other nonnletallic materials. -Because of these higher HRRs, a car fire in a parking garage can ignite an adjacent car. 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

Figure 2.20 Crib made of geometrically arranged sticks.

was used for tests of the smoke management system at the Plaza Hotel (Klote 1990). This crib was made of 144 wood sticks, 1 .S in. (38 mm) by 1.5 in. (38 mm) by 2 ft (0.61 m) long, and it had a peak HRR of aboet 1400 Btuls ( l 500 kW) when burned in free air. The stack of nine wood pallets shown in Figure 2.2 1 has a peak HRR of about 3,500 Btu% (3,700 kW) when burned in free air. Gross (1 962), Block (197 l), and Walton (1988) have burned wood cribs of various sizes and stick spacings. Babrauskas (2002) provides heat release data of cribs and pallets.

VENTILATION-CONTROLLED FIRES

<h3 d d )

As already stated, the HRR of a ventilation-controlled fire depends on the amount of air that reaches the fire. Further, the HRR can be expressed as a function of the openings to the fire room as

where

Chapter 2-Fire and Heat Release

Q

= = =

heat release rate of fire, kW (Btuls); area of ventilation opening, f? (m2); height of ventilation opening, f?(m);

Example 2.1 Ventilation-Controlled Fire

A,

H ,
C ,,

= 61.2(1260).

Equation (2. l) appiies to rooms of normal construction and size with only one rectangular opening. Figure 2.22 shows the HRR of a ventilation-controlled fire as a function of width of the door or other opening. Equation (2.1) provides useful estimates for rooms made with normal construction materials (drywall, concrete, wood, etc.), but it is not appropriate for metal rooms, such as on a ship with steel decks and bulkheads. For large rooms (over 300 ft2 [30 m2]), the appropriateness of Equation (2.1) is questionable. For information about the effects of construction materials and room sizes, see Walton and Thomas (1995). For a number of rectangular openings with the same bottom and top elevations, the heights are the same, and the effective area is the sum of the individual areas.

I

For a room with a single doorway opening that is fully how involved in h, big will the fire be? The doorway opening 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*) , Because the tire is ventilation controlled, Equation (2.1) is applicable.
Q = 6 1 . 2 ~ ~ ~ : / = 61.2(21)(7)1'2 = 3400 Btuk (3600 kW) '

SPRINKLERS
Figure 2.24 illustrates t-squared fire growth with the three possible responses to sprinkler spray: (a) sprinklers overpowered by fire, (b) constant HRR, and (c) reduction of HRR. Sprinklers can be overpowered by an extremely fast growing fire due to burning materials that exceed the sprinkler design. Sprinklers can also be overpowered when the smoke reaching the sprinklers has cooled due to plume entrainment, as-can happen with fires in spaces with ceilings that are relatively high compared 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 esrimate 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 conservative estimate for many applications. Fire decay after sprinkler actuation is more realistic. Fire decay can be expressed as

where

Q
t to,,

=

post sprinkler actuation HRR, kW (Btuls); HRRat sprinkler actuation, kW (Btuls);

= =

time from ignition, s (S); = time of sprinkler actuation, s (S);

Time (a) Sprinklers Overpowered by Fire
I

Conservative Estimate of Constant HRR

= time constant of fire suppression, s (S). 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 gpm/ft2 (0.07 m d s ) of water. They determined that a 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:

r

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
Time (c) Fire Decay After Sprinkler Activation While the information in this section is primarily about sprinklers, it also applies to vents actuated by fusible links and fixed temperature heat detectors. The responsiveness of sprinklers is tested by the plunge test, where a sprinkler is "$mgedWinto a heated 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 fluid, as described in heat transfer texts (Kreith 1965: Incropera and DeWitt 1985). This analysis is based on the assumptions that ( I ) the internal resistance of the sprinkler is negligible, (2) the sprinkler is instantaneously put ill the oven, (3) the convective heat transfer coefkient is constant, (4) the gas temperature i n the

Figure 2.24 Interaction between fire and sprinklers.

flame height is typically less than the ceiling height, and room air entrainnient cools the gases in the w o k e plume. Methods of calculating the plume temperature are in Chapter 13. If the sprinklers do activate, the spray could evaporate before the droplets reach the fuel.

Chapter 2-Fire and Heat Release

oven is constant, and (5) the only heat transfer is from the sprinkler to the gas. The temperature of the sprinkler increases exponentially, as shown in Figure 2.26. The time constant, r, of the sprinkler is

calculated from Equation (2.6). The RTI of standsrd sprinklers varies from about 140 to 280 fill2 s1I2 (77 to 155 m'I2 sln), and the RTI of quick-response sprinklers (QRS) varies from about 50 to 100 fill2 slR (28 to 55 ,lR ,lR). The response time index does not account for conductive heat transfer from the sprinkler. To account for conduction, a virtual RTI can be calculated as
RTI, =
RTI - CRTI

.

where
Z

=
= =
=

time constant, s (S); mass of the sprinkler, Ib (kg); specific heat of the sprinkler, Btuflb"F (Jkg "C); convective heat transfer coefficient, ~ t d f t s "F 2 (w/m2 "C); surface area of the sprinkler, ft2 (m2). where

m

'+1/2

C
h,

RTI, = virtual RTI, fill2 slR (m1' slR);

A
.

=

CRTI= conductivity factor, f i l n / s " ~(m'l2 IS'").
.

The time constant, r, is the time at which the temperature of the sprinkler has reached 63% of the way to the gas temperature. The convective heat transfer coeficient 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.

.

where u is the velocity, Ws (mds). In the plunge test, the time to actuation and the gas velocity are measured. Then the time constant can be calculated from the time to actuation, and the RTI is

:
I

': is time constant I

Time Figure 2.26 Temperatutasfot~ spr-ir~kler-plztr?ge n 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 Wood Cribs

.
X

-

0

200

400

600

800

Time, t -

t,,

(S)

Figure 2.25 Filr decuj' due to spri~ikler aclio/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~rinkler Actuation
Actuation depends on gas temperature and velocity near the sprinkler. In a fire, a jet of hot gases flows radially from where the smoke plume intersects the ceiling. Computer programs have been developed that use correlations for such a ceiling jet to predict actuation time. The program DETACT-QS (Evans and Stroup 1986) assumes that the thermal device is located in a relatively large area, that only the ceiling jet heats the device, and that there is no heating from the accumulated hot gases in the room. The required program inputs are the height of the ceiling abo:!e the fuel, the distance of the thermal device from the axis of the fire, the actuation temperature of the thermal device, the response time index (RTI) for the device, and the rate of heat release of the fire. The program outputs are the ceiling gas tkmperature and the device temperature, both as a function of time and the time required for device actuation. DETACT-T2 (Evans et al. 1986) is similar to DETACT-QS, except it is specifically for t-squared fires. Several zone fire models (such as FAST, LAVENT, and JET) are capable of calculating ceiling jet temperatures and predicting actuation (Chapter 8).

are not intended to be located in the space are referred to as tramientfuels. A few examples of transient fuels are Christmas decorations, paint and solvents in stairwells during redecorating, unpacked foam cups in cardboard boxes after delivery, cut up cardboard boxes awaiting removal, and closely stacked upholstered furniture after delivery. Sometimes, transient fuels remain in place for long periods. Some examples are (1) a number of polyurethane mattresses delivered to a dormitory and waiting for distribution in the next school year, (2) automobiles on display in a shopping mall, (3) boats and campers on display in an arena, and (4) a two-story colonial house built for display inside a shopping mall. Transient fuels must not be overlooked when selecting a design fire. One approach to incorporating transient fuels in a design fire is to consider the fire occurring over 100 ft2 (9.3 m2) of floor space with a heat release rate density of 20 Btuls ft2 (225 kw/m2). This amounts to an allowance for transient h e l s of 2000 Btuls (2100 kW).

Steady Fires
It is the nature of fires to be unsteady, but the steady fire is a very useful idealization. Steady fires have a constant heat release rate. In many applications, use of a steady design fire can lead to straightforward and conservative designs.

DESIGN FIRES
A design fire curve is the description of the development of a design fire that can be used in a fire scenario. The curve is for HRR as a function of time. This curve can be as simple as a constant, and it can also be a simple function of time. The design fire curve can also be a complicated sequence of lesser cunles for some or all of the stages of tire development described at the beginning of this chapter. A fire scenario includes more than just the design fire curve. The word sce17nrio means an outline of events, as in a play or other theatrical production. A fire scenario can be thought of as the outline of events and conditions that are critical to detemiining the outcome of alternative designs. In addition to the HRR and fire location, a scenario could include the type of materials burned, airborne toxicants and soot produced, and people movement during fire.

HRR per Unit Area
Morgan (1979) suggests a typical rate of heat release per unit floor area for mercantile occupancies of 44 Btuls ft2 (500 kw/m2). Fang and Breese (1980) determined about the same rate of heat release for residential occupancies. Morgan and Hansell (1987) and Law (1982) suggest a heat release rate per unit floor area for office buildings of 20 Btds f? (225 kw/m2). For smoke management applications, a heat release rate per floor area of 20 Btuls ft2 (225 kw/m2) is suggested for restricted fuel spaces, and 44 Btuls ft2 (500 kw/m2) is suggested for spaces with furniture, wood, or other 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 Table 2.2 can be useful in determining design fires.

In many spaces, the fuel loading is severely 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 --. objects made of similar materials plus plants. Even for such a /ire1 reswicmf space, there can be an almost unlimited number of combustiblc objects that are in the space for short periods. Such combustible materials that

Unsteady Fires
Fires frequently proceed through an incubation period of slow and uneven growth, followed by a period of established growth as illustrated in Figure 2.27 (a). 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
Material Burned I. 2. 3. 4. 5. 6. 7. 8. 9. 10.
11. 12. 13. 14. I . 16. 17. IS. 19. 20. 21. 22. 23. 24.

Heat Release Density, q kwlrn2 Btuls f$
1400 4,000 - 6,800 10,000 400 1,700 8,500 2,000 1,400 6,200 2,000 1,900 . 14,000 5,400 2,000 3,300 3,400 4,400 6,200 2,000 400 400 420 745 90 90 940 900 125 350 .600 900 35 150 750 175 125 550 175 170 1,250 475 180 290 300 390 550 IS O 35 35 37 66

Wood pallets, stacked 0.46 m (1.5 h) high (6-12% moisture) Wood pallets, stacked 1.52 m (5 ft) high (6-12% moisture) Wood pallets, stacked 3.05 m (10 ft) high (6-12% moisture) Wood pallets, stacked 4.88 m (16 ft) high (6-12% moisture) Mail bags, filled, stored 1.52 m (5 fi) high Cartons, compartmented, stacked 4.57 m (15 fi) high PE letter trays, filled, stacked 1.52 m (5 ft) high on cart PE trash barrels in cartons, stacked 4.57 m ( l 5 ft) high PE fibeglass shower stalls in cartons, stacked 4.57 m (15 ft) high PE bottles packed in item 6 PE bottles in cartons, stacked 4.57 m (15 ft) high PU insulation board, rigid foam, stacked 4.j7 m (l5 ft) high PS-jars packed in item 6 PS tubes nested in cartons, stacked 4.27 m ( l 4 ft) high PS toy parts in cartons, stacked 1.57 m (l5 ft) high PS insulation board, rigid foam, stacked 4.27 m (14 ft) high PVC bottles packed in item 6 PP tubes packed in item 6 PP & PE film in rolls, stacked 4.27 m (14 ft) high Methanol pool, 0.16 m (0.52 ft) diamcter Methanol pool, 1.22 m (4.0 ft) diameter Methanol pool, 1.74 m (5.7 ft) diameter Methanol pool, 2.44 m (8.0 ft) diamc~er Methanol pool. 0.97 tu (3.2 ft) square Silicone transfornler fluid pool, 1.74 m (5.7 fr) diameter Silicone transformer fluid pool, 2.44 m (8.0 ft) dianletcr Hydrocarbon transformer fluid pool. 1.22 nl (4.0 ft) diameter Hydrocarbon transformer fluid pool, 1.74 m (5.7 ft) diameter Heptane pool, 1.22 (4 ft) diameter Heptane pool, 1.74 (5.7 ft) diameter
I . Abbreviations are: PE = polytl~ylenc. = polyslyrsnc. PVC PS 2. Items I tlirough I 0 frorn~fl'~0 (2000). 42 3. ltenis10 tl~rdugh30 rrolii Hcskcs~ad(IYS4). 4. ltcms 25 tlirot~gli28 arc proprietary products
=

.

25. 26. 27. 28. 29. 30.

8
8 83 80 270 280

3.000
3.200
pulyvinyl cliloride. PP = polypropylene. P U = polyurethane.

Nn~rc.

represented by an idealized parabolic equation (Heskestad 1984).

where
Q

It is generally recognized that consideration of the 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 fires. Ncg-lecting the incubation pcriod, the t-squared fire can bc written as

=

heat rclcasc rate of fire, kW (Btuls); firc growth coefficient, k w k 2 ( ~ t u l s ~ ) ; time aficr ignition, S; cfl'cc[ivc ignition time,
S.

a
1

= =
=

*<,

Principles of Smoke Management

where t is considered the time from effective ignition. For I-P units, the following form of Equation (2.9) is often used:

tains, etc;). The key to selecting the items that make up the base h e 1 package is that the radiant flux from buming 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) considers the flame as a small thermal source such that the intensity of thermal radiation is proportional to the inverse of the square of distance from the source. Thermal radiation also is called radiant heatflux. The intensity of thermal radiation is

I

where
Q = heat release rate of fire, Btuk;

t tg

=
=

time after effective ignition, S; growth time, S.

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 this book, the terms slow, mediznn, fast, and zdlra fast fire growth refer to the NFPA 92B values.

where
4 ; Qr
= =
=

intensity of thermal radiation, ~ t u / f s (kw/m2); t~ radiant heat release of the fire, B d s (kW); distance from the center of the fire, ft (m).

Fuel Package Approach
The base fuel package is the maximum probable size of h e 1 package that is likely to be involved in fire for a specific application. A fuel package can be made up of a number of fuel items (sofa, chair, bed, table, cur-

R

Table 2.3: Fire Growth Constants for T-Squared Fires
NFPA 9 2 8

I
) g

NFPA72
Range of

u (8tuls3) cx (l;w/s2)
Slow Medium Fast Ultra Fast 0.002778 0.01 1 1 1 0.04444 0.1778 0.002931 0.01127 0.04689 0.1878

(S) 600

(S)

Ig2400 150 5 tg < 400 150
NIA

300 150 75

Thin Plywood

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

Time, t (a) Typical HRR curve

Time, t (b) Idealized Parabolic curve

0

200

400

600

800

Time From Ignition (S)

Figure 2.27 Fire grorvlh clil.~;cs.

Figure 2.28 Relalion of r-sq~iaredjires 10 sotlze fire 1es1.7(adapled )I-OIU Nelsoti [l 9871).

Chapter 2-Fire and Heat Release

Fire

Oriented Normal

have high radiative fractions. However, for design applications, values of X, = 0.3 and X, = 0.7 are common. The idea of separation distance is useful fo; evaluation of what items should be in the base fuel package. Using the point source radiant model, the separation distance is

The point source model is a good approximation provided that R > 20. Figure 2.29 Point source radiation model.
where The point source radiant model is appropriate provided that the distance from the center of the flame is greater than twice the diameter of the fire (R > 20). The radiant heat release of the fire is

RSD

=

separation distance from the center of the fire to a target, ft (m); intensity of thermal radiation needed for nonpiloted ignition, ~tu/ft2 (kw/m2). S

qi.

=

where
Q
X,.
=
=

heat release rate of the fire, Btu/s (kW); radiative fraction.

Fuel items less than RSD away from the fire would be expected to ignite, and fuel items farther than RsD away would not be expected to ignite. The radiant flux needed for nonpiloted ignition varies from about 0.9 13tu/ft2S (10 k ~ l m ' ) for thin easy-to-ignite materials to ~ 1.8 ~ t u / f St (20 kw/rn2) for thick materials. For a fire, the heat release rate, Q,.. that results in ;, ignition of an object at a distance of R away is

Heat transfer from a flame is by conduction, convection, and radiation. For most fires, conductive heat transfer from the flame is negligible. The radiant fraction can be expressed as

where X , is the convective fraction. The radiative fraction depends on the material burned and the diameter of the fire, and the radiative fraction varies from about 0.1 to 0.6. Low sooting fuels, such as methanol, have low radiative fractions, and high sooting materials, such as gasoline and polystyrene,

For radiant heat transfer where R is less than twice the diameter of the fire, a method other than the point source model is needed. Several texts have general information about radiant heat transfer (Siege1 and Howell 1992; lncropera and DeWitt 1985; Kreith 1965). For information about radiant heat transfer of fire, readers are referred to Quintiere (1998), Drysdale (1985), and Mudan and Croce (1995).

Principles of Smoke Management

11

1)

-

'. "Ruarnnle 2.2 'r-- --a

- - "v

Race Fuel Packaoe --- m-

-

1

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?

Part I: Initial Estimate of Base Fuel Package
Use a radiant flux for nonpiloted ignition of qr,
= 0.3(2960) = 888 = 1.8

I(11

~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) , . ~

For the sofq Q = ,

Bhds (937 kW).

1 l

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

I

I1
1l

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
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)

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 s sofa 2 is expected when the chair I fire also reaches 293 B t ~ d(309 kW). Calculations of the HRR are done graphically on Figure 2.3 1: (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 calorimeter 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).

1

For the base fuel package,

8,. =

%,.Q = 0.3 (3600) = 1080 Btds (l 140 kW).

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~tld not be expected. So the base fuel package and its HRR curve can be ~lscd directly 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 Chair 2 Note:

Sofa 3

7 7
4000

R, = R ,= 3 . 6 f t ( l . i m)
Figure 2.30 Arrangemen! offurni~ure the aft-iutnof in 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 of Examnple 2.2. o

CHAPTER 3

Smoke and Tenability
n this book, the term srnoke is used in accordance .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 when a material undergoes pyrolysis or combustion, together with the quantity of air that is entrained or 0therwise mixed into the mass. The products of combustion usually include particulates, unburned fuel, water vapor, carbon dioxide, carbon monoxide, and some other toxic and corrosive gases. As smoke moves through a building, air mixes into the smoke mass and the concentration of combustion products in the smoke decreases. Including air that is entrained or othenvise mixed facilitates discussions about fire smoke management in atriums and other iarge spaces. Generally. smoke is thought of as being visible, but the above definition includes "invisible smoke" produced by burning of materials that produce little or no particulate matter, such as hydrogen, natural gas, and alcohol. Information about smoke hazards is useful in evaluating'the effects of small quantities of smoke migrating into "protected spaces," and it is useful in evaluating the consequences of smoke migration without smoke protection. This chapter concentrates on smoke hazards due to toxicity, temperature, and smoke obscuration. The hazards of temperature consist of hear exposwe, which can occur when a person comes into bodily contact with e-vposur-e, n.hich can hot gases, and thermal racliafiot~ occur when a person receives thermal radiation from flames or hot smoke that are some distance away from the person. Exposure to toxic gases, heat. and thermal radiation can be a direct hazard to life, and reduced visibility due to smoke obscuration can be a significant indirect haz-

I

ard. Frequently, people become disoriented in fire situations 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. Further, in buildings with balconies, smoke obscuration can result in fatal falls. Smoke management systems can be designed with the objective of providing a tenable environment in the means of egress or at other locations during evacuation. Such a tenability system needs to be designed to meet tenability criteria. Such criteria need to include exposure to toxic gases, heat, and thermal radiation. Further, the criteria often include visibility. As discussed at the end of this chapter, th-e criteria for a tenability design depend on the specific application.

oBSCURAT1oN
Many different methods of expressing smoke obscuration are used in fire science and fire protection engineering, and this section discusses the common methods. There is a lack of uniformity concerning smoke obscuration, and some engineering publications use different terminology or have different mathematical definitions for the same terms. These differences could result in significant errors, and readers are cauticned to take care to verify the exact meanings of obscuration terms used in other publications. The terminology that follows was selected with the intent of being consistent with most technical publications in this field. The fraction of light transmitted through the pathlength of smoke is called the transniittance and is written as

Chapter 3-Smoke and Tenability

where T = transmittance, dimensionless; I, = intensity of light at the beginning of the pathlength; I, = intensity of light remaining after it has passed through the pathlength. The units for light intensity are arbitrary, and such units are unnecessary for discussions of smoke obscuration 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 laser beam, but light meters using less exotic light sources (such as incandescent bulbs) have been used extensively for fire tests. When the atmosphere is "smoke free," the intensity of light remaining after it has passed through' the pathlength 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 beam passing tliroi~ghL L ~ i ~ i b l e is less than one. smoke" Neutral density tiltcrs, wliich allow only a specific fraction 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. O/~fical de17xitj:, is delined as 6

p-tight

Photo
+- Li9hiBe, ! =?

Source

-+- - - -

Wlres To Power Source and Data Acquisition System Figure 3.1 Smoke meter used to measure smoke obscuration.
Wires From

I

Power Source

where a is the extinction coefficient per unit distance in units of ft" (m-'). The extinction coefficient is sometimes called the attenuation coefficient. Percentage obscut-ation is occasionally used and is defined as

where R is the dimensionless percentage obscuration. The specifc optical dens@ is measured in some laboratory smoke tests and is defined as

where

6,
6

= =

specific optical density (dimensionless); optical density per unit distance, ft-' (m-'); volume of the smoke test chamber, ft3 (m3); decomposed surface area of the test sample burned ft2 (m2).

I/, =

Substituting Equation (3.1) and rearranging results in an equation for optical density in terms ol'transmittance,

A

=

where optical density per unit distance. K' (m-'); = transmittance. dimensionless; X = distance of light travel or the pathlength, ft (m). Thc e.vti17ctiori co@cit'rit pcr unit distance is defined as
=

6 T

The specific optical density is a practical measurement of smoke obscuration only when the decomposed area of the sample is well defined. For laboratory tests where the mass loss of the sample is measured, the mass optical densiry is an appropriate measure of obscuration. The mass optical density is detined as

where

4,
Substituting E q i ~ a t i o ~ ~) and rearranging yiclds (3. l

=
=

mass optical density, ft2/lb (ni2/S); optical density pcr unit distance, K ' (m-');

d

Principles of Smoke Management

Table 3.1: Comparison of Different Methods of Expressing Smoke Obscuration
Pathlength Transmittance Percentage Obscuration
X

Optical Density

Extinction Coefficient a

6

V, AM

= =

volume of the smoke test chamber, ft3 (m3): mass loss oftest saniple, Ib (g).

The mass concentration of fuel burned in the test chamber is
nl

- / ' - VC

AM

cern that a disoriented person could fall from a balcony. Because a person falling 5 ni (16 ft) has about a 50% chance of fatality, falls are a serious concern for buildings with balconies. Based on the work of Jin (1974, 1975, 1985), the relation between visibility and smoke obscuration is
S = -

(3.9)

K a

(3.11)

where nyis the mass concentration of fuel burned in units ). of lb/ft3 ( g / ~ ~ 3Substituting this density into Equation (3.8) yields

where S = visibilitj, fi (m); extinction coefficient ft-l (m-'); = proportionality constant (Table 3.3). The visibility is the obscuration threshold, which is the distance at which an object c,n just be seen. The proportionality constant is dependent on the color of smoke, the illumination of the object. the intensity of background illumination, and visual acuity of the observer. Jin conducted tests determining visibility of light-emitting and -reflecting signs. Signs in a smokefilled chamber were observed from outside through a glass window, and the results for illuminated signs are shown in Figure 3.2. White smoke \\as produced by smoldering fires, and black smoke \\.as produced by flaming tires. Visibility through the \vhite smoke was less, probably due to higher light scattering. It is well
=

a K

Table 3.1 lists some values of optical density, extinction coefficient, and percentage obscuration for different path lengths. Equations for conversion between differefit smoke obscuration terms are listed in Table 3.2.

VISIBILITY T H R O U G H S M O K E
When people cannot see because of smoke from a building fire, they walk slowl!.. \vhich can significantly lengthen evacuation time, and they can become disoriented and lost, thus prolonging their exposure to toxic gascs. In atrium fire situations. there is the added con-

Chapter 3- Smoke and Tenability

Table 3.2: Conversion Equations for Smoke Obscuration
Convert To

'

From
Extinction Coefficient Percentage Obscuration

Equation

Optical Density Optical Sensity

6 = 0.4343 a 6 = - log,, (1 - A1100)

Optical Density Optical Density Extinction Coefficient Extinction Coefficient

Specific Optical Density Mass Optical Density Optical Density Percentage Obscuration

6 = 6,mf

a = 2.303 S
a = -log, ( l - A / 100)
X

Extinction Coefficient Extinction Coefficient Percentage Obscuration Percentage Obscuration Percentage Obscuration Percentaze Obscuration

Specific Optical Density Mass Optical Density Optical

a = 2.3036,,m,

ensi it^

2. = 100(1- 10-~")

Extinction Coefficient Specific Optical Dznsity Mass Optical Density

Specific Optical Density Specific Optical Density Specific Optical Density

Optical Density Extinction Coefficient Percentage Obscuration

6, = 6 v , / A

6 =Specific Optical Density Mass Optical Density Mass Optical Density Mass Optical Density Mass Optical Density Optical Density Extinction Coefficient Percentase Obscuration

K Io&~(I-A/100)
Ax

6, = 6,,,m

lA

S,,, = 6 /.m,

Mass Optical Density

Specific Optical Density

6
,,l

=A

6 A

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): ? s = specific optical dcnsity (din~cnsionlcss):& = mass optical density, liZllb (n121g): 1; = volumic oflhe snloke tssr chambcr, lij i (111'): :\.V
m') [ m ,

-

= 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:

h.tl! 1;

I:

.Y

=

distancu of'lipht m x c l or lllc ~ . ;I I; L

l (111). i

~ r i n c i ~ lof s e Smoke Management
.

-

Table 3.3: Recommended Proportionality Constants for Visibility Based on Research of Jin (1974,1975, and 1985)
Situation Illuminated signs Reflecting signs Building components in reflected light
K 8 3

Example 3.2 Visibility of Doors and Walls In Example 3.1, what is the visibility of walls and doors? From Table 3.3, K = 3. Extinction coefficient = 0.207 m-'. From Equation (3.12), S= 31.207 = 14 ft or 4.3 m.

3
Kind d Smhe Black Smoke Black Smoke Whde Smoke Whmte Smoke

Brightness of S i n

0 Irritating Smoke a NonirritatingSmoke

a 2000 cdlm'
a !
SO0 cdlm' 0 2000 c d d

0 Mo cdlm2

E >
OWOQ

>

0.4

I

o 4 p
I

I

I

I

0.5

1 1.5 Ednction Coefficient,a (lh)

0.7

2

21

I

,

,

I

I

I

I

l

0.2

0.3

15 2 . Extinction Coefficient, U (llm)

0.5 0.7

1

3

Figure 3.2 Relatiomhip between the visibility of lightemitting signs and smoke obscuration (adaptedfi-on7Jiu [l 9Sj1).

Figure 3.3 Relationship between visibility of liglitemitting signs ar7d smoke obscuratioi7 forir-r-itating and 17onir-r-itating smoke (adaptedfiotn Jin [ l 9851).

known that scattering of background lighting can significantly reduce visibility of lighted signs, but quantitative data about the effect of background illumination are needed. Jin found that the proportionality constant ranged from 5 to 10 for light-emitting signs. For reflecting signs, the constant ranged from 2 to 4 . Jin indicates that the minimum value of visibility for reflecting signs may be applicable for the visibility of other objects, such as walls, floors, doors, and stairs. Based on Jin's research,.the values of K are listed in Table 3.3. Example 3.1 Visibility of an illuminated

The above information about visibility does not take into account the irritating effects of smoke on the eyes. Jin (1985) conducted tests correlating the visibility and walking speed of subjects exposed to irritating smoke with the extinction coefficient. There are shortcomings with correlating pl~ysiologicaleffects with an optical property of smoke since the effects would seem to be primarily caused by chemical components of smoke. However, the effects of eye irritation are so significant that Jin's work on the topic is discussed below. Figure 3.3 shows the relation between visibility and obscuration for irritating and nonirritating smoke for a light-emitting sign. The irritating smoke was white smoke produced by burning wood cribs; the less irritating smoke was produced by burning kerosene. The visibility 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 their eyes open long enough to read the sign. Figure 3.4 shows the relation between smoke obscuration and \valking speed of people walking down a corridor in irritating and nonirritating smoke. Both eye irritation and smoke density affect walking speed. Walking speed decreases with cxtinction coefficient for both smokes,

1

Fo r:

Table 3.3, K = S.

11 Extinction coelficient is a=2.303 d so a=7.303(.09) = 0.207 11
From Equation (3.1 1 ), S = 81.207 = 39 li (1 2 m). the distance

I

Chapter 3-Smoke and Tenability

but it is much worse for irritating smoke. For an extinction coefficient of 0.4 m-', the walking speed through irritating smoke was about 70% of that through nonirritating smoke. For extinction coefficients greater than 0.5 m-', the walking speed decreased to about 1 ttlsec (0.3 &S)--the speed of a blindfolded .person. The drop in walking speed was because subjects could not keep their eyes open, and they walked in a zigzag or went step-bystep as they held the side wall. Jin (1985) developed an empirical relation for visibility in irritating smoke:
K S = -(Cs- 1.471og,,a) a

chapters on compartrnentatioridesign and atrium design. For laboratory smoke test chambers and simple room calculations, the mass concentration of particulate, my can be calculated from Equation (3.9). The extinction coefficient can be expressed as

where

a

=
=

extinction coefficient, fi-' (m-'); specific extinction coefficient, f&lb (m2tg);

a,
(3.12)

{only for a 2 0.076 ft-' (0.25 m-' ) where

rnp = mass concentration of particulate 1b/ft3(g/m3).

a = extinction coefficient, fi-l (m-'); S- = visibility, f (m); t K = proportionalityconstant (Table 3.3); CS = -0.6255 (0.133).

The specific extinction coefficient depends on size distribution and optical properties of :he particulates. Seader and Einhorn (1976) obtained values for a,,, of 2.1 x 104 ft2/lb (4.4 m21g) for smoke from pyrolysis of wood and plastics and 3.7 X lo4 ft2/lb (7.6 m2/g) for smoke froc: flaming combustion of these same materials. 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-'. From Equation (3.12),
S S = -[-

S

=

visibility, fi (m);

K = proportionality constant (Table 3.3):

.207

.6255 - 1.47 log(.207)] = 15 f (1.6 m) t

a,,, specific extinction coeflicient, ft2/lb( n ~ ' / ~ ) ; =
n7/,

An alternate approach to calculation of visibility from the mass concentration of particulate is obtained from combining Equations (3.10) and (3.1 l ) with the conversion from optical density to extinction coefficient (Table 3.2).

=

mass concentration of particulate lb/fi3 (glm3).

Equation (3.15) relates visibility to the mass concentration of particulate. The comment concerning the utility of Equation (3.13) also applies to Equation (3.15).

0 Irritating Smoke Non~mtating Smoke

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 are in Table 3.4. Equation (3.13) can be useful because the mass concentration of fuel burned can be calculated from a smoke transport model as discussed later in the

0

0.2

0.4

0.6

0.8

1.0

1.2

Extinction Coefficient, a (m")

Figure 3.4 I+hlking spcctl it7 irrim1it7got7d not7it-rila1i17gs117oke (crthp!ed.fi-orx [I9S5]). Jir~

Principles of Smoke Management

Table 3.4: Mass Optical Densities (adapted from Mulholland2002)
Mass Optical Density,
- 4"

Sample Combustion Conditions
Pyrolysis Pyrolysis Flaming1 Flaming2 Pyrolysis Pyrolysis Pyrolysis Pyrolysis Flaming 1 Flaming1 Flaming1 Flaming1 Flaming1 Flaming1 Flaming1 Flaming2 Flaming 1 Flaming2 Flaming1 Flaming2 Flaming1 Flaming3 Flaming1 Flaming3 Flaming1 Flaming3

Material
Natural Materials: Plywood Wood (Douglas fir) Cotton Cotton Synthetic Materials: Polymethylrnethacrylate (PMMA; ~ l e x i ~ l a s ~ ~ ) Polyvinylchloride Polyvinylchloride (with plasticizer) Neoprene Polypropylene Polyethylene Paraffin wax Polystyrene Styrene Polyvinylchloride Polyurethane Polyurethane Latex Latex Neoprene Neoprene Polystyrene Polystyrene Polystyrene foam Polystyrene foam
Acrylonitrile-butadiene-styrene (ABS) Acrylonitrile-butadiene-styrene (ABS)

ft%b

m21g

Thickness in. cm
0.24 0.24 0.6 0.6

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 primarily of soot, and the production of particulates can be estimated as

Example 34 Visibility Due to a Pillow Fire . If smoke from the burning of a 0.50 Ib (230 g) polyurethane ' 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

MP = mass of particulates produced, Ib (g);

My
yp

=
=

mass of he1 consumed, Ib (g); particulates yield (dimensionless).

Approach 1: From Table 3.5, the particulate yield of flexiblt polyurethane foam is 0.1 88. From buation (3.16), the mass 0) airborne particulate is

Values ofyp are listed in Table 3.5 from small-scale experiments of turbulent flaming combustion for a number of materials. While it is expected that particulate production will vary with the size of the fire and the orientation of the fuel, the data of Table 3.5 are rccommended h the absence of data from the kind of large fires for \vhicli smoke management systems are designed. Considering a \veil rnixed space, the mass concentration of the pa~ticulatesis

From Equation (3.17), the mass concentration of the particulates is

, o4 Using a = 3 . 7 1~ fiZAbfor flaming combustion and K = 8 Fable 3.3), visibility is calculated fi.orn Equation (3.15) a s

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). From Tablc 5.4, the mass optical densiy 4,.of polyurethmc Foam from a flaming mattress fire is 1600 ft'nb (0.33 m' g). Visibility is calculated from Equation (3.13):

Equation (3.17) can be used for a laboratory test where l/,. is [he volume of the test chamber. This equation also can be used for a tire in a room or atrium where 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 of fuel consuliied by a fire can be expressed as

kibility of a light-emitting sign.

WCsee that this is different from !he 9 fr (7.7 m) estimated in
~pproachI, and this is indicative ofthe limitati~ns this techof iology, including availability of a," and d,,, data. where
M/
= =

EXPOSURE T O GASES
mass of Cue1 consumed, Ib (g); total heat rdcase rats Gtuis (kW); chemical hcat of cornbustion Btullb (klkg); timc li.om ignition, S (S); 1 (1000).

AHch
1

=
=

K/-

=

Values of' I'or some materials are listed in Tablc 3.5. Iri tires. combustion is never complete. Combustion efficicnc is the ratio of tlie chemical heat of combustion LO the ncl Iieat of combustion. Using AH,.,, eliminates thc nccd to consider conibustion efliciency.

I n the following sections, information about human responses to exposures to toxic gases applies to an a w r age person. A person's response to an exposure to toxic gases primarily depends on age, metabolism, health history, and respiratory rate. Carbon monoxide (CO) poisonin,o accounts for the majority of total fire fatalities (.Berl and Halpin 19SO; Harland and Woolley 1979). Table 3.6 lists toxicity dsta for several gases, but only a few gases have been incorporatcd in predictive toxicity models. The toxic efficts of CO are probably the most well known, but some o h e r gases included in toxicity models are hydrogcn cyanide (HCN), hydrogen chloridc (HCI), and hydrogsn 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. Yield Yp Chemical Heat of Combustion, AHch Btulib
kJ/kg

Material
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~lrethane Foam (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: Lethal Concentration of Some Gases
Gas
Co2
C2H40
C2H402

LCS0for 30-Minute Exposure (ppm)
,

carbon dioxide acetaidehyde acetic acid ammonia hydrogen chloride carbon monoxide hydrogen bromide nitric oxide carbonyl sulfide hydrogen sulfide hydrogen fluoride acrylonitrile carbonyl fluoride nitrogen dioxide acrolein

470,000

NH3 HCI CO HBr NO COS H2S HF C3H4N COF2 NO2 C3Hj0 CH20

due to building fires tend to change with time. Thus, Haber's rule has limited use for tenability calculations. In the past few decades, tenability limits have been expressed in terms of time integrated values. Time integrated values account for the effect of exposure to a changing concentration of a particular gas over a period of time rather than an instantaneous exposure. The E parameter in Haber's rule can be considered a time integrated value with a constant gas concentration. If the concentration is variable in time, then an integration must be conducted to obtain the area under the concentration-time curve in order to determine a time integrated value.

FED from Animal Test Data
While most animal toxicity tests have been conducted on rats, other animals include mice, guinea pigs, hamsters, and rabbits. Because of concern for animal rights, the toxicity research programs used the minimum of animals, and most laboratories stopped animal testing near the end of the 20th century. These tests determine the concentration of airborne combustion products that is lethal to 50% of the test animals exposed for a specified time, and this lethal concentration is referred to as the LCjo. The specified time for animal tests is usually 30 minutes, and the number of fatalities consist of animals that die during the test and during a post-exposure time, usually 14 days after the test. Using extrapolated animal test data, the fractional effective dose is

fonnaldeliyde hydrogen cyanide HCN C9H602N2 toluene disocyanate phosgene COCl, perfluoroisobutylene CAFX Hyperventilation due to carbon dioxide (COz) 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 the combined effects of these gases. The effect of exposure to toxic gases on a specific individual depends on the physiological characteristics of the individual.

where FED = fractional effective dose (dimensionless);

C
f

= =

concentration, 1blft3( & I ~ ) ; exposure time (min);

Exposure and Time
Haber (1924) proposed that the effect of an exposure to a gas is related to the product of the gas concentration and time duration of the exposure. Haber's rule is expressed as

LCf 50 = lethal exposure dose from test data, Ib ftJ min (g m'3 rnin). An FED greater than or equal to one indicates fatality. The concentration, C, is the density of materials that started as fuel that have accumulated at a location at time I. This concentration has units of mass of the material burned per unit volume. The lethal exposure dose, LCI ,o, is the product of the LCso and the exposure time. Table 3.7 lists some values of LCfSOfor a number of common materials. The above equation is the time-integrated form of the FED equation. For most applications, the time functional relationship of concentration is not known, and the following expression can be used for discrete pairs of concentration and time intervals.

where E = effect of exposure (ppm-min), C = concentration (ppm), and I = duration of exposure (rnin). This elementary equation assumes a constant ingestion rate of the tosin. The effects of some gases do not follow Haber's rulc, and concentrations of toxic gases

Principles of Smoke ~ a n a ~ e m e ' * t

Table 3.7: Approximate Lethal Exposure Dose, LCtSO, Common Materials (adapted from Purser 1995) for
Material Cellulosics C, H, 0 plastics PVC WooVNylon (low N2) Flexible Polyurethane Rigid Polyurethane ~odacrvlicl~~~'
I.
PAN is polyacr).lonitrile.

Ib

Nonflaming Fire min g m-3 min 0.046 0.03 1 0.03 1 0.03 1 730 500 500 500 680 63 160

f lb- n i n ft-3 - 0.19
0.075 0.0 19 0.057 0.087 0.0062 0.0087

( Fuel-Controlled Fire '
-3120 1200 300 920 1390 100 140

-

\
d

g md min)i

Fully Developed Fire lb ff3 min g m-3 min 0.047 0.033 0.012 0.0044 0.012 0.0034 0.0028 750 530 200 70 200 54 45

0.042 0.0039 0.0 10

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 Ci Afi LCf n
=

concentration for time interval I, lb/ft3 (eJni3);

= time interval i, min (rnin);
=

Flexible polyurethane foam would be expected to bum very rapidly compared to the 20-minute exposure time, so the concentration in the room can be considered constant.
C =
6 mass of fuel burned volume of space - (8)(12)(8)

lethal exposure dose from test data, Ib (g m-' min);

rnin

=

number of discrete concentration time pairs. From Table 3.7, LCt 50 = 0.087 Ib ft-3 rnin (1390 g m-3 min) for a fuel-controlled fire. Because the concentration is constant, the FED is calculated as Ct FED = - - 0.0078(20) = l LC,,, 0.087 This indicates that fatality would be expected.
,

When the concentration is constant, Equation (3.2 1) written as Cr FED = LCt50 (3.22)

Many references use the term corzcentrafion rime pmdzicf, Cf, to mean the integral term of Equation (3.22), and this meaning of C will be used for the rest o f f this book. The question arises, should incapacitation or fatality be used as the design criterion for gas exposure. A person who is incapacitated due to exposure to toxic gases will continue to be exposed to those gases. Unless the person is rescued or the gas concentrations improve dramatically, such exposure will result in fatality. Incapacitation often is used to mean the condition that self-evacuation is very difficult or impossible. Usually an incapacitaling dose is less than a fatai dose, but this is not always the case. It is possible that a person could walk out of the smoke-filled environment only to die some time later. While a FED of one indicates fatality, Bukowski et al. (1 989) state that an FED of 0.5 can be considered an approximation to the incapacitaling dose. I L is possible that this approximation is a conservative criterion for smoke nnanage~ncnt design analysis.

Table 3.8: Components of Air
constituent' Nitrogen (Nz) Oxygen (01) Carbon Dioxide (CO2) Argon (Ar) Trace Gases (He, Kr, Xe, H?,CH,, and N20)
1.
Handhook oj Clrerrrisr~)~ Plryics (CRC 1985) alrd
O h

by Volume 78.084 20.946 0.033 0.9?4 0.003

Components of Air
Calculations using predictive toxicity gas models involve the components o f air, and these components are listed in Table 3 3 . The small concentration of CO2 is essential to control normal breathing, but it does not have a significant impact on toxicity calculations.

Chapter 3-Smoke and Tenability

For the fire protection purposes of this book, the small quantities of CO2, argon (Ar), and the trace gases are neglected, and air is considered to be composed of 20.9% O2 and 79.1% NZ by volume. Some sources use 21% 0 2 and 79% N2 by volume, which also yields useful engineering results. :

CO and CO2
Exposure to CO results in carboxyhemoglobin uptake (COHb) in the blood, which results in decreased oxygen-carrying capacity of the blood. Stewart et al. (1973) conducted a series of experiments on humans and, based on this research, C O H b uptake can be expressed as

is seen for any time. In the former case, this would represent such physiological effects as breath holding and the time required for the gas to be transferred to the blood and then to the tissues. In the latter case, this represents an exposure concentration for which the equilibrium concentration of carboxyhemoglobin (COHb) in the blood is below the level that causes lethality (Levin et al. 1987). Following the work with CO, the effect of CO2 on the observed CO toxicity was studied. The result of this work was the observation that the "effective toxicity" of CO increases with increasing CO2 concentration, doubling at a level of about 5% (50,000 pprn), as shown in 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%; CCOHb,O = CCo
V
=

concentration of COHb in the blood at time zero,%; concentration of CO in air, pprn; volume of breathed air per minute, Llmin; exposure time interval, min.
"0

1 I I

Asymptote l minute

-

= =

Ati

Asymptote 1700 ppm

-

3350 PPm

Equation (3.23) does not include the effects of oxygen depletion, increased breathing rate due to CO2 exposure, or exposure to other toxic gases. The volume of breathed air, V , is called the respiratory minute volume (RMV). The typical RMV of a 150 Ib (70 kg) person 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 calculations, a value of CCONb,O 0.75% can be used, and = incapacitation and lethality are approximately 25% COHb and 50% COHb, respectively. However, calculation of the COHb level from Equation (3.23) is not a 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 models 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 necessary to produce deaths of 50% of the exposed animals (the LC50)for each exposure time was determined. The plot of these data (Figure 3.5) shows that the curve has two asymptotes-an exposure time (about I minute) below wliich no cl'fect is seen for any concentration and a concentration (about 1700 ppm) below which no effect

at 60 m - - - - - - - _ in* _ _ _ _

10

20

30

40

50

Time (minutes)

Figure 3.5 Carbori I I I O I I O S I ~COI~C~I~II-a1io11 to ~ VS. time letlinlitj~($SO% of exposed rate (odnyied

..

I

..&&-EL. , Deaths ,

I

..?E%

0

0

l

,

.

1000

. 2000

1

, . 3000

'.'..,.."-.m

4M)O

5003

E

Carbon Monodde (ppm)

Figure 3.6 Dearlls,fi.o~n e.vposi~re CO alone and CO to p1zr.s COz (udoptcd ,%.on7 Lcvin er ul. ( 1 9s 71).

. . .,... .
.,A: ;: ;

> .

.

.

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

CO and CO2 is apparent from the formulations of the Ngas and FINmodels that follow.

N-Gas Model
The N-gas model was developed at the National Institute of Standards and Technology (NIST) and relates fatality with animal test data of exposures to pure gases and mixtures of gases (Levin 1996; Levin et al. 1995; Babrauskas et al. 1991). For mixtures of gases, including NO2, the N-gas model can be stated as

and for mixtures not including NO2, the N-gas model can be stared as
NGas =
+

The model incorporates the ~ncreased breathing rate due to CO2 exposure. It is apparent that there is a unique interaction between HCN and NO2. For many of the gases, the contribution to lethality is expressed as the ratio of the gas exposure to the LCS0. This is how O2 is treated, except that it is in terms of oxygen depletion. The toxicity of CO2 is not included in the N-gas model because fire-generated atmospheres do not contain 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 expected to die. The time-integrated average exposure to CO is
[CO] =
1 -1,'= o Ccodf
=
l..

,,,[CO] [ C O 2 ]- b
+

20.9 - [ 0 2 1 20.9 - L C p ( 0 2 ) ,

where
N ~ a ~=
n 7
= =

N-Gas model indicator (dimensionless); -18 for CO2 S 5% and 23 for CO2 > 5%; 122,000 for COz < 5% and -38,600 for CO, > 5%;

B

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] [CO?]
L021
=
=

time-integrated average exposure to CO, PPm; time-integrated average exposure to COz, PPm; time-integrated average exposure to O-,,

=

W);
[HCW [NOz] [HCI]
[I-IBr]
=

=

=

=

time-integrated average exposure to HCN, PPm; time-integnted average exposure to NOZ, PP'T time-integrated average exposure to HCI, ppm; time-intcgrated average exposure to H Br, ppm.

+

[HCN] + LC,,(HCN)

[HCf] LC,,(HCf)

[HBI-] LC,,(HBr)

(3.25)

fe

where I, is the exposure time. The other time-integrated averages can be expressed in a similar manner. For discrete concentratio11 data, the time-integrated average can be written as follows:
1
e

[CO] = f

2 Cco, ;At;
i: l

I [HBr] = l'>

2 C,,,.,

;Ati

i= l

&apte; 3 -Smoke and Tenability

Cco,i CCm,i C02,i

=
= =

concentration of CO, pprn; concentration of CO2, ppm; concentration of 02,%;

CHCNi = concentration of HCN, ppm; Cm,,
=

concentration of NO2, ppm;

CHCISi= concentration of HC1, ppm; CHB,+ = concentration of HBr, ppm;
fe

= =
=

exposure time, min; time interval i, min; number of concentration values for each gas
and time interval.

At
n

Equation (3.27) can be used where the time intervals are either uniform or nonuniform. For uniform intervals, the time-integrated average terins of these equations become mean averages. When the concentration of any of the gases other than 0 2 is zero, the contribution of that gas to the NW value is also zero. This is to be expected, but it is not so for the fractional incapacitating dose method discussed later. Equations (3.24) and (3.25) apply when the exposure time is the same as the duration of the LCS0data. Example 3.6 demonstrates the use of the N-Gas model for four gases, but Table 3.9 has LCSo values for all of the gases in this model for many exposure times. For exposure times between those listed in this table, LC50 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
I

.

C0.i

CCOZ.;
.

GO. i
PPm 0 40 60 120 120 I60 500 600 600 600 600

c~c~~. i
PPm 0 2 3 6 6 8 25 30 30 30 30

(m in)

0
I

0 2
4

Yo 20.90
20.72 20.30 19.80 19.70 19.60 19.60 19.60 19.60 19.60 19.60

PPm 0 5SO 1900 3200 3600 3800 3800 3800 3800 3800 3800

2 3
4

6

8
10 I2 14 I6
IS

5
6 7

S
9 10

20

The time-integrated avcragc exposures can be calculated from Equation (3.27). Bccausc the intervals are unifomi, the timeintegrated average ternis are mean averages of the concentrations as listed below. [CO] = 340 [02] 19.8 1 = [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: Lethal Concentration, LCSO, Various Gases of
Exposure Time min
1 2 5 10 15 20 25 30 45 60
I.

'

FIO, i

=

hction of an incapacitating dose of ~ O W -

HCN PP"'
3000 1600 570 290 230 170 160 150 120 90

Oz
4.0 4.8 5.0 5.2 5.3 5.4 5.6 5.8

HCI PP"' 15900 8400 6900 6400 5900 3800 3300 2800

'

HBr PP"' -

NOz PP"'
1450 830 510 380 320 290 200 150 100

oxygen hypoxia per unit time (min-l); FImSi = hction of an incapacitating dose of CO2 per unit time (min-l); Ati = exposure time interval i (min);

n

12600 6600 5400 5100 4700 3000 2600 2200

number of concentration values for each gas and time intervals. The following terms are calculated as
=

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.

Fractional Incapacitating Dose
Purser (2002) developed a model to calculate a fractional incapacitating dose for exposures to CO, HCN, CO2, and reduced Oz. The notation in this section has been modified from that of Purser to facilitate computer programming.

whichever is greater, where

FIN FIco,;

=

fractional incapacitatingdose of all narcotic gases (dinlensionless); fraction of an incapacitating dose of CO per unit tinie (min-l); fraction ofan incapacitatingdose of HCN per unit time (min-l); factor for CO1-induced hyperventilation;

where Cco,; = concentration of CO (ppm); CHC.\r; = concentration of HCN (ppm); CC02,; = concentration of CO2 (percent); Co., = concentration of Oz (percent). A value of FlIr of I or more indicates incapacitation. and the incapacitation time based on can be 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 applications where there are no nonfire sources of CO1, Equation (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 test measurements that are used for input to calculations of F/,, should be consistent with this O2 concentration. Examination of Equation (3.30) sho\vs that
1. for zero CO, Flco,i has a value of zero;

=

2. for zero HCN, FICjV,, a value of about 0.0045 nii11-l; has and has 3. for zero 20.9% O1, FIO,, a value of about 0.002 1 For item I , it would be expected that a zero concentration of CO would result in a zero contribution to the

=

Vcoz,,

=

Chapter 3-Smoke and Tenability

FIN. However, items 2 and 3 were unexpected. A z e r o concentration of HCN results in a positive contribution to the FIN,and no oxygen depletion ( 0 2 = 20.9%) also results in a positive contribution. For the short exposure times characteristic of most fire protection applications, these positive contributions are small and should not be of concern. some are measured in hours as was the case for the World Trade Center explosion. From Equation (3.28) and (3.30), an incapacitation time of

-

.

about 3.3 hours can be calculated for exposure to an atmosphere of normal 0 2 and zero concentrations of CO2, CO, and HCN. This exposure can be thought o f a s breathing normal air, and no such exposurewould result in incapacitation. This indicated that the FIN approach is inappropriate for long exposures. However, the FED and the N-gas model are based predominantly on test data with 30-minute exposure times, and applying these models for long exposure times is also questionable.

Example 37 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) 0 2 4 6
8

FICO.i NIA

F ~ ~ ~ . i

'~02,;

Floe i NIA

FIN

0
l

NIA

NIA

0 0.013 0.029 0.050 0.072 0.097 0.155 0.223 0.291 0.359 0.427

0.00 13 0.00 19 0.0039 0.0033 0.0053 0.0 1 73 0.0209 0.0209 0.0209 0.0209

0.00475 0.00486 0.0052 1 0.0052 1 0.00545 0.00806 0.00904 0.00904 0.00904 0.00904

1.053 1.080 1.107 1.115 1.119 1.119 1.119 1.119 1.119
1 119

0.000325 0.000407 0.000534 0.000563 0.000594 0.000594 0.000594 0.000594 0.000594 0.000594

2 3 4 5
6

10 12 14 16 18 20

7 S 9 10

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

hlodels Example 3.8 Comparison of To~icitv For the gas concentrations listed below, calculate NG,,, FIAi,and COHb. Time
i
c~?,i

Cc02.i

Cc0.i

C ~ i ~ ~ . i

(min) 0 2 4 6 8 10 12 14 16
IS

%

PP"' 0 2320 7600 12800 14400 15200 15200 15200 15200
1 5700

PP111 0 320 480 960 960 1280 4000 4800 4800 4800, 4800

PP"' 0 8 12 24 24 32 100 120 120 120 120

0
1

20.90 20.18 18.50 16.50 16.10 15.70 15.70 15.70 15.70 15.70 15.70

2 3 4 5 6 7
S.
0

10
Part

20

15200

I: In thc snnic manner as Example 3.6, Nh

= 1.1 is calculated. This means fatality \\xiuld he expected

from rh~s csposurc.

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) 0 2 4 6 8 10

F l ~ ~ , i

0 I 2 3 4 5

N/A 0.0109 0.0166 0.0340 0.0340 0.0458

F ~ i N/A

~

~

vN ~ 2 . i ~

N/A
1.OS8 1.203 1.328

Flo.i N/A

0.000 0.036 0.093 0.210 0.333 0.496

0.00545 0.00598 0.00788 0.00788 0.00947

0.00043 0.00 108 6.00317 0.00393 0.00488

1.369 1.390

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

A C ~

= 3.3 I 7 X

- 1.036 . . I O-'CCO, lfAt and CCoHh= CCOHb, + o
i= l

ACCOHb. l

c'

=

18 Urnin; At = 2 rnin: CCOH&,= 0.75%. Time
(min)
CCO. :

ACco~b.i

c,,,,
O/a

* .;,
0.0Oil 0.032 0.048

%

I/

0.00000 0.7500 0.00003 0.7500 0.00005 0.750 1 G 0.096 0.000 1 1 0.7502 S 0.096 0.7503 0.000 1 1 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 calcularams sl~ow C@!{!, c~lculakd h ~ Equation (3.23) is not a reliable indication of incapacitation or fatality. [hat l n burns can be expected to be the dominant effect for d q air temperatures greater than 250°F (1 21°C). The effect o.f esnosure to elevated tem~eratures depends on the of the a i r and the type and extent of clorhing worn. Physiologically, exposure to an elevated tcrnperature environment Can cause an increase in body or blood temperatuce. Also affecting the thermal

0 2 4

EXPOSURE T O HEAT
Exposure to elevated temprl.ratul-e atmospheres can lead to skin burns and hypenhermia (heat stroke). A temperature limit of 250°F (1 2 1 'C) for d ~ air is used as y a rule of thumb to dctermins \\.hicl1 of thcse two possible efl'cc~s will dominate. Generally, csposure to hcatcd . . dry a i r a, a ,empcratllrc less approximately 2jO"F (121°C) lcads only to Iiypmhc.rmia. Pain from skin

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. However, clothing may inhibit the efficiency of this natural cooling process. Conversely, clothing provides insulation 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 considered to be beneficial. However, at the lower temperatures, where hyperthermia is the dominant effect, clothing is detrimental. As in the case of exposure to toxic gases, consideration 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, naked, with low air movement is shown in Figure 3.7. Purser recommends the following relationhip for time to incapacitation based on averaging the time to incapacitation for exposures to huniid air arid dry air:

Exposure Time (minutes) . Figure 3.7 Ther-~nal~olerance for humans at rest, nnked, with low air movement (adapted fi-on7 Blockley [1973]).

The cumulative dose is the sum of the doses for each of the intervals:

where where
1~1,
=

F1,ll = total cumulative dose (dimensionless);

time to incapacitation due to thernial exposure, min;
F,,,,, =

incapacitating dose for time interval i (diniensionless).

C, C2
T

= = =

5.670 (5.185); 0.0 152 (0.0273); temperature of air, "F ("C).

Incapacitation would be expected for FI,l, greater than or equal to one. Substituting Equations (3.31) and (3.32) into Equation (3.33) yields

Equation (3.3 l ) applies when the teniperature is not changing witli time. To deal witli changing temperatures, 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 incapacitating dose is given as
G

whcrc
Fl,/,
= = =

total cumulative dose (dini~nsionless); cxposure time interval i, n:in; temperature of air in interval i, "F ("C); 5.670 (5.185): 0.0 157 (0.0373).

A/;
= [/A. i

Flrh

(3.32)

Ati

where
F =

C,
incapacitating dose for the time interval (di~nensionless); exposure time intcrval i, min; timc to incapacitation for tcmpcralureof'intenral
i, min.

= =

C?

Ati

= =

tlh,,

Equation (3.34) is in a forin uscful for calculation with lcmpc'raturcs ~ I - O ~ L I byCa~smoke transport model C or tcmpcrarurcs f.1-on1 lire tcsls 1,ccorded with a data acq~risitiorl ?.stem.

Principles of Smoke Managemelit
- ---

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 temperature 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

1

C
"F

11, 11 i

F I Ii ~

Ft~h

min
1

min 0.02 0.06 0.11 0.18 0.28

125 43 0.02 2 l50 30 0.03 3 175 20 0.05 4 200 14 0.07 5 225 9 0.11 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 is often not addressed in discussions of tenability for smoke control applications because of the limited smoke temperatures for such designs. Gas temperatures 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 observers be able to approach the fire? From Chapter 2, the separation distance for nonpiloted ignition due to thermal radiation can be adapted for the separation distance to prevent skin pain as

where
Q, = radiant heat release of the fire, Btuls (kW);

RsD = separation disrance from the center of rhe fire to a person, f (m); i q,, = limit of radiant flux to prevent pain, ~ t d f t 's (kW1
Ill2).

III
I

where
I,:[,
= = = = =

Calculare Q, exposure time to pain, s (S); exposure time to blister, s (S); intensity of thernial radiation, ~ t u &s ( k ~ l m ' ) : 3.20 (85); 8.39 (223).

=

1000 (0.3) = 300 Btds (320 kW), and use

;,,h

RsD

=

qr
C,:, C,,

d4d.22)

1'0" = 10 ft (3 m) separation distance. -

TENABILITY AND PERFECT DILUTION
It is cornnion to encounter situations where the dilution necessary to meet some visibility criterion results in sucn ION gas concentrations that toxicity. is not an issue. Generally, such dilution also results in smoke temperatures so low that heat exposure and thermal radiation exposure are not issues. However, this is not so for fuels that produce low amounts of soot.

The above relationships are shown in Figure 3.S. .A value of q): = 0.22 ~ t u l f t 's (2.5 kwlrn2) is often used as the value that can be tolerated for a few minutes \vithout unbearable pain.

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); mass concentration of fuel burned, l b l g (g/ m3);

m/
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
Klote (1 999a) developed equations based on perfect dilution that allow relative comparison of visibility, toxicity, and temperature for a particular fuel. This section presents a similar but more straightPonvard approach. The analysis considers that the products of conibustion (particulates, gases, and heat) are diluted by air. This analysis neglects smoke panicle aging (agglomeration 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 of toxic esposure, and smoke temperature in terms of a common variable so comparisons can be made. The variable selected is the mass concentration of fuel burned, 11.;: Equation (3.13) already has visibility in such terms:

=
=

smoke temperature, "F ("C); ambient temperature, "F ("C);

To

Q = heat release rate of fire, Btu (Id); M = mass of smoke, Ib (kg); ,
Cp = specific heat ofsmoke, 0.24 BtuAb "F (I .O I d k g
"C);

The follo\ving equations are needed in order to get the desired expression for the smoke temperature:

where Mj = mass of fuel bumed, Ib (g);
AHc11=
PS

chemical heat of combustion, Btu4b (kJ/kg); density of smoke, lb/fi3 (kg/m3); ambient pressure, lb/$ (Pa);
gas constant of smoke, 53.3 ft Ibfllbm O R (287 JI

= =
=

i"2 R

C7.
visibility. ft (m); proportionality constant (Table 3.3); mass optical density, li2/lb (m2/%); mass concentration of furl burned 1b/ft3(g/rn3).

= = =

kg K); 460 (273); volume of smoke, h3(m3);
1 ( 1000).

v, 9

Substitu~ing Equations (3.39) into Equation (3.38) and rearranging yields

Principles of Smoke Management

+ T,=- CT -

l-a

where a =

m

Rmch L .

K/ CpPa

(3.40)

Equations (3.36), (3.37), and (3.40) are in terms of the mass concentration of fuel burned, m/: Equation (3.36) can be solved for mass concentration of fuel burned:

The design criterion for visibility can be put into Equation (3.41) to get the maximum value of the mass concentration of fuel burned to meet the visibility ciiterion, and Equations (3.37)-and (3.40) can be used to calculate the upper limits of the FED and T, resulting from 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. . 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),

I

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
In the most general sense, the criterion for all tenability systems could be stated as: tenable conditions are to be maintained in spaces where people are expected to be for the expected duration of their time in those spaces. However, such a criterion is too general to be useful for design applications, and more specific criteria are needed. More detailed criteria deal with one or more of the following: exposure to toxic gases, exposure to heat, exposure to thermal radiation, and visibility 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 movement that can be used to calculate this time. For the conditions of Example 3.1 1, the exposures to toxic gases, heat, and thermal radiation are insignificant provided that the system was designed to meet the visibility criterion. For such insignificant exposures, detailed tenability criteria have no real purpose. Whenever possible, this approach can sgnificantly simplify design analysis. For applications where exposure to toxic gases is significant, it might seem that the tenability criterion should be based on prevention of both incapacitation and fatality. Because a person who is incapacitated will

remain in the toxic environment until fatality or rescue, it seems that the criterion should at a minimum be based on incapacitation. Exposures to some gases (for example HC1 and HBr) can result in post-exposure fatality, such that a person might not be incapacitated while being exposed but die some time after exposure. Considering both the dominance of CO among toxic fire gases and that CO does not result in post-exposure fatalities, incapacitation could be a sufficient criterion for most applications.

The visibility distance for exit signs depends on the distance between the exits in a specific building. The visibility distence for seeing balcony walls and railings might be taken as two or three times the width of the balcony. In many applications, the criterion for seeing the exit signs wodd be expected to be the more stringent of the two. For additional material about survival of exposure to fire produced environments, see Gann (2001).

CHAPTER 4

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

THE MYTH OF PANIC
Often, movies, television. and the press present the unrealistic image that panic bchavior in fire situations is common. However, extensi\.t. research supports the conclusion that panic behavior in fire situations is \.cry rare. Even in large building fires resul~ing multiple deaths, in people experiencing fear still usually act in pi~rposeful ways.

Chapter 4-Evacuation Analysis

Table 4.1: Types of Fire Alarm Signals Used in Drills in London Subway Station
Type Bell only Staff Public Address Staff + Public Address Directions + Public Address Description Alarm bell rung, no staff or PA Alarm bell rung, two staff members gave PA announcement to "evacuate station"and then directed evacuation Each 30 seconds, PA announcement said twice, "please evacuate the station immediately" PA announcement instructing people to leave via trains or exits, with staff directing people following the directions of the announcement 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 Concourse Escalator Comments 8:15 9:OO Delayed or no evacuation 2:15 3:OO Occupants directed to concourse . 1:15 7:40 Occupants stood at bottom of escalator 1:15 1 :30 Occupants evacuated 130 1:00 Occupants evacuated difficult to estimate reliably, being a function of the fire scenario, building characteristics (compartmented versus open-plan), and thc proximity, alertness, and mental abilities of the occupants. PRE-MOVEMENT Interpretation o f t h e alarm signal as an indication of a threatening fire by building occupants is dependent on the type of signal provided by the alann system (Ramachandran 1991; Proulx and Sime 1991; Prouls and Fahy 1997). Bells and horns arc often ignored, being considered to indicatc a drill, test, or false alarm. In a laboratory exercise, Ranlachandran found that only 13% of 96 individuals considered bells to signal an actual alarm. Similarly, Pauls' survcv of occupants of office buildings indicatcd that only 17% of occupants responded to traditional fire alann signals in high-rise office build1r:gs. The response of people to various types of fire alarm signals was observed by Prouls and Sime in drills at mid-afternoon in a London subway station. Cameras recorded the responses of the individuals, with interviews conducted to supplement the video recording. The five types of alarms used in thc study are described in Table 4.1. Alarms were initiated tivc seconds after a train arrived at the station. It can be seen from Table 4.2 that pre-movcment ~ i r n ewas as much as nine minutes

Evacuation Alarm Bell Only Staff investigates, makes P announcement. A directs evacuation A Plain "recorded" P announcement. repeated every 30 seconds P directive + staff directing evacuation A P;\ directive plus status Extreme frustration Chaotic/antisocial behavior Entrapment Flight Contagion

-

The panel indicated that all nine characteristics may not be evident for every individual who does engage in panic type behavior (Quarentelli 1979b). However, they also caution against quickly labeling any particular action as panic behavior that has only a few of these characteristics. B E C O M I N G A W A R E O F T H E FIRE Bryan (2002) discusses several ways that occupants become aware of'a fire. In most cases, the initial cues of a fire are ambiguous, involving a different odor, a slight haze, or strange noises. In some cases occupants may observe the flames. In still others, occupants may be alerted by an alartii system. Evaluating the rime to become aware o f the tire via an audible or visual fire alarm signal actuated by a fire detector or sprinkler waterflow switch may involve an analysis of the response time of automatic detection equipment or sprinklers. Several computzr models discussed in Chapter 8 are capable of calculating sprinkler dctection. In contrast. manual detection is ~iiuchmore

Principles of Smoke Management

for an alarm bell only, but the pre-movement time was much less with verbal. announcements. For guidance on the use of verbal announcements, see Keating and Loftus (1977). Given the predominance of ambiguous cues during the early stages of a fire, building occupants often investigate these cues or ignore the initial cues completely, thereby delaying initiation of evacuation. Pre-movement time may also be dependent on the time of day. Proulx and Fahy measured the pre-movement time to be up to 10 minutes long in a mid-rise apartment fire drill during the day. During an early morning high-rise apartment fire, the pre-movement time was 15 minutes for numerous occupants and up to five hours for others.

Evacuation Analysis Using Hydraulic Analogy
Evacuation analysis using the hydraulic analbgy assumes that people follow a directed route of travel to their destination, which is typically outside or an area of refuge. As such, the occupants are assumed to travel along a route where the distance to the destination is continuously decreasing, neglecting the possibility of traveling in circles, proceediag in the "wrong" direction, and retracing steps, etc. Consequently, an "efficiency" factor may be applied to evacuation times estimated using this approach to account for the possibility of an evacuation process where the occupants may divert from a directed route. Evacuation modeling following the hydraulic analogy requires information on the people movement characteristics of velocity, flow rate, and specific flow.

EVACUATION TIME ANALYSIS
There are three principal approaches for estimating the evacuation time for a building: I . ~&&ical correlation of total evacuation time for building. 2. Model movement applying hydraulic analogy, simulating people as fluid particles. 3. Model movement applying hydraulic analogy, with consideration of the behavioral aspects of the people.

-

Veloci~: of travel along a corridor, ramp, stak4 rate Flow rate: number of persons passing a particular segment of the egress system per unit time (for example, persons per unit time passing through a doorway or over an imaginary line drawn across a corridor). Specificflow: flow rate per unit width of the egress component (for example, per unit time per unit width through a doorway).

Empirical Correlations
The first method consists of correlations that were developed from a regression analysis of evacuation data from 50 fire drills in high-rise office buildings ranging from 8 to 15 stories in height. The two correlations developed by Pauls (1980) (one from a linear regression analysis and the other from a nonlinear regression analysis) are

The movement of people has been examined for travel on stairs (mostly downward travel), in corridors, and through doonvays. Virtually all of the information on people movement has been collected from observations of fire drills or normal movement.
Population per Effective WidM (plm)

and

where . T = evacuation time (win); C, = constant, 0.193 (0.08 1 );
Cz
= =
=

constant, 0.0394 (0.01 2);
"0 50 190 150

Equations: - - - (4.1) (4.2)

-

population using the stair (p); effective width of stair, ft (m) (see discussion on efective width later in this section). The unit of population above is persons, and the symbol used in this chapter for persons is p. The predictions of Equations (4.1) and (4.2) are very close to each other, as shown in Figure 4.1. Becausc Equation (4.2) is the simpler form, i~ is IUOI-c commonly uscd.
P
W

200

250

Population per Effective Width (plft)

Figure 4.1 Estinlared evncmtion time jor- high-rise buildings (Pauls 1980).
4.

; di;lgonal I

Thc v c l o c i r on stairs rcSers to the rate o f ~ r a v calong l p;trh obtaincd by con~wcting tips ofthc stairs. the

Chapter 4-Evacuation Analysis

Considering that people tend to move faster in emergencies than they do in fire drills (Figure 4.2), it might seem that evacuation time estimates based on fire drill data would be conservative. However, this does not account for the possibility of exit routes being blocked by smoke or fire. An "efficiency" factor also may be applied to account for blocked exits routes.

v = 0.85k

where v = velocity, @m (mk); a = constant, 2.86 (0.266); k = velocity factor, fpm (mls); and

Velocity
The velocity has been shown to be a function of the density of the occupant flow, type of egress component, and mobility capabilities of the individual (Gwynne et al. 1999; Nelson and MacLennan 2002; Predtechenskii and Milinskii 1978). Nelson and MacLennan propose correlations of velocity for mobile individuals considering the available data collected by numerous researchers. For a density greater than 0.05 1 p/ft2 (0.55 p/m2),

-D = density of occupant flow, @/m2). Equations (4.3) and (4.4) apply to flow on horizontal surfaces and on stairs. For horizontal surfaces and the stair tread and riser types listed in Table 4.3, the velocity factors are listed in Table 4.4. On stairs, the distance of travel is the diagonal of the stair (Figure 4.3), which is

where

LD Lv

= =

diagonal distance of the stairs, ft (m); vertical distance of travel, ft (m);

For densities less than 0.05 1 p/ft2 (0.55 p/m2), other occupants do not interfere with the walking speed of an individual. The maximum walking velocity for level walkways and stairways is

B = angle of the stairs. The dependence of the velocity on density, as predicted by Equations (4.3) and (4.4): is presented in Figure 4.2. The velocity correlations prejznted in Equations (4.3) and (4.4) principally relate to adult, mobile individuals. 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

Figure 4.2 Cornpar-ison o f nor-nzal velociq and velocity during emergencies (P,-edtechenskii and Milirukii 1978).

Area Density. 6

Egress Component Corridor, aisle, ramp, doorway Riser and Tread Type 7.5110 711 1 6.5112 6.511 3

k (fpm) 275

k (mls)
1.40 1 .OO 1 .08 1.16 1.23

196 212 229 242

Table 4.3: Dimensions of Stair Risers and Treads
Riser and Tread Type 7.5110 in. 7.5 Riser, LR mm
190

Tread, LT in.
10

Stair Angle, 8
36.9"

mm 254

Sin, Q
0.600

~ r i n c i ~ l e s 'Smoke Management of

Figure 4.3 Stair geometry.
also appreciably less than the maximum velocity noted in Equation (4.4).5 Table 4.5 lists mean velocities for impaired individuals.

Density,.,
Density is the ratio of the number of people in a 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

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 occupant load factors included in each of the referenced codes are similar and these occupant load factors represent average maximum density of occupants. Occupant load factors from the NFPA Life Safety Code are listed in Table 4.6. Predtechenskii and Milinskii use a definition of e density based only on areas. rea density & ratio of the floor area occupied by each individ~ualperson in the group divided by the tdal floor area occupied by the gouk(including the area between individuals). This can be expressed as

where where P = population, p (p);
A

S
A,,
7

= =

area density (dimensionless); average area occupied by an individual, ft- (m-).
7

7

total floor area occupied by the group, ti- (m-). Typical densities of people nlovenlent range from 0.1 to 0.2 p/ft2 (1.0 to 2.0 p/n~2) (Predtechenskii and Milinskii 1978; Frantzich 1996; Pauls 2002; Fruin 1987). The. normal occupant loading may not be considered an appropriate population for evacuation calcula=

7

The average area occupied by an individual includes the floor area directly under the individual and the floor space around the individual. The relationship between these two density tenns is

5. An encumbered adult is an individual c a v i n g packages, luggage, o r a child.

For the areas that people occupy. see Tables 4.7 to 4.9.

Table 4.5: Mean Velocity for Impaired Individuals (Shields et al. 1996)
- - .. lnipairment Electric wheelchair Manual wheelchair Crutches Walking stick Walking fialne Rollator No aid No disability
Level Walkway fpni nits Stairs down f ~ m nils Stairs up fpni nils

260 200 280 160 100 I10 I S0 2-10

0.89 0.69 0.94 0.8 1 0.5 1 0.6 1 0.93 1.24

43 63

0.22 0.32

43 67

0.22 0.3

65 140

0.33 0.70

SI I40

0.4 l 0.70

Chapter 4-Evacuation Analysis

Table 4.6: Occupant Load ~actors'
Occupant Load ~ a c t o ? Space Use Assembly Less concentrated use without fixed seating Concentrated use without fixed seating Waiting space Library-stack areas Library-reading areas Mercantile Street floor and sales basement Multiple street floors Other floors Storage, shipping Educational Classroom area Shops Daycare centers Business (offices), industrial Hotel and apartment Health care Sleeping departments In-patient treatment departments Detention and correctional perslf? 15 net 7 net 3 net l00 gross 50 net 30 gross 40 gross 60 gross 300 gross 20 net 50 net 35 net l00 gross 200 gross 120 gross 240 gross 120 gross pers/m2 1.4 net 0.65 net 0.28 net 9.3 gross 4.6 net 2.8 gross 3.7 gross 5.6 gross 27.9 gross 1.9 net 4.6 net 3.3 net 9.3 gross 18.6 gross I l . l gross 22.3 gross 11.1 gross

0 l Data from Table A-S-3.1.l of NFPA l 1(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 indicated above.

Table 4.7: Area Occupied b y people1
.

Age

10 to 15 Walking Female Male All All ,411'
I . Data are from Kendik (IYSj). 2. Wih coats

15 to30 m= 0.126 0.122 0.124 ft' 1.63 1.78 1.72 1.57 2.00 m2 0.151 0.165 0.160 0.146 0.186

Crcater than 30 ft2
.,l

ft 1.36 1.3 1 1.33

2

Standing

2.08 1.87

0.192 0.174

Principles of Smoke Management

Table 4.8: Area Occupied by People in IP units1
Person Type Adult Youth Child Encumbered ~ d u l t ~ Horizontal projection2 1,I-1.4 Shoulder Breadth 1.5-1.6 Body Depth 0.92-1.1

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 Adult Youth Child Encumbered ~ d u l t ~
Data are from Predtcclienskii and klilinskii (1978). 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.
I.
2.

Horizontal projection2
m2

Shoulder Breadth
m

Body Depth
m

0.10-0.13

0.46-0.50

0.28-0.32

Densily ( p h i )
3 0 0 ° , 1
I I

2
I I

3

,

,

:d

Density (plrn')
0 30 1 2 3 4

1.50

16 Stair Riserand Tread Type:
12

Stair Riser and Tread Type:

- 1.00

- 25

= '5
0 -

= $
C

F
a Z

- 0.75 .$ - 0.50 3
50 -

- 20
a

15
3 10 a

U

0.8
0 c
0

W

- 0.25
0.4

m

5 0 0

0.4

-0

0.1

0.2

0.3

0.4

0

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
The specific tlo\v, F,, is analogous to the mass flux in hyd;dillic systems. As such. the specific flow is defined as the product of the velocity and density of the flow,

F,. = DV = ( 1 -aD)kD

(4.10)

\\.here F = specific flow, plmin-ft (pls-m). , For densities less than 0.05 1 p/ft2 (0.55 p/m2),

Expr-essions Ibr thc specific flow as a function of density call o n l y be obtained by for [he velocity 1.1-0111 Equations (4.3) and (4.4). FOI- dcnsity yeatcr than 0.05 i p/li2 (0.55 p!m2), a

The specific flow predicted by Equations (4.10) and (4.1 1) is presented in Figure 4.5. The width referenced in the units for the specific flow equations relates to the ..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 people do not generally occupy the entire width of an egress component, staying a small distance away from the walls or edge of the component. Nelson and MacLennan refer to this small distance as a "boundary layer," in keeping with the hydraulic analogy for people movement. The width of the boundary layer for the variety of egress components is presented in Table 4.10. The boundary layer and effective width are illustrated in Figure 4.6.

Flow
Nelson and MacLennan (2002) present a method to obtain a first order approximation of the egress time in buildings. The method involves determining the maximum flow rate for each of the egress components in the egress system. For a density greater than 0.051 p/ff2 (0.55 p/m2), the flow rate for a particular egress component is given as

Maximum Specific Flow
Considering that Equation (4.10) is a quadratic function, a maximum specific flow is achieved at a density of

where

F ,
w

=

flow rate p/min, @/S); effective width, ft (m).

=

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 component, according to this correlation. the specific flow is maximized at the same density for all types of egress components. Predtechenskii and Milinskii provide data that indicate differences in the density where the specific flow is inaxinlized for different types of egress components. 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.
Table 4.10: Boundary Layer Width

The maximum flow rate occurs when the specific flow is maximized (i.e., where D,,,, occurs (see Equa-

L
I l

Effective
width

-;
I

4

/ Boundary

Layer

/ Boundary

Layer

Component Theater chairs, stadium benches Railings, handrails' Obstacles Stainvays, doors, archvays Corridor and ramp walls
I.

Boundary Lager in. mn! 0 0 S9 3.5 -1 100 6 l50 S 200

Water 'cooler

Width?

Effective

\Vllcrc Ilandr;~itsarc present. Nelson and I I x L c ~ l n m sug:cst that the boundxy laycr a-id111i'or Iwndr:~ils should he used i i d i c Iwundary laycr \r id111lirr 11;lndrailsis Ics; [ h ~ n 1 1 Ibr 111c 1 ~ 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 Corridor, aisle, ramp, doorway Riser and Tread Type 7.5110 711 1 6.5112 plmin-ft 24.0 17.1 18.5 20.0 pls-m 1.32 0.94 1.01 1.09

Figure 4.7

Comti-ained flow i7 evacuotion of a fiver story building.

For a particular egress system composed of several components, the maximum flow rate, F,, of each comThe flow time ponent can be determined as FS,,w. associated with each component is P/FS,,,,w, where P is the population passing through the component. The is component with the greatest value of P/Fs,,,,,w defined as the controlling element where the constraint is expected. In many situations, the point of constraint can be identified easily. For example, consider a stainvell discharging directly to the outside that has doors of only 0r.e width (see Figure 4.7). For staiiwell and door widths designed to comply with the Life Safety Code or model building codes in the US., the minirnuni flow will be associated with the doorways. Tlie entire population using this stainvell would have to pass through this exterior door. Because the interior stainvell doors on the upper floors would only serve a fraction of this population, they would be less congested. The exception is the stair that is used for evacuation of only one floor, and this stair would have constraints at both stairwell doors, provided that both doors are of the same width. In such a case, the evacuation analysis could be conducted witli the constraint at either location. When the exterior stairwell daor is the constraint in tlie egress system, the modeled evacuation time becomes

tion (4.12)). Maximum specific flow, F,,,,,,w, for a variety of egress components is provided in Table 4.1 1. Tlie controlling egress coniponent is tlie component with tlie smallest maximum flow rate, relating to where a queue is expected to form if D,,,, occurs ill an upstream component.

EVACUATION TIME

Constrained Flow Approach
The constrained flow approach is based on the assumption that there is a point along the egress system where a queue forms. Tlie evacuation flow envisioned when applying this type ofevacuation model is depicted in Figure 4.7 where the egress system is funneled into a particular point, such as an exterior doorway, before the evacuees depart from the building or affected area. Assuming that all occupants start their evacuation simultaneously at time zero, the niodeled evacuation time using the constrained flow approach can be estimated as

where
I,,,
=
=

nodel led evacuation time for an egress route.

f,
I, I,

time for first person to arrive at constraint, time for population to pass through constraint. time for first person to travel fi-om constraint and proceed to outside (or area of'reli~ge).

=
=

Example 4.1 illustrates tlie constrained flow approach. This example is appropriate for situations where a queue is expected to form at the exterior stairwell door. Generally this happens when an appreciable number of people occupy tlie area of the building being niodeled. Conversely, in buildings with low occupant loads, a queue is unlikely. In cases with low occupant loads, a more complex analysis is needed to examine the occupant flow on a component-by-co~iiponent basis. These analyses also may be applied to provide a more 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 Solution: Component Door into stairway Stairway Landing Door from stainvay
X

2.44 m). Handrails are provided on both sides of the stairways.

Effective Width

ft (m) 1.67 (0.51)
3.08 (0.94) 2.67 (0.82) 1.67 (0.5 1)

Specific Flow plft-rnin (plm-S) 24.0 (1.32) 18.5 (1.01) 24.0 (1.32) 24.0 (1.32)

Flow Rate ptmin @/S) 40 (0.67) 57 (0.95) 65 (1.08) 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 occupants 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). The density on the stairs is taken at D,,,,. From Equation (4.12), D,,
= 0.175 p!ft
2

(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 landing). Because thc velocity on a stairway is less than that for a horizontal component, such as a landing, the velocity on the landing 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 stairways 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.
If the conibined flow rate of egress components leading to the intersection is greater than the capacity of the f l o n rate for the egress component leading from the 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 capacity of the component. Questions are often asked concerning the composition of the queue relative to the incoming flows (i.e., does any one group have a "right-of-way" while most or all of the other group stops). The total evacuation time of the building is not dependent on which group has the right-of-way. Alternatively, if the intent of the analysis is to determine the time required to clear a particular floor level and the merger is nith people from another floor level, then the right-of-nay decision will impact the results. Unfortunately, there is no technical support for establishing any rules co~lcerning right-of-way or the proportion of the tlorvs from the entering streams that Gccurs at a merger. Ho\ve\,er. given the observation from human behavior studies that people tend to react altruistically, it is reasonable to expect that people traveling from other floor levels nould yield to people leaving the fire floor. Where the \vidth o f the egress component changes, as indicated in Figure 4. IOa and 4.10b, the density of the flow also changes. The flow rate of people entering the egress component equals that leaving it:

I
'

I

'(b) Plan Mew

Figure 423 Diagram o f building for- Esati7ple 4.1.

Component-by-Component Analysis
The component-by-component analysis involves a determination of the time for the population to traverse each egress component. In this case, the density of the flow along each egress component must be determined so that the velocity and floiv rate can be determined. The starting point of such an analysis is to assume an initial density of the population. If such a calculation is to be done using algebraic equations (instead of one of the computer nlodels described in the last section of this chapter), a reasonable assumption is to consider all building occupants on a particular floor to be uniformly distributed in the corridors. As the population starts to move, the density of the people may change as a result of t'lree types of transitions: mergers o f flows at corridor intersections or where people entering a stair merge with people traveling in the stairs from other floors, changes in the widtl; of the egress component, changes in specific flan,, resulting in a transition from one type of egress component to another, e.g., a corridor to a stair. The new density after a transition may be determined by applying one of the following principles. The combined flow rat< of people entering an intersection equals the flo!!, rate of people from the intel-section (see Figure 4.9).

For converging flow. as illustrated in Figure 4.IOb, a queue might be espected to form at the transition. When there is a queue, the flow downstream from the transition is equal 1.0 the ~llasi~nurn capacity of the component. 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 of the quadratic equation results in two possible solutions for the density. The lesser value for density should be selected as the correct value. The lower density is correct for reasons indicated in the following example. If an occupant flow at the maximum density was approaching a widening comdor (Figure 4.10a), the solution of Equation (4.13) would yield one density greater than the maximum and one less. However, in the case of a widening corridor, it's unreasonable to expect the density to increase (and velocity to decrease) when proceeding from the narrow to the wide corridor. In either of these types of analyses where multiple egress paths are available to a group of occupants, some

assumption needs to be made of the distribution of occupants among the available paths. Often; an equal proportion of the group is assumed in each of the available paths. Alternatively, the distribution may be determined in propxtion to the respective capacities or other characteristics of the available paths (Predtechenskii and Milinskii 1978; Murosaki et ai. 1986). The following model can be applied if the order of evacuation is arbitrarily determined to proceed from highest floor to lowest floor. At time zero, all people move to the stairs on all floors and travel to the next floor level. If the stairwell capacity is exceeded as a result of the merger, then the maximum flow proceeds in the stairwell with the right-of-way given to the occupants on upper levels. (The total evacuation time is independent of whether people from upper floors have or surrender the right-of-way.) Consequently, the merged flow in the stairwell is composed predominantly of people from the upper level, supplemented by additional people from the next floor to provide the maximum permitted 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. The component-bycomponent approach is illustrated in Example 4.2.

(a) Diverging Flow

(b) Converging Flow
Figure 4.10 T,-ansiliorzin egress componcrit.

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 people in the corridor promptly upon notificaiion. Assume an equal number of occupants use the two stairs. (1.35 p/m2). The density of the people in the corridor is 0.125 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). (1.07 p/m2). Given flow of 40 p/min (0.67 p/s) in stairway, density is 0.099 lea\-ingthe fifth floor approaching the fourth floor is 152 @m (0.77 m/s). The vel&ity in the stair for 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 person from the fifthtloor reaches the founh floor. the flow of the G remaining people from the fourth floor recommences. S 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: 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). paths (where choices are available). The flow distribution between multiple paths may be determined by occupant behavior considerations. Optimization models minimize the evacuation time by considering an optimal distribution of occupants among multiple flow paths. The current optimization models neglect behavioral considerations. The risk assessment models quantify the risk posed to building occupants by conducting a fire Ilazard analysis, combined with an elementary evacuation analysis. The risk assessment models need to be

I

COMPUTER-BASED EVACUATION MODELS
The lbllowing three types of e\ acuation models are available: Sh~lation Optimization Risk assessment Si~nulatio~, modcls predict 1n0\sment and bella\.ior of occupants by assessing the t l o ~ disrribution among

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 networks assume unifonn conditions on each node and a constant traversal time along arcs. Alternatively, fine networks divide each room into several small sections. In some cases, a small grid is created over the entire building space where the size of a particular area may be as small as the area occupied by an individual. Theevacuation models assess movement of the building occupants by two perspectives. A global perspective tracks the occupants anonyn~ously. In this approach, the iiiodel does not distinguish which individual leaves the room or building at a particular time. The global perspective models assume uniform characteristics for the entire building population. In contrast, models with the individual perspective track each person, identifying where any particular person is during the evacuation period. The models with the individual perspective consider individual traits (e.g., mental and physical capabilities, tolerance to smoke, and group interactions). Behavioral characteristics included in the models may be done by several methods (e.g., deterministic equations [functional analogy], pre-established behavioral patterns, and iflthen rules, \vhich may or may not be subject to change by the user). One principal area of concern with the evacuation models relates to the reliability of input parameters. People movement characteristics need to be provided. Where a constant velocity is required, the results of the analysis will be dependent on whether the mean or maximum velocity is included. Some of the niodels require personal characteristics of building occupants (e.g., as

Figure 4.11 Evacuation models (adapted Gwynne and Galea 1999).

from

"patience" and motivation factors) be entered. Justification of such input is subject to much debate. Most of the models assume that occupants only become engaged in evacuation behavior. Neglecting the variety of nonevacuation behavior that occurs will result in a smaller evacuation time, perhaps substantially, than if such behavior is considered. None of the models currently considers the possibility of two-way flow in a corridor, either as the result of emergency personnel or some building occupants moving opposite to the evacuating occupants. As a prerequisite to any evacuation analysis, the number of people in the building must be established. The location of the occupants also needs to be specified, though at varying levels of detail, depending on the model. Location of individual occupants can be "placed" at a specific point for applications involving fine network niodels. For the coarse network models, people only need to be located in a room or floor of a building. When using a first-order approach with hand calculations, the calculations become very tedious when placing people in individual rooms. As such, for firstorder estimates, people may be placed in a queue at the esit door from the floor or large section of the floor to simplify the calculations. The loss of accuracy with this assun~ptionrelates to the time for people to travel from their respective starting points to fonn a queue at the door. In many buildings, this time is relatively short.

CHAPTER 5

Effective Areas and Smoke Movement
n building fires, smoke often migrates to locations remat; from the fire space. Stairwells and elevator shafts can become smoke-logged, thereby blocking evacuation and inhibiting fire fighting. In this chapter, several of the driving forces of smoke movement are discussed, methods of determining the neutral plane Ere provided, and some general comments are made conceming smoke movement. The information in this chapter is also applicable to the migration of other airborne matter, such as hazardous gases, bacteriolog~calmatter, or radloactlve matter in laboratories, hospitals, or indugtrial 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.

I

The total flow, vT,from the space is the sum of the flows through the leakage paths:

The effective area, A,, for this situation is that which results in the total flow, flow can be expressed as

-

vT.Therefore, the total

m

m

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 temperature. Variations in flow coefficients and temperature are addressed later.
7

Parallel Paths
Three parallel leakage areas from a pressurized space are illustrated in Figure 5.1. The pressure difference, 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 =
m
=
=

volumetric flow rate through the path, c h (m3/s); mass flow rate through the path, Ibis (kgk); dimensionless flow coefficient;

C

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 i series. s

The flows V* and v3 can be expressed in a similar manner. Substituting the expressions for PI, V * , and V3 into Equation (5.1) and collecting like terms yields

The effective area for flow paths in series is the 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 with Equation (5.2) yields this

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. as

Solving Equation (5.8) for ApT yields

The pressure difference across A , can be expressed

In Figure S. l. if A I is 1.08 R* (0.10 ni') and A? and A3 are both 0.54 f (0.05 m*), t ' what is the effective flow area ofthe system?

I Fro111Equation (SS),
Series Paths

A , = 2.16 R' (0.20 m').

The pressure differences Ap2 and Ap3 can also be expressed in a similar manner. Substituting Equation into Equa(5.9) and the expressions for ApI, Apz,and tion (5.7) yields anexpression for the effective flow area. (5.11)
AJ.

Three leakage paths in series from a pressurized 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 pressurized space to the outside is the sum of the pressure differences Ap ,. Ap?, and Apj across each of the respective flow areas. .-l ,,A?, and Aj:

This same reasoning can be extended to any number of leakage areas in series to yield

Principles pf Smoke Management

where n is the number of leakage areas,

Ai,in series.

In

smoke control analysis, there are tiequently only two paths in series, and the effective flow area for this case is

These two effective flow areas are in series with A , . Therefore, the effective area of the system is given by

1
11

Example 5.2 Two Equal Series Paths Calculate the effective leakage area of two paths of 0.22 (0.02 m2) in series.

ftL

following flow areas: AI =A2 = A3 = 0.22 = A ~ = A ~ = o 9(0.01 m2). .II

9 (0.02 m2) and A4

For two equal flow areas ( = AI = A?), Equation (5.13) A becomes A, = 0.707, A = 0.707 (0.22) = 0.156 9(0.0145 m'). Example 5.3 Two Unequal Series Paths Calculate the effective flow area of two paths in series, where the flow areas are AI = 0.100 ft2 (0.00929 m') and A2 = l . O f (0.0929 m2). O ;
..

From the equations above, A23, = 0.44 ft2 (0.04 m2), A4& = 0.33 9 (0.03 m2), and A, = 0.17 9 ( 0 316 m2).

Effects of Temperatures and Flow Coefficients

I

From Equation (5.13), A,

= 0.0995

(0.00924 m2).

This example illustrates that, when two areas are in series and one is much larger than the other, the effective area is approsimately equal to the smaller area.
P P

I/ I

II

For most calculations involved in smoke control, the a&umptions o f constant temperature and unifomi flow coefficient are appropriate, but it may be desired in some cases to consider the effects of these parameters. For parallel and series flow paths, the equations for effective flow area are

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

for parallel patlis and

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 .. example: ' This figure shows that A2 and A3 arc parallel; therefore, their effective area is

l

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,
C,

= = =

absolute temperature in effective flow path, "R Q; flow coefficient for effective path, dimensionless; absolute temperature in path i, "R (K); flow area of path i, fl? (m2);

Ai

=

Ci = flow coefficient of path i, dimensionless.
For the case of two areas in series with the same flow coefficients, the effective area is
-

. ,. ,.;

,..'

. ,< . : , ; .

.

. 7 ,.

.<

, .

,,

Normal S@& Effed

, , ,#, :,,;c y,f..?.,9.,>;$...'.;,9 ? ; Reverse Stack Effect

Figure 5.4 Air movement due to no/-nlaland reverse slack effect.

1. What'is the effective area of nvo paths in series, both of 0.22

Stack Effect
Frequently, when it is cold outside, there is an upward movement of air within building shafts, such as stainvells, elevator shafts, du~nbwaiters shafts, rnechanical 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 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 with the upward flow of gases in a smoke stack or chimney. However. a downward flow of air can occur in airconditioned buildings when it is hot outside. For this manual, the upward flow will be called normal stack effect and the downward flow will be called re\srse stack effect as illustrated in Figure 5.4. Most building shafts have relatively large crosssectional areas and, for most flows typical of those induced by stack effect, the friction losses are negligible in comparison with pressure differences due to buoyancy. Accordingly, this analysis is for negligible shaft friction. but shafi friction is specifically addressed later. Pressure within a shaft is due to fluid static forces and can be espressed as

ft? (0.02 m area with one at 70°F (21°C) and the other at ' )

100°F (3S°C)? Use

c of 70°F (21°C).

I

From Equation (5.19),A, = 0.153 9 (0.0 142 m ' ) . With both temperatures the same, the effective area of this system is 0.156 ft2 (0.0145 m) as calculated in Example 5.2. " . Considering the degree of uncertainty associated wit11 flow areas; adjustment of the effectiie flow area is unnecessary. 2. What is the effectix area above if the elevated temperature is 1000°F (538"C)?

I

From Equation (5.19). A , = 0.1 I @ (0.0105 m'). 1

DRIVING FORCES OF SMOKE MOVEMENT
The driving forces of smoke movement include naturally occurring stack effect, buoyancy of combustion gases, expansion of combustion gases, the wind effect, fan-powered ventilation systems, and elevator piston effect. This section discusses these driving fcrces and, in particular, addresses smoke movement due to the stack effect process, either naturally occurring or that of combustion gases. Generally, each driving force is discussed here as acting alone in order to facilitate discussion and lead to an understanding of smokc transport.

where = air pressure inside the shall,
g
z
= =

acceleration of gravity, elevation. gas density inside the shafi.

=

For the ele\.ations relltvant to buildings, the accslsration of gravity can bc considered constant. For constant density. Equation (5.70) hc integrated to yield can

..

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 pressure,po, is

where To
=

absolute temperature of outside air, absolute temperature of air inside the shaft.

where p, is the density outside the shaft Pressures inside the shaft and outside the building are graphically illustrated in Figure 5.5 for normal stack effect. This figure also shows the pressure of the building spaces, and methods of calculating this are presented later in this section. The pressure difference: 4,: from the inside to the outside is expressed as

T,

=

in Because i~ariations pressure within a building are very small compared to atmospheric pressure, atmospheric pressure, p,,,,,, can be used in calculating gas density from the ideal gas law.

Equation (5.25) was developed for a shaft connected to the outside. The neutral plane is a horizontal plane located at z = 0, where the pressure inside equals that outside as stated above. If the location of the neutral plane is known, this equation can be used to determine the pressure difference from the inside to the outside regardless of variations in building leakage or the presence of other shafts. Methods of determining the location of the neutral plane are discussed later. Tables 5.1 and 5.2 are comparisons of pressure differences due to various driving forces. For standard atmospheric pressure of air, Equation (5.25) becomes

where p
p,,,,,
= = = =

where air density. absolute atmospheric pressure, gas constant of air, absolute temperature of air. To
=

Aps0

=

pressure difference from shaft to outside, in. H 2 0 (Pal; absolute temperature of outside air, "R (K); absolute temperature of air inside shaft, "R (K); distance above neutral plane, ft (m); 7.64 (3460).

R
T

T,
h

= =
=

Values for the gas constant and of standard atmospheric pressure for several systems of units are given in

K,

Building Pressure.p,

Pressure

Figure 5.5

Pirsswcs nt~dpresszoa diJZwlices dut-ing normal stack efecf

Chapter 5-Effective Areas and Smoke Movement

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

7

Driving Force Stack effect,
Equation (5.26)

Location of'Ap Shaft to outside

CoriciiCions For all stack effect examples, T,= 70 "F and To= 0 "F:
h=30ft h=30Oft . For Tf= 1600 "F and To= 70 "F: h = 5 ft h = 10 ft

(in. H20)

0.07 0.7 0.05 0.1 1

Buoyancy of combustion Fire room to adjacent gases, room at ceiling Equation (5.3 l) Wind effect, Equation (5.34)

Across building (windward to leeward wall)

For all wind examples, p = 0.75 lblf?, C, = 4 . 3 : ,z

c,,

= 0.8,

and

U H = 15 mph

0.12 0.48 0.05 to 0.35

UH= 30 mph
Ventilation systems
Across barrier of Note: Values based on experience. smoke control system . Elevator piston effect, Elevator lobby to For all the examples of the upper limit of pressure differbuilding 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
Stack effect, Equation (5.26)

Location of Ap
Shaft to outside

Conditions
For all stack effect examples, T,= 21 "C and T, = -18 "C:

&(Pal

Buoyancy of combustion gases, Equation (5.3 1) Wind effect, Equation (5.34)

Fire room to adjacent room at ceiling

For Tf=870 O and To = 21 O : C C h = 1.5m h=3m

the (windward to leeward wall)

For all wind examples, p = 1.20 kg/m3, Clp2= - 0.3:

c,,

= 0.8

and

U H =14 m/s
Ventilation systems Elevator piston effect, Equations (5.41) to (5.43) Across barrier of smoke control system Elevator lobby to building Note: Values based on experience. For all the examples of the upper limit of pressure difference due to elevator car motion, p = 1.20 kg/m3, A, = 0.149 m2,

130 12 to90

'

For a single-car shaft with Cc = 0.83, A, 1.80 m :
2

= 5.61

m2, and A,

=

U = 2.03 m/s U = 3.56 m/s
m2, and A,

For a double-car shaft with Cc= 0.94, A, 7.41 m :
2

= 11.22

=

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 f (185 m) t 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 vertical airflow is in the shaft. Of course, the floors of -real buildings have some leakage and there is some airflow through these floors. The discussion of stack effect to this point has been general and it applies to buildings with o r without leakage through floors. To analyze the pressure differences on building floors, an idealized building model is used that has no leakage between 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

-ro

-30

- 20

-10

o

10

10

so

4)

50

Outside Temperature. To('C)

Figure 5.6 Graph of pressure difference due to stack effect. stituting Equation (5.27) into this relation and rearranging, the effective area is eliminated.

where A, = effective leakage area between the shaft and the outside, fi2 (ni2); A,; = per floor leakage area between the shaft and the building, ft2(m2); A;,
=

per floor leakage area between the building and the outside, ft' (m').

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 outside, as can be seen from the examples listed in Tables 5.1 and 5.2. In the event that many windows on the fire floor break due to the fire, the value of A,, becomes very larse on the fire floor. When this happens, the ratio A;;/ Aio becomes very small, and Q approaches , ; Thus, when a large number of windows break on the fire floor, the pressure from the shaft to the building is almost the same as that from the shaft to the outside. The development of Equation (5.28) considered the pressure difference uniform with height at each floor, which introduces an error-the maximum value of which can be calculated by Equation (5.26) for a value of /7 equal to the distance between floors. In the examples of Tables 5.1 and 5.2, if the floors were l0 ft (3. I m) apart, the maximum error of Equation (5.28) is about 0.01 in. H 2 0 (2.5 Pa). In general, this error is not significant. Equation (5.2s) can be rcwritkn for the pressure, p,. at the building space.

The mass flow rate, ril , for a floor can be expressed '/', by the orifice equation as C A , ( ~ ~ A , L I ~ , ) where C is a dimensionless flow coeficient that is generally in the range of 0.6 to 0.7. For paths in series, the pressure difference across one path equals the pressure difference across the system times the square of the ratio of the effective area of the system to the flow area of the path in question. Thus, the pressure difference from the shaft ~ to the building space is A / J , = ~A p , , , ( A , / . - l , i ) - . By sub-

Principles of Smoke Management,

The series flow approach to determining building pressures described above can be used for buildings with multiple shafts if all the shafts are at the same pressures and if all the shafts have the same starting and ending elevations. Pressure measurements on several buildings (Tamura and Wilson 1966, 1967a, 1967b) verify the stack effect theory presented above for conditions encountered in the field. Further, these studies show that the zero floor leakage idealization is generally appropriate for determining pressure differences on building floors due to stack effect. Additionally, Igmura and Klote (1988) have conducted full-scale stack e;Fect experiments at the Canadian ten-story Fire Research Tower near Ottawa, which verified the stack effect theory for a'iange of temperatures and of leakage conditions they considered r-presentative of most buildings. Figure 5.7 shows comparisons of measured and calculated pressure differences due to stack effect for outside temperatures of 12°F (-I 1"C), 27°F (-3°C). and 45°F (7°C). Figure 5.8 shows comparisons of measured and calculated pressure differences for ratios A,, /A,, of 1.7. 2.4, and 7. Further, this stack effect theory provides a useful approximation for buildings in which all of the shafts do not have the same starting and ending ele\ations.

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 lower than the building temperature. The stairwell was the cold column of air and the other shafts within the building were the warm columns of air. Smoke movement from a building fire can be dominated by stack effect. During normal stack effect (Figure 5.4), smoke from a fire below the neutral plane moves with the building airflow into shafts and up the shafts. This upward smoke flow is enhanced by anv buovancy forces on the smoke due to its temperature. Once above the neutral plane, the smoke flows out of the shafbinto theupper floors of the building, as illustrated in Figure 5.9b. As discussed in Chapter 1, this kind of smoke flow can have fatal conseauences. as in the fires at the MGM Grand and other buildings. Leakage between floors ..-results in smoke flow to the floor above the fire floo~. If lkakage between floors is negligible, the floors below -the neutral plane+xcept for the fire floor-will be &sentially smoke-free. For significant leakase b e t w e e ~ flls, s x e floor will be much greater than to other floors below the kutral plane, as is shown in Figure 5.9h For a fire above the neutral ,lane. the buildinn airtlo\vs due to normal stack effect tend to restrict the extent of smoke flow. Airflow from the shafts to the fire floor can prevent smoke infLLtdQnafihasahafts ( ~ i g ure 5.9c), but leakage between floors can result in some smoke movement. If the buoyancy forces of the hot

-

U

U

-

>

2 '

Pressure Difference (in Ho ,)
-.l2
-.cd

0

.d c

.08

.l2 28

24

20

16

g
m .-

s

12

q

Note: Solid lines are calculated valuas.

8

4

-30

-20

-10

0

10

20

0 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.

-15

-10

I -5

I
0

I

1
1s

5

10

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. /a s/ack effect. (0) due 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 b e fire floor, smoke can infiltrate the shafts and flow to 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 smoke, buoyancy forces can be so great that smoke can flow upward even during reverse stack effect. Further information about smoke flow due to stack effect and other driving forces is presented by Klote (1989).

perature. For standard atmospheric pressure, the above relation becomes

where
=

pressure difference from fire compartment to surroundings, in. H 2 0 (Pa); absolute temperature of outside air, OR (K); absolute temperature of gas inside fire compartment, OR (K); distance above neutral plane, ft (m); 7.64(3460).

To

=
=

Buoyancy of Combustion Gases
High-temperature smoke from a fire has a buoyancy force due to its reduced density. The pressures occurring during a fully involved compartment fire are illustrated in Figure 5.10, and these pressures can be analyzed in the same manner as pressures due to stack effect. In the same manner as Equation (5.26) was developed for stack effect, the foIlowing equation for the pressure difference Apfi from the fire compartment to its surroundings can be developed:

Tf
h IS

= =

Fang (1980) has studied pressure differences caused by the stack effect of a room fire during a series of full-scale fire tests. During these tests, the maximum pressure difference reached was 0.064 in. H 7 0 (163a) across the bum room wall at the ceiling. Observation of Tables ..5.1. and 5.2 can pro_~~.@e 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- bbU.oyan.c wi lr-di,-iriinxe ~ - w i
-

where To = absolute temperature of gases surrounding the fire compartment; Tj = absolute temperature gas \vithin the fire compartment; h = distance above the neutral plane. The neutral plane is a horizontal plane where the pressure inside the fire compartment equals that outside. Equation (5.30) is for a constant fire-compartment tem-

f@rlarge values of As; /Aio at almost any location from the neutral olane. For low values of-A~/d;, at locations a. fa_rfrom the neutral plane,>&ck effect can dominats even when windows are unbroken. When windows are broken, stack~effectis even more likely to dominate. Stack effect p .only be the dominant driving-force durcan - - ing times of significant inside-toloutside temperaxe di tlerence. ------Much larger pressure differences are possible for tall fire compartments where the distance, h, from the neutral plane can be larger, as illustrated by the follow-. ing example.
S"
,

-.

Chapter 5- Effective Areas and Smoke Movement

I

II

Example 5.8 Buoyancy Pressure in a

l

Fire ~o&ar&ent For a firecompartment temperature of 1470°F (800°C), what is the buoyant; pressure difference at 6 It (1.83-m) above the neutral plane?
-

11

l
I
Fire CunpmenlTfmw&m

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

1)

11

Example 5.9 Buoyancy Pressure Difference for Very Tall Fire Compartment if the fire compartment temperature is 1290°F (700°C),what is the pressure difference at 3 5 R (10.7 m) above the neutral plane?.

(TJ
700
800 900

I(

IOC

2w

300

U10

SM

WO

Fire C-Temperahrre(T)

Figure 5.11 Graph of pressures dueio buoyancy.
pressure difference across these openings due to expansion is negligible because of the large flow areas 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 expansion in such a closed space subject to the exhaust o f zoned smoke control, is addressed in Chapter 12.

)I

l

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 illusrrarr the exr-nt to which Equation (5.31) can be applied.
.

I

I(

Expansion of Combustion Gases
In addition to buoyancy, the energy released by a fire can cause smoke movement due to expansion. In a fire compartment with only one opening to the building, air will flow into the fire compartment and hot smoke will flow out of the compartment. Neglecting the added mass of the fuel, which is small compared to the airflow, 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 ternperatures.

Wind Effect
Wind can have a pronounced effect on smoke movenient. The pressure, p,,, that wind exerts on a wall of a building can be expressed as

where
p,,. = wind pressure, in. H20(Pa);

v,,,,, v;,, where

TO,/,

Tit,

C,,. = dimcnsionless pressure coefficient;
p,
=

outside air density, lb/$ (kgh3); (mfs); 0.0 129 (1 .OO).

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);

,

= volumetric flow rate ofair into the fire compart-

K,,.

=

ment, chm (m3/s); To,, = absolute temperature of smoke leaving the fire conipartnient, "R (K); T,, = absolute temperature of air entering the fire compartment, "R (K). For a smoke temperature of I 1 10°F (600°C), the gas will expand to about three tinies irs original volume. For a tire conlpartnlcnt with open doors or windows, the

I t is thc nature of wind to be variable with peak values &at can be two or three times that of the average. The peak values arc important for structural loads, but - - ___theX5aTaverage wind-velocity is more appropriate for . - . . -. the calculation OS smoke transport and evaluation of -smOEnlanagenlent s y ~ e r n In . this discussion of wind ~ . .- effects, the tern1 \*cloci~y s ~ u s e dto indicate the nreo,~ i
-- W -

-

Principles of Smoke Management

The pressure coefficient depends on building geometry 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 symmetrical, 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 negative for the other walls. For a tall building without local wind obstructions, typical distribution of the pressure coefficient is shown in Figure 5.13. As with Figure 5.12, distribution of these Pressure coefficients is also symmetrical. Values of oressure coefficient C.... averaged over the wall area. are li$ed in Table 5.3for rectangular buildings, which are free of local obstructions. An approximation of the overall pressure difference from one side of a building to another due to wind effect can be obtained from
L

__

.v

Figure 5.12 Typical distribution ofpressure coefficient over a low-rise building free of local obstructions.
W2d

U

L

where

L-0.6 '
average pressure coefficient for windward wall;

-

C,",

=

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 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,

I

where
U U .
= =

wind velocity, @m ( d s ) ; velocity at reference elevation, @m ( d s ) ; elevation of velocity, U, ft (m); reference elevation, ft (m);
Side

/
\\\\\\\
Front

z

=

zo =

a = wind exponent, dimensionless.

Some general values of the wind exponent, a , are

Figure 5.13 q ~ i c adistribution ofpressure coefficient l over a tall buildingj.ee of local obstt-uclions., the city center and category 2 (Figure 5.14) for winds Trom the opposftedirection. I nere has been a la%of consistency among authors regarding recommended val-

-

edge of a large city center could be considered terrain category 1 (Figure 5.14) for winds from the direction of

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])
Building Height Building Plan Ratio Ratio Elevation Wind Angle Plan

c,,,for Surface
A B

a

C

D

Note: h = height to eaves or parapet P = length (greater horizontal dimension of a building); (lesser horizontal dimension of a building).

W

= width

Principles of Smoke Management,

-

Terrain Category 1: Large City Center 50% of Buildings Higher Than 70 ft (21 m); Over at Least 6600 ft (2000 m) Upwind

n

Terrain Category 2:. . Urban, Suburban, Wooded Areas & Other Areas With Closely Spaced Obstructions Compared tc or Laps: Than 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

m

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

-

-7

rc

Wind Velocity Profile

a = 0.10 6 = 7OOft (210m)

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 were chosen to be consistent with those o f the 1997 ASHRAE Handbook-Fundamentals, Chapter 6, "Airfl;w Around Buildings." Using Equation (5.35) with z = H (where H is t h e upwind height of the wall of a building), the average velocity of the gradient wind can be expressed as

For building and wind measurement sites that are near each other, the velocities of the gradient winds are equal. Equating Equations (5.36) and (5.37) and rearranging results in
Smet
'H

=

'met met

(:)

H

a

(5.38)

Substituting this into Equation (5.33) yields where

UgeH velocity of the gradient wind above the building, = fpm (&S); = wind velocity 2t the top of the wall, fpm (&S); UH = upwind height of the wall, fi (m); H 6 = boundary layer height in the vicinity of the building, ft (m); a = wind exponent in the vicinity of the building, dimensionless. General values o f boundary layer height, 6, are listed inFigure 5.14 for the terrain categories. and these were also chosen to be consistent with those of ASHRAE Fundamentals. The weather service measures typically at 33 wind data at airports and other locatio~is, ft (10 m) above the ground. The average velocity of the gradient wind can also be expressed as

where

where

Ug,,,,,, = velocity of the gradient wind above the wind anemometer, fpm ( d s ) ; U,, = measured wind velocity, fprn ( d s ) ; H,,,,, = height of wind measurement, ft (m); 6,,,,, = boundary layer height in the vicinity of the wind anemometer, ft (m); = wind exponent in the vicinity of the wind anea;,1et mometer, dinlensionless.

It can be seen that Equation (5.39) has the advantage in that it can be used to calculate wind pressures based on measured design wind data. The above discussion is for buildings without large local obstructions. For buildings with such obstructions, specialized wind tunnel tests are needed to determine the pressure coefficients due to the wind. Such tests are routinely conducted for structural analysis of large buildings. For both structural and smoke management purposes, the wind flow around buildings is fully developed turbulent flow, and the flow coefficients are independent of the Reynolds number. Thus, the flow coefticients obtained from wind tunnel tests for structural analysis are applicable for smoke management analysis. While the tern~inologyof a wind tunnel test report may ditkr from that o f this section, the results are applicable to smoke management analysis. For tnforniation about wind and smoke management, readers are referred to Kandola (1986a, 1986b) and Klote (1995). For additional information about wind pressures on buildings see Aynsley (19S9), Shaw and Tamura (1977). and Kandola ( 1 9 8 6 ~ ) .Several civil engineering tests provide useful information about wind engineering-for example, Dyrbye and Hansen (1997); Liu (1 99 I): MacDonald (1 975); and Simiu and Scanlan ( 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~eight the windward wall is 120 fr. the wind coetticient is 0.8, and tlie outside air density is 0.075 of lb/m3 (1.2 kg;lm3). Calculate the wind pressure.

a,

a the From Figure 5.14, the city center is terrain category 2 with a = 0 . 2 and 6 = 1200 FI (370 IN), ~ l d 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).

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
11

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 a e is terrain category 1 with o = 0.33 and 6 = 1500 ft (460 m). ra time, 2amct 2a g00 2(0.14) 120 2(0.33) From Equation (5.40), C = = 0.476. h - Hme 6 33 l500 From Equation (5.39),pW 0.09 in. H 2 0 (22 Pa). is 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.

(-,

(9 (-1

(1 -

11

II

Forced Ventilation Systems
Heating, ventilating, and air-conditioning (HVAC) systems frequently transport smoke during building fires. When a fire starts in an unoccupied portion of a building, the HVAC system can transport smoke to a space where people can smell the smoke and be alerted 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 control mode of operation. The advantages and disadvantages 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 shutting down the HVAC system prevents it from supplying oxygen to the fire, system shutdown does not prevent smoke movement through the supply and return ducts, air shafts, and other building openings due to 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 continued 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 theoretically verified, it seems that it may have the potential to minimize smoke transport through the HVAC system. Computer 'simulation of smoke movement through HVAC systems is discussed by Klote (1987). Machinery Room Lobby Building
Space
4--

+-

+-

'

"/ /

/v h '

/h <L ' I '

/'/

//

L

Note: Arrows indicate the direction of flow. /,y / ,b /

Figure 5.15 Airflow due to the downward moveinenf of an elevator cal:

where
P
=
= = = = =

air density within the shaft, lb/ft3 (kg/m3); cross-sectional area of shaft, ft2 (m2); velocity of elevator car, Fpm (mls); number of floors above the car, dimensionless; number of floors below the car, dimensionless; flow coefficient for building leakage paths, dimensionless; effective flow area per floor batween the shaft and the outside, (m2); flow coefficient for flow around the car, dimensionless;

A,

U
N, Nb

Elevator Piston Effect
When an elevator car moves in a shaft, transient pressures are -produced. A downward-moving elevator car forces air out of the section of shaft below the car and into the section of shaft above the car, as illustrated in Figure 5.15. Klote and Taniura (1986) developed the following analytical equation for the pressure difference, 41,,, to elevator piston effect from the outside due to the elevator shaft above the car:

C A,

=

c,

=

Chapter 5-Effective Areas and Smoke Movement

A, = free flow area in shaft around car, or cross-sectional area of shaft less cross-sectional area of the car, I(m2); ?

bp,- = pressure difference h m the building to the lobby, in. H20 (Pa);

dp,, = pressure difference from the outside to the shaft,
in. H20 @a);

Kpe= 1.66x10~(1.00).
The coefficient C, was cvzluated at 6.94 for a twocar shaft with only one car moving and at 0.83 for a twocar shaft with both cars traveling side-by-side together. The value for the two cars moving together is believed to be appropriate for obtaining approximations of pressures produced by the motion of a car in a single car shaft. For the sake of simplicity in the analysis leading to Equation (5.41), buoyancy, nind, stack effect, and effects of the heating and ventilating system were omitted. Omitting stack effect is equivalent to stipulating that the building air temperature and the outside air temperature are equal. For the system of three series flow paths from the shaft to the outside illustrated in Figure 5.15, the effective flow area, A,, per floor is

A, Air

=

effective flow area between shaft and the outside,

ti? (m2);
=

leakage area between the building and the lobby,

rt' (m2).
This series flow path analysis does not include the effects of other shafts, such as stairwells and dumbwaiters. Provided that the leakage of these other shafts is relatively small compared to AOi, Equation (5.42) is appropriate for evaluation of A, for buildings with open floor plans. Further, Equation (5.43) is appropriate for closed floor plans, provided all the flow paths are in series and there is negligible vertical flow in the building outside the elevator shaft. The complicated flow path systems probably require case-by-case evaluation, which can be done by using the effective area techniques presented later in this manual. To test the above theory, experiments were conducted in a hotel in Toronto, Ontario, Canada. Figure 5.16 shows measured pressure differences across the top floor elevator lobby while a car was descending. Also shown is the calculated pressure difference, which is in good ageement with the measurements. This experiment is described in detail by Klote and Tarnura (1986).

where
A, = effective flow area, ft2(m2): A,, = leakage area behveen the lobby and the shaft, l? i

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 provided later in this text. in a similar manner to the development for stack effect, the pressure difference from the lobby to building interior can be expressed as
Time (S)

where

Figure 5.16 Pressure difference across elevator- lobby 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 singleshaft? 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) 0.54 (0.050) AOi,area between outside and building 60.4 (5.61) As, cross-sectional area of shaft 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 difference kom the building to the lobby is 0.18 in. H20 (45 Pa).
The pressure difference, Mri, cannot exceed the upper limit of

K,,

= 1.66

X

I O -(1.00). ~

where (Wri),

'

=

upper limit of the pressure difference from the building to the lobby, in. HzO (Pa); air density within the shaft, 1b/ft3(kg/m3);

=

2 = cross-sectional area of shaft, ft' (m );
=

effective flow area per floor between the shaft and the outside, ft2 (m2): velocity of elevator car, fpm (rnls); free flow area in shaft around car, or cross-sectional area of shaft less cross-sectional area of the car, ft2 (m2); leakage area between the building and the lobby, ft2 (m2); flow coefficient for flow around the car, dimensionless;

This relation is for unvented tlevator shafts o r shafts for which the vents are closed. The pressure difference, (M,;),,,is strongly dependent upon U, As, and A,. For example, Figure 5.17 shows the calculated relationship k t w e e n (W,;),, and U due to one car moving in a single-car shaft, a double-car shaft, and a quadruplecar shaft. As expected, (Wri), is much greater for the single-car shaft. It follows that the potential for smoke problems due to piston effect in single-car shafts is hhi greXeXi~i~E~arSLftJ.CbXparisOn< f stack effect induced pressure differences indicates that they can be larger than those of other driving forces-(Tables 5.1 and 5.2).
a

= =

Air

=

=

Operation of . -- --by the fire ssrvice during a elevators , --. - . fire can result in smoke being pulled into the elevator shaft by piston effect. It seems a safe rzconmendation that fire fighters should favor the-"se o f elevators in mult~ple-car shafts over ones in singe-car shafts. Klote -(1988) developed another analysis of piston effect. including the influence of elevator smoke control, and experiments conducted by Klote and Tamura (1987) 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 & 600 fpm (3.05 d s ) 0.83 0.075 lb/@ (1.20 kg/m3) p, air density in shaft 0.325 ft2(0.0302 m2) A,, effective area between shaft and outside 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 A,, free flow area around car shaft For cyadru~le-car A,, cross-sectional area of shaft A,, free flow area around car 120.8 ft2 (1 1.22 m') 79.8 ft2 (7.41 m')

U, car velocity C flow coefficient for flow around elevator car ,

241.5 ft2 (22.44 m ' ) 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.
o.2or

; , ,

Car Velocity (rnls) 2 3
1

l,

1

4

,

l

5, 50
40

these neutral plane locations, the flowv rates and pressures can be evaluated.

g

-

Shaft with a Continuous Opening
The flow and pressures of normal stack effect for a single shaft connected to the outside by a continuous opening of constant width from the top to the bottom of the shaft is illustrated in Figure 5.18. The following analysis of this flow, and the resulting location of the neutral plane, \\.as developed by McGuire and Taniura (1975). The pressure difference from the shaft to the outside is expressed by Equation 15.25). The mass flow\. rate, dm;,,, through tlie differential section, tlh. of the shaft below the neutral plane is

Single Car Shaft

Quadruple Car Shaft Double Car Shaft

Q

ioo

200

400

600

800

1000

n

Car Velocity (fpm)

Figure 5.17 Calczclated q m r limit ofpi-esszcre d(lj^eretice,Ji.onithe elevator lobby to the buildi ~ l g to pis1017effect. due

LOCATION OF NEUTRAL PLANE
In this section, methods of determining the location o f the neutral plane arc described for a single shaft connected 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

where

Principles of Smoke Management

Example 5.14 Location of Neutral Plane 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).

II

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 opening.

a continzro~rs

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 flow sate from the shaft can be developed, where H is the total height of the shaft.

hvo openings.

Shaft With Two Vents
Normal stack effect for a shaft with two openings is illustrated in Figure 5.19. The pressure difference from the shaft to the outside is expressed by Equations (5.25) and (5.26). To simplify analysis, the distance? H , between the openings is considered much greater than the hzight of either opening. Thus, the variation of pressure \\h11 height for the openings can be neglected, and the mass flow rate into the shaft can be expressed as

For steady flow, the mass flow rate into the shaft equals that leaving it. Equating Equations (5.46) and (5.47), canceling like terms, rcarrangins. and substituting, Equation (5.24) yields
I H,, H l+(T~/T0)""

(5.45)

where

H,,

=

distance from the bolt0111of the shaft to the neutral plane, ft (m);

and the mass flow rate out of
~ j l ~ =, ~ ,

is
(5.50)

H = height ol'shal't. fi (m);
Ty
= =

c~,, ,I )

absolute temperature of air in shaft. "R ( K ) ; absoluw temperature ofoutside air. "R ( K ) . where A , and A* are the areas above and below the neutral plane. Equating these two flo\vs as was done above yields

r,

Chapter 5-Effective Areas and Smoke Movement

where

H, H

=

distance from the bottom of the shaft to the neutral plane, ft (m); heightofshaft, ft(m); absolute temperature of air in shaft, "R (K); absolute temperature of outside air, "R (K); area above neutral plane, ft2 (m2);

small in comparison to the shaft height, H. Thus, a constant 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, expressed as Equation (5.47), plus the flow out of the vent of area A, located a t an elevation of H, above the shaft bottom. .

=

T,

= =
=

T,
A,

Ab - = area below neutral plane,

ft2(m2).

The conservation of mass equation for the shaft can be written as

The location of the neutral plane is highly dependent 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 function of temperature. Example 5.15 Location of Neutral Plane with Two Equal Openings 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). From Equation (5.5 l), the neutral plane is located 46.4 fi (14.1 m) above the bolton1area. This is only a little less than Example 5.14 with the continuous opening (48.8 ft [14.9 m]).

Canceling like terms and incorporating Equation (5.53) results in

1
As would be expected, this equation reduces to Equation (5.48) for A,. = 0. Equation (5.54) can be rearranged as

1

Example 5.16 Location of Neutral Plane with Two Unequal Openings 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 f r m the building's mid-height.

l

1

Vented Shaft
The flow and pressures of normal stack ellkcl lor a shaft connected to the outside by a vent and a continuo u s opening are shown in Figure 5.20. The tbllowing analysis is tor a vcnt above the neutral plane, but a similar 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 simplicity of analysis, the height of the vent is considered

Neutral Plane

-

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 the neutral plane, the neutral plane will be located between the height described by Equation (5.4) for an unvented shaft and the vent elevation, H , Further, the larger the value of A, /ArH, the closer the neutral plane will be to H ,

For relatively large vents, the ratio A1H/AV approaches zero. As A1H/AVapproaches zero, the first and third terms in the above equation approach zero, and the equation is reduced to H, = H , Thus, the neutral plane is at or near the vent elevation, for a vent area very

CHAPTER 6

Principles of Smoke Management
he term "sn~okemanagement," as used in this manual, includes all methods that can be used singly or in combination to modify smoke movement for the benefit of occupants or firefighters or for the reduction of property damage. The use of barriers, smoke vents, and smoke shafts are traditional methods of smoke management. The effectiveness of barriers is limited to the extent to which they are free of leakage paths. The effectiveness of atrium smoke vents and smoke shafts is limited to the extent that smoke must be sufficiently buoyant to overcome any other driving forces that could be present. Fans are used with the intent of providing smoke protection by means of pressurization. The mechanisms of compartmentation, dilution, pressurization, airflow, and buoyancy are used by themselves or in combination to manage. smoke conditions in fire situations. These mechanisms are discussed in the sections below. criteria for the construction of smoke barriers, including doors and smoke dampers in these barriers. The extent to which smoke leaks through such barriers depends on the size and shape of the leakage paths in the barriers and the pressure difference across the paths. Hazard analysis (chapter 9) can be used to evaluate the performance of con~partmentation.

Dilution Remote From a Fire
Dilution of smoke is sometimes referred to as smoke purging, smoke removal, smoke exhaust, or smoke extraction. Dilution can be used to maintain acceptable gas and particulate concentrations in a room subject to smoke infiltration through leakage paths from an adjacent space. Tliis can be effective if the rate of smoke leakage is small compared to either tlie total volume of the safeguarded space or the rate of purging air supplied to and removed from the space. Also, dilution can be beneficial to the fire service for removing smoke after a fire has been estinguished. Sometimes, when doors are opened, smoke will flow into areas irltended to be protected. Ideally, such occurrences of open doors will only happen for short periods of time during evacuation. Smoke that has entered spaces remote from the fire can be purged by supplying outside air to dilute the smoke. The following is a simple analysis of smoke dilution for spaces in which there is no tire. At time zero (I = 0), a compartment is contaniinated with some concentration of smoke and no additional smoke flows into the compartment or is generated within it. Also, the contaminant is considered uni fol-mly distributed throughout tlie

SMOKE MANAGEMENT

Compartmentation
Barriers with sufficient fire endurance to remain effective throughout a fire exposure have a long history of providing protection against fire spread. In such fire compartmentation, the walls, partitions, floors, doors, and other barriers provide some level of smoke protection to spaces remote from tlie fire. Tliis section discusses the use of passive compartmentation, while tlie with pressurizause of compartmentation in co~ijunction tion is discussed later. Many codes, such as the NFPA Code (NFPA 2000c), provide specific 10 1 L{/b S a j i ! ~

Chapter 6 -Principles of Smoke Management

space. The concentration of contaminant in the space can be expressed as

Example 6.2 Smoke Dilution in a Space Remote from the Elre A space is isolated from a fire by smoke barriers and selfclos-

This equation can be solved for the dilution rate and the time.

ing doors so that no smoke enters the compartment when the doors are closed. However, when a door is opened, smoke flows through the open doonvay into the space. If the door is 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

CO = initial concentration o f contaminant
C = concentration of contaminant at time, t a = dilution rate in number of air changes per minute = time after smoke stops entering space or time after I which smoke production has stopped, minutes e = constant, approximately 2.7 1S The concentrations COand C must be ekpressed in the same units, and they can be any units appropriate for the particular contaminant being considered. In reality, it is impossible to ensure that the concentration of the contaminant is uniform throughout the compartment. Because of buoyancy, it is likely that higher concentrations would tend to be near the ceiling. Therefore, exhausting smoke near the ceiling and supplying air near the floor will probably dilute smoke even faster than indicated by Equations (6.2) and (6.3). Caution should be exercised in the location of the supply and exhaust points to prevent the supply air from blowing into the exhaust inlet and, thus, short-circuiting the dilution operation
Esnmple 6.1 Smoke Purgin!: After the Fire is Extin~uished 1. After the fire department puts out a fire, they want to clear

Many people have unrealistic expectations about what dilution can accomplish in the fire space. The analysis of the previous section is not applicable to spaces in which there is a fire. There is no theoretical or experimental evidence that using a building's heating, ventilating, or air-conditioning (HVAC) system for smoke dilution will result in any significant improvement in tenable conditions within the fire space. I t is well kno\vn that HVAC systems promote a considerable degree of sir mixing within the spaces they serve. Because of this and the fact that very large quantities of smoke can be produced by building fires, it is generally believed that dilution o f smoke by an HVAC system in the fire space will not result in any practical improvement in the tenable conditions in that space. Thus, it is recommended that smoke purging systems intended to improve hazard conditions within the fire space or in spaces connected to the lire space by large openings not be used.

Pressurization
Systems using pressurization produced by mechanical fans are referred to as stnoke contt-01 in this book and in NFPA 92A (NFPA 2000a). A pressure difference across a barrier call control smoke movement, as illustrated in Fizure 6.1. Within the barrier is a door. The high-pressure side of the door can be either a refuge area or an egress route. The low-pressure side is exposed to smoke from the fire. Airflow through the gaps around the door and through construction cracks prevents smoke infiltration to t!le high-pressure side. When the door in the barrier is opened, airflow through the open door results. When the air velocity is low, smoke can flow against the airflow into the refuge area or egress route. as shown in Figure 6.2. This smoke backflow can be pre\mted if the air velocity is sufficiently lar,ne, as shown in Figure 6.3. The magnitude of velocity necessary to preLrent backflow depends on the energy release rate ofthe fire. as discussed in the liext scction.

the smoke quickly so that they can make an inspection to determine if the fire is completely out. If the HVAC system is capable of a dilution rate of 6 ach. how long will it take to reduce the smoke concentration to I% of the initial value?
, The dilution rate, a, is 0.1 changes per minute, and C /C is 100. From Equation (6.3), the time to get the concentration to I% is 46 minutes. Considering the desire of the tire department to quickly inspect the area, such a long purging time will probably be excessive.

2. If the tire department wants the space to be purged in 10 minutes. what dilution rate is needed?

The time, I, is 10 minutes, and C/ is 100. From Equation ,C (6.2).the dilution rate is 0.46 changes per minute. or about 28 changes per hour.

Principles of Smoke ManagemeM

High Pressure Side

Low Pressure Side

'\ \\Y\';

\\\\

c u d

..\\\\\.\\\\

\ \\\ ...

Figure 6.1 Pressure differ-ence across a barrier of a snzoke control system can prevelir smoke infiltratioii to r/7e high-press~rre side of h e barriei

In this case, the appiopAate physical quantity is' pressure difference. Consideration of the two mechanisms as separate has the added advantage that it emphasizes them different considerations that need to be given for opened and closed doors. To ensure that expansion pressures are not a roblem, pressurization systems should be designed so that a patin exists for smoke movement to the outside. This path could be as simple as relying on a top-vented elevator shaft, f i ed e exhaust. It is important that some be provided. The pressurization systems most com=ly used are pressurized stairwells and zoned smoke 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

---+ +
..,\\.;..,

Airflow
Airflow has been used extensively to manage smoke from fires in subway, railroad, and highway tunnels. 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, airflow is not used so extensively in buildings. The control problem consists of having very small flows when a door is closed and then having those flows increase significantly 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, which demonstrate why this technique is not recommended, except wlien the fire is suppressed or, in the rare cases, when fuel can be restricted with confidence. Thonias (1970) determined that airflow in a corridor 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 is believed to result in transfer of trace amounts of 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 follon.ing enlpirical relation for this critical velocity:

,\

\,,\'...\

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

..

%\

...%, \...~,,

.>.,,\V: \,\\...

. ..... ,

Figure 6.2 Smoke bnc&fIowagainst low air veloci/J! tlirorrgli an open door-rvny.

.

.

,.

"-.,

. . ,:~ ..'..\'..'\"\.',. '\

Caution: Because it supplies oxygen to the fire. airflow needs to be used with great care.

, .-.,..................... ........................ ........... ................ ....;:..........................................
>

. '\,\\\\\.\'\\ '

'.

\,\

, . . . ., . . . . , ...... , . . . . . . .................................. . . . .

----+
High Air

BVelocity

Figure 6.3 High uir- velocip 1171arrghan open doorway pre1~er71s suioke backj7ow.

Pressurization results in airflows of high velocity in 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, considering these mechanisms as separate is advantageous for discussing smoke management systems. For a barricr with one or more large openings, air velocity is the appropriate physical quantity for both design and nicasurcment. Ho\i.cver. wlien there are only small cracks, such as thosc around closcd doors, designing to and measurement of ail- velocities is impractical.

critical air velocity to prevent smoke backflow, enerzy release rate into corridor, corridor width, density of upstream air, specific heat of downstream gases, absolute temperature of downstream gases.

Chapter 6- Principles of Smoke Management

K
g

constant on the order of 1, acceleration of gravity. The units are not given for Equation (6.4), as it is valid for any homogenous system of units (Appendix A). The downstream properties are considered to be sufficiently far downstream of the fire for the properties to be uciforrn across the section. Note that_ T is for the downstream gases, and p is for upstream gases. This means that p is not calculated from T. The critical air velocity can be evaluated at p = 0.08 1 lb/ft3 (1.3 kg/m3), Cp= 0.24 Btullb O (1.005 W k g 'C), T = 81°F (27OC), F and K = I.
= =

where

Uk = critical air velocity to prevent sn:.?ke backflow, fpm
(m/?);
= energy release rate into cor~idor, Btu's (kW); W = corridor width, fi (m); K, = 86.3 (0.292). Equation (6.5) can be used when the fire is located in the corridor 01 when the snioke enters the corr~dor

0

through an open doorway, air transfer grille, or other opening. The critical velocities calculated from Equations (6.4) and (6.5) are approximate because an approximate value of K was used. However, the critical velocities from this relation are indicative of the kind of air velocities required to prevent smoke backflow from a fires of different sizes. As c : be see;; from Figure 6.5, the critical velocity is less for wider corridors. Examples 6.3 and 6.4 illustrate the flows needed for different fires. The equation of Thoinas can be used to estimate the airflow rate necessary to prevent smoke backflow 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. While the critical velocity can be calculated, the oxygen supplied is a concern. Huggett (1980) evaluated the oxygen consunied for combustion of numerous natural and synthetic solids. He found that for most materials that are involved in building fires, the energy released per unit of mass of oxygen consunied is approximately 5630 Btu/lb (13.1 MJJkg). Air is 23.3% oxygen by 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) 0 800

-

0.4

0.8

1.2

1.6

2.0

2.4

g
.9 m
X

600

-

0

400

-

.-

.-

6

-

200

Principles of Smoke Management

sumed, 1300 Btu of heat is liberated. Stated in the S1 system, if all the oxygen in a kg of air is consumed, 3.0 MJ of heat is liberated. As can be seen from Example 6.3, the air needed to prevent smoke backflow can support an extremely large fire. In most locations of commercial and residential buildings, sufficient fuel (paper, cardboard, furniture, etc.) is present to support very large fires. Even when the amount of fuel is normally very small, short-term or transient fuel loads (during building renovation, material delivery, etc.) can be significant. Because of the concern about supplying combustion air to the fire, caution is recommended when airflow 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 justifi-kd by the lack of appropriate smoke management alternatives. The use of fuel restriction or fire suppression t6 limit the size ofthe fire for a smoke mana,oement system relying on airflow has the potential for catastrophic failure. Therefore, the use of airflow is not recommended for smoke management in buildings except when the potential for failure of fuel restriction or fire

suppression is evaluated to be acceptable. The methods of tenability analysis discussed in Chapter 9 can be used to evaluate the consequences of such failures.
Example 6.3 Airflow to Prevent Smoke Backflow from a Small Fire 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 is needed to prevent smoke backflow h m such a fire in a corridor 4 ft (1.22 m) wide and 9 ft (2.74 m) high? 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 (1.22 x 2.74 = 3.34 m2). The flow rate is .the cross-sectional

11

Example 6.4 Airflow to Prevent Smoke Backflow from a Large Fire An energy release rate of 1420 Btu/s (1.5 MW) would result in a large portion of the corridor beins completely involved in fire. What flow rate of air is needed to prevent smoke backflow from such a fire in the corridor of Example 6.3? From Equation (6.5), the critical velocity is 616 fpm (3.13 nds). The flow rate is about 22,200 c h (10.5 m3/s).

1

1

chapter 6-Principles of Smoke Management

Example 6.5 Airflow Through a Doorway and Fire Growth

I

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

fire could have an energy release rate on the order of2270 Btu/s (2.4 MW). w a estimate of critical v&city is obtaded 6bm the ht 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). 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)

II/l

IIl

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). 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)

1

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 not recommended except when the fire is suppressed or in the rare cases when fuel can be restricted with confidence. doors. The flow through a crack or other opening can be represented by the general function,

Buoyancy
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 malls_A employed for atriums and covered s h o p p i ~ ~ ~ concern with atrium smoke management systems is that the sprinkler flow will cool the smoke, r e d u c w y %cy and, thus, system effectiveness. There is no question that spr~fiklETlow does-cooi..mke. but i t j s unknown to what extent that cooling reduces&&.veness of fan-powered venting. Further research is needed i z t h i s area. However, the existing information can be used to develop new design information for fan-powered venting systems. NFPA 92B (NFPA 2000b) provides methods of design analysis for smoke management systems in large spaces, such as a t r i u m and shopping malls.

-

where
V
=

volumetric flow rate through the path,

< -

_-

A = pressure difference across path, p

f

= general functional relation.
.

The particular form of the functionf depends on the geometry of the opening and Reynolds number. The Reynolds number is

where

AIRFLOW AND PRESSURE DIFFERENCE
For a crack, gap, or other opening with a pressure difference across it, a flow will result from the higher pressure to thc lower pressure. Many different equations have been used to express the relation between fluid flow rate and pressure difference with regard to air and smoke flow in buildings. This section contains a discussion of some of the more common equations, as well as a detailed discussion of flows through the gaps around

R,

=

Reynolds number, dimensionless;

D, = hydraulic diameter of flow path, in. (m);

U = average velocity in flow path, fpm (mts);
v
2 = kinematic viscosity, $/S (m /s);
=

KR

1.39

X

1U3(1 .OO).

Values of kinematic viscosity are listed in Tables A.8 and A.9 of Appendix A. The hydraulic diameter is four times the cross-sectional area of the path divided by

Principles of Smoke Management

the "wetted perimeter" of the path. For example, the hydraulic diameter of a circle is the diameter of the circle, and the hydraulic diameter of a square is the side of the square. For the long rectangular gaps around doors, the hydraulic diameter is the gap thickness (D,, = 2a, where (z is the gap thickness). The Reynolds number is usually thought of as the ratio of-kinetic forces to viscous forces. Later sections discuss different approaches that apply for flow dominated by viscous forces, kinetic forces, or both. The pressure difference above can be expressed as

p

density gas in path, lb/ft3 (kg/m3); KO = 776. (1.00);
=

KO? = 12.9 (1.00). Dynamic forces dominate flow with Reynolds numbers greater than about 2000 or 4000, depending on path geometry. At these Reynolds numbers, the flow becomes turbulent. For turbulent flow, the velocity at a given point fluctuates rapidly in an apparent random manner. 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 extensively to orifice flow meters that it is often referred to as the orifice equalion, and Equation (6.10) also is referred to by the same name. The orifice equation is also commonly used for analysis of airflow in buildings and for analysis of smoke management systems. Because the orifice equation is based on Bernoulli's equation, it strictly applies to steady, frictionless, incompressible flows. However, the flon. coefricient was introduced to account for friction losses due to viscosity and for dynamic losses. The flow coefficient depends on the Reynolds number and the geometry of the flow path. For flows through gaps around doors and through construction cracks, the coefficient 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.

where pi = pressure at path inlet,

p, = pressure at path outlet,
p = den&

gas in path,

Zi= elevaiion of the path inlet, Z = elevation of the path outlet, ,
g
acceleration of gravity. Equation (6.8) is for constant density in the flow path and for flows where the values of the inlet pressure, 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. Thus, this representation is appropriate for smoke control analysis, as well as general considerations of airflow in buildings.
=

-

For standard air density of p = 0.075 1b/ft3(1.20 kg/ 111') and for C = 0.65, Equation (6.9) can be expressed as

Orifice Equation
For large Reynolds numbers- flow is directly proportional to the square root of the pressure difference across the path: where

i/ =

K+A

&

( G. I I )

k = volumetric flow rate through the path, cfm (m3/s);
A
=

Ap =

flow area (also called leakage area), ft2 (m2); pressure difference across path, in. H 2 0 (Pa); 26 10 (0.839).

hj

=

where

i/
ti~

= = = = =

volumetric flow rate through the path, cfni (ni3/s); mass flow rate through the path, Ibis (kgls): dimensionless flow coetlicient; flow area (or leakage area). fi- (m-): pressure dillerence across path, in. H 2 0 (Pn):
1

C
A
4 3

?

Equation (6. I l ) gives flow at standard temperature 70°F (21°C) and standard atnlospheric pressure of .7 psi (l 0 l kPa). Frequently, volunietric flows are adjusted to standard \,ol~~metric llow rates. The mass flow rate is divided by the standard density to obtain the standard volunietric tlow rate. This is convenient because it allon.s engineers to think in terms of the familiar volumetric flow rates. Further. these standard flows can be treatsd as mass flow rates because they only deviate 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 manual. 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 this section are included for the unusual cases of very tight construction. 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)

' , 1
:\\\\\\\\\\\\\\\\\\' \\\\L\\\\\\\\\S\\\\

VHQ

r

Figure 6.6 Parabolic velociy profile for- Poiseuille flow between two parallel plates.

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). 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).

Fully Developed Lammar ~ i o w

Figure 6.7 Developnzenf oflaminar-flow in a gap.
there are inlet and outlet losses due to flows just outside the gap. These deviations from plane Poiseuille flow can be significant and are accounted for in methods of analysis presented later.

Plane Poiseuille Flow
For low Reynolds numbers, flow is directly proportional to the pressure loss. Viscous forces dominate flow with Reynolds numbers below about 100 to 1000, depending on particular path geometry. Plane Poiseuille flow is an exact solution to the Navier-Stokes equations for the flow of a viscous fluid between nvo parallel and infinitely long plates. The velocity distribution between the plates is parabolic, as illustrated in Figure 6.6. The 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 is proportional to pressure loss (dp/d~).

Exponential Flow Equation
In order to accommodate the flows, which are between viscous dominated and kinetic dominated, the following exponential relation has been used extensively in analysis of airflows through buildings:

where
V

C,
where a = distance benveen plates (gap thickness); p = dynamic viscosity; p = pressure Real gaps in buildings are not infinitely long, and some distance is needed for the parabolic flow profile to become established, as illustrated in Figure 6.7. The pressure losses (dphh) over this inlet length are greater than those of f ~ l l y developed parabolic flow. Further,

-

=

volumetric flow, cfni (rn3/s); flow coefficient for exponential flow equation, ft3 min-' (in. H20)-"(m3 S-' Pa-");

Ap = pressure difference across the path, in. H,O (Pa);

flow exponent, dimensionless. For a flow exponent of n = 0.5, Equation (6.13) is essentially the same as the orifice equation. For 11 = I , Equation (6.13) describes viscous dominated tlow. As \vould be expected from the above discussion, the flow exponent n varies from 0.5 to l . Equation (6.13) only approximates the relation bettveen flow and pressure difl'erence, and the values of
11

=

Principles of Smoke Management

C, and n depend on the range of Ap. This equation has
proven useful for the evaluation of flows through many small cracks in buildings at low levels of pressure difference. However, this equation is not directly related to the geometry of the flow path, and the values of C, for particular flow paths must be determined empirically. F_or analysis of buiiding airfiow, h e exponents of interior paths are often taken at 0.5, and exponents of exterior walls often are considered to be about 0.6 or 0.65.

4
Dh
p v

= = = =

pressure difference across gap, in. H20 (Pa); hydraulic diameter, in. (m), Dh = 2a, density of gas in gap, lb/@ (kg/m3);

kinematic viscosity, f?/s (m2/s); KNp = 0.108(1 .OO)

Gap Method
Gross and Haberman (1988) developed a generalized approach, the gap method, for determining the leakage through gaps of different geometry such as those of door assemblies. They developed a functional relationship between the dinlensionless variables NQ and NP.

Gross and Haberman used an analytical method of Miller and Han (197 1) to account for the pressure losses in the entrance region before fully developed flow is achieved in a straight-through slot. Their relation for flow versus pressure difference is shown in Figure 6.8. Three regions of flow through the straight-through slot were identified, and equations for these regions are: Region 1 (%scous dominated region- NPs250): for

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

where NQ = dimensionless flow rate; NP = dimensionless pressure diference; R, = Reynolds number, dimensionless(Equation (6.7));
a
s
= =

thickness ofgap in direction perpendicular to flow, in. (m); depth of gap in flow direction, in. (m);
--

The equations for regions I and 3 were developed by Gross and Haberman, and the exponents are as expected; considering that region 1 is dominated by viscous forces and region 3 is dominated by kinetic forces. Region 2 is a transition between the other two regions,
Region 2

Region l

+-

-.

Region 3

10

1o2

1o3

104

105

106

10'

NondimensionalPressure Difference, NP

Figurc 6.8 Flow and pressure ~rlo/io~ishi,u S//-aigI71-/lit-o~lgli for gaps (aclc~p/ecl~fiut~t atid Habernimli [ l 9881). Gross

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 Pressure Difference, NP Less than or equal to 4,000
7,000 10,000 15,000 20,000 40,000 100,000 200,000 400,000 1,000,OOCi 2,000,000

Flow Factor Flow Factor for Single-Bend for Double-Bend Slot, F, Slot, F2
1.OOO 0.98 1 0.972 0.960 0.952 0.935 0.9 10 0.890 0.872 0.848 0.827
1 .OOO

0.939 0.908 0.880 0.862 0.826 0.793 0.772 0.742 0.720 0.700

10'

10'

106

10'

NondimensionalPressureDifference. NP

Figure 6.9 Flow factors for gaps (adapted from Gross and Habeman [1988]).
Gap T h i i n e s s , a (mm)

0

1

2

3

4

5
straqht Gap 1 Bend

7

0

and an approximation developed by Forney (1989) is used for this region. This approximation is particularly attractive for computer applications because it is continuous with the expressions for the other two regions. Equations (6.7) and (6.14) can be combined to obtain a relation for volumetric flow rate through a straight-through slot. (6. 9) where voltrmetric flow rate, cfin (m3/s); dimensionless tlow; depth of gap in flow direction, in. (111): hydraulic diameter, in. (m),(D,, = 20); length of gap, fi (m); kinematic viscosity, ft2/sec (~i?/s); 60 (1 .OO). Frequently, slots around doors have one or more bends. For single- a:ld double-bend slots, the dimensionless flow, NP, can be obtaincd by multiplying values ; for a straiphr-thr.ough slot by flow [actors F , and 1 2 (where F , is for single-bend slots, and F2 is for a double-bend slors). These flow factors are presented in Table 6. I and Figure 6.9. Figure 6.10 shows the tlow predicted by the gap mctliod fol- a stmight gap and gaps with bends. As would bc cspcctcd, rhc Ilow incrcascs \\.it11 gap tliick-

.g 0.6 E

- 0.8
C

-

2 Bends
Note: Gap depth in flow direclion is 2 in.(50.8 mm). pressure difference i 0.04 in. H>O(10Pa). and air s temperatureis 7OSF(21 .C).

6 0.4 6 - U-

0.2

"0

0.05

0.10

0.15

0.20

0.25

0.30

Gap Thickness, a (in)

Figure 6.1 0 Flow coefficients for stl-aigl~tgaps atld gaps with be17o's. ness, n, and the f l o is less for gaps with bends than for ~ straight gaps. I r is espected rhat the gap model predictions for a relatively \ride gap (and relatively large Reynolds number) would be closer to those of the orifice equation than predictions of the exponential tlon equation. Figure 6.1 1 compares predictions of the orifice equation, the 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 predictions of the orifice equation are almost identical with those of the gap method. As might be expected for a 0.1 in. (2.54 mm) gap, the predictions of the exponential flow equation with 11 = 0.65 are much closer to those of the gap method (Figure 6.12). The design book of Klote and Fothergill (1983) used Equations (6.9) and (6.1 1) for all smoke control analysis because it was bcliwcd that the orifice equation was sufficiently.accurate for design analysis. Klote and Bodart (1985) reevaluated this use of the orifice equation and the exponential flow equation. They experimentally determined flow coefficients and exponents for the leakage paths of' rhc French Firz Research Tower
~

Principles of Smoke Management

using regression analysis. Computer flow simulations using the exponential flow equation with experimentally determined exponents were in good agreement with simulations using the orifice equation. It can be concluded that use of the orifice equation for all flow paths in normally constructed buildings yields acceptable for pressurization smoke control design purposes, No similar study was conducted for smoke management systems without pressurization.
Example 6.7 Gap Method for

I

For most of the applications of this book, flows are represented and calculated by the orifice equation. Two approaches to prescribing values for C and A are:

1-

Use the cross-sectional area for A, and C is chosen to obtain the desired value of the CA product.

2- Arbitrarily choose C, and choose A to obtain the desired value of the CA product.

FIG Through Door caps

A door has the dimensions shown in Figure 6.13. What is the flow through the gaps between the door and the door frame at a pressure difference of 0.15 in. H 2 0 (37.3 Pa)? Use the following properties of air at 70°F (21 "C):

The first approach is used with orifice flow meters and many other flow paths for which the cross-sectional area can readily be determined and for which C values are available. For flow coefficients of many items. readers are referred to Idelcnik (1 986)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 areas listed in Table 6.2 for C = 0.65. It is believed that actual leakage values for walls and floors are primarily dependent on workmanship rather than construction

For the slot at the door bottom:

a = 0.50 h'(0.0 127 m) Dh = 2a = l . O in. (0.0254 m) O L=3ft(O.914m) x = 1.75 in. (0.0445 m) Ap = 0.15 in. H20 (37.3 Pa)
106. From Equation (6.1 S), NP = 28.2~ From Equation (6.18);NQ = 2950. From Equation (6.19), V through slot at door bottom. For slots at to^ and sides:
=

-

Pressure Difference (Pa)

152 cfni (0.0718 m3/s) flow

.Ei 6 0 0

a

- 0.6 c
E
Note: Gap thickness is 0.5 in (12.7 mm). gap depth in flow direction is 2 in. (50.8 mm).pressure difference is 0.04 in. H,O (10 Pa). and air lemwrature is 70 'F(21 ' l . C..
l I

a = 0.12 in. (0.00305 m) D,, 2a = 0.24 in. = (0.006 10 m) L= 17 ft(5.18 m) s = 2 3 7 in. (0.0602 m) Ap = 0.15 in. H 2 0 (37.3 Pa) From Equation (6.1S), NP = 51000. From Equation (6.17), NQ = 109.8.
From Equation (6.19), V
had been straight.
=

0.3

2

3

1

0l

0

01 .

0.2

oS !

Pressure Difference (in H,O)

181 cfin (0.0855 m3/s) if the slot

Figure 6.11 Co~npnrison val-ioz/sflo\c~jln~ctio,u of for a 0.5 ill. (12.7 mm) wide goy.
Pressure Difference (Pa) Exponential
m

From Figure 6.9, F , = 0.93 for a single-bend slot.
V = 181 ,(0.93) = 168 cfin (0.0792 m3/s) flow through slots at top and sides. Total flcw: 152 + 168 = 320 c h (0.15 l ni3/s)

with n = 0.65

..
0.015 0.010

F L O W AREAS AND C O E F F I C I E N T S In the design of smoke control systems, airflow paths must be identified and evaluated. Some leakage paths are obvious, such as gaps around closed doors, open doors, elevator doors, windo\vs, and air transfer grilles. Construction cracks in building walls and floors arc less obvious but no less important.
Y

Gap Method of Gmss and Haberman,(1988)

0

-0.1

0.2

0 0.3

Pressure Difference (in

H,O)

Figure 6.12 Co~qmrison var-iousJ o I ~ . ~ ~ / I I c ~ ~ o I ~ s of fol n 0.I ill. (2.54 nrnl) wide gnp.

Chapter 6 -Principles of Smoke Management

0.62 in

Door
(b)

F;Y~725;7k T
(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: Typical Leakage Areas of Walls a n d Floors of Commercial ~ u i l d i n ~ for C = 0.65 s' Area ~atio~

materials, and, in some cases, the flow areas in particular buildings may vary from the values listed. The second approach above also was used for the flow areas of elevator doors listed in Table 6.3. The gap method can be used to determine values of C and A for flow through gaps around doors. Tables 6.4 and 6.5 provide this flow information using approaches 1 and 2, respectively. The flows ca!culated by these tables are equivalent to each other, and users can select the approach convenient to their application. Additional data concerning building components are also provided in Chapter 25, "Ventilation and Infiltration, of the 1997 ASHRAE Handbook-Fundamentals." The leakage flow rates of door assemblies can be measured and rated at ambient temperature and elevated temperatures in accordance with UL 1784 (1990). For open stairwell doorways, Cresci (1973) found that stationary vortices form in the doorways a;~dthat the resulting flow through those doorways was about half of that which would be expected without such vortices. Using approach I , Table 6.6 lists flow areas of open stairwell doorways for C = 0.35. Alternatively, approach 2 can be used where C = 0.65 and the flow area is about half the cross-sectional area. The determination of the flow area of a vent is not always straightforward because the vent surface is usually covered by a louver and screen. Thus, the flow area is less than the vent area (vent height times width). Because the slats in louvers are frequently slanted, calculation of the flow area is further complicated.

'

Construction Element Exterior Building Walls (includes construction cracks, cracks around windows and doors)

Stairwell Walls lTight 0.14 X (includes construction cracks but not cracks around windows or doors ~~~~~g~ 0. I I Loose 0.35 X Elevator Shaft Walls Tight 0.18 X (includes construction cracks but Average 0.84 X not cracks around doors) loose 3.18 X Floors (includes construction cracks and gaps around penetrations) Tight3 Average

lo4

1 0-3
Io-~
IO-~

Io-~
1o

-~

AIAj

0.66 X 1 o - ~ 0.52 X 104

I

loose3

1 0. I 7

X

Io

-~

l . Flow area ratios for C = 0.65 at 0.3 in. H20(75 Pa). 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 by Tamura and aiilson(1966). Tamura and Shaw (1976a: 1976b; 1978) and Sham et al. (1993). 3. Values exlrapolated from average floor lightness based on rangs o f tiglitness o f other constructionelements.

Principles of Smoke Management

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

m

Tightness Tight Average Loose

ft2

m2

3.0

0.914

0.34 0.48 0.60

0.032 0.045 0.056

Tight Average Loose

Tight Average Loose

.

0.37 0.53 0.66

0.035 0.049 0.06 1

Tight Average Loose

Tight Average Loose

Opened Doors I. 2.

3.5

1.07

Avcrage

6.0

0.56

This table is for clc\-ator doors 7 fi (2.13 m) Ihiph. Flow areas t'or C = 0.65 at 0.1 in. H1O( 2 5 Pa). 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 '
Width in. Gap Thickness at Top and Sides Cap Thickness at Bottom in. nlm Cross-Sectional Area ft' m' Flow Coeflicient

m

in.

nlni

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 '
Width in. 36
m

Cap Thickness at Top and Sides

Cap Thickness at ~ o t t o m

Flow ~ r e a '
fi? 0.079
m2

m.
0.02

mm

in.
0.25

mm

0.914

0.508

6.35

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 C = 0.65. The tlow will1

the gap method.

Table 6.4: Areas and Flow Coefficients for Open Stairwell ~ o o r s '
Door Width Condition of Door Propped Fully Open in. 36 36 44 44 m 0.9 14 0.914 1.118 1.118 Flow Area
ft2
m'

Person in Doonvay2 Propped Fully Open Person in Doonvay'
1.

21.0 10.5
25.7

1.95 0.78
2.3s

Flow Coefficient C 0.35 0.35 0.35 0.35

12.5

1.19

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 ~ wall is 0 . 3 5 10' (520) = 0.182 ft2 (0.0169m*). . ~ 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
The pressure losses due to friction in ducts and shafts is represented by

where

A

= =

area of the duct or shaft, f (m 2 ); ; perimeter of duct or shaft, fi (m).

P

where

&J =
f
L

pressure loss in shaft or duct due to friction, in. H20 (Pa);

Equation (6.20) is ths Darcy-Weisbach equation for pressure loss in ducts and pipes. For ducts and pipss, the friction factor can be obtained from the traditional Moody diagram (Figure 6.14), or it can be calculated from the Colebrook equation.

= dimensionless friction factor of shaft or duct;
= =
=

shaft or duct length, ft (m); hydraulic diameter of shaft or duct, ft (m); density of gas inside shaft or duct. 1b/ft3 (kdm3); average velocity inside shaft or duct, fpm (rids); where
E
= =

D,,
p U

roughness of the inside surface of the duct. Reynolds number (sse Equation (6.7)).

fi (m);

=

R,

Kyl, =

1 .G6 X 10-"(1.00).

The hydraulic diameter of shaft or duct is

Some categories of duct roughness, E, are listed in 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
Duct Material Uncoated carbon steel, clean PVC plastic pipe Aluminurn Galvanized steel, longitudinal seams, 50 in. (1200 mm) joints 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 Fibrous glass duct, rigid Fibrous glass duct liner, air side n-it11 facing material Fibrous glass duct liner, air side spray coated Flexible duct. metallic Flexible duct. all types of Ljbric and wire -. Medium Smooth 0.0003 0.09 Roughness Category Smooth Roughness, e ft 0.000 1
mm

0.03

Average Medium Rough Rough

0.0005 0.003 0.0 1

0.15 0.90 3.0

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 Institute of Standards and Teclinolopy. the clliciency oTthis numerical solution is signilicnrltly i ~ n p r o ~ d by suhstitutinp.~ /-I1' and solving Ibr .v. = In network computer flow tnodels (Chapter 8), it can be useful to use the equivalent orificc area for a duct 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 orifice.is

P
=

The flow also can be expressed as

P = A,U
where

where volumetric flow rate through the duct or shaft, cfm (m /s);
3

As = cross sectional area of the duct or shaft, ft2(m2);

U

=

average velocity in the duct or shaft, @m, (mls).

C
A,

=
=

dimensionless flow coefficient; equivalent area,

Considering Ap = Apj and combining Equations (6.20), (6.24), and (6.25) results in

ft2 (m2);

Ap = pressure difference across path, in. H20 (Pa);

p K,

= =

density gas in duct or shaff lb/ft) (kg/m3); 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) by 12 ft (3.66 ni) with a length equal to the floor height of 12

I(

ft 11

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
A, = 12.5AS = 12.5(8.6 x 12) = 1290 ft (120m ) .
"0

2

2

20

40

60

80

100

Hydraulic Diameter, Dh(R)

This large equivalent area is indicative of a duct section with a small pressure lossdue to friction.

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) Figure 6.18 Area ratio for medium rough ducts.

Values of K are listed in Table 6.8. ,

Hydraulic Diameter. D, (m)

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 l. FromTable6.8, the friction factor, K,, is32,andA,.JA, The equivalent area is A,
= 0.28 (8 X

= 0.28.

18) = 40 ? (3.72 m'). f

2. An alternate approach is below.

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

From Equation (6.28),

Hydraulic Diameter, D, (R) Figure 6.19 Area ratio for rough ducts.

SYMMETRY Stairwells
Tamura and Shaw (1976b) showed that the pressure losses due to friction in stainvells is similar to that of shafts, and this pressure loss is The concept of symmetry can be used to simplify flon networks, thereby simplifying analysis. While advances in network modeling (Chapter 8) have reduced the need for such simplifications, symmetry can srill be useful. Figure 6.20 illustrates the floor plan of a multistory 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 pressures are solved for one side, those on the other side are also known. To apply symmetry to a building, ei.ery floor must be such that it can be divided in the same manner by the plane of symmetry. If wind effecrs are included in the analysis, the wind direction must be parallel to the plane of symmstry. It is not necessac that the building bc geometrically synlnletric, as shown In

where pressure loss in stairwell due to friction, in. H20 (Pal; K,,, = dimensionless friction factor of stairwell;
AQ=

L

= height of section of stairwell, ft (m);

D,, = hydraulic diameter of stairwell, ft (m);

principles of Smoke Management

'

Table 6.8: Typical Friction Factors and Area Ratios for ~taiwells'
Stairwell Type ft Floor Height Friction Tread people
2

AdAs
Per Floor
-

m

Factors, K,

Conventional Conventional Conventional Conventional Conventional Conventional Scissor

12 12 8.5 8.5 8.5 8.5 14

3.6 3.6 2.6 2.6 2.6 2.6 4.3

Open Closed Open Open Closed Closed Closed .

None None None High None High None

29 32 61 104 71 170 15

0.30 0.28 0.24 0.19 0.22 0.15 0.32

I. Based on data from Tamura and Shaw (1976a) and Achakj~ Tamura (1988). and 2. "High" is high density of 0.18 person/ft2(2.0 persodm2).

I

F
Knob,

1

LWPressure Side

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

Figurc 6.20 Building floor- plan illzafr~arirz,o sjx1n7efr?. concepf.
Figure 6.20; it must be synimetric only with respect to flow.

F A(.
Fff

= total door openins force, Ib (N);
=

moment of the door closer and other friction, Ib fi

=

PJ m); door width, ft (m);
pressure difference across the door, in. H 2 0 (Pa); door. fi (m); 5.20 (1.OO).

A

= door area, f? (ni2);
=

DOOR-OPENING FORCES
The door-opening forces due to the pressure differences produced by a smoke control system must be considered in any design. Unreasonably high door-opening forces can'result in occupants having difficulty or being unable to open doors to refuge areas or escape routes. This is addressed in the next section. The following analysis is for a door hinged at the edge with a door knob, as shown in Figure 6.2 1. Users need to adapt the analysis to fit other conditions, such as pi\ ots inset froni tlie edge. The forces on a door in a smokc control system are illustrated in Figure 6.21, and tlie sum of the nionients about the hinge is

n
Kd

= distance froni the doorknob to the knob side of the
=

The moment to overcome the door closer and friction consists of all moments about the hinge due to the door closer or friction forces such as friction in the hinges or rubbing of the door against the door frame. The force at the knob needed to overcome hinge friction is about 0.5 to 2 Ib (2.3 to 9 N). Some poorly fitted doors rub against the frames, resulting in extremely high dooropening forces. Ideally, such poor workmanship will be identified and corrected during building commissioning. The component fo,rce, F,, at the knob to overcome the 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).
and this can be solved for the pressure difference as
2 ( W - d ) (F- F,) KdWA

@ =

where

F F,

= =

total door opening force, Ib (N); force to overcome the door closer and other friction, Ib (N);

W = door width, ft (m);
A = door area,

(m2);

Ap = pressure difference across the door, in. H20 (Pa);

This relation assumes that the door-opening force is applied at the knob. This force to overcome the door closer is usually greater than 3 Ib (13 N) and, in some cases, can be as large as 20 Ib (90 N). Caution should be exercised in evaluating the door closer force because the force produced by the closer when the door is closing is often different from the force required to overcome the closer when opening the door. Many door closers require less force in the initial portions of the opening cycle than that required to bring the door to the full open position. For this discussion, the force to overcome the door closer and other friction is that force at the very beginning of the opening process. The pressure difference component of the door-opening force can be determined from Figure 6.22 for a door 7 ft (2.13 m) high 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-opening Force. 1. What is the door-opening force for a door 7 ft by 3 ft (2.13 m by 0.9 1 m) subject to a pressure difference of 0.25 in. H 2 0 (62 Pa)? The force to overcome the door closer and other friction is 10 Ib (44 N), and the knob is 3 in from the door edge.
W= 3 ft(2.13 m)
Ap = 0.25 in. H 2 0 (62

d

= =

Pa)

F ,

0.25 ft (0.076 m) lOlb(44N)

A = 3 X 7 = 21 Kd = 5.2 (1.00) From Equation (6.3 l), the door-opening force is 25 Ib ( l l 0 N). Alternately, Figure 6.22 gives 15 Ib (66 N), and adding this to the door closer force gives 25 Ib ( l l0 N).

ft2 (1.95 m2)

2. What is the pressure difference across a door that has a 30 Ib (133 N) door-opening force and a frictional and door closer force of 5 Ib (22 N)? The door is the same size as in part 1 above.
+

F=3O lb(l33 N) F,= 5 Ib (22 N) From Equation (6.32), Ap is 0.42 in. HzO (104 Pa).

DESIGN PRESSURE DIFFERENCES
It is appropriate to consider both a maximum and a minimum allowable pressure difference across a barrier o f a smoke control system. The values discussed in this section are based on the recommendations in NFPA 92A (NFPA 2000a). The maximum allowable pressure difference should be a value that does not result in excessive door-opening forces. The force that a particular person can exert to open a door depends on that person's strength, the location of the knob, the coefficient of friction between floor and shoe, and whether the door requires a push or a pull. Read and Shipp (1979) studied door-opening forces, and they present strength data for the very young (age S to 6 years) and the elderly (age 6 0 to 75 years). From Tables 6.9 and 6.10, the five perce~tilepushing force for the very young females is only 6.5 Ib (29 N), and the five percentile pushing force for the elderly

females is only 2 0 Ib (91 N). The five percentile push force of healthy male adults is 4 5 Ib (200 N). These forces are gradually applied, and a 'jerk" method o f suddenly applying the force results in a peak force o f 175 Ib (780 N). These push forces are one handed, and the subjects are not leaning forward; the push force increases to 146 Ib (652 N) for a forward leaning twohanded push. The Life Safe@ Code (NFPA 2000c) states that the force required to open a n y door in a means o f egress shall not exceed 30 Ib (133 N). Based o n the data of Read and Shipp, it seems that this 30-lb (133 N ) limiting force is appropriate for most occupancies, but care should be exercised when building occupants are likely to have low levels of pushing and pulling strength. For a 30-lb (133 N) limitation on door-opening force with a side-hinged door with a singe knob, the maximum allowable pressure differences are listed in Table 6.1 I . The fire effect of buoyancy of "hot" smoke can be incorporated in the selection of the minimum design pressure difference. Unless otherwise stated, the minimum design pressure differences used in this manual incorporate buoyancy and are based on the idealization that the mass flo\v through the leakage paths is constant for the duration of the fire. A method for handling variable mass flon. through these paths is presented in Chapter 9. The smoke control system should be designed to maintain this minimum value under likely conditions of stack effect and wind and when there is no building fire (such as during acceptance o r routine testing). NFPA 92A (NFPA 20COa) suggests minimum design pressure differences, and these values are listed in Table 6.12. The values for nonsprinklered spaces are those that will not be overcome by the buoyancy forces of hot gases. These values for sprinklered buildings were calculated from the equation for buoyancy of combustion gases (Chapter 5) for a gas temperature of 1700°F (927OC), for a neutral plane located at a height of two-thirds of the ceiling height below the ceiling and with a safety 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 ' Mean, Ib (N) 20 (90) I6 (73) 27 (120) l9 (86) Maximum, Ib (N) 26 (155) 2s (l 26) 41 (184) 32 (141)
I l n ~ i d .Suddsnly

minimum,

Function Push Pull
I.

.

Gender
M F M F

Ib (N) 7.2 (32) 10 (46) 18 (82) l l (48)

Fifth Percentile, lb (NI 8. l (36)6.5 (29) 17 (77) 8.7 (39)

Note: :\dapkd ( i o l i ~llcad and Shipp (1979). Sul$cctr i w d only one puilics would h a w rcsultcd ill frcnwr Ibrccs.

applied ''jerk" ~ u s h e and pulls o r two-handed forward lesning s

Chapter 6-Principles of Smoke Management

Table 6.10: Functional Strength Values for Age Group 60 to 75 years1
Mean, Function Push Pull Gender M
111 (NI

Maximum,

(N)
121 (540)

53 (237)

Minimum, Ib 0 2 1 (92)

Fifth Percentile, lb 0
23 (101)

F
I.

45 (201)

9 1 (407)

22 (100)

21 (95)

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) 10 (14.5) I2 (53.4) 14 (62.3)
I.

0.41 (102.) 0.37 (92. l) 0.34 (84.5) 0.30 (74.6)

0.37 (92.1) 0.31 (81.5) 0.30 (74.6) 0.27 (67.2)

0.34 (84.5) 0.30 (74.6) 0.27 (67.2) 0.24 (59.7)

0.3 1 (77. l) 0.28 (69.7) 0.25 (62.2) 0.22 (45.7)

0.28 (69.7) 0.26 (64.7) 0.23 (57.2) 0.21 (52.2)

i\:olc: Adnpted from NFPA (2000al. Total door opening force is 30 lb (133

N . and the door l~eiehr 7 fi (2.13 m). ) is

Table 6.12: Suggested Minimum Pressure Design ~ifference'
Building ~~~e~ AS NS NS Ceiling Height,
ft (m)
Any

Design Pressure ~ifference? in. HzO (Pa) 0.05 (12.4) 0.10 (24.9) 0.1; (34.8)

9 (2.7) l 5 (4.6)

I. Adnpted from NFPA (2000n). For d a i g n purposes, a sn~oks control systeni should maintain these minimum 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 systems tend to fluctuate due to the wind, fan pulsations, doors opening, doors closing, and other factors. Shortterm deviations from the suggested minimum design pressure difference may not have a serious effect on the protection provided by a smoke control system. There is no clear cut allowable value of this deviation. It depends on tightness of doors, tightness of construction, toxicity of smoke, airflow rates, and on the volumes of spaces. Intermittent deviations up to 50% of the suggested minimum design pressure difference are considered tolerable in most cases.

WEATHER DATA
The indoor to outdoor temperature difference has an impact on building airflows and pressures. For some analyses, wind data may be needed. The 1997 ASHRAE Handbook-Fundamentals, Chapter 26, "Climatic Design Information," provides weather data for locations throughout the world. NFPA 92A and NFPA 92B suggest that the 99.6% heating dry-bulb (DB) temperature and the 0.4% cooling DB temperature be used as the winter and summer design conditions. NFPA 92A and NFPA 92B also suggest that the 1% extreme wind velocity be used as the design condition.

CHAPTER 7

Air Moving Equipment and Systems
he National Board of Fire Underwriters examined the NFPA fire data from January 1936 to April 1938 to determine the extent of the smoke hazards due to heating, ventilating, and air-conditioning (HVAC) systenls (NBFU 1939). Of 25 fires recorded, 19 had conlbustion of parts of the air-moving system. Ducts, duct linings, and filters bunied. In five cases of no fire in the HVAC system, smoke was distributed by the system. This report has had a nlajor impact ,on tlie materials and consiruction of modern HVAC systems, as is apparent from examination of current codes and standards. The report recommended that HVAC systems be shut d o h during fire situations to prevent them from spreading smoke and supplying combustion air to the fire. System shutdown became the standard response to fire. However, operation of the HVAC system In a s G k e controi mode has become a common alternative in recent years, as discussed in later chapters. The information in this chapter is provided as a broad and general background on air-moving systems. The material was selected to aid in tlie understanding of the smoke control systems discussed in later chapters. This information should help tire protection engineers, firefighters, and code oRicials to commur.icate with HVAC designers and to recognize and understand HVAC equipment. Because energy conservation is a major concern, energy efficiency of systems and equipment is addressed in this chapter. This chapter is not an exhaustive treatment of the firc safety requirements of HVAC systems, and the design or such systems should be done by experienced professionals. Many publications provide more detailed information about these systems and equipment (for example, ASHRAE 2000a;

T

SMACNA 1990, 1987; Handbook of HVAC Design 1990). The simplest systenls consist of a fan in a housing, such as a roof-mounted atrium exhaust fan. Most systems are more complicated, with ductwork and some of the following components: supply air outlets, return air inlets, fresh air intakes, humidifiers, filters, heating and cooling coils, preheat coils, and dampers. Ductwork is constmcted of a variety of materials, including steel, aluminum, concrete, and masonry. Duchvork of fiberglass, gypsum board, and fabrics is used with some restrictions. Discussions of fans and dampers are provided later. The air-moving systems that are discussed later are primarily intended for maintaining comfort conditions. Exhaust systems for toilets, laboratories, and kitchens are not discussed, but they are generally less complicated and use many of the same conlponents.

--

HVAC LAYOUT In large buildings, the heating and cooling loads often vary considerably from one location to another. Heat is transferred to or from the spaces near the exterior walls depending on outdoor weather conditions. Solar radiation affects each of the exterior zones differently. It is common to divide a building into four perimeter zones and a core zone as shown in Figure 7. l a. The heating and cooling capacities c the perimeter zones d agdesigned to accommodate outside temperatures and solar loads. Because of the heat produced by occupants, hghting, and equipment. the core zones often need cooling even in the winter. The perimeter zones can be conditioned by a variety of means, including fan coil units, air conditioners, and heat pumps. Generally, fan coil units are supplied

-

Chapter 7- Moving Equipment and Systems Air

with hot and cold water to allow both heating and cooling. Often, air conditioners and heat pumps are located through-the-wall. Both fan coil units and through-thewall equipment can receive ventilation air directly from the outside or from a ducted ventilation systeni. In large commerc~al buildings, ventilation air is needed to control the odors due to cooking, smoking, perspiration, and other processes.7 The perimeter zones may be served by 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 needs simultaneously and are used to serve both perimeter and core zones. The different types of forced air systems are discussed later. ~istributionon a floor is often through ducts located above a suspended ceiling. Return air is often 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), on one floor (Figure 7.lc), or on several floors (Figure 7. l d). The arrangements above are but a few of those possible. There may be several forced air systems on each floor. There may be several units located in a penthouse, each serving its own vertical portion of the building. Sometimes, several air systems are used and the areas served are selected on the basis of having similar heating and cooling demands. These demands depend on occupancy, the presence of heat-releasing equipment, electrical lighting levels, and heat transferred to or from the outside. For a complicated building (such as hospitals, laboratories, and hotels), the duct systems can be intertnrined to such a level that considerable study is needed to understand which systems serve which areas.

Floor

29

25
(a) Perimeter and Core Zones

20

15
(b) Ducied Supply and Plenum Return

I Mechanical Penlhouse 1
R e q - i
Duct

1

/

Dun
1

1

(C )

Central System in Penlhouse

(d) MuRipleMechanical Floors

Figure 7.1

Some HVAC a1-1-or7ge11zazts.

system economics but little effect on airflow. The forced air systems discussed in the following sections can be completely built in the field, factory-fabricated subsections can be field assembled, or completely factory fabricated systems can be installed.

Constant Volume, Single Zone
Figure 7.2 is a representation of a single-fan, constant volume system. The term "constant volume" is used in the HVAC industry to indicate that the system pcoduces a constant or nearly constant volumetric flow rate of air. This system is used in residences and some small commercial applications. In this esample. return air from the living quarters is drawn i n at one location, flows through filter, fan, and coils, and is distributed back to the residence. This system does not ha\.e the capability of providing fresh outside air. These systems are intended for applications where there is sufficient natural air leakage through cracks i n walls and around windows and doors for odor control. Single-zone systems are so called because they serve only one HVAC control zone, For esample: a residential system is controlled by a thermostat to maintain the temperature in the l i v i n ~ quarters. Generally, the residential system has a two-position control system, allowing only "on" and "olt" operation to maintain tempel-ature and humidity conditions. Frequently in commercial buildings. constant volume systems have two fans and are capable ol'pro\.iding

FORCED AIR SYSTEMS
Four common types of forced air systems are constant volume, single-zone systems, constant volume systems with terminal reheat, variable air volume (VAV) systems, and dual-duct systems. There are numerous variations on these systems. Generally; the heat source for heating coils is hot water. However, other sources, such as steam or electrical resistance heating, are possible. Cooling coils can be supplied with chilled water or with refrigerant. The source of heating or cooling has signiticant effects on
7. In small b u i l d i n g and residences, such odor col]trol is achie\.cd by ~iaturally occurring air inl?l[ration through construclion gaps and cracks.

Principles of Smoke Management

ventilation air as illustrated in Figure 7.3a. The return fan permits lower supply fan speeds and quieter operation. The return air fan provides positive return and exhaust from the conditioned space. During cold weather, many large commercial buildings have so much heat generated by equipment and people that cooling is required. To save energy, cold outside air can be used for this cooling. The system of dampers and controls that maximizes the use of outdoor air for cooling is called an economizer. For systems with an economizer, the humidifier and cooling coils need to be protected from freezing. Thus, 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 freezing. The preheat coil and reheat coil can be used when heating is required. The reheat coil used with the cooling coil allows precise humidity control. the supply fan and fan have the same flow rate, the system is said to be in a "balanced condition." Many designers size the exhaust fan at about 80% or- 90% o W flow of th fan provide 'light building pressurization (about 0-05 in. H70 [ l 2 PA1). -- The intent is to prevent normal infiltration of airborne dirt, odors, and pollen from the outside into the building. Figure 7.3b is a line diagram illustrating the same system as that of Figure 7.3a. In the rest of this chapter, line diagrams will be used to illustrate systems. The components of the following systems are the same as those shown in Figure 7.3a and 7.3b.

Constant Volume, Terminal Reheat
The constant volume, terminal reheat system is intended to serve many HVAC control zones, as illustrated in Figure 7.4. This system can have an economizer as can all the following systems. The supply fan provides cooled air to each zone, where it is reheated to the temperature required to maintain comfort conditions within that zone. The airflow rate through the system is constant, and control is achieved by varying the heat input to each reheat coil. This system is capable of achieving a high level of temperature and humidity control for each zone. However, terminal reheat is not very energy efficient.

Variable Air Volume
The variable air volume system varies the supply rate of conditioned air to the space to maintain comfort conditions. Additionally, the temperature of the supply a'' may be varied. There are many a ~ ~ r o a c h e s for achieving variable flow. In the system depicted in Figure 7.5, flow to each zone is controlled by a damper or other flow control device in the VAV unit. This unit is sometimes referred to as the VAV terminal box. Generally, the supply and return fans are capable of variable flow rates alld are controlled by the static pressure sensors. Some of the approaches that are used to achieve variable flow rates through fans are variable pitch inlet
Exhaust Damper Relum Fan
C

Louver Exhaust Air +-

\

/

C

(a) Diagram showing duct thickness
Exhaust

,/ , n
Exhaust AirC Return Air

Return Fan

/ ,Cooling
/

Humidifier

Coil

Outside

Outside Filter Preheat Air Damper . , Coil

\ Reheat
Coil

Bu113ing Spaces

(b) Diagram with line representation of duct

Figurc 7.2 Si17gIe-jbr7sysletr,.

Figure 7.3 Corisfanfvolume, single-zone sysfem.

Chapter 7- Moving Equipment and Systems Air

Exhaust Exhaust Air

1 Reheat Coil -

Retum Air Damper

Humidifier

I

-U

Outsid
Air

Supply Fan Outside Air Damper Filter Preheat
Coil

Figure 7.4 Constant volrtnze systern with terminal reheat.

Figure 7.5

Vasiable-air-volunze(VAC.') system
systems have been used in multi-room buildings to accommodate highly variable heating and cooling loads. A dual-duct system can be constant volume or VAV. Operating costs of VAV dual-duct systems are less than those of the constant volume systems.

vanes, discharge dampers, variable pitch motor sheaves, eddy current couplings, variable speed DC motors, and variable frequency AC motor speed controllers. As with constant voiunie systems, VAV systems can be designed to provide building pressurization.

Dual Duct
The dual-duct system co~lditions the air at a cenall tral location and distributes it to the conditioned spaces t h r o ~ ~ gtwo supply ducts. One duct conveys cold air, h and thc other warm air (Figure 7.6). A mixing box supplying each zone combines the two airstreams in the proper proportions to achieve comfort conditions. These

FANS
ANSIIASHRAE Standard 149 (ASHRAE 2000b) establishes methods o f laboratory testing and documentation for fans used for smoke exhaust. There are two general fan classifications-xentrifugal and axial. Figure 7.7 illustrates the basic parts of a centrifi~galfan. Flow within a centrifugal fan is prima-

Principles of Smoke Management

rily in a radial directiod to the impeller. Figure 7.8 illustrates the basic parts of an axial fan. Flow within an axial fan is parallel to the shaft.

Centrifugal Fans
Centrifugal fans used in the HVAC industry are generally classified by impeller design as forward curved, backward curved, and airfoil (Figure 7.9). Forward-curved centrifugal fans rotate at a relatively low speed and are generally used to produce high flow rates and low static pressures. Backward-curved fans rotate at about twice the speed of forward-curved fans and have a higher efficiency. The higher rotational speed requires more expensive fan construction. Both forward- and backward-curved impeller blades are single width, stamped from sheet metal. Airfoil fans are basically backward-curved fans with blades of varying thickness to improve fan efficiency. Airfoil blades are designed using the same airfoil technology that is used to design airplane wings. Required performance and economics are major factors in the selection of a fan type for a particular application. However, the following generalizations can be made concerning application. Fonvard-curved fans are used for low-pressure HVAC applications, including
c l ?

,

. ,

these fans have single-width impeller blades and straightening vanes to direct air parallel to the shaft. Tubular centrifugal fans are primarily used for low-pressure HVAC applications, particularly as return air fans. These fans have significant space savings over other centrifugal fans. - Backward impeller rotation is a common problem with systems with centrifugal fans. It is important to note that backward rotation of centrihgal fans results in reduced flow in the normal direction. This problem is often not recognized because of the mistaken belief that backward rotation of these fans results in backn.ard flow. The normal direction of airflow and the direction of rotation of centrifugal fans is shown on Figure 7.7.

Axial Fans
The common types of axial fans used in buildings are propeller fans, tubeaxial fans, and vaneaxial fans (Figure 7.11). For propeller fans, a variety of impeller designs are employed with the intent of achieving high flow rates at low pressures. The irnpellers of propeller fans have two or more blades and are usually of inexpensive construction (for example, these blades are often stamped from sheet metal). Propeller fans are used for low-pressure, high flow rate applications, including kitchen exhaust, toilet exhaust, stairwell pressurization, and space ventilation. Tubeaxial fans have a higher efficiency and can operate at higher pressures than propeller fans. Vaneasial fans have still higher et'ficiencies and operating pressures. Blades of tubeasial and vaneaxial fans can be

equipment. Airfoii and backward-curved fans are used for general purpose HVAC applications, and airfoil fans are usually limited to large systems where the enernv savings are significant. ~ubuGcentrifugal (Figure 7.10) are an excepfans tion to the classification by impeller type. Generally,

Exhaust Exhau

Return

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 24 to 64 Blades About 65% Efficiency

Backward-Curved 10 to 16 Blades About 75% Efficiency

Airfoil -10 to 16 Blades About 80% Efficiency

Figure 7.9 Itq~ellertypes for centrifugal fans.

-.

. .

Principles of Smoke Management

single thickness or airfoil design. Adjustable pitch blades' are used on some vaneaxial fans to obtain high efficiency. Both tubeaxial and vaneaxial fans have the advantages of straight-through flow and compact installation. Tubeaxial fans are used for low- to medium-pressure HVAC applications, and vaneaxial fans are used for low- to high-pressure HVAC applications. Unlike centrifugal fans, backward rotation of an axial fan normally results in backward flow. This backward flow is at a reduced airflow rate. More information about both centrifugal and axial fans is provided by Jorgensen (1 953), ASHRAE (2000~) and AMCA (1 990a, 1987).

Multi-blade dampers operated by electric motors or pneumatic pistons to vary the flow rate are called control dampers. Dampers used to resist the passage of fire are called fire dampers, and these can be multi-blade dampers (Figure 7.13) or curtain dampers (Figure 7.14). Dampers used to resist the passage of smoke are called smnke dampers, and these can also be either multi-blade or curtain. Combination dampers can be used to balance airflow, control airflow, resist the passage of fire, and resist the passage of smoke.

Fire Dampers
Generally, multi-blade fire dampers are held open by a fusible link and are spring loaded. In a fire situation, hot gases cause the link to come apart, allowing a spring to slam the blades shut. In place of fusible links, some manufacturers use other heat responsive devices. In the United States, fire dampers are usually constructed and labeled in accordance with standard UL 555 (UL 1999). Prasad (1995) tested the ability of fire dampers to close under conditions of still air, airflow, ambient temperature, and elevated temperature. In response to the findings of Prasad's findings, the 1999 version of UL 555 includes closure tests for static system (with no airflow) and dynamic systems (with airflow). The dynamic tests can be at ambient temperature, 250°F (120°C) or 350°F (180°C).

DAMPERS
In air-moving systems, dampers are used to balance airflow, control airflow, resist the passage of fire, or resist the passage of smoke. Balancing dampers are used in supply ducts and return ducts to adjust the airflow to the design values. These dampers can be of simple construction (Figure 7.12) or of multi-blade construction (Figure 7.13).

Centrifugal Impeller

J r .
Propeller Roof Exhaust Fan About 25% Efficiency

1

I

<

Propeller Wall Fan Emciency

-

Propeller Roof Fan About 25% Efficiency

Straightening Vanes

Tubeaxial Fan About 55% Efficiency

-

amVaneaxial Fan About 70°h Efficiency

Figure 7.10 fitbular cent/-ifugalfan.

Figure 7.11 Types of axial fans.

. Duct

Y

Arm'

c
Round Damper Rectangular Damper

Splitter Damper

Figurc 7.12 Dotuper types ~tscdjor Dalot~ci/~g.

Chapter 7-Aii Moving Equipment and Systems

Channel Frame

\
Shafl

Shafl Extension

7
Blade

/

Angle stop \
l

J
Section Note: Horizontal (floor) type curtain dampers must have spring closure, but vertical (wall) type curtain dampers can have either spring or gravity closure.

Opposed Action Damper

, Channel Frame

-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 -- -- be should selected based on the requirements of the __-._.. application. - For example, the dampers i n the supply and return ducts can have some leakage v-ithour adversely afi'scring smoke control system performance. Thus, a de&a-cr might select class I I or I l l smoke dampers for such an application. However, a designer might choose clzss 1 dampers for applications that require a very right damper, such as a return air damper (Figure 7.3).
~-

Smoke Dampers
In the United States, smoke dampers are usually constructed and classified for leakage in accordance with standard UL 555s (LL 1999a, 1999b). The standard includes construction requil-ements and tests for cycling, temperature degradation, dust loading esposure, salt-spray exposure. air leakage, and operation under airflow. These dampers are classified as I, 11, or

Y

2__

-

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)
m3s-'m-'

At 4.0 in. H 2 0 (1000 Pa)

Classification
I II II

cfmtft? 4 10

cim/it2
S

,,,Zs-l n,-3

I II II

0.020. 0.05 1 0.203 40 At 8 in. H20 (2000 Pa) II 0.056 2s 0.147 112 0.569

0.04 I 0.102 0.406 A t 1.2 in. 1{20 (3000 Pa) 14 0.07 1 20 S O
>3 .
7

-

130

0 .L 7 s 0.7 1 1

CHAPTER 8

Computer Modeling
moke management applications of computer modeling have increased dramatically in the last few decades. Many computer models have been developed for fire science and fire protection engineering applications by a number of organizations. Many of these applications are very useful for smoke management design. The Smoke 1\4anagen7ent Progt-am CD that accompanies this book contains a number of computer applications that can be useful for smoke managenlent (Table 8.1). Most of these programs were developed at the National Institute of Standards and Technology (NIST). The NIST computer applications are in the public domain, \\ hich means that they are not covered by copyright protection and they can be freely copied and used by anyone. The computer applications on the CD can be classified as building airflow models, zone fire models, detector actuation models, CFD models, elevator evacuation model, 2nd collections of engineering tools. This chapter is a g e ~ e r a l discussion of these classes of models except for CFD models and the elevator evacuation model. The CFD models are dealt with in Chapter 17. The elevator evacuation model, ELVAC, is discussed in Appendix C. The treatment in this chapter is of a general nature. For details and equations of particular models, readers should see the documentation for the model. The equations in this chapter are only intended to describe some of the more important concepts of computer modeling, and these equations are not ~ntendedto be used for calculations. Accordingly, units are not given for variables of this chapter. However, all of these equations are valid for S1 units or any other I~omogcneous unit system (see Appendix A)

S

BUILDING AIR AND SMOKE n o w MODELS
Computer programs that simulate building airflow can be useful for analysis of pressurization smoke control systems. Airflow programs that can simulate contaminates or smoke concentrations throughout a building can be useful tools for hazard analysis. The CONTAM program that is on the CD accompanying the book has air and contaminate flow capabilities. and it also is used for some of the examples of this book. A discussion of the earlier models provides a background for CONTAM. All of the airflow programs also calculate the pressures throughout the building. The National Research Council of Canada (NRCC) developed airflow programs (Sander 1974; Sander and Tamura 1973). The ASCOS program (Klote 1982) simulated airflow and was specifically developed as a research too1 for analysis of.smoke control systems. ASCOS was extensively used for smoke control design for much of the 1980s and 1990s. Yoshida et al. (1979); Butcher et al. (1969); Barrett and Locklin (1969): Evers and Waterhouse (1978); and Wakamatsu (1977) developed programs that also simulate smoke concentrat;ms.

Network Models
These models represent a building by a net\vork of spaces or nodes, each at a specific pressure and temperature. The stairwells and other shafts can be niodeled by a ne vertical series of spaces- for each floor. Air flows through leakage paths from regions of high pressure to regions of low pressure. These leakage paths are doors and windows that may be opened or closed. Leakage can also occur through partitions, floors, and exterior

Chapter 8- Computer Modeling

Table 8.1: Computer Software ~ ~ ~ l i c a t i o n s Smoke ~anagement in the progmms CD'
Software Classification Building Air and Smoke Flow Zone Fire Name CONTAM ASET-C CFAST LAVENT JET AZONE Detector Actuation DETACT-QS DETACT-T2 CFAST LAVENT JET Elevator Evacua- ELVAC tion Collection 01' Engi- ASMET neering Tools FAST I. Comments Airflow analysis including contaminants Available Safe Egress Time - C++ Language Version is part ofthe ASMET package of engineering tools. Consolidated Fire and Smoke Transpott Model Model for the Prediction of Detector Activation and Gas Temperature in the Presence ofa Smoke Layer Model for the Prediction of Detector Activation and Gas Temperature in the Presence 3f a Smoke Layer Atrium zone fire model includes plugholir,g anddelayed smoke exhaust fan activation (Cha~ter 141 Detector Actuation - Quasi Steady Detector Actuation - Time squared Detector actuation is one feature ofthis zone fire model Detector actuation is one feature of this zone fire model Detector actuation is one feature of this zone fire model Elevator Evacuation Atria Smoke Management Engineering Tools A collection of equations and fire protection engineering tools including CFAST

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 function of the pressure difference across the leakage path. In this model, air from outside the building can be introduced by a pressurization system into any level of a shaft or even into other building spaces. This allows simulation of stainvell pressurization, elevator shaft pressurization, stairwell vestibule pressurization, and pressurization of any other building space. In addition, any building space can be exhausted. This allows analysis of zoned smoke control systems where the fire zone is exhausted and other zones are pressurized. The pressures throughout the building and steady floii7 rates through all the flow paths are obtained by solving the airflow network, including the driving forces, such as wind, the pressurization system, and inside-to-outside teniperature difference. The assu~nptionsof the ASCOS model are similar to other network nlodels, and these assumptions are: 1. Each space is considered to be at one specific pressure and one specific temperature.

3. The net air supplied by the air-handling system or by the pressurization system is assumed to be constant and independent of building pressure.
4.

The outside air temperature is assumed to be constant. The barometric pressure at yound level is assumed to be standard atmospheric pressure (1 0 1325 Pa).

S.

The results of the program are not very sensitive to changes in atmospheric pressure. For altitudes considerably different from sea level, a more accurate value of barometric pressure can be substituted by changing a statement in the subroutine INPUT and one in the subroutine CORR. The following is a simple overview of a nehvork model. This overview only considers one flow path between any two nodes, but mar,y network models allow a number of flow paths between the same two points. The mass flow in a path between two nodes can be represented as

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 between nodes i and j, pressure difference fiom node i to node j . A number of functional relationships for flow are discussed in Chapter 6. Possibly the orifice equation and the exponential equation are the most ccmmon such functions. A function can also be used to represent the flow of a fan, which is an exception in that fan flow is from a node of lower pressure to a node of higher pressure. The pressure difference can be expressed as
Ap.. = p;-p.+p;g(Z;-Zj) 'J J

The solutior?to this set of equations is the pressures @,,P*,

... pN) for which all the right-hand side is zero. From these
pressures, all of the pressure differences and flows throughout the building can be calculated. Because of the difficulty in solving these equations, the numerical routines of many of the above models were slow and would sometimes fail to converge to a solution. Such convergence failures seemed to happen more often with large and complicated networks. An ASHRAE-funded research project (Wray and Yuill 1993) evaluated several algorithms to find the most appropriate one for analysis of smoke control systems. They selected the AIRNET routine developed by Walton (1989) as the best algorithm based on successful convergence, computational speed, and use of computer memory. None.of the routines of this study take advantage of the repetitive nature of building flow networks, so data entry for these routines is difficult and time consuming.

I

~,

(8.2)

where
pi = pressure at node i,

fi
pi

=
=

pressure at node j, density gas at node i, elevation of node i, elevation of node j,

Z,

=
=

5

CONTAM Model
There are two versions of this model: CONTAM96 (Walton 1997) for use with the DOS operating system and CONTAMW (Dols et al. 2000) for use with the Windows 95, 98, or NT operating systems. The technical aspects of these models are the same, and they are referred to in this section simply as CONTAM. A simple user guide for getting started with CONTAM is provided in Appendix D. CONTAM uses an improved version of the AlRNET algorithm that was selected as the best algorithm in the study mentioned above. Further, CONTAM has a method of graphical data input that reduces both learning time and the likelihood of input errors. CONTAM was developed for indoor air quality applications, but it has been extensively used for smoke management applications. This model simulates contaminant flow, as well as airflow throughout a building. For smoke management applications, the contaminants can be the products of combustion. The CONTAM documentation considers the model 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.

g = acceleration of gravity.

For steady flow, conservation of mass at node i can be stated as the sum of the mass flows leaving node i are zero. In equation form, this is
,it

C/;J(A~rJ)= 0
/ = I

(8.3)

where M is the number of flow paths between node i and other spaces. The mass flows entering node i have negative values. Writing the conservation of mass equations for each node in the building results in f l ~ ( & ~+fl2(Aplr) + ... +fl.\i(A~lh.) 0 l) =
1

f2,(&2l) +f22(Ap22)+ ... +J;v(@2,v) = 0 ,
(8.4)

Substituting Equation (8.2) into the ab0L.e set of equations yields

where Fiis the functional relationship for flows into node i. Equation (8.5) is a set of sin~ultaneous nonlinear equations.

The pressure in room i is considered hydrostatic, 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,

pi, = pressure at the floor ( = 0) of zone i, z

N

Fire

I

C C

g = acceleration of gravity, pi = density of air in zone i,

\ \ \ \ \ \ \ \ \ \ \ L
(b) Zone model idealization of a room fire

-

Airflow

z

=

elevation above the floor of zone i .

The representation of pressure allows for simulation 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 flow path (Figure 8.1). This is not relevant for smoke control systems that rely on pressurization, but it could be significant for simulations of smoke transport that does not include pressurization. For flow paths specified at midheight of the floor, airflows, and pressures calculated by CONTAM are the same as, those of ASCOS within the limits of numerical convergence.

Figure 8.2 Rooni .fire(a) sketch m d (b) zone model ideafiinrion.

ZONE FIRE MODELS
Zone fire models have proven utility for many fire protection applications, including hazard analysis. The concepts behind this type of fire model are the basis of most of the engineering approaches to smoke management design for atria. Early zone fire models include the Harvard Code (Mitler and Emmons l98 l), ASET (Cooper 1985), the BR1 Model (Tanaka 1983), and CCFM (Cooper and Forney. 1990). The University of Maryland has made modifications to CCFM specifically for atrium smoke management design (Milke and Mower 1994). The models ASET-C, CFAST, LAVENT, and JET are discussed below. Because zone models were originally developed for room fires, this discussion will start with room fires. In a

room fire, hot gases rise above the fire, forming a smoke plume. As the plunle rises, it entrains air from the room so that the diameter and mass flow rate of the plume increase with elevation. Accordingly, the plume temperature decreases with elevation. The fire gases from the plume flow up to the ceiling and f o m ~ hot stratified a layer under the ceiling. The hot gases can flow through openings in walls to othcr spaces, and such flow is The referred to as a doo~jer. doorjet is similar to a plume in that air is entrained and the mass flow rate and crosssectional area of the jet increase with elevation, and the jet temperature decreases with elevation. Ths difference is that the doorjet is tlowing through an opening in a wall. Figure 8.21 is a sketch of a room fire. The concept of zone modeling is an idealization of the room fire conditions, as illustrated in Figure 8.2b. For this idealization, the temperature of rhs hot upper layer of the room is unifonn and the temperature of the lower layer of this room is also uniform. The height of the discontinuity bttween thcse layers is the same everywhere. This discontinuity i s called the smoke layer inre&ce. In the idealized modcl, at an infinitesimal distance above the intcrfice, the temperature and contaminant concentrations are thosc of the smoke laver. At an

Principles of Smoke Management

infinitesimal distance below the interface, the temperature and contaminant concentrations are those of the lower layer. However, in real fires, there is a gradual transition rather than an interface. The dynamic effects on pressure are considered negligible, so that the pressures are treated as hydrostatic. Other properties are considered uniform for each layer. Algebraic equations are used to calculate the mass flows due to plumes and doorjets. Many zone computer models allow exhaust from the upper layer, and this capability is essential for simulation of atrium smoke exhaust systems. Many of the computer zone models estimate heat transfer by methods ranging from a simple allowance as a fraction of the heat released by the fire to complicated sin~ulation, including the effects of conduction, convection, and radiation. Zone model application to an atrium fire is illustrated in Figures 8.3a and 8.3b. Rockett et al. (1987) compared measured data with data computed by the Harvard Code for a series of fires 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 approximated 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 models, readers are referred to Karlsson and Quintiere (2000), Friedman (1992), Jones (19S3), Mitler and Rockett (1986), and Mitler (1984) and Quintiere (l989a).

m,
m,,.,,

=mass in the upper layer, =mass flow rate into the upper layer, =mass flow rate out of the upper layer.

r ; i , out

The mass flow rates in Equation (8.7) depend on the specific computer model. ASET-B only simulates the plume flow into the upper layer with no allowance for mass flow out of the upper layer. For this model, ril,. in is the mass flow of the plume and ril,, is zero. For more complex multi-room zone models, nz,, is the sum of all mass flows into the upper layer (plume, doorjet from another room, HVAC flow, etc.) and ril,, is :he sum of all mass flows out of the upper layer (doorjet from another room, HVAC flow, etc.).

..,

The conservation of energy equation is also known as the first law of thermodynamics. Because potential energy and kinetic energy are relatively small, they are neglected, and the energy equation for the upper layer is

Mathematical Description
Many of the early zone fire models were quasisteady systems of algebraic equations, and the atrium zone fire model, AZONE, discussed in Chapter 14, is 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 equation-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. ASET-B is an exception in that equations are only written for the upper layer. The equation of conservation of mass for the upper layer is
I

(a) Sketch of a n atrium fire

Plume

(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

F
-

Corridor

i
A

Temperature Rise CC)

' 8 6
E

25
I

50
I

75
1 .

100
1 .

Burn Room

Note: 8 indicates thermocouple tree. Temperature Profiles: +Measured - - Calculated from Zone Fire Model

I

0

50 100 150 Temperature Rise CF) Temperature Rise ("C)

200

Temperatxe Rise)C"(

Temperature Rise ('F)

Temperature Rise CFj

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

ipition (rrdapted

Control Volume Boundaries
I I

..............................
Upper Layer

I

l
I I

I
I

I

L, hUJe u
TJ h,, e,
Lower Layer Opening in Room

I I

Figure 8.5 C'orlrinl ~ ~ ) h i i i m , / u i aiitl loi\vi- 1cg~er.s zq)pei. uj'ci

.viirplc inoirl zoilejiir iiroclcl.

Principles of Smoke Management

where

Q
W

=
=

heat transferred to the upper layer, For an ideal gas, Cp, C, R, and y are constants (Appendix A). The time derivative of Equation (8.10) is

h , i,, h,
ell

work done by the smoke layer on the surroundings, = enthalpy of the mass flow into the upper layer, enthalpy of the mass flow out of the upper layer, = internal energy of the upper layer.
=

,

Substituting this into Equation (8.9) results in

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 to a temperature difference. Because the upper layer is 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, (in differential form, work is dW = Fdl). I The displacement for the upper layer is the moving smoke interface, which is also the surface of the control volume. The force is the product of absolute.pressure at the interface and the area of this surface ( F = pAJ. The work term is

Combining Equations ( 8 4 , (8.1 l), (8.12), (8.13): and (8.16) yields

Equation (8.17) is a form of the conservatix of energy equation for the upper layer. The following conservation of mass and energy equations for the lower layer can be developed in a similar manner:

Most zone fire models consider the gases to behave as ideal gases, and an ideal gas is one that has the following equation of state:
PI.' = I I I R T

and

(S. 10)

where p = absolute pressure, V = volumeofgas, m = mass of gas, R = gas constant, T = absolute temperature of gas. The enthalpy of an ideal gas can be expressed as

The conservation equations can be rearranged as

where C,, is the constant pressure specific heat. The internal energy of an ideal gas can be expressed as

where C,.is the constant volume spzcific heat. The gas constant 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,

V = room volume (V = V + ,

5).

Equations (8.20) through (8.23) were developed by Jones (1983), and readers should see that reference for a detailed description of the net energy release rate terms. Information about solution of such systems of differential equations can be found in many texts on numerical methods (for example, Burden et al. 198 1). Equations for plume mass flow and temperature are discussed in Chapter 13. Flow through doors and other openings in walls or partitions are calculated in much the same way as horizontal flow through an opening, which is treated in Chapter 13, except that the pressures are complicated by the possibility of both air and smoke on both sides of the opening.

controlling vents and sprinklers in compartments bounded by walls, draft curtains, or combinations of walls and draft curtains. The JET model incorporates the conductivity factor to account for the effects of heat conduction from the sprinkler head.

DETECTOR ACTUATION MODELS
Fire-driven ceiling jets can have a significant impact on the performance of ceiling-mounted detection hardware. The plume rises above the fire. As it impinges on the ceiling, the plume gases turn and form a relatively high temperature, high velocity, turbulent ceiling jet, which flows radially outward (Figure 8.6). The temperature and velocity of the ceiling jet are described by Albert's (1972) correlations. The detector actuation model, DETACT-QS, calculates the actuation time of thermal devices below unconfined ceilings (Evans and Stroup.1986). The unconfined ceiling assumption is appropriate for large spaces, such as open plan ofice spaces, but it does not account for the effects of the smoke layer on the ceiling jet in a confined space. Figure 8.7 is a sketch of a ceiling jet in a room with a smoke layer. For a fire below the smoke layer, the plume penetrates thc smoke interface, continues to rise toward the ceiling, and entrains smoke from the smoke

ASET-C Model
ASET-C (Available Safe Egress Time-C Language) is a program for calculating the temperature and position of the hot smoke layer in a single room without smoke flow to other spaces. ASET-C is one of the simplest and easy to run zone fire models. As stated above, ASET-C only simulates the upper or smoke layer. The lower layer is considered to remain smoke free and at ambient temperature. ASET-C is an adaptation of the ASET-B (Walton 1985), and it is one of the engineering tools in the ASMET package. Documentation is provided in Appendix E.

.....

CFAST Model
CFAST is a multi-room zone model that predicts the effect of a specified fire on temperatures, various gas concentrations, and smoke layer heights in a multi-compartment 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 description of CFAST see Jones et al. (2000).
Detector

(a) Sketch of ceiling jet and detector

LAVENT Model
LAVENT (Davis and Cooper 1989) is a single room zone fire model that predicts plume centerline temperature, 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.

Ceiling

JET Model
Like LAVEKT, JET (Davis 1999) is a s~ngle room
7one fire model that prcdicts plume centerline tempera-

Note: The ceiling jet flows radially from the point wher the plume impinges on the ceiling.
(b) Idealized ceiling jet flow

ture, ceiling jet temperature, and ceiling jet velocity. SET also can determine activation times of fusible links

Figure 8.6 Ceilitig j e ~ rmder a Jar ceilirrg. mid (b) idealizcd,/lo~c!.

(0)

skerch

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 turns downward. The effects of the smoke layer on ;he ceiling jet'are taken into account in the zone fire models CFAST and JET.
I

I

The detector models account for the thermal lag of detectors by use of the response time index (RTI), as discussed in Chapter 2. The RTI approach is appropriate for the fusible links of sprinklers and smoke and heat vents.

CHAPTER 9

Hazard Analysis
ost smoke management systems provide smoke protection by minimizing people's contact with smoke or by keeping smoke completely away from people. As the name implies, tenability systems provide smoke protection by maintaining tenable conditions. Tenability svstems allow smoke contact, but the systems are designed such that the temperatures and concentrations of combustion products are limited. An analysis of these systems is called a hazard analysis, in that the level of hazard to life is evaluated. Technological advances have made hazard analyses feasible, and tenability systems based on such analyses have gained a level of acceptance in the last decade. NIST developed a group of computer programs, HAZARD I (Peacock et al. 199l), for hazard analysis in spaces consisting of a relatively few rooms, such as residences. Bukowski and Spetzler (1992) used HAZARD I to reconstruct the fire at the Happyland club in the Bronx, New York, that killed 87 persons. Klote et al. (1992) extended hazard analysis to large multi-story buildings for the study of staging areas for persons with mobility limitations. Hazard analysis is a powerful fire protection tool that has application beyond smoke management. This tool can be used to evaluate alternative building materials and furnishings. The most common smoke management applications are compartmentation and atrium protection. Compartmentation systems can be with or without pressurization. The atrium systems can have any con~binationof smoke filling, smoke exhaust, or natural smoke venting. Jt is also possible to use hazard analysis to evaluate the efect of component failure.

M

HAZARD ANALYSIS CONCEPT
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 vision and cause incapacitation or fatality. A hazard analysis can be used to calculate such smoke flow and the consequences for building occupants. A hazard analysis can consist of one or a number of fire scenarios. Hazard analysis consists of the following components: (I) fire scenario, (2) smoke transport, (3) people movement, and (4) tenability. Fire Scenario. As stated in Chapter 2, a fire scenario can be thought of as the outline of events and conditions that are critical to determining the outcome of alternative designs. In addition to the fire location and heat release rate, 0 , the fire scenario includes the status of the doors, the HVAC systems, the smoke management system, and other systems. For details about design fires, see Chapter 2. Species ( 0 2 , N2, CO, CO2, etc.) generation can be included in the fire scenario. The scenario may also include specifics about the fuel, ignition of multiple fuel packages, and the effect of fire suppression. The selection of the fire scenario can be based on professional judgement, analysis of historical fire data, or cods requirements. Smoke Transport. Smoke can flolv far from a fire and threaten life. The major driving forces that cause smoke movement are naturally occurring stack effect, buoyancy of combustion gases, expansion of combustion gases, the wind effect, fan-powered ventilation systems, and elevator piston effect. These driving forces are discussed in Chapter 5 .

Chapter 9-Hazard Analysis

As discussed in Chapter 3, smoke consists of the airborne particulates and gases evolved when a material undergoes pyrolysis or combustion, together with the quantity of air that is entrained or otherwise mixed into the mass. The evolved gases are part of the species mentioned above. Generally, when smoke flows away from a fire, the concentrations of particulates and evolved gases decrease. Conlputer models for smoke transport analysis are discussed later. People Movement. People movement in fire situations is complicated. Some people will fight the fire. Others move against the flow of evacuating people in an attempt to find or rescue loved ones. Scme computerbased evacuation models are capable of simulating the movement of individual people. As people move through the building, they are exposed to smoke. This time-integrated exposure can be used in tenability calculations. For iilfonnatioii about calculation of building evacuation 'time and a discussion of computer-based evacuation models, see Chapter 4. In many applications, consideration is made for people who are immobilized due to an accident or physical disability. Such a person would need to wait to be rescued, and the wait could exceed the-time needed for evacuation of the rest of the building. Tenability. Tenability calculations estimate the hazard to life of a scenario. Tenability calculations address one or more of the following: exposure to toxic gases, exposure to heat, exposure to thernlal radiation, and visibility through smoke. For calculation of exposures and visibility, see Chapter 3. The exposures are time-integrated doses of toxic gases, heat, and thermal radiation. These doses can be based on the smoke concentrations at several locations as people move out of the building. Alternatively, the doses can be based on the smoke concentrations at one locatio~~ while an inlmobilized person waits for rescue.

Eliminate Heat Exposure Calculation. Detailed heat exposure calculations are not needed if the maximum temperature is relatively low. For exposure times, Figure 3.7 can be used to make such an estimate. Eliminate Radiation Calculation. If exposure to heat does not cause incapacitation, exposure to thermal radiation wiil not cause incapacitation. Exposure to heat consists of direct bodily contact with hot smoke, and exposure to thermal radiation consists of receiving the radiant flux from hot smoke. If the smoke temperature is insufficient for heat exposure to be an issue, the smoke temperature is also insufficient for thermal radiation exposure to be an issue. Eliminate Toxic Gas Exposure Calculation. For many hazard analyses, visibility is the controlling tenability condition. The method described in Chapter 3, "Tenability and Perfect Dilution," can be used to determine if exposure to toxic gases is of concern for particular tenability criteria. (This same method can also be used to help determine if heat exposure is of concern for particular tenability criteria.) Alternatively, toxic gas exposures can be estimated by a simple method, such as the FED approach, to denionstrate that exposure to toxic gases is of concern. SMOKE TRANSPORT For niost applications, smoke transport calculations are done by computer. A wide range of computer models can be used, including ( l ) zone fire models, (2) network flow models, and (3) computational fluid dynamic (CFD) models. The choice of the model depends on the specific application. Smoke transport can also be evaluated by physical modeling (Chapter 15). Use of the zone tire model FAST and the network flow model CONTAM for hazard analysis is discussed later. For general information about zone fire models and building air and network tlow model:, see Chapter S. For general information about CFD n~odeling,sze Chapter lb. Many of theses ~nodels simulate production and can transport orspecific gases (02, N2, CO1COZ,NO2, HCI. HCN, HBr, etc.), but sinlulation of specific gases is not generally necessary for design applications. Generation of the specific gases requires detailed knowledge of rhe fuel, \\.hicl1 is usually not available in desig~iapplications. The approach presented in this chapter is one of many possible zeneral methods of calculating tcnabilir?.. The mass of fuel consumed by the fire is

Level of Complexity
The level of complexity of a hazard analysis depends on the particular application. Analysis of sowe of these co~ilponents consist of straightfonvard reacan soning, and others require detailed c.~lculations. A feu. ways that a hazard analysis can be simplified are discussed below. Elinii~~ate Evacuation Simulation. For esposures based on proteering immobilized people, the need for a detailed c\:acuarion simulation can be eliminated, provided thar the design wiring is sullicie~ltlylong. This is because the esposurc iimc considered Ibr a \vaitirlg person \\.ould be rnuch greater than that fbr a person e\.acuaring 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); total heat release rate Btuls (kW); chemical heat of combustion BtuAb (kJ/kg);

1 (1000).
The heat release rate, Q , and the mass of fuel consumed, riz , are entered into the computer smoke transport model, which calculates the concentrations of material burned, C, at every location and each time interval in the simulation.

to 02, N2,CO2, and CO. This allows simulation of the synergistic effects of CO production and O2 depletion on toxicity of CO. Considering this and that the CO is the dominant toxic gas in building fires, limiting the gases to 02, N2,CO2, and CO is appropriate for many applications. For information about CO production in fires, see Table 2.1. For any instant, the visibility can be calculated from

TENABILITY CALCULATIONS The following is one of many approaches to tenability calculations, and more extensive information can be found in Chapter 3. Tenability analysis addresses visibility, gas exposure, and heat exposure. can The mass concentration of material burned, Ci, be obtained from zone fire models. The fractional effective dose.(FED) can be used to obtain an approximation of the effects of exposure to toxic gases.

where

Si
K

=
=

visibility at the end of interval i, ft (m); proportionalityconstant (8 for illuminated signs, and 2 for non-illuminated signs);

6, = mass optical density, &lb (m21g);
Ci = concentration ofmaterial burned in interval i, lb/ft3

where FED = fractional effective dose at the end of interval i (dimensionless); ci = concentration of material burned at interval i, Ibl At
=

(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 hyperthern~ia.For hyperthermia, heat exposure can be estimated from

f? (gm3); i time interval, rnin (min);

LCt,, = lethal exposure dose from test data, lb f- min t' (g min). This equation is for unifornl time intervals, as calculated by computer models, and it evaluates the FED for the exposure time at the end of interval i (expos~ve time is nAt). An FED greater than or equal to one indicates fatality. The concentration, C;, is the denisty of materials that started as fuel that have accumulated at a location during the interval i. The concentration has units of mass of the material burned per unit volunle. The lethal exposure dose, LCtjO, is the product of the LCjO and the exposure time. The LCjO is the concentration of airborne con~bustionproducts that is lethal to 50% of the subjects exposed for a specified time. An FED of 0.5 can be considered a rough indication of incapacitation. When a more accurate evaluation of tosic effects is desired, the methods discussed in Chapter 3 can be used. The fractional incapacitating dose (Fl,\;) method is generally considered to be more accurate. Unlike the FED method. calculation of F,,-,! requires calculation of species concentration. The gases considcsed can be lilnited

where F,,/,= total cumulative dose (dimensionless); At = time interval, minutes;

?;. C,
C2

=
= =

temperature of air in interval i, "F ("C); 5.670 (5.185); 0.0152 (0.0273).

Incapacitation due to heat exposure would be expect<d...*.:,>.... . fo,r.F"4""?"ig~~:Fr..I~6 to ,one.rr,.o:f e,qital. . . ,, ,,.., ,: :.*.w;..!r!~~.II~... , , .: .,,,,,. .. If contact with gases does 'not r e d t in incaoacitation due to 11qdt-exposure,thermal radiation from those g&.es &&id not result in incapacitation for the same exposure time. Generally, exposure to thermal radiation a is not- ,,! jsye ,[or <atja, app,licatip,n>. For situations where thermal radiation is significant, see Chapter 3.
- g ->:, . , I ;, , .:

EXAMPLE HAZARD ANALYSIS

Hazard analysis has a wide range of applications, and this example was selected to illustrate some of the capabilities and limitations of this technology. The example is a six-story hotel (Figure 9.1) with a lire in one ol' the guest rooms on the ground 1100s. Both tlic

i

Chapter9- Hazard Analysis

VEND R01 R02

HK

-

(b) Plan for Floors 2 6

-

R03

R04

R05 STG ..
d

R06 ELEV

R07

R08

R09

- +

-

R10

Note: Floor 7 has mechanical penthouse (not shown).

. .

f c o R HK
R01
-

. . . .

.

.

R02

-

R03

R04

R05

OFF ;W

Rm , Fire R07 .:ELEV Room

R08

R09

R10

. J C S L -

(a) Ground Floor Plan

Symbols: Rn Room Number n COR Corridor ELEV Elevator -....- Office Window C W

-

STG OFF

Stoiage Room HK House Keeping Ofke VEND Vending Window Door

t

N

Table 9.1: Roughly Steady Temperatures for First Floor Rooms of Hotel Fire Example Based o n FAST Simulation Temperature Location Fire Room Corridor Section Open to Fire Room Corridor Section Open to the Oflice West Section of Corridor 1.
See F~gure for location ofroorns. 9.1

~oum' R06 COR l COR COR2

OF

"C

1700 470 280 180

927 243 138 82

window and door of the fire room are opened. The open window is large enough to allow combustion air to support a fully developed 5 MW fire (Chapter ?).

CONTAM was used for the smoke transport analysis. The temperature of most of the building locations is 73°F (23OC), and the outside temperature is 20°F (6.7OC). Becausc CONTAM does not include encrgy equations, the temperature of the tire room and that of the other spaces open to the burn room needs to bc spccitied. FAST was used to calculate these tc~i~pcmturcs The parar-ncters for the tenability calculations are listed in Table 9.3. The valuesof AH,,,, K, and 6,,,are con(Table 9. l).

The flow areas used in the CONTAM simulation are listed in Table 9.2. For this simulation, the integrity of the door is considered to be maintained, and warping of the door is considered. When subjected to elevated temperatures, some doors experience burn-through at the edges or may warp to increase the flow area of the doors are susgaps around he, door edges. Con~bustible ceptible to burn-through, and warping is more pronounced with steel doors.

Principles of Smoke Management

Table 9.2: Building Flow rea as' Used for Hazard Analysis of Hotel Fire Example
Flow Path

Path Name WALL-EL WALL FLOOR

ft?lf?

(0,

m2/m2\

Hoistway Wall Other Walls Floor Other Openings: Double Ext. DOO? Single DOO? Open Single DOO? Elevator Doo? Office window' Open window2 ~lev%or Vent Equivalent Areas of Shafts: Elevator Shaft Stairwells
I.

-

ft2

m2

DOOR-DB DOOR-SG DOOR-WP DOOR-E WIN-OF WINLOP VENT E FLOOR-EL FLOOR-SW

0.30 0.17 21.0 0.60 20.0 36.0 4.0
ft2

0.028 0.016 1.95 0.056 1.86 3.33 0.37 72 3.0

770 32

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
MC,, Chemical heat of con~bustion,

5000 Btds (4220 kW) 10,700 Btu/lb (25,000 kJlkg) 8 (for illuminated signs) 1600 ft2/lb (0.33 m21g) 0.033 ~b~3 min (530 g m-3 min) 30 minutes exposure to 180°F (82°C) for about 15 minutes, and a person could withstand exposure to 280°F (138°C) for about 4 minutes, after which they would suffer skin bums. Tolerance to higher temperatures would be much less. From this, it can be seen that the probability o f fatality due to heat exposure is high for many spaces on the first floor during this fire. The results of the tenability calculations are listed in Table 9.4. Graphic presentation of tenability results can be useful. The results of the toxicity calculations are shown graphically in Figure 9.2, and it can be seen that the FED exceeds one for many spaces on the first floor. For these spaces; the probability of fatality is very high. The visibility is shown in Figure 9.3, and i t can be seen that the visibility is less than 25 ft (7.6 m) throughout the ground floor. On all floors, the visibility in the stairs is less than 25 ft (7.6 m).

Proportionality constant, K Mass optical density, 4, Lethal exposure dose for a fully developed fire, LCI 50 Exposure time servative for most applications, and the value of LCf is applicable for fully developed fires. For appropriate parameters for other fires, see Chapter 3. An esposure time of 3 0 minutes was used for this example. For a specific application, the exposure time would depend on a number of factors. If this were a failure analysis for a design study, evacuation time might be taken as 15 to 30 minutes, including the time before people ssx moving. For a fire reconstruction, the exposure time might be taken from the estimates of the people movement based on the fire event time line developed as part of the fire investigation. The tenability calculations did not explicitly include heat exposure. Because the temperatures in this exalnple are consrant, the effects of temperature can be obtained from Figure 3.7. A person could withstand an

Chapter 9 -Hazard Analysis

Table 9.4: Summary of Tenability Calculations for Hotel Fire Example
Time (minutes) to Visibility of Floor G G Room COR CORl COR2 ELEV HK OFF R01 R02 R03 R04 R05 -- - R06
.

Table 9.5: Summary of Tenability Calculations for Hotel Fire Example-Continued
Time (minutes) to Visibility of Floor Room R13 R14 R15 R16 R1 7 R18 R19 R20 R2 1 R22 R23 STG SW1 SW2 VEND CO R ELEV HK R06 STG SW1 SW2 VEND COR ELEV R1 1 R73 SW I SW2 CO It ELEV R1 I R23 s\v I SW2 COR ELEV 14 K RI I R25 SW1 SW2 SW2 MECH ELEV 200 ft (61.0 m) 19.5 FED for 30 min Exposure 0.008

200 ft (61.0 1.9

m)

25 ft (7.6 m)

FED for 30 min Exposure 2.497

2.0

25 ft (7.6 m) . NA

G G G G G G G G G G

G
G

. R07

G G G G G G G G G G G G G G G G G 2 2 2 2 2 2 2 2
2

2 2 2 2 2

R08 R09 R10 R11 R12 R13 R14 RI5 RIG R17 R18 R19 R20 R21 R22 R23 S\VI SW2 COR ELEV HK ROI R02 R03 R01 R05 RG O R07 R08 R09 RIO R11

.

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

U Shading indicates FED of 1.0 or more

Figure 9.2 Toxicity for- 30-nzitzufeexposure o ho fel fire exanzple. f

Chapter 9 -Hazard Analysis

4th Floor

A and 200 ft (61.0 m).

Cross hatching indicates visibility between 25 ft (7.6 m)

2nd Floor

A

Shading indicates visibility less than 25 R(7.6 m)

Ground Floor

Principles of Smoke Management

This analysis needed to use both a zone fire model (FAST) and the network flow model (CONTAM). Using a zone fire model to simulate smoke transport for such a large building would be impractical, and a network flow model lacks the desired fire simulation and temperature

prediction. The combined approach above works around these limitations to produce meaningful results. Hopefully, a combined zone fire and network flow model'will be developed to produce even more realistic predictions in the future.

CHAPTER 10

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

PRESSURIZATION SYSTEMS
It is in~possible provide detailed design methods to for the almost infinite number of possible stairwell pressurization systems. The intent of this book is to discuss, in general, some .systemic considerations and alternatives and to provide detailed analyses of a few systems. 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). Centrifugal

Fan

m'-

Duct
/ Shaft

Roof Level

7

r
/

/

Duct

Figure 10.1 Stairwell presszirizafion bv top injec!!otl.

Figure 10.2 Sfair-wellpressurizafion by mulfiple injection \*pith the fa17 located at the ground level.

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 pressurization 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 these open doors, and the system will then fail to maintain 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 can be extended to twelve stories. Careful design is recommended for single injection stainvells in excess of eight stories. There is the potential for failure of a bottom injection system when the exterior door is opened. Some of the supply air can short-circuit the system by floning directly out the opened doorway. It is recommended that supply inlets be at least one floor above or below euterior doors.

Figure 10.3 Sfair-\i,ellpt-essuriiafiot~ by multiple injectiou ~ i t roof-mountedfan. h

Figures 10.2 and 10.3 are two examples of many systems that can be used to possible multiple i~~jection overcome the iirnitations of single injection systems. In Figures 10.2 and 10.3, the supply duct is shown in a separate shaft. However, systems have been built that have eliminated the expense of a separate duct shaft by locating the supply duct in the stairwell itself. If the duct is located inside the stainvell, care must be taken that the duct does noi become an obstruction to orderly building evacuation.

P ~ C i p l of~ e Smoke Management
i

l
1

1

!

Many multiple injection systems have been built with supply air injection points on each floor. These represent the ultimate in preventing loss of pressurization air through a few open doors; however, that many injection points may not be necessary. There is some difference of opinion as to how far apart injection points can be safely located. Some designers feel that injection points should not be more than three floors apart, while others feel that a distance of eight stories is acceptable. For designs with injection points more than three stories apart, the designer should determine by computer analysis that loss of pressurization air through a few open doors does not lead to loss of stairwell pressurization.

lated buildings, where total building evacuation by the stairwell is planned in the event of a fire. Compartmentation can be an effective means of providing stairwell pressurization for very tall buildings, when a staged evacuation plan is used and when the system is designed to operate successfully when the maximum number of doors between compartments are open. This maximum number of doors open between compartments would need to be determined by an evacuation analysis. Compartmentation does have a disadvantage from an architectural standpoint in that it probably cannot be achieved without increased stairwell landing space at the location of the compartmentation doors.

Compartmentation
An alternative to multiple injection is compartmentation of the stairwell into a number of sections, as illustrated in'-~igure 10.4. The stairwell is divided into a number of sections or compartments, each compartment being from one to about eight floors high. The compartments are separated by walls with normally closed doors. Each compartment has at least one supply air injection point. The main advantage of compartmentation is that it allows satisfactory pressurization of stairwells that are otherwise too tall for satisfactory pressurization. A disadvantage is the increase in floor area needed for the walls and doors that separate the stairwell sections. When the doors between compartments are open, the effect of compartmentation is lost. For this reason, compartmentation is inappropriate for densely POPLIRoof Level

Vestibules
A number of pressurized stairwells have been built with vestibules, which can be either pressurized or not pressurized. Vestibules provide an additional barrier around a stairwell and, to some extent, a vestibule can reduce the possibility of an open-door connection existing between the stainvell and the building. An evacuation analysis can be performed to determine the extent to which both vestibule doors are likely to be opened simultaneously. Analysis of a pressurized stairwell with an unpressurized vestibule can be performed using the same methods employed for analyzing a system without a vestibule except that the effective leakage areas from the stainvell to the building would be used. These effective areas can be detemiined by methods presented in Chapter 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 manual.

Supply Air Intakes
In the pressurization systems illustrated in Figures 10.1, 10.2, and 10.3, centrifugal fans supply pressurization 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 propeller fans should have wind shields as illustrated in Figure 10.5. Because the horizontal component of wind is genei-ally about ten times greater than the vertical component, wall-mounted propeller fans are estreniely 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. 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

I

1

1/

Each com~artment has at least one supply injection point.

I

1 I

Ground Level

Chapter 10 -Stairwell Pressurization

from the building, wind speed and direction, and on the temperature difference between the smoke and the outside air. At present, no formal method of analysis has been developed for this complex outdoor airflow. However, 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 building 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 on the other side and on the roof. PRESSURE PROFILES The pressure differences across a stain\;ell normally vary over the height of the stainvell. Analysis of the pressure profiles of unpressurized shafts was presented in Chapter 5. The analysis of pressure differences in stairwells presented in this chapter is slightly more complicated in that pressurization is incorporated. To facilitate analysis, the following discussion is limited to buildings that have the same leakage areas on each floor. Figure 10.6 shows pressure profiles for pressurized stainvells located in three buildings with diferent leakage characteristics, all of which have the same stairwell and outside temperatures. These profiles represent winter conditions; that is, an outside temperature less than the inside temperature. In a building without vertical leakage between floors or througli shafts other than the stainve!l, the pressure profile of a pressurized stain\-ell is a straight line. The slope of that straight line depends on the temperature difference between the stairwell and the outside and on the building leakage areas. This relation is discussed later in this chapter. Figure 10.6 shows typical pressure profiles of pressurized stainvells in a building with leakage between the floors a~id a building without leakage between tloors in that are similal-except at the top and the bottom of the buildings. The extent of the deviation depends on the magnitude of the leakage area between floors. The pressure profiles depend on the leakage areas of the stairwell, the elevalor shati. and the exterior \\-ails, as well as the tcmpcratures of tlie building, the stairwell, and tlie outside air. Analysis of sucl~ building is cornplicatcd a and 1 generally kasible only with the aid of a computer. s

Figure 10.5 Stairwell pressurization by roof-nzo~cnred propeller fan.

Top of Stairwell Building N t h VerticalLeakage Between Flmffi (Except at the ends. this curve is the same as that for a building without vertical leakage behveen floors.)

Building Wfih Vertical Leakage Through an Elevator Shafl

,

P" /\
Bottom of Stairwell

Building Wlthout Vertical Leakage Between Floors

I

,

I

Pressure Difference

Figure 10.6 PI-essrtr-e profile for presszlrized stai~x.e/ls iti three buildings wit17 different leakage
characteristics.

The pressure difference across a stairwell at one height can be much larger than at another height. Therefore, in addition to being concerned with the average pressure difference across a stain\dl, a designer should also be concerned with both the minimum and the ma.ximum pressure differences. STAIRWELL ANALYSIS
In this section, a method of analysis is presented for a pressurized stairwell in a building without vertical leakage between floors. This is the same zero floor leakage idealizatiou that was used for the analysis of stack 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. Neglecting the effects of leakage through floors and other shafts increases the spread between the minimum and maximum pressure differences. In this sense, the analysis is conservative. This analysis considers only one pressurized stairwell in a building; however, it can be extended to any number of stairwells by use of the concept of symmetry, as discussed in Chapter 6 . The initial analysis does not include consideration of open stairwell doors, but they are addressed later in this chapter. This analysis is for buildings where the leakage areas are the same for each floor of the building and where the only significant driving forces are the stairwell pressurization system and the temperature difference between the indoors and outdoors.

where
Apso = pressure difference at elevation y, in. H,O (Pa); ApSob = pressure difference at the bottom of the stair-

well, in. H20 (Pa). The above analysis assumes no change in densities, po, with elevation resulting in a slight overprediction of pressure difference. The magnitude of this overprediction increases with elevation and, for a 100story building, the resulting error would be less than 4%. For purposes of this book, this overprediction is 'insignificant. By substituting the ideal gas law into Equation (10.3), bpSocan be expressed as a function of temperature.

ps and

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 doors. Therefore, the absolute pressure in thestairwell is considered hydrostatic and can be represented as
Ps = p,, - li',Ps Y

and where

b

=

temperature factor, in. H20/ft (Palm);

(10.1)

To = absolute temperature of outside air, O R (K); Ts = absolute temperature of stainvell air, O R (K);

where
ps
=

K, absolute air pressure in stairwell at elevation y, in. H20 (Pal;

=

7.64 (3460).

psb = absolute air pressure in stairwell at stainvell bot-

The effective flow area from the stainvell through thc building to the outside is expressed on a per floor basis as

tom, in. H20 (Pa);
ps
y
= = =

air density within the stainvell, 1b/ft3(kg/m3); elevation above stainvell bottom, ft (m); constant, O.lg2 (9.8). \vhzre
Asso, = effective flow area behveen the stairwell and the

Kp

For the case where the wind velocity is essentially zero, the outside air pressure, po, is also hydrostatic and can be expressed in the same manner.

outside, @ (m2); f
ASB = flow area behveen the stainvell and the building,

fi!(m2);
where
p .
=

ifBO

=

flow area behveen the building and the outside,

pob = p .

absolute air pressure at elevation y, in. H 2 0 (Pa); absolute air pressure at stainvell bottom, in. H 2 0 W ; air density outside the stainvell, 1blft3(kg/rn3).

=

ft'(m'). The areas in this equation are those of the entire floor. In such a case, the pressure difference, ApsB, bcnveen the stairwell and the building can be expressed as
~

The pressure difference from the stainvell to the outside can be expressed as ApSO = p, - p O , and substituting Equations (10. I) and ( l 0.2) this is

Chapter l 0 -Stairwell Pressurization

The pressure differences Apm and ApsB are related as follows: QSB =
Qso
1+ ( A ~ B / A ~ ~ ) * '

= Kq-(

NASB

APZ - APSB~
Qs~t-Qs~b

J;;

'1

(10.14)

(10.8)

where
VsB
=

volumetric flow rate of air frorn~stairwell to building, cfm (m3/s);

which can be rewritten as

Pressurization Air
For the case where a stairwell is positively pressurized throughout (i.e., the direction of air flow is from the stairwell to the outside ovzr the entire stairwell height), the flow from the stairwell to the outside can be written in differential form as
y d d~ = CA,, J - y .

AsB

=

flow area between the stainvell and the building per floor when stairwell doors are closed, ft2

N
&I

= =

(m2); number of floors; pressure difference between the stairwell and the building at the stairwell top when all the stairwell doors are closed, in. H20 (Pa);

hBb difference between the stairwell and = pressure
the building at the stairwell bottom when all the stairwell doors are closed, in. H20 (Pa);

The term AI,, is the distributed effective flow area per unit height, which is uniform vertically. This distributed flow area is expressed as

where AI,, = distributed effective flow area per unit heieht, fi (m); H = stairwell height, fi (m);
h '
=

number of floors.

Substituting Equations (10.4) and (10.1 1) into Equation (10.10) gives

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:

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, bSBb at the stairwell bottom and &l at the top. In a building with vertical air leakage, the exact evaluation of the system would require that the effect of three or more colunlns of air at different temperatures be included. Such an analysis is cumbersome and, for practical purposes, a computer is needed. For this reason, the method of analysis presented in this section is based on 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

( l 0.13)
where ApSo, is the pressure difference between the stairwell and the outside at the stainvell top (1. = H). Because the Apss is a linear function of Apso as expressed in Equation (1OX), Equation (10.13) can be writtsn in temls of the pressure from the stairwell to the building. For C = 0.65, this becomes

where To = absolute temperature of outside air, "R (K); TB
=

absolute temperature of t!?e air in the building, "R

K~

=

(K); 7.64 (3461)).

For a building temperature of 70°F (2 1°C) and for \\.inter conditions, the temperature factor b can be 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 Equations (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, Apt = ApSOp Equaand tion (l 0.17) can be approximated by

The maximum error in this relation is approximately 6 % and occurs when Apb = 0.

HEIGHT LIMIT
- 20 - 10 0 10 20 S O 40 50

Cutside Temperature, To CF)
:SO

.............................

- 25

-20

-I5

-10

-5

0

5

10

. .:.. ,I

Outside Temperature, To

PC)

Figure 10.7

Ternper.ntur-e fnctor:

Average Pressure Difference
The average pressure difference can be defined as a pressure difference unifomi over the stairwell height that would result in the same total flow as a nonuniforni pressure profile. The flow from the stairwell can be expressed as

where

As stated before, two problems with pressurized stainvells are that the minimum pressure difference may be too low to prevent smoke infiltration and that the maximum pressure difference may be too high, making 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 stairwell can be impossible even when all the stairwell doors are closed. By satisfactory pressurization, it is meant that no\vhere over the stainvell height is the pressure difference greater than the maximum allowable pressure difference or less than the minimum allowable pressure 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 pressurization is possible:

N
A,

= number of floors;
= = =

erective flow area, ft2 (m2); flow coefficient; average pressure direrence across the effective flow area, in. H 2 0 (Fa); density of air, lb/ft' ( k g h 3 ) ; 776 (1 .OO).
Ap,,,;,, =

C
4'011

where
H,,,
=
=

height limit, ft (m); niaximuni allowable pressure difference between the stainvell and the building, in. H 2 0 (Pal; minimum allo\vable pressure difference between the stairwell and the building, in. H 2 0 (Pal . , outside design temperature, "R (K); building temperature, "R (K); flow area between the stainvell and the build.
7 7

P K,

=

& ,

=

The effective area can be either the area between the stairwell and the building or between the building and the outside. The average pressure difference needs to be consistent with the effective area. Equations (10.13) and (l 0.16) can be combined and solved for Apa,.to give

To
TB AsB

= =

A,).,. =

;[

/\/),

-

A/),,

(10.17)

=

Ing. fi- (m-);

cnapter 10-Stairwell Pressurization

ABO

= flow area between the building and the outside,

SIMPLE STAIRWELL SYSTEMS
A simple stairwell system is one for which no design provisions have been made to overcome the diop in pressurization when one or more stairwell doors are opened. Analysis of the simple stairwell system forms a foundation for the analysis of systems with open docrs. Some of the stainvell doors must be opened during evacuation if the stainvell is being used. No consensus exists concerning appropriate applications of simple stairwell systems. A possible criterion for such an application is that smoke leakage during times of low pressurization will not adversely affect the use of the stairwell during evacuation. In a lightly populated building (for example, telephone exchanges, luxury apartments), the stairwell doors may only be open for a few short intervals during a fire evacuation. Applications of the simple stairwell have so far been based on engineerins judgment because no formal method of analysis has been developed for evaluation of effects of intermittent smoke infiltration. Such an analysis would need to consider tenability conditions, evacuation analysis, and flow analysis. The simple stairwell system can use single or multiple injection. One or more fans are used, which can be centrifu~al,axial, or propeller. When all the stainvell doors are closed, the system must maintain satisfacton' pressurization. When stainvell doors are open, the pressure difference across closed stairwell doors usually drops to low levels [in the range of 0.01 in. H 2 0 (3 Pa)]. These low levels are not sufficient to prevent smoke infiltration into the stairwell, and simple stairwell sj-Stenis are only appropriats for applications for which stairwell doors are closed for almost all of the time during fire evacuation. Example 10.2 is for two 20-story stairwells in the same building. Syn1nietr~- used so that calculations are is needed for only one s~ainvell. The same approach can be used for three or more stainifells. The flow rate of pressurization air is highly dependant on the leakage area. Because these areas can only be roughly estimated in most situations, the fan needs to be sized conservatively so that the fan flow can be adjcsted tc xceptable It\-els of pressurization during system commissioning. This fan sizing can be by choice of high values of building leakags or of safety factors. The calculations of Ezample 10.2 are based on winter design teniperaturcs. This is appropriate when the inside-to-oulside dcsign temperature difference for winter is greater than thc outsidc-to-inside design temperature difference I'or summc'r. Othenvise, summer design data should be uscd.
'

(m2);

Km

=

0.13 1 (0.000289).

Ts was replaced by TB in Equation (10.1 g), s o that the equation would yield conservative values of H, for buildings with'vertical leakage. 1n' such buildings, the actual pressure profiles depend on three or more columns of air at different temperatures. If the stairwell temperature is between the outside temperature and the building temperature, then Equation (10.19) will yield conservative results. The absolute value of the temperature term is used in Equation (10.19) s o that the equation will apply to both winter conditions (TB > To) and summer conditions (To > TB). In many cases, ASB is much smaller than AB*, and, in such cases, Equation (10.19) can be ~implified~to

The units for this equation are the same as those for Equation (10.19). For a building temperature of 70°F (21°C) and for winter conditions, the height limit, H,,,, can be obtained from Figure 10.8. Example 10.1 illustrates the use of height limit.

180 165 150

.

v

135

-

E

. .
'05

I
.A
a

E
.-

-

E 120-

: -

.

:

E m

90
,

. .

=

.

75. 60 45 30

- 20

- 10

0

10

20

30

40

50

Outside Temperature. To(OF)

Outside Temperature. To(T)

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 -Qmin0.40 - 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 f (0.1 18 m2). i ? Calculate the height limit from Equation (10.19).
H
= 3

0'4 - 0.05 [ l 1
474 530 1

I

+

('6)
2]

= 2 19 ft (67 m).

The height limit is greater than the height of the stainvell, so the equations presented in this chapter can be used for analysis. 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
As discussed in the preceding section, when any stair door opens in the simple stairwell pressurization systems, the pressure differences across closed doors drop significantly. However, opening the exterior stairwell 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 doorways 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 constructed buildings. Thus, the exterior stairwell door is the greatest cause of pressure fluctuations due to door opening and closing. For densely populated buildings, it can be expected that many stairwell doors will be open during fire evacuation. Accordingly, stairwell pressurization systems in such buildings should be designed to operate with some number of open doors. This design number of open doors depends heavily on the evacuation plan, and specific guidance about this number is beyond the scope of this manual. Four types of systzms intended to maintain acceptable levels of pressurization with all doors closed and with some doors opened are discussed in this section. System with constant-supply air rate and an exterior stairwell door that opens automaticall!. upon system activation (Canadian System). System with constant-supply air rate and a barometric damper. System with variable-supply air rate. System using stairwell pressurization i n combination with either fire floor venting or fire floor exhaust. The following is a discussion of these systems. Field tests of these and other systems for stairwell pressurization were conducted by Butchzr et al. (!97 1); Dias (1978); and Taniura (1994, 1990a. l99Ob, and I'BOc).

Note: Canadian system can be single or multiple injection.

n
Roof Level

Figure 10.9 Canadiarz system has exterior door that operzs automatically on system activation.
small. However, the term constant-supply is used to differentiate this system from the ones with variable-supply air rates, but constant supply systems actually have some variation in flow due to the pressure-flow characteristics of the fan. Supply air can be introduced at one location, or the system can be multiple injection, as illustrated in Figure 10.9. By eliminating opening and closing of the exterior stairwell door during system operation, the Canadian system eliminates the major source of pressure fluctuations. This system is simple to design and relatively inexpensive. Accordingly, thissystem is recommended whenever it can meet the design requirements.

Systems with Barometric Dampers
This system has sufficient supply air when a design number of doors are open. When all the doors are closed, part of the supply air is relieved through a vent to prevent excessive pressure buildup. Barometric dampers that close when the pressure drops below a specified value can, be used to minimize air losses through a vent when doors are open. There are two approaches to location of barometric dampers: (I) in exterior stairwell walls or (2) in walls to other building spaces. Venting to the building has the advantage that the barometric dampers maintain the pressure difference of interest, which is from the stairwell to the building. However, venting to the building has the disadvantage that the air vented can supply oxygen to the fire. Exterior venting eliminates this disadvantage, but exterior vents can be subjected to adverse effects of the wind. Figure 10.10 illustrates a system vented to the building at each Iloor. In systcms built with vents

Canadian System
The system with constant-supply ;ir rate and an exterior stairwell door that opens automatically upon system activation is essentially the same a, that in the Sirppler~~er~rthe Nrrriontrl Bliilcli~rgCnrlc. of C m a h ro (1985). The supply air rate is not actually constant. but it varies to some extent \\.ith the pressure across the fan. For centrifugal fans this variation in flou. rate can be

Principles of Smoke Management

between the stairwell and the building, the vents typically have one or more fire dampers in series with the barometric damper. As an energy conservation feature, these fire dampers are normally closed and open when the pressurization system is activated. This arrangement can reduce the possibility of annoying damper chatter that frequently occurs with barometric dampers.

Systems with Variable-Supply Air Rate
Systems with variable-supply air can be used to provide overpressure relief. The variable flow rate can be achieved by using one of the many fans commercially available for a variable flow rate. Alternatively, a fan bypass arrangement of ducts and dampers can be

Notes: 1. Vents have barometricdamperand one or two fire dampers in series. 2 A system with vents can be single or multiple injection

Roof Level

I '

used to vary the flow rate of supp!y air to the stairwell. The variable-flow fans are controlled by one or more static pressure sensors that sense the pressure difference between the stairwell and the building. When doors are opened, the stairwell pressure drops and the flow rate of supply air is increased to achieve at least the minimum design pressurization. When all the doors are closed, the stair pressure increases and the flow rate is reduced to prevent excessive pressure differences. In the bypass system, the flow rate of air into the stairwell is varied by modulating bypass dampers, which also are controlled by one or more static pressure sensors that sense the pressure difference between the stairwell and the building. The system operates in essentially the same way as the variable-flow fan systems to prevent excessive pressure differences and provide at least the minimum design pressure. The response times of these sysiems depend on the particular components used for the pressurization system including the feedback controls. Figures 10.1 1 and 10.12 show response times of systems tested at the experinxntal fire tower of the National Research Council of Canada (Tamura 1990b).

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, exterior wall vents allow an increased pressure difference across the closed staincell door on the fire floor. Venting the fire floor can also aid tire fighters in smoke purging after the fire has been put out.
II

Figure 10.10 Stair-~.ell pressurization system with vents to the bziildi17gat eachjlooc

Peak Pressure

Time (minutes)

1.47 in H O , (365 Pa)

"

0

5

1 0

15

20

25 30 35 Time (minutes)

40

45

50

55

60

Figure 10.12 Response time of staitwell pressurization system with bjpass system.

Smoke shafts are similar to external wall vents except that smoke from the fire floor is vented through a shaft. The venting is aided by buoyancy forces of hot smoke. Smoke shafts should be constructed in accordance with local codes; specific. engineering data regarding sizing of smoke shafts are available from Tamura and Shaw (1973).

ANALYSIS OF SYSTEMS WITH OPEN DOORS
The analytical approach developed for simple stairwell syste~ns can be extended to pressurized stairwells with open doors, provided that the frictian losses due to airflow in the stainvell are negligible. Friction losses can be minimized by having a multiple injection system designed to minimize vertical airflow in the stairwell. Because the pressure losses due to friction are considered insignificant for this analysis, the pressure differences described by Equations (1 O.4), (1O.7), (1 0.8), and 10.15 apply for both summer and winter conditions, as is illustrated in Figure 10.13. When all of the doors are closed, the pressure differences 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 pressure difference across it increases, as shown in Figures 1 0 . 1 3 ~ 10.1 3d. This means that the flow through i n and open esterior doorway can bc very large. This is especially true during the summer when the pressure difference is greatest at the shaft bottom where most exterior doors are located (Figure 10.13d). When doors are opened to the buildins, the pressurc difference across

the open doorway drops significantly, as illustraied in Figures 1 0 . 1 3 ~ 10.13d. However, the flow through and the large area of an opened doorway can be very large, as can be seen from the examples discussed later. In the winter, the pressure difference across opened doors increases \i.ith elevation. The greatest amount of pressurization air is needed when the design number o f opened doors are located in a section at the top of the stairwell, as illustrated in Figure 10.13e. This forms a conservative winter design condition. The conservative summer design condition is for the opened doors to form a section at the bottom of the stairwell, as in Figure 10.13f. Equation (10.14) applies when the effective tlo\v area between the stairwell and the building is thesams for each floor. When some doors are opened and others closed, this flow area varies from floor to floor. Equation (1 0.14)'can be applied piecewise to vertical stairwell sections, where the values of Ass and the values of ABO are the same at each floor. Both of these areas are used to calculate the pressure differences and the effective flow area. Equation ( l 0.14) can be written in a general form for C = 0.65 and p = 0.075 Ib/ft (l .20 km/m) as

where p = volumetric flow rate from the section, cfni (mds):

N
A,

= =

number of floors in section;
7 7 effective flow area per tloor from stairwell, R- (m-):

G = the tlow factor, fpm (m/s). 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
AP

0
AP

(f) Summer With Design Condition (e) Winter 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 these examples is only of one stainvell, but it can be thought of as being applicable to any number by application 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 remperature of 94OF (34OC). It is an unusual occurrence that the total pressurization air calculated for both design temperatures is the same [17.500 cfm (5.26 m3/s)]. As expected from observation of Figure 10.13e and 10.13f, the tlow through the open exterior doonvay is greater for summer than \\inter (9.200 cfm [4.3 rn3/s] in summer and 6,800 cfni [3.2 m3/s] in winter). For a taller stairwell, the flou through the exterior doonvay in summer would be even greater. As with the simple stain\.ell system, safety facrors are needed to size the supply air fan or fans-the fan needs to be sized conservati\.ely so that the fan flow can be adjusted to acceptable levcls of pressurization during system commissioning. This fan sizing can be by choice of high values of building leakage or of safety factors.

Ap, - Ap, (in:H,O)

Figure 10.14 Flol~, fnctol:

where
Qb

=

pressure difference at the bottom of the section, in.
f

H 2 0 (Pal;
Ap, = pressure difference at the top ofthe section, in.

H 1 0 (Pal;

Kg

=

1740 (0.559).

Equations (10.7 1) and (1 0.22) can be used to calculate either irSB lfsBO.where i s BO is the flow from or the stairwell through the building to the outside. When calculating vss,A, and the two pressure differences are from the stainvell to the building. When calculati~ig i?,, , A, is the effective flow area from the stairwell through the building to the outside, and the two pressure 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 from those through the building. For the exterior doors, exterior vents, or other openings directly to the outside, the flow can be espressed as

NONUNIFORM BUILDING FLOW AREAS
Flow areas that differ from tloor to floor can result in significant challenges to stairwell pressurization. Possibly 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 open garage. lt can be seen that the pressure differences are much greater on the floors where tlie stainvell opens into the parking garage. This figure is based on the same assumptions as in rhs preceding discussions except rhat

where

vso

=

volun~el~-ic rate from stairwell to outside, flow cfi11 (n?/s);

Principles of Smoke Management

11 11 A ,

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 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,
= 10.5

I

11 11

I1

(0.975 m2) with stainvell door opened, To = 14 "F-(-IO°C) or 474OR (263 K), TB= 70°F (21°C) or 530% (294 K), Apm, = 0.40 in. H20 (100 Pa), and hi,0.05 in. H20 (12.4 Pa). Because the design temperatures are the = 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 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 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 stairwell 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)')
=

? f

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, hBbe ~ must evaluated. From Equation (10.9), bSOb [l+(0.32/1.2712]= 0.053 in. H20 (13.2 Pa). The pressure differences, Apb = 0.05 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
l11

(10.21), PS,

= GNAsBoe=

1020 ( S ) 1.26 = 10,300 cfm (4.9 m /s).

3

Exterior Stairwell Door d Estimate the flow through the cpened exterior doonvay with air density ofO.075 lwft3(1.20 kglm3)~ n aty = 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 Equation (1 O.23),

i/sO = 776(0.6j)(l0.5)[2(0.062)/0.075]"

= 6800

cfm (3.2 m3/s).

Total Flow Needed During Winter 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 differences 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 The data and computer o u t i t for this example are provided in Appendix G It is usually inipractical to determine the values 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 Across Closed Stairwell Doors in. HIO 0.053 Fa
13.2

11

l

Stairwell Supply
cfm

Maximum Ap Across Closed Stairwell Doors in. HzO 0.6 1 Pa
15.2

m3/s

20,500

9.67

1

13,900 6.56 0.107 26.6 0.110 27.4 For run I, the stairwell supply 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 floor. When these floor-to-floor pressure difference variations are unacceptable, approaches to dealing with them include: modify the building flow network (possibly by use of partitions or pressure relief vents), eliminate the doors into stairwell on garage floors, and use other stairwells for the open garage, use hardware on the stairwell doors to the garage floors that assists door opening by reducing door opening forces.

COMPUTER ANALYSIS USING A NETWORK MODEL
Except as noted otherwise, the preceding sections were based on the simplifying assumptions of(1) the only pressurization system being the pressurized stairwell, (2) the flow areas of the building being the same from floor to floor, (3) the leakage between floors being negligible, (4) the flow through other shafts (elevators, mail chutes, etc.) being negligible, and (5) the friction pressure losses in the stairwell being negligible. Network computer models (Chapter 8) can be used to account for all of these and many others. Example 10.4 uses the computer model CONTAM to analyze two pressurized stairwells in a 15story 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 1 1

Elevator Smoke Control
his chapter addresses two very different kinds of 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. Most elevators worldwide do not have smoke protection, fire protection. and other features necessary for them to be considered as a means of fire evacuation. Elevator systems not specifically designed and built for fire evacuation should not be used in fire situations (Sumka 1988). Honever, the use of elevators for fire evacuation is a topic that has received considerable attention in recent years. Because the concept of elevator evacuation is so new, this chapter provides a general overview of the topic in addition to the smoke control considerations. The other kind of elevator smoke control system addressed in this chapter is intended to prevent smoke flow to other floors by way of the hoistway (elevator shaft). The problems that can result from snioke migration through lioist\vays are illustrated by the fire at the MGM Grand Hotel (Best and Demers 1982). The fire occurred on the ground floor, but smoke migrated to the upper floors where the majority of the fatalities at occurred. The hoist~..ays this hotel did not have any special smoke protection, and they were one of the major paths of smoke ~nigration the upper floors. to

T

TOP VEhT
The requirement for vents at the top of the hoistways has been in codes for so many decades that the original intent of the vents is uncertain. The most common reasons that the authors of this book have heard for these vents are that they ( l ) vent smoke during a building fire, (2) vent odorous gases, and (3) prevent esces*

sive pressures at the top of the hoistway due to a rising elevator car. The idea that the vents are needed to prevent excessive pressures is doubtful for two reasons. First, vents would also be needed at the bottom of the hoistway if the pressures from moving elevator cars needed to be relieved. Second, the pressures produced by moving elevator cars are very small, as described in the following section about piston effect. To understand the of idea venting odorous gases, a historical perspective is needed. In 1853, an elevator safety device to prevent elevator cars from falling was developed by Elisha Otis. By the 1880s, elevators gained wide acceptance in many large cities. During the 19th and early 20th centuries, the standards of sanitation were not as advanced as those of today, and it is likely that open elevator hoistways were used as trash chutes by some people. Further, it is possible that vents were required at the top of elevator hoistways to relieve some of the malodorous gases emanating from garbage at the bottom of the hoistway. Regardless of the original purpose for these vents, the idea that they can significantly improve smoke conditions during a building fire has gained wide acceptance even in the absence of supporting data. Research is needed to evaluate the effect of vents on the hazard of smoke exposure during building fires. For most of the elevator pressurization systems discussed in this chapter, there is either no top vent or the top vent is closed. For energy conservation, these-top vents are often normally closed. Such normally closed vents should remain closed during elevator pressurization unless the open vent is part of the pressurization system design. The capability of remote operation of top vents may be desired by the fire service.

Chapter 11-Elevator Smoke Control

(a) System Wlth Lobby Used for Elevator Evacuation

(a) System Whotit Lobby Used for P ~ t i o of Smoke n
Migration

Figure 11.1 Elevator cc]-motion ardpistotz effect in pressurized elevator- shafis.

PISTON EFFECT
The transient pressures produced when an elevator car moves in a shaft are a concern for elevator smoke control. Such piston effect can pull smoke into a normally pressurized elevator lobby or hoistway. Analysis of the airflows and pressures produced by elevator car nlotion in a pressurized lioistway was developed by Klote (1988), based on the continuity equation for the contracting control volume in the hoistway above a moving elevator car. The elevator system can be with or without enclosed lobbies (Figure 1 1. l). Piston effect experiments (Klote and Taniura 1987) were conducted on an elevator of a hotel in Mississauga, Ontario, Canada. This elevator served each floor of the 15-story building, and the hoistway was pressurized by a vane axial fan. Figure 11.2 is a coniparison of measured and calculated pressure differences due to an elevator car ascending fro111 tlie ground floor to the top floor. The general trends of the calculations are in agreement with the measurements. On the ground floor, piston effect causes a rapid drop in pressure followsd by a gradual pressure increase as the car moves away from the ground floor. A reduction in pressure is expected below an ascending car. This pressure reduction decreases as the car moves away due to the etYect of increasing leakage area of the shaft below the car. On the top floor, piston effect due to the ascendin,0 car causes a gradual pressure increase with distance traveled until the car gets close to that floor. On a middle floor (the 8th) the pressure increases as tlie car approaches, drops suddenly as the car passes, and increases after it travels away. For thc ground and Stli floors, tlie extremes of the calculated c u n e s deviate from those of

the measured curves by only about 0.004 in. H 2 0 (I Pa) and, for the 15th floor, the extremes deviate by about 0.03 in. H 2 0 (S Pa). From the analysis by Klote, equations wers developed for the critical pressure difference at which piston cfrect cannot overcome the elevator pressurization sgsten1 both for systenis intsnded to prevent smoke migration tlirougli the hoistway and for systems intended for elevator evacuation.

Without Enclosed Lobby
This section is limited to elevators without enclosed lobbies, and the elevator pressurization systems discussed in this section are intended to prevent smoke migration through the hoistway. The critical pressure difference, 43,,.,,, is froni the shaft to tlie building:

where
= = = =

critical pressure difference, in. H 2 0 (Pa): air density in hoistway, lb/ft3 ( k g & j; cross-sectional area of hoistway, fi (m2): leakage area betiveen lobby and building. f? ~ (m2); 7 7 free area around the elevator car, ft- (m-):.
.

= =

stfective area bet\vsen hoistway and outside, (m2); elcvalor car velocity. Ipm (m/s);

fi

=

Principles of Smoke Management

C,

=

flow coefficient for flow around car, dimensionless; 1.66
X

where Aio is the leakage area between the building and the outside in fr2 (m2). Example 11.1 illustrates calculation of the critical pressure difference for an elevator pressurization system without enclosed lobbies.

Kp,

=

lo6 (1.00).

The flow coefficient, C,, was determined experimentally (Klote and Tamura 1986a) at about 0.94 for a multiple-car hoistway and 0.83 for a singlecar hoistway. The effective area from the elevator to the outside is

With Enclosed Lobby
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

; 0.20
0.10 0.05 0 0
.

-50 -40

a 8
C

a

G

G

g
0

1 t Floor 5h
a

l
V)

-20

lobby in ft2 (m2). Equation (11.3) is the same as that for the effect in an upper limit of pressure difference due to p~iston unpressurized hoistway in Chapter 5 even though the two equations were derived differently. The effective area between the hoistway and the outside is

a -10 .
1 0 20 30

lime (seconds)

0.30r

where

A,, A,,
(U

= =

leakage area between lobby and hoistway, ft2 (m'), leakage area between the building and the outside,

Z 0.15-

-40

f
3

ft2 (m2).
Example 11.2 illustrates calculation of the critical pressure difference for an elevator pressurization system with enclosed lobbies. S M O K E CONTROL FOR PREVENTION O F S M O K E MIGRATION These systems consist of supplying air to the hoistway 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 procedure is for elevator cars to be taken out of nornial service 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 other localities, the elevator doors are closed after sufficient time to allow passengers to leave the car. The fire service has elevator keys allowing them to operate clevators for rescue and for transportation of personnel and equipment to fight the fire. As stated for stairwell pressurization, the flow rate of air is highly dependant on the leakage area. Because

a

a, 0.10V)

-20 2 8 h Floor t -10 0 1 0 20 30

,! $

0.05 0

lime (seconds)

0.30r

0

-

0

0

10

20

30

lime (seconds)

- - - - - - Calculated

Measured

-

Figure 11.2 Meas~tr-ed calc~tla~ed a17d p~sss~rt-e differEl7CES d ~ t eto the pis~ot~ e/f^ect of an ascet7dit7g eleiator cat-.

Chapter 1 -Elevator Smoke Control 1

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 pressure 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). ,= Therefore, piston effect will not pull smoke into the hoistway. The hoistway is pressurized at a level above

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 'f (1 1.2 m2), A, r

=

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,,.~,= in. H 2 0(6.0 Pa). 0.024 is The hois~\vay 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~ple abo5.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

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

From Equation (l 1.3),&C,.I,= 0.13 in. H 2 0 (33 Pa). The Iio~stway pressurized at a level below Q,,.;,.Therefore, piston effect may pull smoke into the elevator lobby. Posis 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 piston etkcr lasts only a few seconds, and a hazard analysis could be used to evaluate the still not a problem. Fu~ther, ef'fect on life safety

Principles of Smoke Management

(a) Wlnter

40

(b) Summer

AP

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 situations, the fan needs to be sized conservatively so that the fan flow can be adjusted to acceptable levels of pressurization during system commissioning. This fan sizing can be by choice of high values of building leakage or of safety factors. the simple equations. This approach has the advantage of being able to account for complicated building flow networks. Network models including CONTAM are discussed in Chapter 8. Example 11.4 illustrates the use of CONTAM for analysis of a pressurized elevator system. Because of flow through the stai~~vells floors and friction losses and in the shafts, the pressure profiles for this example (Figure 1 1.4) differ froni those for the ideal building without vertical leakage (Figure 1 1.3).

Analysis by Simple Equations
The equations for analysis of pressurizzd stairwells presented in Chapter 10 can be adapted for use with pressurized elevators by redefining the subscript S in the analysis from stairwell to hoistway. Such an analysis is then applicable for the idealized conditions listed below. The only pressurization system is the pressurized elevator. The flow areas of the building are the same from floor to floor. The leakage between floors is negligible. The flow through other shafts (stainvells, mail chutes, etc.) is negligible. T h e friction pressure losses in the hoistway are uegligible. Figure 11.3 shows the pressure difference profiles of a pressurized elevator in a building \\.ith exterior ground floor doors open With the exterior doors open, the pressure on the ground floor is nearly the same as that outdoors. For a mathematical description of this, readers should see the section on effective flow areas in Chapter 5. Example 1 1.3 is based on the pressure difference, from the elevator to the buildinp being equal froni the elevator to the to the pressure difference, outside at the ground floor.

With Stair Pressurization
Often elevator hoistways are pressurized in conjunction with stairwell pressurization, as in Example I 1.5. This example is the same as Example 1 1.4 except for the pressurized stairwells. Because of stainvell pressurization, the pressure profiles of Example 1 1.5 (Figure ! 1.5) are closer to those of the ideal building without vertical leakage (Figure l l 3 ) than those of Example I 1.4 (Figure 1 1.4).

SMOKE CONTROL FOR ELEVATOR EVACUATION
Throughout most of the world, there are signs next to elevators indicating that they should not be used in fire situations and that stairwells should be used for fire e\.acuation. Thcse elevators are not i~dendedas means of fire egress, and they should not be used for fire evacuation. However, some peopie cannot- use stainvells because of physical disabilities, and for these people, fire evacuation is a serious problem (Pauls 1988; Pauls and Juillet 1989). This section discusses smoke control systems that can be used to proyide smoke protection for elevators as a part of an ovcl-all elevator protection scheme to allow fire evacuation by elevators. The information in this chapter is based on a joint project of the National Institute of Standards and Technology (NIST) in the United

Analysis by Network Model
Network computer programs can be used for analysis of systems without thesc simplifying conditions ol'

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 open. There i no vent at the too s - . f the shaft (or the vent is &tly closed), and the flow area of the vent can be neglecied. I i

-

Elevators

m

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 O (32OC) F 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 (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)

ft'

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). ipsoi

= MsB,[~ + (AsB/ABo)]

= O.OS[I

+ (1.2/0.8)

2

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
" 0

= 7.64(&-A)

= -0.000418 in. H201m.

Ihe heightaltlieshaR is H = 6x12 = 72 fi(21.9 m). tearnn~e Equation (10.4) to g t bpsob = NSO,- = G.163 -(-0.000418)(72) hH 41 die ground floor, the en'ecdve area is

= 0.193 in. H 2 0 ( 4 8 Pa). .

Ihe density ofair in the building is p = 0.075(530/533) 'ram Equation (10.16). the flow rrom the ground lloor is
V = N K ~ A ~ C

= 0.0746 lblfi (Ililm

3

3

).

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 4'o,,

=

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

+

Q',

-2

+

'.lG3

0'188

= 0.176 in. H 2 0 (43.8 Pa).

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 Location of hoistway supply air inlet Height between floors Outside winter design temperature Outside summer design temperature Building design temperature Winter stairwell temperature Summer stairwell temperature Minimum design pressure difference
Areas:

14 Penthouse (l 5th floor) 12.0 ft 3.66 m 14°F -10°C 93OF 34°C 70°F 2I0C 45°F 7°C 82°F 28°C 0.05 in. H 2 0 12.4 Pa

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 0.1 1 0.0102 Leakage area of stainvell walls to the outside 0.25 2.32 Leakage area around closed single doors 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 1360 126 Equivalent orifice area for friction losses in hoist~r-ay Chapter 6) (see 0.65 Flow coeficients for all leakage and flow areas, except open doorways 0.36 Flow coefficients for open doonvays 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

Winter

18.0SO

Lls 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. As with h e previous example, flaws and pressures w e n calculated using CONAM. The flow rates are: Elevator Hoistway Each Stairwell

I

cfm
Winter Summer 15,900 13;380

Us
7.500 6.320

cfm
5,853 4,660

U s
2760

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

II(I

Typical Floor Plan
The d e s i g paramerer; and flow areas are: Number o f floors sensed by elevator Location o f hoisruay supply air inlet Hsiglit between tloors Outside winter design telnpcrature Outsidc sunimer design trinperature Building design tenipcra~ure Winter slainvcll Iemperature Sumnlcr stainvell tcmperaturc hlasimum design pressure dilkrsnce xross IL,r\by doors Mininwndesi~n pressure dilkrcnce across k5h!. doors Design doors open durmp w i n m Dcsign doors opcn during summcr
Arus

I1
Penthouse (15th floor) 12.0 ft 14'F 93°F 70°F 4j°F 82°F 0.52 in. H?O 0.05 in. H?O

3.66 m
-10°C 54:C 21°C 7°C ?SzC 79.6 Pa 12.4 Pa

Ground floor and floors 12, 13, and 14 Ground floor and lloors 2 5. and 4 .

ft'
2.26 S4 0.850 0.11 0.11 0.25 0.071 0.63

m0.110 7.8 0.0790 0.0102 0.0102 0.0Zj 0.00657 0.0585 0.0465 3-90 0.0186 5.72 126

Lcakase area o f cslcrior building u-alls psr I-Ion area oi1u.o open es~cnor ground Iloor 2cuhlc doors I.cakagc area bcrwen tloorr o f ~ l ~ s buildinp Lcakagr area dslainvcll aalls to the bulldm< Leakage area orst:tinwll u i l s m h c outside I.cakagc arc8 aroul~d closed iingls doors L c a h g c area ol'lwistu%iy walls 10 ~ h building c

L c l a g c area arou~id closed clcvauir dams Leakage
are;^

around closed lobby doors

0.50
42 0.20 40
l .X0

I'low area o f o p n lohby d w r s Leakqe arca orclused vent at top ol'li0ist\\.3! Equivalent arilicc arm fbr iric~ion losscs i n r:~in\.r.ll (*cc i-li:~ptcr ] h Equivalcnl orilice arsa Tor friction losses i n bc:r.laa). (\cc Ci~:~plcr 6) F l o cocllicicnt.; (or all Ic3La:s ~ and h v ; ~ r c s c\ccpt o l x n do\m\ays . Flow cocllicicnts for open doonra).s

0.65

.

0.16

The progran CONTAhl was used forthis anal! >is (output II~IIl a ~ u i ~ ) . with li3n1plcs 11.4 arid 11.5. COWAM had to h run aad rerun a numher s As : ofti~ncs obtain the suppl!. rat t l m rcsul~cd 5 s desired i ~ i i n i ~ i ~ w r l u1 in pressure diltcrcnec. l l l e !h\\ I\vithY d c s i y nun1h.r ul'doors open are: ~ L lhc cfm \\-in~c: Summcr For I 6.200 cliii. tlic ~ ~ t i n i ~ nand i111;1~inwni un pr:?wrc ill. H . 0 \Vinicr S111111iicr 0 . lL : . 0.25: 3linin1u111 I'a in. 1 1 > 0 ll..XJl 1l.luS 16.200 I(l.000 dill>renccs 3cross tllc Ik>hb> J<I(TS :re: Xl;lsin~a~r~
Pa
IS ! .

7.h5U 5.WlO

.,>.,

5 . .

74.9 74.1

02.7

-

IIIL.C prcwlrc d i l k r c ~ ~ c :ire tt i11li11 c c s i :~ i ~i i t ~ w ~;dI I I : ~ ~ ~I ~ ~ I I I I cs tl~ d : i ~ KU I C .

164

Principles of Smoke Management

L 2

U

0

0.1 0.2 0.3 Pressure Difference (in. H,O)

0.4

0

0.1 0.2 0.3 Pressure Difference (in. H,O)

0.4

(a) Elevator pressure differences in winter

(a) Elevator pressure differences in winter

Pressure Difference (in. H,O) (b) Elevator pressure'differences in summer

Pressure Difference (in. H,O) (b) Elevator pressure differences in summer

Note: Stairwells are not pressurized.

I

L I

~;0'12

-0.06 0 Pressure Difference (in. H,O)

0.06

G]
0

0.05 0.10 0.15 'Pressure Difference (in. H,O)

0.20

I

(c) Pressure difference from stairwell to building

(c) Pressure difference from stairwell to building

Figure 11.4 P~zssure difference profiles calculated by CONTAM for a presszrrized elevator in bzrilding with outside exterior doors open (Esa/nple 11.4).

Figure 11. Pressure difference profiles calculated S by CONTAM for a presszrrized elevator in building with outside exterior- doors open (Examnple I 1 S).

States and the Natio~ial Research Council of Canada (NRCC) to evaluate the feasibility of using elevators for the evacuation of the handicapped during a fire (Klote and Tarnura 1987, 1986a, 1986b; Tamura and Klote 1988, 1987a, 1987b). Before this joint project, Klote (1984, 1983) conducted field tests of several elevator pressurization systems. It shoi~ld be emphasized that conventional elevators do nor ha\-e any protection

scheme for fire evacuation, and fire evacuation by these conventional elevator systems is not recommended.

Concerns about Elevator Evacuation
This section provides a description of many concerns about elevator evacuation, and the nest section discusses these concerns along with one approach to

Chapter 11-Elevator Smoke Control

deal with them. The 1976 edition of the Life Safity Code (NFPA 1976) listed the following "problems" involved with the use of elevators as fire exits:* Persons seeking to escape from a fire by means of an elevator may have to wait at the elevator door for some time, during which they may be exposed to fire, smoke, or developing panic. Automatic elevators respond to the pressing of buttons in such a way that it would be quite possible for an elevator descending from floors above a fire to stop automatically at the floor involved in the fire and open automatically, exposing occupants to fire and smoke. Modern elevators cannot start until doors are fully closed. A large number of people seeking to crowd into an elevator in case of emergency might make it impossible to start. Any power failure, such as the burning out of electric supply cables during a fire, may render the elevatois~noperative might result in trapping or persons in elevators stopped between floors. Under fire conditions, there might not be time to'pemiit rescue of trapped occupants through emergency escape hatches or doors.
I t is common for elevators serving more than three floors to dsscend autoniatically to the ground floor in the e\;ent of a fire. Fire fighters have keys to control elevators rnaiiually during building evacuation and fire fighting. However, smoke infiltration into hoistways frequently threatens lives and hinders use of elevators by fire fighters. In addition, there are three other concerns. First, water from sprinklers or fire hoses could short out or cause other problenls with electrical power and control wiring for the elevator. Second, shah pressurization cocld result in elevator doors jammhg open, limiting movement of the car. Third, piston effict could pull smoke into the elevator lobby or thc hoistway, and a method of preventing this has already been presented in this chapter.

Elevator lobbies, hoistway, a i d elevator machinery room must be protected against fire and smoke. Elevator equipment and electrical power must be protected from the water exposure of sprinklers and fire hoses. Elevator machine room must be protected from overheating. Reliable electric power must be supplied. In areas of high seismic activity, elevator equipment must be protected from earthquakes. The likelihood that elevators will be available during fires needs to be ensured by use ofmultiple cars or by quick response maintenance contracts. Elevator control must ensure safe and efficient evacuation. Communications capabilities are needed between people waiting for elevators and the fire service or appropriate building personnel. Evacuation capacity of the elevator system must be adequate for the number of people intended to use the system. As previously stated, elevator cars are controlled so that they go to the ground floor in the event of a fire alarm. In the event of fire on the ground floor, the elevator cars go to an alternate floor. The fire department or other authorized personnel can then use the elevators for evacuation. Firefighters, police, and uniformed guards have positions of authority in our society. With the elevators controlled by such authority figures, the likelihood of a large number of people crowding into the elevator and making it impossible to close the doors will probably be reduced. Of course, there may be other approaches to elevator control that could allow orderly evacuation by elevators. Reliability of electric power consists of ensuring a source of power and ensuring continued distribution of power to where it is used. Considerable experience exists in ensuring the supply of electrical power for critical functions in hospitals, communication facilities, computer facilities, and the like. For these applications, a major concern is providing backuppower when power supplied by the local utility is interrupted. These applications operate most or all of the time, and they need highly reliable power for all the time that they operate. Fire evacuation by elevators is different in that this mode of elevator operation is only needed during a building fire. At most, the fire evacuation mode of an elevator would be expected to operate for a few hours per year. Thus, the probability of simultaneously having a fire and having the utility company's power interrupted is relatively small. However, the probability of having a power distribution failure during a fire is relatively high. This is because lire frequently damages

.

Conceptual Solution for Elevator Evacuation
The feasibility of elevator evacuation for office buildings and air traffic control towers is discussed by Klote et al. ( 1992. 1994). In order to overcome the concerns discussed in the preceding section, an elevator system used as a lire exit needs to have the following

S. This c'dition ol'the L i f i So/./,r Code \\as the last cditioii to list rhcsc "problen~s."

Principles of Smoke Management

electrical distribution within- buildings. Therefore, the power distribution to the elevator and associated smoke control fans should be such that it is highly unlikely that a fire could interrupt electrical power to this equipment. T h x e are numerous applications of electric power and electronic systems being designed and built to function when in contact with water. Street lighting and traffic lights operate during rain, -and swimming pool lighting operates underwater. In fact, some elevators operate on building exteriors where they are subjected to rain and the other elements. It is beyond the scope o f this manual to examine specific approaches to making these systems resistant to water; however, it is obvious that the technology exists to make elevator systems function when they are subjected to water. Considerable information is available concerning the fire resistance of walls, partitions, floors, doors, etc. The ability to design and build elevator lobbies and hoistways-that can withstand severe building fire has existed for years. Smoke protection for elevator systems is the topic of the next section. Elevator doors jam open wl:x the force of the door opener is insufficient to overcome the force .of friction. The friction force increases with the pressure difference from the hoistway to the lobby. In tall buildings, elevator doors frequently jam open during extremely cold weather. This is caused by stack effect induced pressure differences. Elevator mechanics commonly adjust the door-closing forces to prevent door jamming. During elevator smoke control operation. the possibilih of door e jamming may decrease or increase. If the leaka,m area of the elevator lobby doors is less than that of the elevator doors, the pressure difference across the elevator doors can be less than that normally occurring. In field

tests conducted by Klote (1 984), no door jamming was encountered at pressure differences as high as 0.3 in. H 2 0 (75 Pa). When door jamming was encountered in an elevator without smoke control, it was found that only a small additional force applied by the palms of the hands was sufficient to prevent jamming. Fire fighters can be taught to overcome door jamming this way, and elevator doors could be fitted with grips or handles to aid in this effort.

Smoke Control Considerations
Smoke control systems for elevator evacuation n u t provide smoke protection for elevator lobbies, hoistways, and machinery rooms. Protection of lobbies is essential so that people will have a safe place to wait for the elevator. Protection of the machinery room is important to prevent damage to elevator machinery. Figlire 11.6 illustrates a system that pressurizes the hoistway directly and indirectly pressurizes the elevator lobby and the machinery room. As stated for other pressurization systems, the flow rate of air is highly dependant on the leaka,oe area. Because these areas can only be roughly estimated in most situations, the fan needs to be sized conservatively so that the fan flow can be adjusted to acceptable levels of pressurization during system commissioning. This fan sizing can be by choice of high values of building leakage or of safety factors. Pressurization air can also be supplied to the elevator lobbies. Examination of the relative leakage areas of the elevator system provides insight into both hoistway and lobby approaches to pressurization. Considering the leakage from the elevator lobby to the outside to be negligible,

Machinev Room , ,Lobby

I

Machinery /Lobby Room
/
P

I

%

$ .I

Car

11: '
,

- I
Fan

3: 4I
5

Building Space

Pit

, ; 4.. ,., ':.:. ; ' , ,

,,;/ ,<', ,,y A Note: The supply duct on the.floor needs to be in a fire rated enclosure.

(a) Shaft Pressuriiation

(b) Lobby Pressurization

Chapter 11 -Elevator smoke Control

where &sr
= =

pressure difference from hoistway to lobby, in. H20 (Pal; pressure difference from lobby to building, in. H20 (Pal;

&. 11A ,

= leakage area between the building and the lobby

? in f (m2);
A. ,
=

leakage area between the lobby and the shaft, ft2

(m2). For elevator doors with wide gaps that are common in most buildings, the gap areas are generally in the range of 0.34 to 0.72 ft2 (0.032 to 0.067 m2), as shown in Table 6.3,. Based on general experience with building leakages, Ai,/A, is about 0.4 for construction of average tightness and about 0.1 for tight construction. From Equation (1 1 .S),&&,.l is therefore 0.16 and 0.01 for Api;. average and tight construction. Thus, the pressure in the elevator lobby can be expected to be close to the pressure in the hoistway, provided that the construction is not unusually leaky. Pressurization air can be supplied to the elevator lobbics (Figure 11.6b). However, from the above discussion, it seems that this direct lobby pressurization does not resuh in any significant improvement in pressurization over supplying the air into the hoistway, as illustrated in Figure I I h a . Direct lobby pressurization has some advantage over direct hoistway pressurization in purging small amounts of smoke from the lobby. Part of the pressurization air to an elevator smoke control system goes from the hoistway to the outside, and the rest goes from the lobby through the building to the outside. With direct lobby pressurization, both of these amounts flow through the lobby. Such an increased tlow rate tends to better purge any small amounts of smoke that would get into the lobby-before smoke control activation or when a person is entering the lobby. The relative benefit of this improved purging con~paredto its cost has not been evaluated. The following discussions have been focused arbitrarily on the hoistway pressurization systems.

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 pressure fluctuation. . . During a fire, it is expected that several exterior doors will be propped open, and the elevator doors will open and close as elevators are used for evacuation. Further, stairwell doors are likely to be opened and closed as people use them for evacuation. It is envisioned that lobby doors will close automatically upon smoke control system activation. However, lobby doors can be inadvertently blocked and the closing mechanism can fail. It is anticipated that occupants will close any such opened lobby doors to prevent being exposed to smoke. Doors may not be closed on floors where there is no smoke danger or there are no people waiting in the elevator lobby. The smoke control system should be designed to maintain pressurization when some elevator lobby doors are open on floors away from the iire. The examples presented later deal with pressure fluctuations due to doors opening and closing.

Smoke Control Systems
Elevator smoke control systems can incorporate features to deal with pressure fluctuations due to opening and closing doors. An alternative is a pressurized hoistway without provisions for these fluctuations such that smoke transport through the hoistway is insignificant when evaluated by a hazard analysis. Features for dealing with pressure fluctuations include pressure relief vents, vents with barometric dampers, variable-supply air fans, fire floor venting, and fire floor exhaust.

Pressure Relief Vent System
This system has a constant supply air rate fan and a pressure relief vent to the outside, as illustrated in Figure 11.7. The area of this vent is fixed and sized for operation in the smoke control system. The vent can be fitted with automatic dampers if it is desired for it to be normally closed. The supply rate varies to some extent with the pressure across the fan, but the term "constant supply" is used to differentiate this fan from one that has a variable supply rate. The vent must be large enough that the maximum allowable pressure difference is not exceeded when all doors are closed. When paths to the outside are opened, air flows through then1 and the hoistway pressure drops. This system must maintain at least the minimum allowable pressure difference when some design combination of paths is open.

Pressure Fluctuations Due to Open Doors
Elevator systems must be designed to maintain design pressure differences under the likely conditions of opened and closed doors. Klote and Tamura (198Ga) showed that opening a large flow path from the pressurized spaces to the outside can result in a signiticant loss in pressurization. For example, opening the elevator doors, elevator lobby doors. and exterior doors resulted

Baromekic Damper System
This system is siniilar to the one above except that the vent has a barometric dampcr that closes when the pressure drops bcloiv a specified value. The use of these 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 relief vent.
hoistway are opened, and the pressurization fan can be sized smaller than for the above system. A normally closed automatic damper in parallel with the barometric damper can prevent damper chatter caused by the wind.

Figure 11.8 Elevator smoke control with fire jloor exhaust.

System with Fire Floor Venting or Exhaust
Smoke venting and smoke exhaust of the fire floor can improve system performance. The venting or exhaust increases the pressure difference from the lobby to the fire floor. The vents can be exterior wall vents or nonpowered smoke shafts. Figure 11.8 shows a fan-duct system intended to exhaust the fire floor. Upon detection of fire or smoke, the damper opens on the fire floor and the exhaust fan is activated. The detection system must be configured to identify the fire floor.

Variable Supply Air System
Variable supply air can be achieved by using one of many fans commercially available for variable flow rate. Alternatively, a fan bypass arrangement of ducts and dampers can be used to vary the flow rate of supply 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. There are two approaches for use of the sensors. The airflow rate can be controlled by the average of all signals from the sensors or it can be controlled by the signal from the fire floor. Using the average of all the signals has the advantage that no information is required about where the fire is located. u s i n g the fire floor sensor signal requires intormation about the fire location. This information can come from smoke detectors, heat detectors, or sprinkler water flow indicators. Using the fire floor signal has the advantage that the system maintains a set pressure difference at this most critical location.

Design Analysis
There are many different approaches that can be taken to the design of the systems discussed above. The design of an elevator smoke control system includes the selection of a system for dealing with pressure fluctuations, determining appropriate values for leakage areas andother parameters, as well as calculating the performance of the smoke control system. The objective of the design analysis is to determine a flow rate of pressurization air that ~ v i l lresult in acceptable pressurization with a minimum and a maximum design number of large open paths from the hoistway to the outside. Example 11.6 illustrates design of a pressure relief system.

CHAPTER 12

Zoned Smoke Control
he stairwell systems and elevator systems discussed in previous chapters were primarily intended to prevent smoke infiltration into these shafts. However, smoke can flow through cracks in floors and through unpressurized shafts to damage property 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 movement within a building.

Open and Divided Spaces
Zoned smoke control works for smoke zones that are open spaces, such as open plan offices. Zoned smoke control is more complex for potential smoke zones that are divided into a number of separate spaies. Examples

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 produced by mechanical fans are used to limit the smoke spread to the zone in which the fire initiated. The concentration of smoke in this zone goes unchecked. Accordingly, in zoned smoke control systems, it is intended that occupants evacuate the smoke zone as soon as possible after fire detection. Frequently, each floor of a building is chosen to be a separate smoke control zone. However, a smoke control zone can consist of more than one floor, or a floor 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 sandwich."

Note: In these figures. the smoke zone is indicated by a minus sign and pressurized spaces are indicated by a plus sign. A smoke zone can consist of one floor as in (a) and (b) or of more than one floor as in (c). All the nonsmoke zones in a building may be pressurized as in (a). or only the nonsmoke zones adjacent to the smoke zone may be pressurized as in (b). (c) and (d). A smoke zone may be part of a floor as in (d)

(d)

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 conjunction with passivi smoke control (compartmentation). 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. Another example is a hotel where the corridors are the smoke zones, and the guest rooms are protected by cornpartmentation. Analysis of compartmentation smoke management is discussed in Chapter 9.

Figure 12.2 Flowpattern due to smoke venting.

SMOKE ZONE VENTING
vent& of smoke from the smoke zone is important because it. prevents significant overpressures due to thermal expansion of gases as a result of the fire. Venting can be accomplished in the folloning three ways: exterior wall vents, smoke shafts, and mechanical venting (or exhaust). When the first two methods of venting are used, it is essential that adjacent zones (or all nonsmoke zones) be pressurized in ordzr to maintain prsssure differences at the boundaries of the smoke zone. Mechanical exhaust by itself call result in sufficient pressure differences for smoke control. However, in the event of window breakage or a large opening to the outside from the smoke zone, n~echanicalexhaust might not be able to ensure favorable pressure differences. Smoke purging, consisting of equal air supply and exhaust rates, is not considered here because it does not produce pressure differences that control smoke movement. It is generally believed that such purging at the airflows available with HVAC systems cannot significantly reduce smoke concentrations in a zone where a large tire is located. Dilution away from the fire is discussed in Chapter 6.

all of the exterior walls. For buildings that are much longer than wide, the vents can evenly be divided between the two long sides. Exterior wall venting is most appropriate for buildings with open floor plans and least suitable when the floor plan is divided into man!. compartments. Because the flow of hot gases through a wall vent can be substantial, precautions should be taken in the design of exterior walls to minimize the possibility of exterior fire spread to floors above the vent.

Vent Areas
The following is a method for evaluating the size of exterior wall vents presented in essentially the sams form as originally developed by Tamura (1978a). In this analysis, each floor consists of a smoke zone. For the analyses presented in this and the following section, ths effects of fire are indirectly incorporated in the selection of minimum design pressure difference (Chapter 6). For this system, the fire floor (smoke zone) is vented to thc outside, supply and exhaust fans serving the fire floor are shut off, and the floors above and below the firc floor are pressurized. Air flows from floors adjacent to the fire iloor and through the vent to the outside, as illustrated in Figurs 12.2. Because the supply and exhaust fans are shut off on the fire floor, the total airflow rate through the wall vents equals the total flow rate into the vented floor from the surrounding smoke control zones:

EXTERIOR WALL VENTS
Exterior wall vents can consist of n.indo\vs or panels that open automatically when the smoke control system is activated. The system considered here consists of a ventcd smoke zone without any mechanical exhaust and zdjacent zoncs that are pressurized. In order to minimize adverse efkcts of wind, the area of wall vents should be evenly distribirted among

where
A,, =
=

flow area of the exterior vent, ft2 (m'); etrective flow area of the enclosure of the smoke .. zone to the other zones, ft2 (m'); smoke zone pressure, in. H20 (Pa); outside pressure, in. H 2 0 (Pa);

A,

p~ = p.
=

Principles of Smoke Management

pB = building pressure on nonsmoke zcnes, in. H20

(W. 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
\

Substituting the above into Equation (12.2) and rearranging yields

where ApBo = ApBF
ApFo
=

pressure difference fi-om rhe nonsmoke zones to the outside, in. H20 (Pa): pressure difference from the nonsmoke zones to the smoke zone, in. H20 (Pa); pressure difference from rhe smoke zone to the outside, in. H20 (Pa).

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 exercise. Opening doors in a stairwell on both a nonsmoke zone floor and the smoke zone floor results in considerable airflow to the smoke zone, which is accompanied by reduced pressure difference across the boundary of the smoke zone.

11

Esample 12.1 Vent Areas and Pressure Differences I . If the ratio of A,/& is 1, what is the ratio of
Q g ~ / A p g o?

=

Then

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

l
I

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

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,
11

=

etfective flow area of the enclosure of the smoke zone to the other zones, ft2(m2); number of adjacent nonsmoke zones;

=

Chapter 12-Zoned Smoke Control

ABFi

=

flow area between nonsmoke zone i and the (m2).

smoke zone, f? Considering steady flow conditions, the mass flow rate of pressurization air entering a nonsmoke zone equals the flow rate of air leaving the zone:
tn,;

= rizBFi

+ rilBOi

( 12.5)

For an unsprinklered fir.e, the gases leaving the smoke zone are likely to be relatively hot. However, the flows in question are both from the nonsmoke zones, which are probably near building temperature. Considering the very approximate nature of flow area estimates, the errors involved in using volumetric flow rates at standard conditions are not significant. Such equations are

where
mti

=

mass flow rate of pressurization air into zone

i, Ibis (kg$);
mBFi

where

=

mass flow rate from zone i to the smoke zone, lb/s (kg/s);

vBFi = volumetric flow rate from zone i to the smoke
zone, cfm (m3/s); ABF;
=

z B 0

=

mass flow rate from zone i to the outside, Ib/s

flow area between nonsmoke zone i and the
'

(kds). The flow rate from zone i to the smoke zone can be expressed in the fomi of the orifice equation,

smoke zone, ft2 (m2); QBF pressure difference from the nonsmoke zones to = the smoke zone, in. H20 (Pa);
Kj.
=

coefficient, 2610 (0.839);

and where

'GFi- mass flow rate from zone i to tlie smoke zone,
Ibis (kds);
C
A ~

where

=
= ~

dimensionless flow coefficient; flow area between nonsmoke zone i and the i smoke zone, ft2 (m2); density of air in flow path, lb/ft3 (kglm3); pressure difference from the nonsmoke zones to tlie smoke zone, in. H 2 0 (Pa); coefficient, 12.9 (1.00).

ifBOi

=

volunietric flow rate from zone i to the outside,

cfm (n13/s); ABOi = flow area between nonsmoke zone i and the
=

P

=

WBF=
K,,,
=

outside, ft2 (m'); pressure difference froni the nonsmoke zones to the outside, in. H 2 0 (Pa); coefficient, 2610 (0.839).

=

Similarly, tlie mass flow rate to the outside is

The pressure difference from the nonsrnoke zones can be obtained from Equation (12.3) as

where
IilBOi

-

mass flow rate from zone i to t11eoutside, Ibis (kds); dimensionless flow coefficient; flow area between nonsmoke zone i and the outside, ft 2 (m2)

where
qjBo

=

C

=

pressure differencefrom the nonsmoke zones to the outside, in HzO(Pa); pressure ditkrence from the nonsmoke zones to the smokc zone, in. H 2 0 (Pa); tlow arcn of the exterior vent ofthe fire floor, ft2 (111~); efl'cctive flow area of the enclosure of the snioks zonc to the other zones. ft2 (ni 2).

ABO~ ~ 0 =

MsF
A,.
4 ,

=

P
4

density of air in flow path, lb/ft3 (kglm3); pressure difference from the nonsmoke zones to tlie outside, in. H 2 0 (Pa); coeflicient, 12.9 (1.00).

=

=

K,,,

=

.

.

Principles of Smoke Management

Example 12.2 Supply Air and Exterior Wall Vents

MECHANICAL EXHAUST
Mechanical exhaust 'of the smoke zone can be accomplished by either a dedicated exhaust system or by the exhaust fans of the HVAC system. Generally, such exhaust is done in conjunction with pressurization of nonsmoke zones. These systems can also include stairwell pressurization. Mechanical exhaust by itself can result in sufficient pressure differences to control smoke. However, in the event of window breakage or another large opening to the outside from the smoke zone, the pressure differences can decrease significantly. For this reason, mechanical exhaust alone does not constitute an adequate smoke control system when there is a significant probability of window breakage or an opening from the smoke zone to the outside. In the smoke zone, the location of the exhaust inlets is important. These inlets should be located away from exit stairs so that smoke in the vicinity of the shaft inlet does not pose an increased hazard during evacuation or fire fighting. Because hot smoke frequently stratifies near the ceiling, it is recommended that smoke exhaust inlets be located in or near the ceiling. Exhausting air from the smoke zone results in air from the outside and from other zones being pulled into the smoke zone. This air flowing into the smoke zone can provide oxygen to the fire. Most commercial airconditioning systems are capable of moving about four to six air changes per hour, which probably accounts for the popularity of six air changes in smoke control applications. Current designs are based on the assumption that the adverse effect of supplying oxygen at six air changes per hour is insignificant in comparison with the benefit of maintaining tenable conditions in zones away from the fire. Thus, six air changes is recommended as the upper iimit for exhaust airflow. In any analysis of a smoke control system, the fire effects of buoyancy and expansion need to be addressed. This can be done directly as part of the analysis or indirectly. As discussed in Chapter 4, the indirect approach consists of establishing a minimum design pressure difference that will not be overcome by buoyancy pressures resultilig from smoke at design temperatures. This indirect approach is much simpler, and so human errors in analysis, other aspects of design, construction, and commissioning are less likely. The following sections present both methods. When the temperatures on both sides of the bound-. ary of the smoke zone are the same, the pressure difference across the boundary is the same over the height of the barrier. This is the condition under which smoke

The smoke zone of a zoned smoke control system is to have exterior wall vents and two adjacent nonsmoke zones. Supply and return are shut off to the smoke zone, and the adjacent zones are pressurized. The nonsmoke zones have the same flow areas: ABol = ABO2 = 4.5 f (0.42 m2) and ABFl = ABR = 3.0 : (0.28 m2). Use AJA, = 2, and ApBF = 0.10 in. H20 (25 Pa). How much pressurization air is needed? From Equation (12.4), 4 = 3.0 + 3.0 = 6.0 ft2 (0.56 m2). From Equation (12.10), ApBo H20(3 1 Pa). From Equation (12.8), (1.2 m3/s).
=

_

0.10((1+2~)/(2~))0.13 in =

GFI= 26!0(3)(0.1)% = 2500 c h
=

From Equation (12.9), = 2610(4.5)(0.13)" cfm (2.0 m3/s). The supply air is 2500 + 4200 = 6700 cfm (3260 Us).

4200

SMOKE SHARS
A smoke shaft is a vertical shaft intended to be a path for smoke movement from the fire floor to above the level of the roof. Generally, the driving force of smoke movement is buoyancy, although the flow through some smoke shafts is aided by mechanical fans. This mechanical exhaust is addressed in the next section. A smoke shaft can serve one floor, a group of floors, or all the floors in a building. Smoke shafts have openings above the roof level and on the floors they serve. These openings are fitted with dampers that are nomially closed. In a fire situation, only the damper on the fire floor and the top outside damper open to vent smoke outside. Smoke shafts should be constructed in accordance with local codes. Tamura and Shaw (1973) provide inforniation concerning sizing of smoke shafts. Smoke shafts used in conjunction with pressurization of nonsmoke zones can produce pressure differences to restrict smoke to the smoke zone. Smoke shafts lend themselves to use in buildings with open floor plans. The air movement caused by smoke shafts operating during normal siack effect tends to pull smoke toward the smoke shaft inlet on the fire floor. It is recommended that smoke shafts be located as far as possible from exit stairwells, so that smoke in the vicinity of the shaft inlet does not pose an increased hazard during evacuation or fire fighting. Because hot smoke frequently stratifies near the ceiling, it is recommended that snioke shaft inlets be located in or near the ceiling.

Chapter 12-Zoned Smoke Control

Exhaust Fan Temperature
The mass flow through the exhaust fan is

where
=
me

mass flow rate of exhaust fan, Ibls &$S); density of gases in exhaust fan, 1b1@ (kg/m3);

p&
(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 system 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 reduction in smoke control system pressurization. The maximum allowable fan temperature can be calculated as

where

qa,=
T,.

absolute temperature of the gases in the exhaust fan, "R (K); = 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 oases in the snioke zone are '%ot," the buoyancy of the . , hot gases results in a nonuniform pressurz difference. Figure 12.4a is a uniform pressure difference at the niinimuni design value. This minimum value is selected such that positive pressurization of the snioke zone continues, pro\.ided that the mass flow from nonsmoke zones to the srnoke zone remains unchanged and that the snioke zone does not exceed its design temperature (Figure 12.4b). However, if this mass flow decreases. srnoke may flow into the "protected" spaces, as illustrated in Figurc 1 2 . 4 ~ . method of anal\.sis presented Thc i n the follon'ing scction allows evaluation ol' this decreased Inass ilow rate.

Example 12.3 Fan Temperature If a reduction of 20% in the mass flow rate is acceptable, what is the maximum allowable fan temperature?
I

The paraiiieters are: T,. = 70 "F + 460 = 530 "R, b= 0.2. From Equation (1 2.13), = 5301(1 - 0.2) = 663 O or R

G,,,

From Example 12.3, a 20% reduction in mass flow through the fan occurs at 203°F (95OC). When man!. I-IVAC systems are used for smoke control, they exhaust air from all or niost of the rooms on a floor. Thus, hot tire gases and lower temperature air from remote rooms are mixed, and the fan temperature is much lower than that of the firc gases. Also, heat transfer. from the exhaust duct lowers the fan temperature.

Principles of Smoke Management Table 12-1: Typical Gas Temperatures and Densities for Severe Building Fires
Temperature
Fire space is a room or corridor h l l y involved in fire Communicatirrgspace is a room or other space connected to the fire space by an open door or other large opening Removedspace is a room or other space connected to a communicating space by an open door or other large opening. The removed space is not connected to the fire space or is only connected to it by very small cracks or gaps Separated space is a room or other space not connected to any ofthe three spaces above, or it is only connected to them by very small cracks or gaps

~ ~ ' 7 ' ~
Density

"F 1700
800

"C 927
427

lblfe 0.01 84
0.0315

kg/m3 0.294 0.504

A fire space is a room or a corridor that is filly involved in fire. A communicating space is one that is connected to the fire space by an open door or other large opening. A removed space is a room or other space connected to a communicating space by an open door or other large opening. The removed space is not connected to the fire space or is only connected to it by very small cracks or gaps. A separated space is a space that is not connected to any of the three spaces above, or it is only connected to them by very small cracks or gaps.

400

204

0.0462

0.739

80

27

0.0736

1.18

To determine the extent of each type of space, a floor plan should be evaluated in light of likely locations of fires, doors likely to be opened, and doors likely to be Example 12.4, it can be seen that cool air closed. ~ r o m from the separated spaces mixes with the hot gases from other spaces and cools them. If the fan temperature is too high, the zone can be increased in size so that air from separated spaces will further dilute the hot gases.
Example 12.4 Fan Temperature and Smoke Control Zone Size A smoke control system has exhaust rates from the following spaces: cfrn Fire space: 400 0.189 0.378 Communicating S00 space: l SO0 0.850 Removed space: 6000 2.83 Separated space: 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.

The temperature of the gases in the fan can be conservatively estimated by considering dilution of hot gases with cooler gases and neglecting heat transfer. Considering constant specific heat, the fan temperature can be expressed as

where

Tbr7 temperature of the gases in the exhaust fan, "F =
("C);
pi
V
=
=

density of gases in space j, lblfi? (kg/m3); volumetric flow rate of exhamt from space j, cfm temperature of gases in space j, "F ("C);
From Equation (l 2.14).

=

= number ofspaces. Table 12.1 lists typical gas temperatures and densities for severe building fires that can be used in Equation (1 2.14). The following idealized types of spaces are addressed in this table:

n

C
Tfan

=

' ,,

'= I

- 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
In many buildings, the HVAC system serves many zones, as illustrated in Figure 12.5. For such a system, smoke control is achieved by the following sequence upon fire detection: The smoke damper in the supply duct to the smoke zone is closed. The smoke dampers in the return duct to nonsmoke zones are closed. If the system has a return air damper, it is closed. Precautions must be taken to minimize the probability of smoke feedback into the supply air system. Exhaust air outlets must be located away from outside air intakes. To conserve energy, most HVAC systems in modem commercial buildings have the capability of

recirculatidg air within building spaces. During normal HVAC operation, the return damper .is completely or partially open to allow air from building spaces to be mixed with outside air. This mixture is conditioned and supplied to building spaces to maintain desired temperature and humidity. This process is shown in Figure 12.6. During smoke control operation, the return damper must be tightly closed to prevent smoke feedback into the supply air, as is illustrated in Figure 12.7. As discussed in Chapter 7, smoke dampers are supplied in several leakage classifications. The particular class of damper specified should be selected based on the requirements of the application. For example, the dampers in the supply and return ducts shown in Figure 12.5 can have some leakage without adversely affecting smoke 'control system performance. Thus, a designer
k.::.'..::

~xhaust Air2

Return Duct\' \

-I
A -

Mechanical Penthouse

:--

L

Outside Air

L..'....

4::::.-. ,:
c<

Mechanical Penthouse

Stside :Air

-T-,Supply

I - I --+

-T/i
&

Duct

LT

Smoke

,/Damper

-

- .
C

lP
L

p am er

_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

~rsing HVAC qare17i tliat selves a11

Exhaust Air

-

Return Fan
/

/
1'

I
I

I-{

From Return

Supply Fan

.

T , .

Outside ~ir-

Equipment: Filters, Heating & Cooling Coils, etc.

Figurc 12.6 HKAC .v~rrerri~c'itlr ~.ecit~crrlario~i capahilir~ tlrc rio1-17~al iri HKAC lllotic.

Principles of Smoke Management

Return Fan

Outside
~ir-

c /

Heating & Cooling

Figure 12.7 HIQC syslenz with r.ecirculation capability i7 the sinoke control t mode.
might select class 1 , 111, or IV smoke dampers for such 1 an app!ication. Further, a designer might choose class I dampers for applications that require a very tight damper (for example, the return damper illustrated in Figures 12.6 and 12.7). Some designers have eliminated the smoke dampers from the return air system in the mistaken belief that the resulting system would still be effectke. This idea consists of shutting a smoke damper in the supply to the smoke zone and relying on the return air beins pulled from the zone to produce a significant pressure difference. However, shutting the supply to the smoke zone lowers the pressure there and. for these supply-damperonly systems, the return airflo\v from the smoke zone is also reduced. Field tests on such systems sponsored by the U.S. Veterans Administration have indicated that these supply-damper-only systems produce insignificant pressure differences (Klote 1986). Thus supply-damperonly systems are not recommended. In a fire situation, these small pressure differences can be overcome by buoyancy of hot smoke, stack effect, or other nomially occurring building airflows. Figure 12.8 illustrates the failure of a supply-damper-only system to control smoke movement with resulting smoke flo\v to the floor above the fire floor due to buoyancy or stack effect. For systems where the HVAC system serves only one smoke control zone, smoke contrd can be achieved by putting the KVAC systems in t!le modes below.
Srrioke Zorze: return fan-on, supply fan off, return damper closed, and exllaust damper open (optionally the outside air damper may bc closed). A1onsrriokeZoric: return fan 011; suppl!- f i n on, return damper closcd, and outsidc air dnmpcr open (optionally tllc exhaust air dmlpel. Ins! bc closed).

:I?;.:.<
.i

...................:.*- . . . ....... . . . . . . . ... ........ ..

Mechanical Penthouse ....................

Outside
...

Return Duct \ ...
' \

-

I

-7A

Smoke /Darnper

-C

Supply-Damper Only System -- CAUTION --THIS SYSTEM MAY NOT WORK AND SHOULD NOT BE USED. Notes: 1. This system is not recommended because it generally does not achieve satisfactory pressure differences to control smoke movement. 1. For simplicity, distribution ducts on each floor and equipment in the penthouse are not shown. Figure 12.8 Sche~naticof Jailwe /o achieve sinoke co/i/rolL only shutting a smoke damper y in /he supply duct lo the snioke z o ~ e .

This kind of system was tested at two new Veterans Administration hospitals (Klote 1986), where each floor of each wing was a smoke control zone supplied by a separate HVAC system. This performed well, was especially simple, and required no expensive dedicated equipment.

-

ANALYTICAL APPROACHES
Most zoned smoke control systems are in buildings with a numbcr of floors, shafts, and rooms per floor. As with other smoke control systems, the flows at the

Chapter 13 -Fundamental Concepts for Atria

Mass Flow with Virtual Origin Correction
Heskestad's equation for the mass flow of an axisymmetric plume is
tiz

Q

= =
=

heat release of the fire, Btuls @W); diameter of fire, ft (m); 0.278 (0.083).

D/
C ,,

. 113 5/3 2/3 = C,, QC (Z - L,) [ l + cU2Qc (z - Z J - ~ ' ~ ] (13.1) , , for ( z 2 z l )

In Figure 13.1, the virtual origin is shown above the 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 (kds); convective heat release rate offire. B d s (kW); height above fuel, ft (m); virtual orisin correction of the axis~mimetric plume, fi (m); mean tlame height, ft (m); 0.022 (0.07 1 ); 0.19 (0.026); 0.0 126 (0.0051); 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 being smoke. It follows that these equations can be thought of as equations for the production of smoke from a tire. Simplitied axisymmetric plume mass equations will be presented later, and the same comments also apply to them. A condition of the axisym~netric plums is that the tire is circular, and many experimental studks have used liquid pool fil-es in round pans. However, a tire need not be round for the axisymmetric plume equations to be useful. The effective fire diameter can be expressed as
D,- = 2 J z

where X, is the convective fraction of heat release as discussed in Chapter 2. The convective fraction varies fiom about 0.15 to 0.9, and using a value of G = 0.7 is common for most design applications.

Flame Height
Equations (13.1) and (13.2) are dependant on the mean flame height of the fire. The flanie height depends on the fire geometry, the ambient conditions, the heat of combustion, and the stoichiometric ratio. A relationship (Heskestad 2002) for flame height that can be used for many fuels is

where

z,

= =

mean flame height, ft (m); 0.788 (0.235).

C,,

This flame height is the elevation where tlie niaximum plume temperature is 900°R (500 K). The ceiling heights of atria are relatively higli, and it is tlie nature of atria smoke management that the elevations, z, of interest are much greater than either virtual origin, zo, or the flanie height, zl. 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 section, and the plume temperature is greatest at the centerline of the plume, as shown in Figure 13.2. The centerline temperature is of interest in the unusual cases when atria are tested by real fires. The centerline temperature equation (Heskestad 1986) is

, r

: . : .. .

(13.3)

where A is tl~c of the fire. At some distance above the area f?e, the plume li-om fire that is not round will bz nearly the same as that from a round fire. Virtual Origin Heskestad's (1953) relationship for the \.irtual origin is
S,,

= C , jQ-

. l /j

-

I .OZ D,.

( 1 3.4)

Principles of Smoke ~ a n a ~ e r n e i t

D -L P - CPD
where

(13.9)

D,,
z

= =

diameter of visible axisymmetric plume, ft (m); height above fuel, ft (m); plume diameter coefficient ranging from 2 to 4.

CpD =

Y

Radial Distance
Figure 13.2

~adiai Distance

This equation indicates that the axisymmetric plume diameter can vary significantly, and it is suggested that the value of CpDbe chosen s o that the results of calculations are conservative.

Radial temperature variation of axisymmeb-ic plume.

AXISYMMETRIC PLUME WITHOUT z ,
Axisymmetric plume equations that neglect the virtual origin are often used for atria applications because z is much greater than z,. The siniplified equations are listed below, and the consequences of this simplification are discussed later.

absolutecenterlineaxisynllnetricplume temperature at elevation z, "R (K); absolute ambient temperature, O (K); R density of ambient air, l b / g (kg/m3); acceleration of gravity, WS' (nds2);

Mass Flow:
riz =

. I ,'3 5/3 ColQC + c , ~ Q ~ f o r r r z , I

(13.10)

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): convectijt heat release rate of fire, B t d s (kW); height above fuel, ft (m); mean flame height, ft (m); 0.022 (0.071); 0.0042 (0.00 18); 0.020s (0.032).

Q,
where

=

=

Tp To

=

centerline axisymlnetric plume ten-iperature at elevation z , O ('C); F ambient temperature, O ('C); F 338 (25).

I:

=

C,, C, C,,,

= = =

= =

C, ,

Plume Diameter
Coosidering that a fire plcnie is composed of eddies, determination of the plume diameter is difficult. The plume diameter has been based on both visual observations and temperatures. One definition of this diameter is that it is the position at which the plume temperature has decreased to some fraction of the centerline temperature. The following equation is a way of expressing the expected range of diameter (Figure 13.1 ) of an axisymmetric plume:

It should be noted that when 1 is less than z l , the condition of z being much greater than zo may not be met. However, the separate equation above for z < z, is included tbr conlpleteness.

Flame Height:

where 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 into Equation (13.3) results in the following equation for the effective diameter:

where

Tcp To
p,
g

=

absolute centerline axisymmetric plume temperature at elevation z, OR (K); absolute ambient temperature,
O R

=

(K);
Table 2.2 lists heat release densities for some warehouse materials and pool fires. In this table, q ranges f from 8 S t d s ? (90 kw/m2) to 1250 Btds ft2 (14,000 kw/m2). The low value is for a proprietary silicone transformer fluid, and the upper value is for polystyrene jars in compartmented cartons stacked 15 ft (4.57 m) 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). Figure 13.3 shows the effect of heat release density, g , on the location of the virtual origin. For 35 Btuls ft2 (400 kw/m2), zo is about -2.6 ft (-0.8 m) at i) of 1900 Btds (2000 kW) and -14 ft (-4.3 m) at Q of 24,000 Btds (25000 kW). The negative values of zo indicated that the virtual origin is below the fire surface. For 880 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 24,000 Btu/s (25000 kW). Figure 13.4 shows the impact of the virtual origin correction on plume mass flow for q = 35 Btuls ft2 (400 kw/m2) and y = 880 Btu/s ft2 (1 0,000 k ~ / m ' ) .Neglecting the virtual origin correction results in overprediction for q = 35 Btuls ft2 (400 kw/m2) and underprediction

=
= =

density of ambient air, lb/fi? (kg/m3); acceleration of gravity. ft/s2(m/s2); 0.0067 (9.1).

C,,

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 m/s2), and C p of 0.24 Btu/lb "F (I .OO kJ/kg "C), Equation (13.13) becomes

where
TV,
=

centerline axisymmetric plume temperature at elevatio!] z, "F ("C); ambient temperature, O F ("C); 338 (25).

To C,,

= =

Example 13.1 Simple Plume Calculations 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 the fucl'? Use a con\wtive fraction of0.7, and the ambient tempcrature is 72°F. From Equation (13.5). thc convecti\.e heat release rate is

DC =

= 0.7(4000) = 2800 Btds (2950 kW!.

From Equation (13.12). the mean flame height is
z l = 0.533&
.2/5

Heat Release Rate. i)(1000s kW)
0 5 10 15 20 25

= 0.533(2800)"-" = 12.8 ft (3.9 m).

Because 2, < z, the plume mass tlow is calculated from Equation (13.10):
NI = 0.022Qc z
. 1 /3 j/3
+

~.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, 117e wir11ud origitz.

011

Principles of Smoke Management

Elevation. z (m)

.
5-2

a000

-

- - - - - ~t q = a80 - - - ~t q = 35
None

Virtual Origin Correction:
B~UIS

ft' (1 0.000 kwtml)

- 4000

B~UIS

f? (400 kwlm')

-E
U

$

6000

d
4000

Heat Release Rate

-

S
2000

0 0 75 Elevation, z (ft) 225 300

.. .-

Figure 13.4

Comparison of nzass ~ I O Wpredictions u ~ t h and rvithozrt correction for virtual origin.

for q = SS0 Btuls ft2 (10,000 k ~ l m ' ) . These over- and underpredictions are with reference to Equation (13. I). An estimate of the uncertainty of Equation (1 3.1) is not available, b u ~ should be noted that the state of it 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 being a strong function of eleva~ion, and a weak funcz, tion of the convective heat release rate, Q,. However, 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 selection af the fire surface would be at the floor. Figure 13.5 compares the predicted flame heights from Equation (1 3.6) tvith the approximaLe relation of Equation (1 3.13). Again, the approsimate relation is in the middls of the range of predicted values. It is apparent that flame height, z l l increases with q. In atria smoke management design, flame height is primarily used to ensure that the plume mass flow equations are appropriate. The flame height, z l , rangcs from about 8 ft (2.4 m) to 14 ft (1.1 ) a 1000 Btuls (2 100 kW) and from about m L 14 f (4.3 m) to 39 ft (12 111) at 25,000 Btuls (26,000 i kW).

Heat Release Rate,

Q (1000s kW)

Heat Release Rate.

Q (1000s Btuls)

Figure 13.5

Cotr~pat-ison 117eatz i m e height u.itll of f and without virtzral origin correctiotl.

WALL AND CORNER PLUMES
A fire that is located next to a wall will entrain air over a smaller perimeter than a fire located far from a wall. The fire and wall plume may be considered half that of the axisymmetric plume (Figure 13.6). Thus, the mass flow rate of a wall plume is half that of an axis)mmetric plume, but the analogous fire for the axisymmetric plume is twice that of the "real" fire creating the \\.all plume. The mass flow rate of a wall plume can be estimated as follo\vs:

-

I.
2.

Let Q be twice that of the wall plume. Apply a plume equation to calculate the mass floi~.,

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 follows: 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 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 these plumes is less than that of the axisymmetric plume.

Figure 13.6

Wall pl~ime.

Fuel@,, ,,,

=

j

Fuelw

=

(a) Fire far from walls

(b) Fire at wall

BALCONY SPILL PLUMES

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):
i
=

C, (

l

7

)

l :3

(z,,

+ 0.25H)

(13.17)

(c) Fire at interior corner

(d) Fire at exterior

wlicrc
~ i r = mass tlow
=

corner

in p l u m at height zb, Ib!s (kgs);

Iicat release of thc lire, Btuls ( d ; : I\) plume width as it spills undcr balcony, ft (m); Iieiglit above balcony, ft (m); height of balcony above top of fuel, ft (m); 0.12 (0.36).

W
3. The smoke production of the wall plume is fir 12.

= = = =

q, H
C,,

This approach is very rough for heights in tlie range of the flame height, but the flow becomes more realistic for higher elevations, z. The approach can be applied to any axisymrnetric plume model, but for consistency with the information above, 1i1 would be calculated 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 become the ge17ernl 11.crllpllcnle ey~rcr/ion.

When draft curtains are used (Figure 13.S), the width, W, of tlie spill plume is the distance between the curtains. In the absence of draft curtains, the following approximation can be used.

W =

width ol'rhc opening liom the tire room, ft (m);

h

where 1 1 is a lire location factor that is sho\\w in F i y r e 13.7. For li~rthcr information about \w11 and corner plumes, sec Mowrcr and Willinnison (1987).

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~lg such that tlic inomenturn of the ceiling jet in tlic lire room Jocs not directly contribute to the flow out ol'thc ol~c~ling. thickness of the ceiling jet Tlic is in the range ol' 10% to 20% of the height from the 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 contributing factor when the top of the opening is not greater than 80% of the distance from the base of the fire to the ceiling. . For spill plumes not consistent with the conditions of Equation (13.18), scale modeling (Chapter 15), CFD modeling (Chapter 16), or other correlations can be used. While Morgan et al. (1999) is a source of other correlations, most of these are in terms of fire perimeter

rather than HRR. Those using the correlations of Morgan et al. will need to convert HRR to fire perimeter. Equation (l 3.17) is extensively used for design analysis, but there is controversy about the extent of its applicability. ASHRAE Technical Committee 5.6 is planning a research project consisting of large-scale fire experiments to check the applicability of Equation (13.17) and to develop information for some spill plumes not consistent 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? 500 Btds (528 kW) Heat release rate, Q 20ft(6.10 m) Height above balcony, zb

10 ft (3.05 m) Height of balcony above top of fuel, H 6 ft (1.83 m) Width of the opening from the tire room, W 12 ft (3.66 m) Distance from the opening to the balcony edge, b 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 Front View Without Draft Curtains

I

Section View

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:
fit

=

1/3 c a l Q c (2,

+a)

5/3

+

ca9~c

(13.2 1)
I

Inserting the same factor into Equation (13.16) yields a general equation that incorporates the presence of walls:

Figure 13.9

M'irdow plume.

where z is the height above the top of the window. Substi, tuting Equations (13.19) and (13.20) into Equation (13.22) results in

WINDOW PLUMES
A window plume is one that flows through an opening such as a window or door to a room with a fully involved fire (Figure 13.9). As described in Chapter 2, a fully involved fire is one where all, the combustible materials in the room are burning. The high intensity of such a fire explains why window plumes are not normally considered design fires in sprinklered buildings. In such a fire, fuel \.olatilized in the room would burn outside the opening. The heat release rate of a fully developed tire is constrained by the combustion air that can reach the lire. and such a fire is referred to as being ve~ililnfioti. cot~~rolleti. Accordingly, the size of the fire depcnds on the size and shape of the opening to the room and the material burning. Based on experin~ental fire data Ior \irood and polyurethane in a room with a single rectangular opening, the average heat release rate iS where C,+.2 = 0.077 (0.68); C,,,
=

0.18 (1 59).

Equations (13.22) and (13.23) apply to wall and corner plumes, and with i 7 = I , these equations become the same as the window plume equations. I t may be noticed that Equation (13.23) does not contain a heat release rate tenn, and this can be so because the fire is ventilation controlled such that the heat release rate depends on the ventilation opening. This analysis of window plumes is based on the assumptions concerning entrainment and the adaptability of the general wall plume model, but it has not been experimentally verified. 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 with the control volume shown in Figure 13.10. Neglecting the small amount of mass added to the plume flow due to combustion, the first law for the plcme is
Q ~ + = til(lte- 11;+ AKE

where

0

=

heat release of the fire, Btuls (kW):

A,,. = area of ventilation opcning, ft2 (m');

H!,. = height of ventilation opening, ft (m):

C,,.,

=

61.2 (1260).

0,

+ A P E ) + l~

((13.21)

Thc equations for tlle axisymmetric plume or wzll and corner plumes can be adapted for the window by plume. This is accon~plislied determining the entrainment ratc at the tip of the flames coming out of the opening and determining the height in the axisymmetric plume cquation that would yield the same entrainment. The height m m in the asisyrnmetric plume equation nceds to bc adjusted by the follo\ving factor:

where

Q~

=

heat generated within the control volume, Btuk (kW),;

Q,
tit

= heat transferred liom surroundings into the con-

trol volume. Blu/s (kW):
=

Inass flow rate. Ib/s (kds);

Principles of Smoke Managemen't

hi
h,

=

enthalpy of flow entering the control volume, Btu/s Ib (kWkg); enthalpy of flow leaving the control volume, Btu/s Ib (kWikg); change in kinetic energy, Btu/s.lb (kwikg); change in potential energy, Btuls Ib (kW/kg); work done by system on its surroundings, Btu/s where
z,,
=

=

AkE
ME
W

=
=

maximum height at which plume is considered buoyant, ft (m);

=

(kW). For the steady plume, the work is zero and the changes in kinetic and potential energy are negligible. The heat generated is the heat release .of the fire (Qg = Q ). Heat is transferred from the plume by conduction and radiation to the surroundings (Q, = -Q, where Q, is the radiated heat), so that (Qc = i), + Q ,). Specific heat can be considered constant (h = Cpr). The first law leads to an equation for the plume temperature.

QC = convective heat release rate, Btu/s (kW); ATnlit7 = minimum temperature rise of plume above ambient, "F ("C); c,,,,, = 189 (14); CznLr2= 19.1 (0.0254). Equation (13.26) applies to axisymmetric plumes. The idea of minimum temperature rise is that it is the smallest temperature rise at which the plume has sufficient buoyancy to continue to be a strongly buoyant 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. plume height for this Figure 13.1 I shows the maxi~num minimum temperature rise. For a total heat release rate of 500 Btuls (530 kW). the masiniuni plume height from Figure 13.11 is about 130 ft (40 m). For a total heat release rate of 2000 Btuls (2100 kW), the maximum plume height is about 220 ft (67 m). While the maximum plume height is not a concern for most designs, it needs to be considered for atria with high ceilings.

where Tp = average plume temperature at elevation z, "F ("C); To = ambient temperature, "F ("C); Cp = specific heat of plume gases, BtuAb "F (kJkg "C). Fire plumes consist primarily of air mixed with the products of combustion, and the specific heat of plume gases is generally taken to be the same as air [Cp = 0.24 Btullb "F (1.00 kJ/kg "C)]. Examplc 13.4 Average Plun~e Temperature
What is the average temperature of the plume in Example

i

at Elevation z w pl~i~plo

,

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 maximum plume height at \\:l~ichthe plume can be considered strongly buoyant.

Chapter 13 -Fundamental Concepts for Atria

5
N

-

Heat Release Rate. Q (1000s kW)
70oO.~~ 10
I

15
1

20
I

25
I

2 600 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.
I

E
135 'G

outside Air T, p,

A.. 4liL

m .-

E 500 -

I

a

.2
2

l00

-

25

45 0

.g
S

5

L
rni

Floor To,p,

I

0 0.5

l

I

I

5

10

15

20

-

Heat Release Rate.

(1000s Btuls)

Figure 13.11 Maximum plunze height.

(b) Pressure Profile of Smoke Layer Figure 13.12 GI-aviv smoke venting.

(a) Sketch of Gravity Smoke Venting

VOLUMETRIC FLOW
The volumetric flow of a plume is
where
p,. = reference density, 1bnl/ft3 (kghn3);

where
tit

T,
mass flow in plume at height z, Ibls (kgls); volumetric smoke flow at elevation z, cfnl (rn3/s); density ofplume gases at elevationz, lb/ft3 (kg/m3); GO(1).

= absolute reference temperature, O (OK). R

= = = =

V

There are an infinite number of possible p,; T,. pairs. and one that can be used for such calculations is p,. = 0.075 lbm/ft3 ( 1 2 0 k g h 3 ) , T,. = 530 O (294 K). R

PP C/,

CONFINED FLOW
As already noted, the diameter of a plume increases with height. For a tall narrow atrium, the plume may contact all of the atrium walls before the plunie reaches the ceiling. Where a plume contacts a wall, i t cannot entrain air. For smoke management purposes, the smoke layer interface should be considered the elevation where the smoke contacts all or most of the atrium walls.

AIR AND PLUME DENSITY
The density of air and plunie gases is calculated from the perfect gas law:

NATURAL VENTING
where

p

=

density of air or plume gases, lbm/f$ (kg/m3);

p = absolute pressure, lbf/ft2 (Pa); R = zas constant, ft Ibfllbm O ( J k g K); R T = absol~ite temperature, O (K). R The absolute pressure is often taken to be standard atmosplleric pressul-e of 21 16 lbflft' (101,325 Pa), and the gas constant is generally taken to be that of air, which is 53.3 ft IbVlb~n R (287 Jlkg K). O At most localions, atmospheric pressure call be considered constant for purposes of calculating air and smoke density. This means that p/R can be considered constanl. and dcrisily can h e n be calculated.

Natural vents consist of openings in the ceiling through which smoke flows due to buoyancy. The hot acts to force smoke out of smoke layer under the ceil~ng the vent and to pull makeup air through other openings into the atrium. The analysis of natural \,enting that follows is adapted from an analysis by Thomas et al. (1963), and it is illustrated in Figure 13.1 2. The ternperature in the atrium below the smoke layer is considered the same as that outside. As discussed in Chapter G, the mass f l o w out of the vent and in the inlet opening can be expressed by the orifice equalion as

Principles of Smoke Management
l

and
nzi = K , ~ c ~ A ~

A, = vent area, fi? (m2);

J

E

i

(13.3 1)

A~

=

inlet opening area, fi? (m2);

,

.

where
ril"

p, = outside air density, 1blft3(kg/m3);

=

mass flow rate through the vent, Ibls (kgts); mass flow rate through the inlet opening, Ibls (kg/ S); flow coefficient of the vent (dimensionless); flow coefficient ofthe inlet opening (dimensionless); area of the vent, f? (m2); i area of the inlet opening, fi2 (m2);

g db To

= =
=

acceleration of gravity, 32.2 ft/s2 (9.80 m, s2); depth of smoke layer below the smoke vent, R (m); absolute temperature of outside air, "R (K); absolute temperature of smoke, "R (K);

i

/izi =

c, c.
A, A;

=

T,

=

=

TRANSPORT TIME LAG
A plume takes time to rise to the ceiling, and a ceiling jet takes time to form a smoke layer under the ceiling (Figure 13.13). The idealized zone fire model 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
P, -P!, = K,,,gdh(~o - P,)

(a) Growing plume

(13.32)

where
y
= =

db

G=
=

acceleration of gravity, ft/s2(nds2); depth of smoke layer below the smoke vent, ft (14; 370 ( 1 .OO).
(b) Fully

The flow coetticients are considered to be equal ( C C,, = C;). The mass flow out the vent equals that through the inlet opening (/ill: = ii~,). smoke density The can be ex~ressed as
P,
=

developed plume and growing ceiling jet

...................................................... .. .. . . .. . .. . . . . . . .. .. .. .. .. .. . . . . .. .. .. .. .. . ....

................................

To PO-.
?Y

(13.33)

Combining Equations (13.30) through (1 3.33) yields
rir ,. =

CA,.P,,[~P/,,(T.~T<,)(T<,/TJl
7

I /2

l/?

(13.34) (c) Fully developed plume and ceiling jet extending under entire ceiling Figure 13.13 Develop~l~enf pl~rri~e ceiling jet. o f at7d

[T,. + (."/A;)- T,,l

where /ir,, = mass llow rate tl~roughthe vent, Ibls (kys);
C
=

discharge cocllicient (dimcnsionless);

Chapter 13- Fundamental Concepts for Atria

Ceiling Height (m)

tpl

= =

transport time lag of plume, s (S); growth time, S (S);

tg

H
Q

= height ofceiling above top of hel,
=

ft (m);

heat release rate, Btuls (kW);

Ceiling Height (ft)
Figure 13.14 Pllrn7e trampor? fag.
form smoke layer forms at the first instant that any smoke reaches the ceilinp. The zone fire models were developed for fire rooms such as ordina~ybedrooms, dining rooms, and recreation rooms. In such rooms, the errors resulting from these simplitications were insignificant. 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 ~mdcrcstimating atrium tilling time and detector activation time. For 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 will have less time to act than indicated by h c prcdiction.

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 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 rational for neglecting the plume lag.

Ceiling Jet Lag
Newman and Mon.rer also developed the follon.ing relationships for the time lag of ceiling jets resulting from steady and unsteady fires. Foi- a steady fire,

For a t-squared fire.

where
1 ,
1.

= transport time lag of ceiling jet, s (S);
=

Plume Lag
Nenmian (1 9S8) and Mowrer ( 1990) devclopcd relationships for the time lag of plumes I'rom s m d y and unsteady fires. Steady and unsrcady tires are discusxd in Chapter 2. Because il is h e nature of tires to g~-o\v. the time lag time for steady l i r a is probably less significant. For a steady fire,
C,,

radius or Ii~rizontal distawe from centerline of plume, fi (m); = 0. l GS (0.833)
=

C$

0.278 (0.72)

As with the plume lag, the t-squared fires ha\.s the greatest ceiling jet lag as shown in Figure 13.15. The ceiling jct of a fast t-squared fire with H = 20 fr (6 m ) takes about 25 s to spread out to a circle with a 25 i't 17.6 m) radius.

Total Transport Lag Time
For a t-squared tire. The total transpon lag time is [lie sun1 of the plume lag rime and thc ceiling jet lag time,

\vherc

whcrc I, is the tolal tralispon Iag rime (S).

Principles of Smoke Management
!

I
i

i l
i
l

PLUGHOLING
When the smoke layer depth below an exhaust inlet is relatively shallow, a high exhaust rate can lead to entrainment of cold air from the clear layer (Figure 13.16). This phenomenon is called plugholing.

maximum mass rate of exhaust without plugholing, lbls W s ) ; absolute temperature of the smoke layer, "R, (K) absolute ambient temperature, "R, (K) depth ofsmoke layer below bottom of exhaust inlet, %(m); exhaust location factor (dimensionless); 0.354 (3.13). In the context of plugholing, the smoke layer depth is always a distance from the smoke interface to the bottom 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 wall exhaust inlet near the ceiling, and 2.8 for a ceiling exhaust inlet far from any walls (Figure 13.18). It is suggested that d/D be greater than 2, where D is the diameter of the inlet. For exhaust inlets, use D = 2ab/(a + b),

I
I

Number of Exhaust Inlets
To prevent plugholing, more than one exhaust point may be needed. The maximum mass flow rate that can be effkiently extracted using a single exhaust inlet is given as [CIBSE 19951

where

Radial Distance (m)
0 100 5
i

10
I

15
I

20
I

"

0

25

50

75

Radial Distance (ft) Figure 13.15 Ceiling jet tra1lsp01-f lug.

Figure 13.17 Depth, d, ofsrnoke layer- beloiv bottorl~ ofe.dmtst inlet.

Ceiling

I 1
(a) Ceiling lnlet Away From Walls Ceiling Ceiling Inlet,

t 0=2.8

Inlet

1 p = 2.0

-+ lnlet

n

p = 2.0

(b) Wall lnlet The Ceiling

Near
:

(c) Ceiling lnlet The Wall

Near

Figure 13.16 Pl11gholi11g air- iuto smoke e.rhartst of i111etsc017 r-es~ilt /bilro.e oj' a snloke in esl~atist s~ste~rl.

Figure 13.18 Vulltes o f exhaltst location,fbcto/:p

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 number.

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 single inlet that would result in arbitrarily small velocity based on sink flow. Using 40 fpm as the arbitrary velocity, the minimum separation distance for inlets located in a wall near the ceiling (or in the ceiling near the wall) is

(b) Exhausting air witil smoke when the depth of

the smoke layer, d, is less than the minimum depth. d,, needed to prevent plugholing.

where S,,,il, = minimum separation between inlets, ft, (m);

i/,
(c) Exhausting only smoke when the depth of the smoke layer, d, is greater than the minimum depth. d,, needed to prevent plugholing. Figure 13.19 Smoke n17dnir exhazat conditions.
1 3

=

volumetric flow rate, cfin (m3/s);

= exhaust location factor (diniensionless)

Minimum Depth to Prevent Plugholing
The approach of the previous section to plugholing considered the number of exhaust inlets needed to prevent plugholing, but the issue can also be viewed as the smoke layer depth needed to prevent plugholing for a specific eshaust flow rats. The minimum smoke layer depth to prevent plugholing can be expressed as

where n and b are the length arid width of the inlet. The results of experiments conducted at the National Research Council of Capada are consistent with this 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 volumetric flow rate as

where

d,,,
,,C ,

=

minimum smoke layer depth to prevent plugholing, ft (m); = 0.78 (0.15).

wherc

v,,~,,,= maximil~n volumetric tlow rate at 7;, cfin (m3/
S);
C, ,! )
=

Satisfying Equation (13.44) is equivalent to meeting the criterion for the number of exhaust inlets of the previous section. This means that meeting either of these criteria is sufticient to ensure that the other criterion will be met.

0.537(0.00887).

Plugholing Flow
Conditions of air eshaust and smoke exhaust are illustrated in Figure 13.19. The mass flow ofcxhaust air

Considering cqual Ilows at each eshaust inlet, the numbcr. 1VjI,,,,,. ol'cshaust irilcts is

Principles of Smoke Management'

is the sum of the exhausted smoke and the exhausted air. For an exhaust inlet, the exhaust can be written as
me = nze0+ mes

13.20. Larger exponents result in predictions of lower smoke layers.

(13.45)

MINIMUM SMOKE LAYER DEPTH
The previous section addressed the smoke layer depth needed to prevent plugholing. This section addresses the minimum design smoke layer depth needed to accommodate the physical flow of smoke under the ceiling. Readers are cautioned about differences in the definitions of smoke layer depth. For plugholing applications, the smoke layer depth is the distance from the smoke layer interface to the bottom of the exhaust inlet. In this section, the smoke layer depth is the distance from the smoke layer interface to the ceiling. When a smoke pll!me reaches a flat ceiling, the smoke is deflected into a ceiling jet flowing radially from where the plume impacts the ceiling. As already stated, the depth of the ceiling jet ranges from about 10% to 20% of the distance from the base of the fuel to the ceiling. This is an oversin~plificationin that the ceiling jet is about 10%, but at the walls, the jet turns back and flows under itself. For an enclosed room or atria 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 design smoke layer depth be at least 20% of the distance from the base of the fire to the ceiling. For rooms and atria with an opening at the ceiling, the smoke of the ceiling jet flows out through the opening (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. For further information about ceiling jets, sec Beyler (I 986).

where
= total exhaust from an me, = lower me,

inlet, Ib/s (kgls);

layer air exhausted from an inlet, Ibls (kgls);

inlet, Ib/s (kgls). When the smoke layer depth is at or below the minimum smoke layer depth to prevent plugholing (d 2 d,,), the mass flows are i e= 0 and riles = o where li, is the total volumetric exhaust in ft3/s (m3/s) and p, is smoke density, lb/ft3 (kglm3). When the smoke layer depth is zero (d = O), the mass flows are ~iz,, = p,& and riles = 0 where p, is the ambient or lower layer air density, lb/ft3 (kg/m3). In the absence of a formally developed and experimentally verified theory of plugholing flow, the following relations are suggested:
i s= p

= smoke exhausted from an

d V( " -)
'"1

' 0

for 0 5 d < d,,!

(13.46)

rile = p

for d,,, 5 d

and

rile, = 0

for d,,, -< d

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 corresponding pressures are shown in Figure 13.23. These pressures are hydrostatic, that is, the dynamic pressure components are negligibly small. The opening 1 consids ered sufl-iciently large that the pressure below the smoke layer can be considered the same as that outside the opening. Further, the top of tlle opening is sufficiently below the ceiling that the ~nomentumof the ceiling jet has no impact on the smoke flow through the opening. The smoke Flow through tlie opening depends on tlie buoyancy of tlie smoke and the thickness of the smoke layer under tlle first level.

Atrium height: 40 fl(12.2 m) Atrium Areq: 10,000 ff(929 m') Exhaust: 120.000 c h (57 m'k) Exhaust Location Factor: 2 No, of Exhaust Inlets: 1 Heat Release Rate: 5000 B t u k
(5300 kw)

- l2

g - 11 .P
Z

;
z

-102 Y a
0

E

9 30 60 90 rime (S) 120 150 180

Figure 13.20 Effecr of' /lie pl~igl~oli17g e.vponenl smoke IO>TI. 11cigI71 sinrrilnled bv OS AZONE 111ocle1.

017

//W

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);
g
=

, Jet Ceiling ,

acceleration of gravity [approsin~ately ~ s (9.8 32 '

minimum smoke layer depth is about ::s::"~': 20% of the distance from the base of the ..... fuel to the ceiling. Fire (c) Formation of minimum smoke layer depth for enclosed room

M

ds2)1;
p, = ambient air density, lb/ft3(kgm3);
p, = smoke density, lb/ft3(kg/m3): C
=

flow coefficient (dimensionless - approximattl>. 0.7);

c/fl= I .oo (l). Equation (13.48) can also be witten as

Figure 13.21 Developrno~toj' tr~iniri71un sriroke layerin an enclosed root11or-atria.

where Cl@ = 3.74 (2.07)

(a) Sketch of horizontal flow
(b) Pressure profiles

Pressure

Figure 13.23 Hor~izor~tulflo~~~ a17 o/m~ing 1 w 7 / . t/ir.o~~gIi or.

Principles of Smoke Management

Airflow for Fire in Communicating Space
Atrium Space

Airflow can be used to prevent or mitigate smoke 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 from flowing to the atrium. As discussed in Chapter 6, Figure 13.24 Airflow used to prevenl or n~ifigafe to prevent such smoke migration, the average velocity in smoke originaling in a cornrnzlnicaling the opening must equal or exceed the limiting velocity space from tnigmting lo [he afriurn. (Heskestad 1989).
Smoke,

Substituting the ideal gas equation into Equation (13.49) and rearranging yields where

v
where

= =

limiting average air velocity, fpln (mls); acceleration of gravity, tVs2 (nl/s2);

T,
To

= =

smoke temperature, 'R (K); ambient temperature, OR (K);

g

H
Ts

= height of the opening, ft (m);
= =

absolute temperature of the fire space, OR (K); absolute ambient temperature, OR (K);

As stated above, this analysis applies when (I) the opening is considered sufficiently large that the pressure below the smoke layer can be considered the same as that outside the opening, and (2) the ~uomentuniof the ceiling jet has no inipact on the smoke flow through the opening. The comments concerning the momentum of the ceiling jet in the above section about the balcony spill plume also apply here. When these conditions are not met, the horizontal smoke flo\\. can be analyzed by 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 flows at openings between compartments of a zone model can b; complex, as is explained by Jones and Bodart ( 1 956).

r,

cf = al

38 (0.64).

Airflow for Fire in Atrium
Airfloxv also can be used to prevent smoke originating in the atrium from flowing into a com~nunicating space. The limiting air velocity can be calculated from

limiting average air velocity, fpm (~n/s); heat rclease rate, Btds (kW); distance above the base of the fire to the bottom of the opening, ft (m); 17 (0.057). Equation (13.52) is not applicable when z is less than 10 ft (3 m). Further, v, should not exceed 200 space fpm (I m/s). If the opening to the comn~unicating is above the smoke interface, the limiting air velocity sliould be calculated from Equation (13.51).

COMMUNICATING SPACES
Communicating spaces are spaces within a building that have an open pathway to an atrium such that smoke from a tire in the communicating space can move unimpeded into the atrium. Smoke from a tire in an atrium can also move unimpeded into the communicating space. Communicating spaces can open directly to the atrium or can be connected through othsr open spaces.

CHAPTER 14

Atrium Systems
t is well known that the ability of sprinklers to suppress fires in spaces with ceilings higher than 35 to 50 ft ( l l to 15 m) is limited. Because the temperature of smoke decreases as it rises (due to entrainment of 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 compartmentation and sprinklers for atriums, the importance of atrium smoke management is not surprising.

I

AZONE are on the CD that accompanies this book. 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.

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 occupants to smoke during evacuation except for the tenability 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 axisymmetric 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 equations or computer zone fire models. For general information about these computer models, see Chapter 8. Computer zone fil-e models CFAST, ASET-C, and

(a) Fire in atrium space producing an axisymmetric plume

(b) Fire in space open to atrium producing a balcony spill plume

Figure 14.1

Locatioti of fire cat7 deter-mine the k i d 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 filling equations and zone fir2 nrodels.

SMOKE FILLING
This approac!i consists of having occupants evacuate the atrium or through the atrium as smoke fills the atrium space (Figure 14.2). Smoke filling applies only to very large volume spaces where the filling time is sufficient for evacuation, including the time it takes to become aware of the fire and to prepare for movement to an exit. Chapter 4 addresses people movement and fire evacuation. Smoke filling calculations can be done by the computer zone fire models or by application of the empirical filling equations presented below. where
2

=

H =
1 =

height of the first indicationof smoke above the fire surface, fi (m); ceiling height above the fire, ft (m); time, s (S); heat release rate from steady fire, Btds (kW);

Q
A

=

= cross-sectional area of the atrium, ft2 (m');
=

c, cl

0.67 (1.1 1).

Empirical Filling Equations
The empirical filling equations are based on smoke filling tests (Heskestad and Delichatsios 1977; Nowler 1987; Mulholland et al. 198 I ; Cooper et al. 198 1 1 Hagglund et al. 1985). Because of the difficulty in determining the bottom of the smoke layer from experimental data, the correlations below use tile unique concept of the first indication of smoke (Figure 14.3). In the idealized zone model, the smoke interface is considered to be a height where there is smoke above and none below. In actual fires, there is 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 section differ froill predictions using a zone fire model.

Equation (14.1) is based on a plume that has no contact with the walls. Because wall contact reduces entrainment of air, this condition is conservative. Equation (14.1) is Tor a constant cross-sectional area with respect to height. For other atrium shapes, the zone fire model AZONE, physical modeling, or CFD can be used.The equation is appropriate for A/H' from 0.9 to 14 and for values oTz greater than or equal to 20% of ff. A value of z/H greater than one means that the smoke layer under the ceiling has not yet begun to descend. These conditions can be expressed as
A =

Constant with resepct to r ,

(14.2)

and

Filling by a Steady Fire
For a steady fire. the smoke filling can be appmsimated as When Equation (14.1) is solved for ?/H, the user
\\,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 from these figures that smoke filling is only appropriate for very large atria. The dashed lines on these figures ~ show the range .of applicability (0.09 A / H 5 14) of the steady filling equation. Example 14.1 Smoke Filling by a Steady Fire 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 and area are 100 ft (30.5 m) and 100,000 ft2 (9.290 m2). The height of the firs: indication of smoke abo\.e the fire surface, z, is 100 - 70 = 30 ft (9.1 m), and z/H = 30/100 = 0.3. From Equation (14.5), the filling time is

where z = height of the first indication of smoke above the fire surface, f (m); t H = ceiling height above the fire, ft (ft); I = time, s (S); tg = growth time, s (S); A
= =

Cd

cross-sectional area of the atrium, ft2 (m2); 0.23 (0.9 1).

As with Equation (l4.1), Equation (14.6) is conservative in that it estimates the height of the first indication of smoke and is for a plume that has no wall contact. Equation (14.6) is also for a constant cross-sectional 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
A = Consrant with respect to H,

Filling by an Unsteady Fire
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 discussed below, the fire continues to grow throughout the 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 completeness.

(14.7)

and

The growth time, lS, has already been discussed, and values of it and characteristic fire growths are dis-

Ceiling Height Above Top o Fuel (m) f

Figure 14.4

Titne for stuokefi.ot~r 5000 Btrds (5280 0 k W)fit.elo./ill 111e 80% of'11te lop olriritt~ ( M = 0.2) eslitmled bv tire etrrpit-icol z ,/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 Time Minutes 15 Slow Fire tg= 600 S Btuls 2,250 Medium Fire fg = 300 S Btuls 9,000
kW 9,500

Fast Fire fg = 150 S Btuls 36,000 kW 38,000

Ultra Fast Fire fg=75s Btuls 144,000

kW
2,370

kW
152,000

30
Notes:

9,000
v

9,500

36,000

38,000

144,000

152,000

576,000

608,000

I . tII is the growth rime for a t-squared fire to reach 1000 Btds (1055 kW); see Chapter 2.
2. Because ofthe laqe fires at the end ofthe evacuation time, the unsleady filling equation has limited

applicability.
60
/ / / --

- 30
H = 100 ft (30.5 m) A = 50.000 f f (4650 m') Q = 5000 Btuk (5280 k W )
- 2 5,

,
E

3O

-

H = 100 fl(30.5 m) A = 50.000 f' (4650 m') t

Zone Fire Models:
CFAST

Steadv -> ', . '

\
1

I
10
1800

g
70

Zone F~re Models

3 0

-ASET-C
CFAST

- 20
10

'
1200 1500

'

0

300

600

900

Figure 14.6

Time ( S ) Conzpar-isoiz of clear heights siti~ulated by d@wnt rnodels.

Figure 14.7

Time (S) Coinpar-isoil of snzoke 10-vet- tettpv-rrtwes sitnirlated!C d i f f e ~ utuodels. t

,a0

0

,e , I

cussed in Chapter 2. As with the steady filling equation, the unsteady filling equation can be solved for time:

where Cej3 is 0.363 (0.937).

Computer Modeling
The height of the smoke layer ajove the fuel is sometimes referred to as the clear- height, and Figure 14.6 shows a comparison of clear heights predicted by different zone fire n~odelsand the steady filling equation. These predictions are for a large atrium o f H = 100 ft (30.5 m) and A = 50,000 ft2 (5780 m2) with a steady fire of 5000 Btuls (5270 kW). It can be observed that the predictions of ASET-C and AZONE are close to each other. CFAST and the steady filling equation predict lower clear heights. The differences in predicted clear height can be attributed to inherent differences in the prediciive tools. These differences include ( l ) the plume models. (2) the definition of clear height, and (3) the approach to heat transfer. For each of the zone models, the mass flow of the plume is calculated from different plume models. As previously stated, the empirical equation is conscrvativc in that it predicts the clear height as the first

indication of smoke above the fire, as illustrated in Figure 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 the zone models. The CFAST simulation calculated heat transfer to gypsum board walls and ceiling based on the temperature difference between smoke layer and the gypsum board. Both ASET-C and AZONE use factors to estimate heat transfer. ASET-C estimates heat transfer by the heat loss fi-action, &which is the fraction of the heat release rate of the fire that is lost to the bounding surfaces of the room and its contents (Appendix F). The heat loss fraction is generally in the range of 0.6 to 0.9. AZONE evaluates heat transfer by the convective fraction, X,, and the wall heat transfer fraction, The convective Laction is the convective portion of the heat release rate; for more information about this fraction, see Chapters 2 and 13. The wall heat transfer fraction is the fraction of the plume enthalpy flowing into the smoke layer that is lost to the walls and ceiling. The smoke temperatures associated with the clear heights of Figure 14.6 are shown in Figure 14.7. For ASET-C, a value of 2,. = 0.6 was chosen. For AZONE, X,. = 0.7 and 11 = 0.4 wcrc used. The factors are rclatcd

v.

~ r i n c i ~ lof s e Smoke Management

as Ac = I - ( I - ?l)and, thus, AZONE was effectively simulated with Ac = I - 0.7(1 - 0.4) = 0.58. It is not surprising that the smoke temperatures are almost the same for the ASET-C and AZONE simulations (Figure 14.7). The smoke temperature of the CFAST simulation was higher, but the convection coefficients upon which the wall heat transfer was based are calculated from general correlations. No convection coefficients have been developed specifically for fire compartments. MECHANICAL EXHAUST j Mechanical smoke exhaust is probably the most common form of atrium smoke management in North America. As with natural venting, mechanical smoke exhaust can be based on either a steady or an unsteady design fire. The equations of the next section deal with a steady fire, and zone fire models can be used to analyze smoke flow due to an unsteady fire.

Steady Conditions
The method of analysis presented in this section is based on the simplieing assumptions below. The only mass flow into the smoke layer is the fire plume. The only mass flow from the smoke layer is the smoke exhaust. The exhaust is removing only smoke and not any air from below the smoke layer. The smoke layer height is constant (Figure 14.10). The flows into and out of the smoke layer are at equilibrium. Heat transfer between the smoke layer and the surroundings have reached equilibrium. Before using this method, designers need to verify that these assumptions are appropriate for their application.

~ i i u r 14.8 e

Katzi~-al smoke 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~n due to the prcse~lcc? f a ~ u lbziilding ~iear-by. l

Chapter 14- Atrium Systems

To calculate the exhaust flow rate, the plume equations from Chapter 13 are adapted with variables redefined for the following application: ~z=CaIQc z and
1 /3 5/3

0 ,

= convective heat release rate offire, ~d~

Cp = specific heat of plume gases,Btuflb "F (kJAcg "C); q = wall heat transfer firaction (diiensionless).
As already stated, the wall heat transfer factor is the fraction of the convective heat release rate that is transf e n d to the waiis and ceiling of the atrium. This factor depends on a number of conditions, including the geometry of the space, the construction materials of the walls and ceiling, and the smoke layer temperature. An atrium with no heat transfer is referred to as an adiabatic atrium (v = 0). The adiabatic assumption is conservative in that it results in high predictions of volumetric smoke exhaust, but it is not conservative with respect to plugholing. In the absence of research about the wall heat transfer fraction, values of q are expected to be in the range of 0.3 to 0.7 for walls and ceilings of normal construction materials (brick, concrete, glass, gypsum board, etc.). The density of the exhaust gases can be calculated from the perfect gas law,

+C,~Q,

forz>zl

(14.11)

where
rir
=

mass flow exhaust of exhaust air, Ibls (kgts);

Q C

- convective heat release rate of fire, Btuk (kW); =

Z1

= =

height of the smoke layer interface above the fuel, fi (m); mean flame height, ft (m); 0.022 (0.071);

c,, c,,

= 0.0042 (0.00 l S);

Col0 = 0.0203 (0.032). The mean flame height is

where C,, l = 0.533 ( 0166). 91-ictlyspeaking, Equations (14. I I ) and (14.12) are for the mass flow rate ofan asisym~nelric plume into the upper layer. M'hen the axisymmctric plume equations at-e not appropriate, other plunle equations may be used. For the balcony spill p l u m equations and the window plu~ne equations, see Chapter 13. l'he convective the heat I-elease rate, G,., is
Q,. =

where
p, = density of exhaust gases, lbrnlf? (kgh3);
p = atmospheric pressure, lbflfi2 (Pa); R gas constant, ft Ibfllbni "R (Jkg K);

-

7;.

= absolute temperature of exhaust gases,

"R (K).

Alter-natively, the density of the exhaust gases can be calculated from

x,.Q

(14.14) where T,. = absolute reference temperature, "R (K);
p,.
=

whcrc
Y,,

=

convcctivc lyaction ol'heal relcasc (see Chap[cr.s2 and 13); total Ilcat rcleasc rate, Btuls (kW).

0

=

density at reference temperature, lbm/ft3 (k9/m3).

For convei~iciicc, tarn sr~okc the layer Iieighr will of be used to mean lllc heigh~ the smoke layer inlerfacc. The term ; snioke layer licigllt above the luel. is The te~npc~-nlure the smoke layer can he of cxprcsscd as
T , = 7;)

There are an infinite number of pairs of T,. and p,. that can be used in Equation (14.17), and one such pair is 530°R (294 K) and 0.075 lbrn/ft3 (1.20 kg!m3). The volu~netric How of exhaust gases in plume is

bc(l - 11 ) + --lil

c,,

whcr-c

i/ = volumetric tlow ofeshaust gases, cfni (m3/s):
s~iiokc Iaycr temperature. "F ("C): a~nbicnt rcmpcraturc, "F ("C):

I;. 7;,

= =

a

= mass Ilow ofeshaust air, Ibls (kgs);

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? 72.0°F (22OC) Ambient temperature Ceiling height 45 ft (13.7 m) Convective fraction 0.7 Oft(0m) Height of top of fuel 2000 Btuk (21 10 kW) Heat release rate 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
z, = 0.533Qc
2/5

= 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 = 0.220, ,

. 1 /3_5/3

+ 0.0042Qc = 0.022(1400'/~)(36~/~) + 0.0042(1400) =

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

'CJnsteady Conditions
Unsteady analysis of an atrium exhaust system may be done to simulate a combination of smoke filling and snioke exhaust, simulate the effects of an unsteady fire, and determine the impact of activation time on smoke layer depth.

A combination of smoke filling and smoke exhaust can be used for an atrium that is not large enough to qualify for smoke protection solel!~ by smoke filling. For this combination approach, the exhaust fans need to be sized so that the smoke filling time is greater than the evacuation time, including the time it takes to become aware of the fire and to prepare for movement to an exit.

rt is the nature of fire that it is an unsteady process Probably the reason that steady fires are used extensively is that they lead to the simple steady analyses like the one above. While large steady design fires can be selected to yield conservative designs, these design fires are not realistic. See Chapter 2 for information about design fires. Zone fire models such as CFAST and AZONE can be used for analysis of atrium smoke exhaust systems \\.it11 unsteady fires. Before smoke exhaust fans can be turned on, the presence o f the fire needs to be detected. There is some delay betwzen detection and activation, and it takes some time for the fans to come up to full speed. Detection time can be estimated from the inforniation about the lag times of plumes and ceiling jets pro\.ided in Chapter 13. When appropriate, detection should takc into account the potential that there could be a stratified layer of hot air under the ceiling, as discussed later.

,

Chapter 14 -Atrium Systems

-9

500

l,=150s 60 120 180 l i m e (S) 240

1

5 0 0

A=l~~)fl~(92.9rn') 60

-2
1 0 3M)

0

300

(a) Variation of smoke layer with atrium area. A

120 180 240 l i m e (S) (b) Variation of m o k e layer witJ~ exhaust adimtion time. l ,

-9 -7
.... -

. c

----

to= 1 9 = 300 , f,=600s

. '5 .

-5

5

0

l, = 90 S A = 1000 ft'(g2.9 m2)

-2 -1
240 0 3M)

Notes: W) . 1. The Are is a 1-squared fire up to 2000 Btuk (2110 k W after that the HRR remains mnstanL 2. The exhaust flow rate was seleded so that lhe midness of lhe smoke layer would be 6 fl(l.83 m) at a sieady HRR of MOO Etuk (2110 k W . 3. As wilh other zone fire models, the details of lhe ceiling jet are not simulated by AZONE. Thus. onty the portions of these graphs where lhe smoke layer is Celow about 24 R(7.3 m) are realistic. 4. Other factors are: Ambient Temperature. T, = 72.0 'F (22.2 %l Ceiling Height. H= 30.0 fl(9.1 m) Height of top of fuel. H = 0 A (0 m) , Exhaust Row rate. V = 49500. h(23.4 m%) Exhaust location factor. P = 2 Exhaust location Delow ceiling, d. = 0? (0 m) i Number of exhaust inlets. , N =6 Wall Heat transfer fraction. q = 0.4

(C)

180 lime (S) Variation of smoke layer fire growth. l, 60 120

Figure 14.11 U17steadylayer- height sit~rdated the zonefire model AZONE. .by It is possible tliat the snioke layer could descend well below the design smoke layer height based on a steady analysis. To check the effect of activation, AZONE allows tlie user to specify the acti\.ation time o f the smoke exhaust fan. Figure 14.1 1a shows the effect of the atrium area on the smoke layer height as calculated by AZONE for an atrium 30 ft (9.14 m) in height with an exhaust activation tinie of 90 seconds. It can be seen tliat for an atriuni area, A, of 5000 ft2 (465 ni2) or more, the delay in activation does not have an adverse effect on smoke layer height for the conditions of the simulations. For A = 2000 ft2 (186 ni2) or less, the smoke layer drops well below the design lieight for the conditions of the simulation~. 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 atrium with A = 1000 ft2 (92.9 ni2). As expected, the smaller the activation time, tlie less the effect on smoke layer lieiglit. At t,,, = 30 S, the smoke layer stays above design lieiglit tliroughout the simulation. Figure 14.1 1c shows the effect of tlie fire growth time, t6" on slnoke layer height for a 30 ft (9.14 m) tall atriuni with A = 1000 ft2 (92.9 ni2). As would be expected, the less the growth tinie (faster the tire), the greater the effkct on tlie smoke layer height. While a study has not been made on the effect of the activation time on smoke layer height. some gcneralizations can be made. For atria with relatively large areas (A/H~> 5 where H is the atrium height), the effect of fan activation at 90 s would not be expected to have an adverse effect on the smoke layer heiglit. For atria with relatively small areas (A/H~< 5 ) , the smoke layer could drop below the design lieight, resulting in smoke contact with people. AZONE can be used to analyze the effects of activation tinie on the smoke layer lieight.

Makeup Air
For steady flow, the mass flow of air or smoke exhausted from the top of an atrium equals the mass flow of air entering below the smoke layer. This airflow entering the atrium is referred to as niakeup ai:, and makeup air can be either supplied naturally or by fan power. Fan-powered niakeup air is often sized at 90% and 97% of the exhaust airflow rate, and the balance of the air needed to acco~nnlodatethe exhaust naturally flows through openings or leakage paths. Natural makeup air flows through openings, such as open doorways and vents, and sometimes makeup airflow paths are complex conlbinations of rooms and corridors. Computer network airflow programs, sucli as CONTAM (Chapter 8), can be used for analysis of these complex flow systems. The velocity of makeup air should not destroy the plume structure or significantly deflect the plume at an angle. It is believed tliat keeping the velocity at or below 200 fpm (l nils) will prevent sucli plume disruption.

Principles of Smoke Management

NATURAL VENTING Natural smoke venting is common in many parts of the world, such as Europe, Australia, and New Zealand. As stated in Chapter 1, natural venting was developed in response to several fire tragedies in the 19th and early 20th centuries. Natural venting relies on the buoyancy of hot smoke to force smoke out of open vents at or near the top of the atria (Figure 14.8). Natural venting can be based either on a steady or an unsteady design fire. The equations in the next section are for a steady fire, and zone fire models can be used to analyze smoke flow due to an unsteady fire.

temperature may be less than the outdoor summer design temperature. The smoke temperature and mass flow of the plume can be calculated from the same equations that are used for mechanical exhaust as discussed later.

Wind
For an atrium attached to a tall building or very near a tall building located in open terrain, wind can produce positive pressures at the top of the atrium, as shown in Figure 14.9. Because such positive pressures can interfere with natural venting, natural venting is not recommended for atria with s ~ wind conditions. h

Steady Conditions
The equation developed in Chapter 13 for the mass flow rate through the vent is

Makeup Air
For natural smoke venting described by Equation (14.19), makeup air flows naturally through the inlet 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. TENABILITY SYSTEMS As already stated, the approaches discussed above have the goal of not exposing occupants to smoke during evacuation. Tenability systems are designed to maintain tenable conditions with occupant exposure to smoke. Hazard analysis consists of evaluation of smoke transport, people movement (evacuation time), and tenability. While smoke transport can be simulated by zone fire models, CFD modeling has the significant advantage of being able to simulate variations of temperature and concentrations of combustion products in the smoke layer. Evacuation time can be evaluated by the methods of Chapter 4. Tenability analysis should address visibility, gas exposure, and heat exposure, and extensive inforrnation about tenability can be found in Chapter 3. Ths calculation method for tenability described in Chapter 9 can be used for atria. STRATIFICATION A N D DETECTION Often, a hot layer of air forms under the ceiling of an atrium as a result of solar radiation on the arriuin roof. While studies have not been made of this stratified layer, building designers indicate that the temperatures of such layers are often in excess of 120°F (50'C). Temperatures below this layer are controlled by thc building's heating and cooling system. and the temperature profile can be considered to increase significantly over a small increase in elevation as shown i n Figure 14.12.

wherc
m, = mass flow rate through the vent, Ib/s (kgls);

C
A.

=

discharge coefficient (dimensionless); inlet opening area, ft2 (mZ);

A,, = vent area, f? (m2);
=

P0 - outside air density, lb/ft3 (kg/m3);
g db
= =

acceleration of gravity, 32.2 ft/s2(9.80 m/ sZ); depth of smoke layer below the smoke vent, ft (m); absolute temperature of smoke, "R (K);

To = absolute temperature of outside air, "R (K);

r,

=

Because buoyancy of hot smoke is the driving force , of natural venting, the mass flow rate, );I,, through the vent increases with increasing smoke temperature, TT. As the size of a fire increases, the mass flow rate of the plume into the upper layer increases and the temperature of the smoke layer increases. For a fire larger than the d-sign fire, the smoke temperature goes above the design value, and the mass flow rate through the vent increases above the design value. This benefit is unique to natural venting, and it helps offset the greater amount of smoke produced by fires that might exceed ths design fire. For air-conditioned atria, it is possible that the smoke temperature may be less than the outdoor summer design temperature. This can result in doivnward outside airflow through the atrium smoke vents. To avoid such downward flow through smoke vents, natural smoke venting should nor be used wlicn the smokc

Chapter 14-Atrium Systems

0

Elevation Abjve Fwl (m) 30 60

90 120

Heat Release Rate:

5.000 Btul S(5.280kW) 2.OM) S(2.110 k W ) Btu1

-1w E
-80 -60 -40

p

f

P c
W

D

l I

- 20
50 0 50 100 150 200 250 300

Q

P

I

Temperature

Elevation Above Fuel (R)

Figure 14.12 Temperature profile of hot layer- of air zrnder atrium ceiling.

Figure 14.14 Average femperature of axisyn~rnetr-ic
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 temperature condition exists. One or more beam detectors 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 function of the specific design but sllould include beams a1 the bottom of identified unconditioned spaces and at or near the design smoke level \\it11 several intermediate beam positions at otlier levels. c. I-lorizontal Beanis to Detect tlie Smoke Plume The purpose of this appl-oacli is to detect [he rising plume rather than the stnokc layer. For this approach, an arrangement ol'beams are installed at a level below tllc lowest expected stratification level. These beams need to be close enough to each other to ensure intersection of the plume, the spacing being based on the width of the beam at the least elevation above a point of fire potential. Tile approaches described above are illustrated in Figure 14.15, and approach (a) has the advantage that i t does not require the location of a number of horizontal beams. Some bean1 smoke detectors are subject to false a ~ t i \ ~ a t i o n sunlight, and alternative (a) min;niizes the by possibility of such false activation bp orienting the rcccivcr at adownward aligle. All of the coniponcnts of a beam s~iioke detector nccd to bc located so they are accessible for nlaintenancc. For thc arrangement sliown in Figure l?. 15. a roof opciiing (not shown) could provide access for mainte~iancc.

Figure 14.13 Smoke stratificatior~ under a layer of hot ail:
When the average temperature of the plume is less than that o f t h e hot air layer, tlie smoke will form a stratified layer under it, as shown in Figure 14.13. Average plume temperatures are sho\vn in Figure 14.14, and it can be observed that the average plume temperature is often less tliali expected temperatures o f tlie hot air layer. Thus, when there is a hot air layer under the atrium ceiling, smoke cannot be expected to reach the ceiling of the atrium; and smoke detectors mounted on that ceiling cannot be expected to go into alami. Beam smoke detectors can be used to overcome this detection difficulty. The follo\\:ing are beam detection approaches that can provide prompt detec:ion regardless of the temperature of the air under tlie ceiling at the time of fire initiation. a. An Upward-Angled Beam to Detect the Smoke Layer The purpose of this approach is to quickly detccl the development of a smoke layer at whatcver tcmperature condition exists. One or more beams arc aimed at an upward angle to intersect thc smoke layer I-cgardless of h e Imel of s~noke stratilication.

Principles of Smoke Management

Plan View

Section

(a) Upward Angled Beams t o Detect the Smoke Layer

Suggested Spacing o Beams: f

Plan View

Section

(b) Horizontal Beams t o Detect the Smoke Layer at Various Levels

Suggested Spacing o Beams: f
X = -B

4

Plan View

Section

(c) Horizontal Beams to Detect the Smoke Plume of beam Figure 14.1 A~-ra~~,oe~nerits smoke detectors.

Chapter 14 -Atrium Systems

Step I: Assign Values to Consfan&
C,, R, P-, U. X,, a

M2 = mass of smoke layer at the end of the interval (kg).
The change in energy of the smoke layer can be expressed as

Step 2: Read Data

t

Z.H.A.H,.,,~,.Q.~.

P. 4. N&". L.4". L

where

Step 3: Assign lnitlal Values
m,

t

AE
Cp

=
= =

change in energy of the smoke layer (H); specific heat of smoke ( k l k g K)

=o; T , =C;

p,=p.l(RT,,);

z=H-H,;

E=O; M = O ;

Q = O ; ~fr,<o. = Q ~ Q

wall heat transfer fraction (dimensionless); Tp = absolute temperature of plume gases entering smoke layer (K);

Step 4: Calculate for each time step

t

T I=
d-;

r=r+ar

; c ; Q ; Q<;d ;
T,; 1

To

=

absolute temperature of smoke layer gases at the beginning of the time interval (K); absolute ambient temperature (K);

m,:z,:mp;

M ;M ; AE;

E;<,;p,; < ; X

Step 5: Check for

Write Oulput

As with the mass flows, Tp is evaluated at the end of the interval. The ambient temperature is considered constant throughout the calculations. The smoke temperature, T,,, at the beginning of the interval was used because the value at the end is unknown. Selection of a small interval makes the resulting error negligible. The energy in the snioke layer is

Tiue

where

El
False
Slep 6: Output

=

energy of the smoke layer at the beginning of the interval (U);

If I is an even multlple

Write Output

E2 = energy of the smoke layer at the end of the interval (kJ). The smoke temperature, at the end of the time interval is

Figure 14.17 Sirr7pl~fiedflo~t~ for- AZONE char-I 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 was used to minimize errors. However, these flo\vs could be evaluated at a time within the interval with the goal of using a larger interval. The mass of snioke in the layer at the end of the interval can be expressed as

where
=

smoke density at the end of the interval (kg/mZ); ambient pressure (Pa);

,vo

=

R

= gas constant (Jkg K). The volume. V?, of the smoke layer is

where

M, =

mass of smoke layer at the beginning oftlie interval (kg),

For an atrium of constant cross section, the height of the smoke layer above thc top of the fuel is

'

Principles of Smoke Management
.

-

Variable Atrium Area
where z2
=

Height and area pairs in descending order are prescribed as hiand A i for i to n. For each height, h , the above that height is atrium volume, Gi, height of the smoke layer above the top ofthe fuel (m); height of he1 (m); cross-sectional area of atrium (m 2). The units of hi, and V; are m, m2, and m3. The Ai, , terms hi,Ai,and V,; are terms of arrays (sequences of numbers), and the subscripts i and j are what is referred to as dummy variables. For example, h; where i = 3 is the third value of the height array. Before calculations are done for the time intervals, the values of Voiare calculated for i = to n. The atrium area at any height X is

,

.
/

H

= height of atrium (m);
= =

H--l
A

It should be noted that H has a different definition in AZONE than it has for the empirical filling equations. The various height temis above are illustrated in Figure 14.17. Determination of z2 for an atrium of variable area is discussed later. The values at the end of the current time step become those at the b2ginning of the next time step.

Plugholing
For each time interval, the exhaust from the smoke layer, i z , , is calculated, taking into account any plugholing that might be happening. The minimum smoke layer depth to prevent plugholing is

where A(x) = atrium area above height X (m2); = height above atrium floor (m). X j
=

dummy variable such that 17,

<X

5 h,-

,.

The volume above any heights is

where
d,,,
V,,
=

minimum smoke layer depth to prevent plugholing (m); volumetric flow rate per exhaust inlet (m3/s); exhaust location factor (dimensionless);

where V(x) is the atrium volume in m3 above height X. The height of the smoke layer above the floor is the value ofx, which satisfies the following equation:

= =

p

Cph4= 0.15.
The volumetric flow per inlet is p,, = I/,/N; where Nin[,, is the number of exhaust inlets. The exhaust from the smoke layer is

The value o f x that satisfies Equation (14.33) can be determined by any of a number of root finding methods. In AZONE the method of bracketing and bisection was used (Press et al. 1986). The height of the smoke layer above the top of the fuel is

Time Interval
The time interva~,~ needs to be selected so as to At, minimize error. Theoretically, errors associated with the interval size are due to inaccuracies of numbers used from previous iiltervals. In AZONE, TI,M,, and E l are calculated in the previous interval, and the values of the heat release rate and exhaust airflow are each evaluated
9. The time interval should not be confused with the output internal. Calculations are made at each time interval. but data arconly written at theoutput intervals.

-

I

=

p l .

for d,,! I tl

where
d
=

depth ofsmoke layer bclow bottom ofeshaust inlet,
(m);

a

=

plugl~oling exponent (dimc~lsionless).

Chapter 14-Atrium Systems

Table 14-2: The Effect of Time Interval on the Accuracy of AZONE ~imulations' Atrium Height, H ft m
30 9.14

Cross-sectional Area, A ft2 m2 Small Atrium
1,000 93

Time Interval,
At
S S

Steady ~ i r e ~ Simulation Time ~rror'
%

Fast t-squared ~ i r e ' Simulation Time
S

~rror~
Y O
l

0.005 0.0 1 0.05 0.20 0.50
1. O O

30 30 30 30 30 30 30 240 240 240 240 240 240 210 4SO 480 480 480 4SO IS0 480

0.0 0.0 0.2 1.2 3.7 7.7 65.0 0.0 0.0 0. I 0. I 0.3
I .j 6.3

90 90 90 90 90 90 90 300 300 300 300 300 300 300 300 300 300 300 M O .O N
.300

0.0 0.0 0. I 0.2 0.6 1.2 GI . 0.0 0.0

5.00

Small Spread-Out Atrium
30 9.14 12,000

1,110

0.01 0.05 0.20 0.50 1. O O 5.00 20.00

0.1
0.1 0.3

I .5 6.1
0.0 0.0 0. I 0. I 0.3 I .A
5.8

Lnrgc Atrium
IS O
45.7 25,000 2,320 0.0 1 0.05 0.20 0.50
1 .OO 5.00

0.0 0.0 0.0 0. I 0.3 I .-I
6.0

70.00

Large Spread-Out Atrium

20.00
I.

1 200

0.7

600

0.7

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(:,,,
tinic intcrval

Co~iditio~is ol'thc si~iiuliltions: l ) alnhlcnt rsmperaturt ol'7OoF ( level. (5) wall Iieal transfer fraction 01'0.3.

(?IT).( 2 )cnnsl:~ntcross-secliotial ;lrc:u. (31 no sliitAs r.h;~ust. top of filet a[ lloor (4) \\-hersI,,, 11icvaluc of: is

-)I :/

Tor that a t r i i ~ ~size. ii

at :>c smallest

at a point in the interval. For calculations made u.ith successively smaller interval sizes. the absolure values o f such theoretical errors also become smaller. In a d d i t ~ o nto tlleoretical errors. round-ofi' errors also can be associated with interval size. The nature o f numerical round-oti'errors is such that predictions made for very small interials can have ven. lalgc errors. S o the time interval needs to be cvaluartd s o that i t is neither too large o r too stiiall.

Table 14.2 lists 21-rors01'~1noke-tilli1lg sinlulations for several values ol'Dt fbr rht at[-ium size categories ( I ) small, (2) small spread out. (3) large, and (4) large spread out. Tliese el-rors pertain to the height o f the smoke layer, and errors (not s h o \ w ) of smoke layer teniperature were less. For the t i l w intervals used, round-off errors due to small il?[cr-valsize did not occur. The largest cl.tor lis~cdin T a b k 14.2 was 65% for a 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 can be meaningless. What is desired is an interval size that has acceptable errors for all atria that might be analyzed. An interval of 0.05 s results in errors less than 0.05% for all the atrium size categories except small.

For the small atrium, this interval resulted in errors of 0.1% and 0.2% for the t-squared fire and the steady fire. Accordingly, 0.05 S was chosen as the time interval 'for AZONE.

CHAPTER 15

Physical Modeling

0

ne option when conventional methods of analysis are inappropriate is fire testing in a reduced scale model, and there is considerable experience with application of physical models to fire technology. Such scale modeling has been used to rsconstruct fires for fire investigations. Two examples of such fire reconstructions are the King's Cross subway station fire in London, United Kingdom (Moodie et al. 1988), and the Hart Albin department store fire (Quintiere and Dillon 1997). Froude modeling is probably the most common kind of physical modeling used for smoke transport, and NFPA 92B recognizes Froude modeling as a method of analysis of smoke management systems for atria. This chapter addresses the fundamentals of physical modeling of smoke movement with special emphasis on Froude modeling. For further information about fire applications of physical modeling, readers are referred to Arpaci and Aganval (199Q), Quintiere (1989b), Heskestad (l 972, 1975), and Hottel (l96 1).

R,

=
= =

Reynolds number, length, velocity,

I
C'

p = density, and p = dynamic viscosity.

DIMENSIONAL ANALYSIS AND SIMILITUDE
The idea of dimensional analysis is to express a complicated process in terms of a relatively few dirnensionless variables. This can simplify the analysis and make physical modeling possible. Many dimensionless parameters can be viewed as being ratios of fluid forces. Three dimensionless parameters that- are of particular interest in this chapter are the Reynolds number, Froude number, and Prandtl number. The Reynolds number can be thought of as the ratio of the inertial forces to the viscous forces. and this number distinguishes between flow regimes such as laminar and turbulent. The Reynolds number is

The above equation is a little dit'ferent from that given in Chapter G for the Reynolds number. In Chapter 6: the Reynolds number was expressed in terms of kinematic viscosity, v, where v = p/p. Also, the units of the variables are given in Chapter G . Most of the equations in this chapter are intended to describe physical modeling and not be used for calculations. Accordingly, units are not given for variables in most equations of this chapter. However, all of these equations are valid for SI units or any other homogeneous unit system (Appendix A). The Froude number can be thought of as the ratio of inertial forces to gravity forces. Because the buoyancy of hot smoke is a gravity force, the Froude iiun~beris very important in physical modeling of smoke movcment. The Froude number'' is
10. An alternate fonn of the Froude number is I;;. =

~ l ( ~ 1 ) This. is 'simply the square root of the Froudc ''~ number. which is used in this book, and basic concepts concerning the Froude number and thc scaling rclalions for Froudc nodel ling are the same rcgardlcss 01' ~rhich form of the Froudc number is used.

Chapter 15-Physical Modeling

Table 15-1 : Quantities and Associated Dimensional Formulas
where
Quantity Length Time Mass Temperature Force
-Symbol(s)

F, g U

= =

Froude number,

L, X, z

Dimensional Formula L

acceleration of gravity, and . = velocity. The Prandtl number is a dimensionless number, which is the following combination of fluid properties:

P,.=
where

Pc -y

Heat Velocity Acceleration Work Pressure Density Internal enersy Enthalpy Specific heat of a solid Constant pressure specific heat Constant volume specific heat Dynamic viscosity Kinematic viscosity Thermal conductivity v=p/p
X-

P, Cp
P k

=

Prandtl number,

= constant pressure specific heat,

dynamic viscosity, and = thermal conductivity.
=

Dimensional Formulas
The system o f primary dimensions (or base dimensions) can be chosen as length L, time t , temperature T, and mass M. The dimensional formula of a physical quantity follows from definitions or physical laws. For example, the dimensional formula oCa doorway width is [L] by definition. The brackets [ ] indicate that the quantity has the dimensional formula within the brackets. The dinlensional formula of velocity is [L /t] and that of acceleration is [L / t 2 ] . For a homogeneous unit system,'' Newton's second law is

L*

/I

ML/$T

Other dimensionless quantities can be evaluated in the same way.

The

n Theorem

where

F

=

force,

m = mass, and

a = acceleration.

The dimensional forn~ulafor force is the dimensional formula of mass times that of acceleration. This is [ML / P ] . Work is force acting through a distance, so the / The units of work are [ M L ~ 81. dimensional formulas of a number of physical quantities are listed in Table 15.1. A dinlensionless quantity has no dimensions; for example, the dimensions of the Froude number can be evaluated as

The n (pi) theorem (Buckingham 19 15) states that, 7 for any phksical application o r process that includes t quantities in which there are m dimensions, the quantities can be arranged into 17 - m independent dimensionless parameters. further, some functional relation of t 7 these 7 - n independent dimensionless parameters exists that describes the physical application or process. Consider an application for which A,, A2, A3, . . ., A,, are the essential quantities involved. such as length, velocity, pressure, mass, etc. A functional relation of these quantities can describe the application, and this can be expressed as
F ( . 4 , , A z ..4;, ..., A , , ) = O.

(1 5.6)

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 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 groups based on the ll theorem is presented in a number of texts, such as Kreith (1965) and Streeter and Wylie (1979). A disadvantage with this technique is that there is no one unique solution for the I groups, and a numl ber of possible combinations of rI groups may need to be evaluated.

and

p, is the ambient pressure distribution.
Conservation of Energy:

Similitude
Physical modeling has been used in many areas of 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 built, and conditions of the tests are maintained such that the rI groups are preserved. This means that at a particular location in the model, each rI group has the same value as it has at the corresponding location in the full-scale facility. For perfect similitude, all the rI groups would be preserved. Fortunately, perfect similitudeis not always necessary. Useful results can be obtained from preservation of only some of the I groups, provided that the l impact of other Il groups is not significant. This is explained later.

Equation of State: The equation of state for an ideal gas is

Variables in the above governing equations are Cp = specific heat,
g k
= = = = =

T
y

p,
u
X

acceleration of gravity, thernlal conductivity, temperature, pressure, ambicnt pressure, con~ponent velocity, of position,

= X =

Q"' = rate of chemical energy per unit volume,

Development of Dimensionless Groups
An alternative to the I theorem approach for devell oping rI groups is the differential equation approach, which is more elegant and provides a high level of understanding. Further, the differentia; equation l approach can be used to develop I groups for physical 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 interest, the governing equations of fluid dynamics that are listed below are in a one-dimensional form. Conservation of Mass:
- ,(

p,

= =

ambicnt density,

I

radiant intensity, K = absorption coefficient, o = Stefan-Boltzman constant, p = dynamic viscosity. For an ideal gas, the gas constant, R, is

where C,, is the constant volume specific heat. Dimensionless variables are defined below.

G/

u.1

(15.8)

Conservation of Monlentum in the Vertical Direction:

where
p' = 11 - p , ,

(15.10)

Chapter 15-Physical Modeling

(15.20) (15.21)

where the characteristic time, r, is chosen to be [ ( .This /I means that Ill = 1. Because this ll group is a constant, it is always preserved and can be ignored. The next ll group is

where

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

I U r

geometric length scale, characteristic velocity, = characteristic time, To = ambient temperature,
=

=

p, = ambient pressure, p, = ambient density, p+
=
3

17; is the Froude number, As previously stated, the Froude nurner can be tliou~ht as the ratio of inertial forces to of buoyancy gravity forces. The four-th n group is

characteristic pressure defect ( =p,U-). p

By substitutiy the dimensionless variables of Equations (15.15) to (15.23) into the governing equations, the following nondirnensional form of the governing equations can be developed:

n ,

is the Reynolds number. The nest ll group is

Mass:
1 5is thc Prandtl nu~nber. ~nany 1 For gases including air, the

Momentum:

Prandtl number is nearly constant with respect to temperature. Smoke is air mised with a relatively sniall amount of combustion products. and the properties of smoke are generally 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 (l 5.26)
//I.

.IZ

+ n3n5n,n,[

1ic,(0-49
"any

+ U + n s21 g
7 he last

n, = ~ , , c , , u v '

(1 5.35)

n group is
11, =

State:

n i= (+--)B?- I
The first

( 1 5.17)

5 c,.

( l 5.36)

n group is
n
= I

l L'r

(15.2s)

n , is tlic commonly used ratio of specific heats, and this ratio is a constant Ibr ideal gases. For air, the mtio of sDecitichears has a ncarly constant value of 1.4. can bc neglected for niodeling that is done in Thus, air.

n,

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 temperature is low enough so that the heat transfer groups do not need to be preserved. However, for higher temperature gases, such as those of flames, neglecting these groups is inappropriate. There has been considerable experience with Froude modeling, and the comparison between fullscale and 117 scale model temperatures (Figure 15. l) by Quintiere, McCaffrey, and Kashiwagi (1978) illustrates the degree of agreement that can be expected. Chow and Lo (1995) used Froude modeling to simulate smoke movement and smoke filling in an atrium.

0

Figure 15.1

600 1200 1800 Time (S) Froude modeling cornparisor7 of corridor gas temperature (adapted from Quintiere, McCafiey, and Kashiwagi [l 9 781).

Saltwater Modeling
The use of one fluid to model the flow of another is called analog modeling, and saltwater has been used extensively to model smoke movement. The idea of saltwater modeling is to submerge the scale model in a tank of fresh water and inject saltwater to simulate a heat source. Because the saltwater has a higher density than fresh water, the saltwater tends to flow down, whereas smoke tends to flow upward. This is accommodated by turning the model upside down in the tank. Frequently, the models are constructed of a transparent polymer, and the saltwater is dyed blue. This helps people to see, photograph, and video the saltwater flow. The major advantage of saltwater modeling is probably that it helps people to visualize smoke P-ow. Saltwater modeling is similar to Froude modeling in that the Froude number is preserved. The concentratiori of salt is adjusted such that the density forces of the saltwater in the model correspond to that of smoke in the full-scale facility. Saltwater modeling has no heat transfer, but the saltwater mixes with the fresh water as it flows in the model. Because of the lack of heat transfer, saltwater modeling is not appropriate for simulations of flow of flames or flow near flames. Chow and Siu (1993) conducted smoke filling visualization experiments on several atria using saltwater modeling. Yii (1998) conducted a series of saltwater modeling experiments of balcony spill plumes. For general information about saltwater modeling, see Steckler, Baum, and Quintiere (1986).

TYPES O F MODELING Froude modeling, saltwater modeling, and pressure modeling have all been used to simulate smoke movement in fire applications. Each of these modeling approaches has less than perfect similitude in that no practical approach can preserve all the H groups. However, these modeling approaches have produced good quantitative results and provided insight into smoke movement phenomena.

Froude Modeling
As previously stated, Froude modeling is probably the most common approach to the physical modeling of smoke movement. A scale model of the room, atrium, or other facility is built. Tests are conducted in the model in air at normal atmospheric conditions. The scaling relations are used to size the design fire and any forced air flows, and these relations are used to translate measurements from the model to the full-scale facility. These scaling relations are discussed later in detail. Because buoyancy is a gravity force and dominates the flow resulting from fires, the Froude number (n3) must be preserved. For reasons already discussed, i l l , H2, n and n9are also preserved. If the size of the model is appropriately chosen, the flow becomes fully turbulent and the viscous effects at solid surfaces are negligible. For this reason, the Reynolds number (n,) can be ignored. Information is pro-

,,

Chapter 15 - ~ h ~ s i cModeling al

Pressure Modeling
Pressure modeling is included for completeness. This modeling preserves both the Froude number and the Reynolds number. The Reynolds number is preserved by changing the ambient pressure. The pressures can be described as

T ,

= =

temperature of gas in model, "F ("C); temperature of gas in full-scale facility, "F ("C).

Tf

Because the model and the hll-scale facility a r e a t the same temperature and pressure, the scaling relation for density is

where where p,, = pressure of the model, psi (Pa);
p f
=

p,, = density of gas in model, lb/& (kg/m3) and

p = density of gas in full-scale facility, lb/fi? (kg/m3). f
Preservation of the Froude number can be expressed as

pressure of the hll-scale facility, psi (Pa); length in the full-scale facility, m (ft);

I,,, = length in the model, m (ft);

If

=

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 pressures are in the same units and both lengths are in the same units. For example, the pressures could be in atmospheres. and the lengths could be in inches. A one-eighth-scale model would need to be tested in a pressure vessel at a pressure of about 23 atmospheres. Probably due to the expense of testing in a pressure 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. S C A L I N G RELATIONS F O R FROUDE MODELING The basic concept of a scalc model is

wherc U,,, = velocity in the model, Ws ( d s ) ; U f
g
=

velocity in the full-scale facility, Ws (mls); and

acceleration of gravity, ft/s2(m/s2). It follows from Equation (15.41) that the scaling relationship for velocity is
=

where U,, = velocity in the model, Ws (mls); U f
=

velocity in the full-scale facility, Ws (m/s); length in the full-scale facility, ft (m).

I,,, = length in the model, ft (m); and

lf
where
X,,,
XJ

=

=

position in the niodel, ft (m); = position in the tidl-scale facility, ft ( ) m;

l,,, = length in the model, It (m); l = length i n the ftlll-scale facility, fi (m). f The ratio (I,,, / l,) is the scale of the model. For example, for a one-tenth-scale niodel, l,,, /lf= 1/10. As already stated for Froude modeling, the temperatures liom the model are the same as for corresponding places in the Sull-scale facility. Thus, the scaling relation for temperature is

The units listed above for the variables of Equation (15.42) were selected as they are ones that might be used for an application, but many other units can be used in this equation. The requirements for units are that U,, and U must be in the same units, and I,, and +must f be in the same units. For example, both velocities could be in feet per minute (fpm), and both lengths could be in inches. All scaling rela.tionships discussed in this section are o f a similar form, and this basic idea about the suitability of a wide range of units is also true for all these other scaling relationships. Volumetric flow is velocity multiplied by area, and the relation beconles

wlicr-c

where

Principles of Smoke Management

V,
V J

= =

volumetric flow in model, @/S (m3/s); volumetric flow in full-scale facility, 91s (m3/s).

where

4 = ,
=

pressure difference in model, in. H 2 0 (Pa); pressure difference in full-scale facility, in. H 2 0 0'4.

Mass flow rate is volumetric flow multiplied by density, s o combining Equations (15.40) and (15.43) results in
h,,, =
"l

J!f

1 5/2 (A)

where
rii,
= =

mass flow in model, lb /S (kg Is); mass flow in full-scale facility, Ibls (kgls).

The use of some of the scaling relations is illustrated in ~ x a m ~ l 15.1 and 15.2. Tsujimoto, Takenoues chi, and Uehara (1990) conducted experiments that verified the above scaling equations for smoke movement in atria. Quintiere, McCaffrey and Kashiwagi (1978) conducted smoke movement experiments that verified these scaling relations for smoke flow in a room and corridor. Example 15.1 The Scaled Fire For a 5000 Btds (5280 kW) fire in a full-scale facility, what is the corresponding fire in a one-sevtnth scale model? Using Equation (15.46),
1 5/2
Q , =

.

hJ
= l,/

Velocity is length per unit time, and substituting U,,, t , and = U $1 t/into Equation (15.43) results in !

where t, = time in model, s (S);

l1
I

=

~ooo(!)~/~=

38.6 Btuis(40.7 kW).

Il

t/

=

time in full-scale facility, s (S). 1.75 ft (0.533 m) above the floor at 42 s after ignition. I-low does this convert to the full-scale facility? Rearrange Equation (15.45) as

Consider the convective heat portion of the heat release rate as enthalpy flows. Q,,, = riz,,,Cp~T and QJ = k/CPAT(AT = AT,,, = ATf), then Equation (15.41) becomes

where
Q,,
=

11 Rearrange Equation (15.38) as
heat release rate in model, Btuk (kW);

QJ = heat release rate in full-scale facility, B t d s (kW). If the convective fraction of the fire in the model: XC,,,,, is the same as that in the full-scale facility, ,ye$ the scaling relation for the convective heat release rate is

l1

.vJ =

X.(+)
111

=

1.75(8) = 14 ft (4.27 m).

This means that at 119 s the smoke layer would descend to 14 ft (4,27 m) above the floor in the full-scale facility.

Approximate Heat Transfer Scaling
Even though the heat transfer groups and the Reynolds number were not preserved, some heat transfer effects can be partially preserved by considering surface heat transfer and solid heat transfer. For a semi-infinite surface, the wall and ceiling materipls can be scaled by

where
Q,, l,l
= =

convective heat release rate in model, B t d s (kW);

/

'

convective heat release rate in full scale facility, B t d s (kW). The ressure difference due to velocity is and substituting this M,,, = p 2 1 2 and A J J ~= into Equation (15.42) yields (remember, p = p,,, = p $

where (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 =
of the hll-scale facility, ~ t in u ~ " F - ~ h-' ff5

The thermal inertia for several materials is listed in Appendix A (Tables A10 and A1 I). Example 15.3 illustrates calculation of the scale thermal properties. In this example, Equation (15.49) was used to scale the thermal inertia of the model to 3.8 ~ t ft" OF-' h-' u ~ (0.44 k w 2 I ~ ~ K - ~Thermal inertia only needs to be S). scaled very roughly, and materials ranging from concrete to plasterboard would be acceptable (Tables AI0 and AI I).
Example 15.3 Scale Thermal Properties The walls and ceiling of a full-scale facility are made of conCrete. How do the thermal properties scale to a one-eighthscale model?

(m k

I

I

From Table A10 (Table All fors1 units), kpC'of concrete is 25 Btu' ft-""~' h-' (2.9 kw2 m4 K-~S). Using Equation
1 0.9

I

3.8 Btu' K4

I
11
J

ll~he thennal properties only need be scaled very roughly, and a wide range of materials would be acceptable. PRACTICAL CONSIDERATIONS FOR FROUDE MODELING

Sometimes it is stated that the scale model needs to be built such that evely dimension is an exact fraction of the full-scale facility, but not every small detail of the full-scale facility needs to be replicated. Little objects such as small light fixtures, light switches, doorknobs, rnoldings, smoke detectors, and sprinklers would not be expected to impact the gross flow of smoke, and these objects can be neglected. In the absence of well-developed criteria about the size of such little objects, it is

suggested that objects less than about 9 in. (0.23 m) can probably be neglected. As already stated, the size of the model needs to be chosen appropriately SO that the viscous effects are negligible and the Reynolds number (n4) can be ignored. The model needs to be large enough so that the flow is fully turbulent at locations of interest. The general rule is that the smallest length that can support such turbulent flow is about 1 ft (0.3 m). The following example illustrates the selection of the scale for a model. Consider that it is desired to realistically determine flows in openings from the atrium to the communicating spaces. These openings are 8 ft (2.4 m) high and 12 ft (3.7 m) wide. Consider that this height is the smallest location where fully developed flow is needed. Then this opening in the scale model should not be less than about I ft (0.3 m) in the model. Thus, the model should be one-eighth scale..The scale for each modeling applicat~on should be determined by consideration of what flows are important to simulate. While some heat transfer effects can be paitially preserved by scaling the thermal inertia using Equation (1 5.49), only very rough scaling is needed as discussed above. Glass is often used for some of the walls to make visualizat~on smoke flow possible. of As previously stated, Froude modeling is appropriate for smoke temperature away from the flame. Froude modeling is appropriate for simulation of smoke transport of an atrium fire where the flames do not reach the ceiling. The flames would not be expected to be modeled realistically, but the smoke flows away from the flames would be expected to be realistically modeled Froude modeling would also be appropriate for simulation of smoke flows in a building from a fully developed room fire. Because of high temperatures, the modeling is not appropriate for the fire room or any flames that might be flowing from that room, but realistic modeling of smoke flow away from the fire room . , would be espected.

CHAPTER 16

Computational Fluid Dynamics
omputational fluid dynamics (CFD) consists of dividing a space into a large number of control volun~esand using a computer to calculate approximate solutions to the governing equations for each control volume. These control volumes are often called cells. CFD is sometimes called field modeling, and a thorough knowledge of this topic requires an understanding of graduate level fluid dynamics. The intent of this chapter is to provide some understanding of the capabilities and limitations of CFD with respect to fire applications and smoke flow in atria for readers who do not have such an understanding of fluid dynamics. Many computer CFD programs have been developed that are capable of simulation of fire-induced flows. Friedman (1992) discusses ten such codes. Several o f these are general purpose codes that are commercially available. For more information about CFD, readers are referred to Anderson, Tannehill, and Pletcher (1984); Abbott and Basco (1 989); Hoffmann (1989); Hirsch (1988; 1990); Kumar (1983); and Markatos (1986). This chapter addresses the basic concepts of CFD modeling, including the co;:ventional approximations for the effects of turbulence. A new approach is also discussed that has the potential to accurately simulate turbulence. This chapter is intended to provide background information fcr people interested in the possibility of using CFD modeling for smoke management applications. Before using a CFD model, people should, at a minimum, learn about the theory, capabilities, and limitations of their specific model. This chapter is not a con~prehensivetreatment of the subject, and persons desiring to write a CFD model need to go to other Soul-ces.

C

Equations are used in this chapter for the purpose of explaining concepts. Accordingly, units are not given for variables in this section. However, all of these equations are valid for S1 units or any other homogeneous unit system (see Appendix A). For details and equations of particular CFD models, readers should see the documentation for the model.

EXAMPLE APPLICATIONS
The atrium in the Lloyds Building is 1 12 ft by 38.1 ft (34.2 m by 1 1.6 m) and about 240 ft (72 m) high (Figure 16.1). For this shaft-like atrium, the plume would contact the sides of the ztrium, making the conventional analytical methods for atria (Chapter 14) inappropriate. Waters (1989) used CFD modeling to simulate smoke movement in the atrium and design a smoke management system. The Luxor Hotel and Casino is a 30-story pyramidal structure, 200 ft (61 m) high with a 500 ft by 500 ft (150 m by 150 m) base, as shown in Figure 16.2. Because of the shape of the structure, the conventional analytical methods are not applicable. CFD modeling was used to design a unique smoke management system for this structure (Evans 2001). The system consisted of supply fans at the base of the structure and an exhaust at the top. The supply fans produced an upward spiral flow that kept the smoke away from the balconies. Other atrium smoke management applications of CFD modeling are presented by Sinclair (2001) and Mills (2001). A fire application of CFD is the NlST analysis of flame blow-down 'at a Navy tire tighter training facility (Forney and Davis 1992). The facility was used to recreate the effects ofjet pool fires on the deck of a ship. A 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 burners that simulated the fires. When there was little or no wind, the flames would be 2.5 to 3 m (8 to 10 ft) high. However, under moderate winds, the smoke and flames would be blown down into the space below the grating. A commercial CFD model was used to evaluate possible alternative solutions to this problem and arrive at a solution. One alternative was a wall intended to shield the facility, and the performance of this is shown in Figure 16.3. The wall did not prevent flame blowdown. The solution consisted of a combination of a fence in place of the wall, plus pressurization of space under the grating. When installed, this solution qualitatively perfomied as predicted.

CFD modeling was used by Klote (1999b) to study the interaction between HVAC airflow and smoke detector activation. A FORTRAN subroutine was written to modify a commercially available CFD model to calculate detector activation time. Figure 16.4 shows the calculated activation time 2 in. (50 mm) below the cei!i~g of an open plan office. As expected, the activation time is delayed in front of the slot diffusers. The surprise was that activation time also was delayed near the ceiling return. Comparisons c f room fire data with CFD simulations 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 comparison 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

(a) Typical Elevation

,Toilet

Capsule

/

Elevat0r

Figurc 16.2 CFD aua!i.sis nns used to d e s i p t17e smoke ~ ~ 1 a 1 7 n g ~ ~ ~ 1 ent rthe Lrr-vor sj,stetir ~ t Hotel atid Cosit70 (ndoptedfi-otn Evnm [200 l]).

Open Boundary

,

Elevator

Exterior Elevators

'
Figure 16.3

Flame Blow-down

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(ax+b,)+j(a,,+by)+k(az+bz) .
The dot product of vectors a and b is

(16.1)

Figure 16.4

Lines of smoke detector activation time (seconds) 2 in. (50 mm) below the ceiling of an open plan ofice (adapedfiom Klote 19996).

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). In Cartesian coordinates, the velocity vector, U, is expressed as

Experimental Data: 0 Simulated Data: 0

R 0 U 0

1
300 6 2

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. The conservation equations are written below.

Mass:

Erne (S)

Figure 16.5

Co~nparison CFD-sitnufated temperof atures and experi~nentaldata for flow under a bea117ed ceiling.

FUNDAMENTAL EQUATIONS This section lists a very general form of the fundamental 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 viscosity. Quantities that have only magnitude are scalars. Examples of scalars are mass, density, area, and temperature. A quantity that has niagniiude and direction is a vector, provided that it obeys the law of addition of vectors illustrated in Figure 16.6. In Cartesian coordinates, the vector a can be represented as a = io, + ja,, + k n l , \\here a, a,, and a, are the magnitudes of vector a in the I,y, and.z directions, respectively, and i, j, and k are unit vectors in the S, !), and z directions, respccti\.ely. In a similar manner, the vector b can be represented as b = i b , + jb,. + kb, . The

In Equations (16.5), (16.6), and (16.7), the terms that include the dynamic viscosity, p, are called the viscous terms. Because the dynamic viscosity is on the order of 1 0 . ~ ft-l S-' ( 2 ~ 1 0 kg~m-Is-'), the viscous Ib terms are relatively small. While these viscous terms can be neglected for many applications, they provide the mechanism for converting kinetic energy to thermal energy.

Chapter 16-Computational Fluid Dynamics

Energy: Equation of State
In addition to the conservation equations, an equation relating pressure, temperature, and density is needed. Such equations are called equations of state. The perfect gas law is frequently used in CFD applications:

The dissipation function, 0,represents the timerate at which energy is dissipated per unit volume through the action of viscosity. The dissipation function tends to cause flows to go to rest, and this function is expressed as

Variables in the conservation equations above are p = density, p = pressure, p = dynamic viscosity, k = thermal conductivity, T = absolute temperature, Cp = constant pressure specific heat,

g"'
I

=

heat release rate per unit volume,

A> V

=time, = body force in the j direction, and
= vector differential operator called del. Del is defined as

V = i-+j,+k-

a

a

a
a 2

2.r

0,lJ

If a scalar function, 9 is operated on by del, it is , written 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 of state, and, in this chapter, the broader meaning is used. It is not possible here to discuss all of the assumptions involved in the derivation of N-S equations. However, two of the more important ones are the continuum assumption and the stress-strain relationship for a Newtonian fluid. For an exhaustive derivation of the N-S equations, readers are referred to Aris (1962). At the level of generality presented above, it is beyond the state of technology to solve the N-S equations exactly. Even with the simplifying assun~ptions of incompressibility or of Boussinesq tlow,'%t is still not possible to solve the three-dimensional N-S equations exactly. Exact solutions have been obtained for a laminar flow in simple geometries, and the most notable application of inexact solutions is to boundary layer flows. Exact and inexact solutions are discussed in several texts (White 1974; Sherman 1990; Schlichting 1960; Schetz 1993). By experimental verification of these solutions, the N-S equations have themselves been
12. Boussinesq flow is an approximation to compressible flow that extends the incompressible flow equations by considering density as a function of ternperature (Sherman 1990, p. 83).

The material derivative acting on y, is defined as

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~tion vectors a and b of

Figure 16.6

Gtq~hical illccs1rc7fiotiof addiliotl of vecfors.

Principles of Smoke Managemen't

verified for laminar flow. There is no such verification for turbulent flow. However, CFD simulations of turbulent flow based on the N-S equations as discussed below often correspond well with experimental data.

Boundary Conditions
In CFD modeling, the conditions at the boundaries of the flow field need to be stipulated. Figure 16.4 is an example of such boundaries. In this figure, the boundary conditions consist of (I) solid wall (and ceiling), (2) plane of symmetry, (3) velocity boundary, and (4) open boundary. The most common condition for the solid walls is zero velocity at the wall surface. At the solid surfaces (walls, floors, and ceilings), the tangential component of velocity is generally considered to be zero. This boundary condition is referred to as the no-slip condition. The symmetry boundary can be compared to a mirror 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 symmetry boundary provided that the direction of such flow is in the plane of the boundary. 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 stipulated at an angle such that the flow \vould become attached to the ceiling. Open boundaries are also called pressure boundaries because the pressures outside the domain are stipulated. The CFD model calculates the flows at these boundaries from the pressures. To improve accuracy, the 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 Figure 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 boundaries and the ceiling return.

vortices on the flow tilrough open stairwell doors is discussed. The turbulent nature of fire plumes is apparent to anyone who has seen one. In CFD modeling, the turbulent effects that are smaller than the cell size cannot be simulated by solving the N-S equations. Turbulence modeling has been developed to account for these small-scale effects, and turbulence modeling is based on Reynolds averaging as discussed below.

Reynolds Averaging
Conventional CFD modeling is based on the assumption that the fluctuations associated with turbulence are random. There is evidence that this may not be true, but the CFD technology that has been developed using this assumption has considerable utility. The timeaveraged quantity is defined as

The "randomly" changing variables are considered to be made up of a time average plus a fluctuation. These are written as

Figure 16.7 illustrates average and fluctuating velocity in the X-direction. The time average of a fluctuating quantity is zero.

I t follows that

and

TURBULENCE
Because air cannot be seen, people are not aware of the turbulent nature of the flow that surrounds them. A cup of coffee can be used as a simple way to illustrate turbulence. Short!y after coffee has been poured, the surface of the coffee in the cup appears still and the coffee 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 applications is turbulent. In Chapter 6, the etfect of stationary

While p = 0, the product of two tluctuating quantities is not equal to zero 0 ). CFD with turbulence modeling was initially developed for incompressible flow, and so Reynolds a\.eraging is illustrated here for that flow. Equation (16.4) becomes

(3

-

This is the conservation of mass equation, and it is also called the continuity equation.

Chapter 16 -Computational Fluid Dynamics

Time

(a) Steady Flow

(b)

Unsteady Flow

Time

Figure 16.7

Velocity components in the X-direction

With the N-S equations, there is one unknown variable for each equation, and this one-to-one correspondence 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 N-S equations. In order to solve the Reynolds averaged This is called the Reynolds averaged .continuity N-S equations, additional equations are necessary. This equation. Observation of Equation (16.20) shows that it addition of equations to the Reynolds averaged N-S is similar to Equation (16.31) except that the variables equations is called turbulence modeling. are replaced by average terms plus some fluctuating The additional equations are empirical, and they components. can be algebraic equations, ordinary differential equaFor 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 twoa = - ag + - 2 p ,E-- - equation model, and Reynolds stress models arc threea~ a X [ 3 ( t r ( 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 models 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 turbulence. and it is

Applying the averaging to Equation (16.20) yields a time-averaged conservation equation:

Turbulence Modeling

-

: E)] ;

a - a , ' uV1" + -[ p[C" + -; ) l--p ' l l ' - p-- - p'11.v a a), ax
)_

l

k = -I( u ' u f 2

+ v'v' + I ~ . ' I I . ' )

.

(16.24)

The rerm E is the turbulence dissipation rate,

Equation (16.23) is the Reynolds averaged xmo~nenturn equation for incompressible flow. In the same manner, the rest of the governing equations can be averaged to form a set of Reynolds averaged N-S equations.

where

I,
C

=
=

dissipation length; constant.

Principles of Smoke ~ a a ~ e m e n ' ;

In addition to the above constant, there are several others, and the values in the 1974 paper by Launder and Spalding are almost the same ones used for most applications today. Nam and Bill (1993) developed modified k-E coefficients that improve predictions slightly. It is well known that CFD models with k-Eturbulence underpredict the diameters of fire plumes. For a discussion about extension of the k-E model to compressible flow and a general presentation about the mathematics of CFD to fire applications, the reader is referred to Kumar (1983).

Governing Equations

-

For decades, the National Institute of Standards and Technology (NIST) in Gaithersburg, Maryland, has been conducting basic research in CFD modeling of fire and smoke transport. This research provides evidence that turbulence is not made up of random fluctuations, but that it has a structure of eddies and vortices that can be simulated by computational methods. The CFD model, Fire Dynamics Simulator (FDS), is a product of the NIST CFD research effort. FDS is in the public domain, and it can be obtained from NIST at no cost. Rehm and Baum (1978) of NIST developed a unique form of the N-S equations that incorporates compressibility effects sufficient for thermally driven flows of smoke and hot gases generated in a fire. Also at NIST, the FDS model was developed (McGrattan et al. 2000; McGrattan and Fomey 2000), which solves the NS equations developed by Rehm and Baum. FDS uses a numeric solwr based on fast Fourier transforms (FFT) that reduces calculation timc to a fraction of that with conventional models. Because FDS does not rely on Reynolds average equations and turbulence modsling, it has the potential to simulate turbulence more realistically. The key to the level of flow detail that can be produced 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 greater the flow detail and the greater the running time. A 100,000 cell applicdtion with a conventional CFD model \ \ i l l take about the same time as a 1,000.000 cell applicatior, with FDS all other things being equal. It is no wonder that FDS is capable of very realistic flow simulations. A limitation of FDS is that tlie cells n!ust be rectangular with aspect ratios that arc not very large. This linitation is due to the FFT solver. This means that the grid cells cannot conform to curved shapes or \valls that are not at righr angles ro the orhcr \valls. Ho\ve\.er. rclativcly small cells can be uscd ro form s~t~-faccs steps \\.irh 10 approsi~nale thesc shnpcs.

FDS solves the governing equations listed earlier plus conservation of species equations. Species equations are used in many CFD programs to simulate the flow of various gases (such as 02, N2, CO, etc.). In FDS, p in Equation 16.13 is replaced by a constant "background" pressure, p,. The use of this "background" pressure in the equation of state is referred to as the low Mach number assumption. The Mach number is 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 the advantage that the time step in the numerical solution is bound only by flow speeds and not the speed of sound. As already stated, FDS does not rely on Reynolds average equations and turbulence modeling. For FDS simulations where the grid resolution is not fine enough to capture the mixing process at all relevant scales, a large eddy simulation (LES) can be used. In the LES, the viscosity is modeled as

where
pLEs viscosity used in LES, =

CS
A
IS1

=

empirical constant, magnitude of the stress tensor.

= length on the order of a grid cell, and
=

The square of magnitude of the stress tcnsor is

In Equation (16.26), ~ L E S the maximum of p or is p ( ~ ~ ~ ) ' lwhere p is the dynamic viscosity of the gas 5l or the \veighted average of the dynamic viscosities of the constituent gases when individual species are simulated. Use of the viscosity of Equation (16.26) accounts for viscous effects on a scale smaller than the cell size.

u s e and Verification of FDS
While it is impossible to completely verify any CFD model, FDS was used extensively for fire applications for several years before i t was released to the public. A sample of these applications of FDS arc McGrattan, Baum, and Rehm (1996, 1999); Davis, Notarianni, and McGrattan (1996); and Baum. McGral-

Chapter 16-Computational Fluid Dynamics

tan, and Rehm (1996, 1997). These simuiations included room fires, warehouse fires, townhouse fires, airplane hangar fires, sprinklered fires, unsprinklered fires, and fires with draft curtains. e An FDS simulation of a fire ~ l u m has realistic pulsating eddies, as shown in Figure 16.8, and average velocities and temperatures of this simulation agree well with experimental data, as can be seen in Figure 16.9.

in all of the large commercial packages. Pre-processing software helps the user generate the grid, specifj. the boundary conditions, and define other input parameters. For geometries that are somewhat complicated, grid generation capabilities can save significant amounts of user time. FDS is an exception in that there is no pre-processing software, and the data are read directly by the processing software before simulation. Because FDS allows only rectangular cells, the lack of pre-processing software is not a disadvantage. some CFD codes allow the user to write FORTRAN routines that become part of the processing software. 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 transport 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.
r
I

SOFTWARE
Typically, software for CFD applications falls into one of the following groups: (I) pre-processing, (2) processing, and (3) post-processing. Not all CFD packages have all of these sofnvare groups, but they are available

Figure 16.8

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 variables 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 simulated fire results in the generation of a file of over 60 million numbers. It is not feasible to examine so many Plume s i t i ~ i ~ I ( ~Le d ty FDS 01 017 i17.510171it7 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
e

- Calculated By FDS Model

o

Correlation of McCaffrey

0.6

3 1 S 0 -

2.0 1.5

ko 2 0.5
C

.2 -

' 0.4 :

'0 1.0

2

0.3 0.2 0.1 0 -3 -2 -1 0 1 2
3

9

3
E F
-2 -1
Q) a

0.5 0 -3

0

Cross Plume Figure 16.9

2 Distance, r/D'

1

3

Cross Plume Distance, r a g

Ratlial p ~ ~ f i l e q - 1.e10c.i~. teri7l)emtlo.e f o r a pool f i r e (ndaptedj.oci7 c/ otrcl Brizoil. McCr~attctr~, Rchrtr [lYY7]). arid

Principles of Smoke Management

Post-proccssing software is used for graphic display of data from the files. This display can vary fiom simple two-dimensional black-and-white contour plots to threedimensional color movies where the view can move around the flow field. Smokeview (Forney and McGrattan 2000) is the post-processing package that was specifically developed for FDS.

the public domain so that it can be obtained at no cost. Because it has been specifically developed for fire applications, it does not require that the user write cdmputer code in order to make routine smoke transport simulations. The general purpose commercial CFD models have typically taken tens or hundreds of person years to develop. These models are rich in features that allow them to be used for a wide range of applications, such as aircraft design and air movement in rooms. To simulate the buoyant flow associated with fire applications, it is generally accepted that the N-S equations need to be capable of simulating compressible flow. This can either be by using the fully compressible version of the N-S equations, or a "partial" compressible form as is done in FDS. Because the Boussinesq approximation does not accurately simulate compressibility effects at high temperatures, it generally is not considered appropriate for fire applications. However, for lowtemperature smoke at a distance away from a fire, the Boussinesq approximation can yield useful results.

CONSIDERATIONS ABOUT CFD
For a CFD simulation, the choice of CFD software, form of N-S equations, grid, turbulence model, fire model, boundary conditions, and other factors need to be chosen so that they are appropriate for the application. For a specific application, there are many CFD approaches that can provide useful information. Successfbl CFD modeling requires experience and an understanding of the technology. However, a few comments can be made. A CFD model developed specifically for fire applications or a general purpose commercial CFD model can be used for smoke management applications. The FDS model is a product of the US. government and is in

CHAPTER 17

Commissioning and Routine Testing
ommissioning and routine testing are needed to ensure that smoke management svstems will function as intended during fire. situations. Many o f the problenls encountered during acceptance testing stem from nlisconceptions about the system's ability to control smoke and misunderstandings about the intended function of a particular system. This chapter deals with determination of \\.hat type of measurements should be made and how to make them. Further, most smoke management s y s t e m should require adjustments of supply airflow rates or pressure relief vent openings to accommodate the particular leakage characteristics of the buildings in which they are located. These adjustments can be made in conjunction with the acceptance test. Commissioning procedures for new systems shouId include

C

Acceptable performance for a new system does not ensure that. years later, the system will perform acceptably during a fire. Components deteriorate with age and can be inadvertently damaged during building modifications. For these reasons, annual testing of smoke management systems is recommended to provide a level of assurance that the system will function as desired in the event of a fire. The methods orroutine testing should be the same as those o f acccptance testing. Deficiencies encountered during routine testing should be corrected as soon as possible. These corrections may include balancing to correct for changes in building leakage and patching of gaps, holes, and cracks in barriers of smoke management systems. Inspection, testing, and balancing o r smoke management systems can be conducted by the building owner, the construction contractor, a testing and balancing contractor, a code official, or some other person. Regardless of who performs the work, all measurements made should be recorded for inspection. Typically, code officials check for compliance with local codes, whereas building oivners and engineering and architectural firms also conduct inspections, checking for compliance with the contract documents. Conlmissioning and routine testing are simplified when conlpliance is checked or measured against some standard. Contract documents can be prepared to reflect agreement between the owner, designer, builder, and code official as to what constitutes acceptable performance. In the following discussion in this chapter and tlle referenced appendices, the phrase "as specified" is used to mean as specified in accordance with a standard or standards that have been agreed upon by the parties involved.

:

inspection of the system components, testing of the system operation. and balancing of the system to ensure perfornlance.

Testing and balancing of the system can be conducted together. Frequently, local authorities want to be present at a formal cxeptance test of a smoke management system. Such a formal acceptance test should be preceded by inspection, testing, and balancing. Before acceptance testing, the owner, designer, builder, and code officials should agree upon what constitutes acceptable perrormance. Acceptable performance should be based on measurements of appropriate design parameters. such as pressure differences, air velocities, and flow rates. If appropriate, the capabilities of the system to prevent smoke feedback into protected spaces should be tested.

Chapter 17-Commissioning and Routine Testing

General information about testing and balancing of HVAC systems is provided by SMACNA (1993) and ASHRAE (1999). Additional information about commissioning smoke management systems is available from ASHRAE Guideline 5 (ASHRAE 1994). INSPECTION Inspection consists of checking smoke management system components, which include barriers, air-moving equipment, controls, and electric power supply. For pressurized stairwells, the barriers consist of the stairwell walls, ceiling, and doors. For zoned smoke control, the barriers are the walls, floor, and ceiling separating the zones. For elevator smoke control, the barriers would be of the elevator shaft and its lobbies. Walls, partitions, floors, and ceilings should be checked for obvious and unusual openings that could adversely affect smoke control performance. Gaps around doors should be as specified. Automatic door closers that are part of the snloke control system should be of the type specified. The air-moving equipment to be checked includes ducts, access openings in ducts, fans, fire dampers, and snloke dampers. The materials and construction of ducts should be checked. Dampers should be the type specified and installed where and in the manner specified. Components of the control system should be checked to determine that they art: as specified. Any special electrical power requirements, such as standby power or dual feeds, should be checked. General inspection procedures are presented in appendix G, and these are only intended as a guide for the development of specific procedures for individual smoke management systems. TESTING A N D BALANCING S M O K E MANAGEMENT SYSTEMS If standby power or other emergency poLver has been provided for the operation of the smoke management system, acceptance testing shall be conducted with emergency power and normal power. During one test started under normal power conditions, the normal power shall be shut off to determine the ability of the system and all associated systems to properly operate under standby power or other emergency power.

activation should be activated by putting an appropriate initiating device into alarm.

Pressurized Stairwells
With all stairwell doors closed, pressure differences across each stairwell door should be measured. Then one door should be opened, and pressure difference measurements made at each closed stairwell door. This should be repeated until the number of doors opened equals the number of doors required by the code authority to be opened.

I,

,

Elevator Smoke Control
The smoke control test depends on the type of elevator smoke control system. In general, the design pressure differences should be measured at the appropriate locations for the particular design. If the intent oi' the system is to pressurize enclosed.elevator lobbies, pressure differences across closed lobby doorsto the building should be measured. If the intent of the system is to pressurize the elevator shaft to prevent smoke flow through it, the pressure differences across the elevator doors should be measured.

Atrium Smoke Management
Generally, designs for smoke management in large spaces will be based on providing specific exhaust capacity from the upper region of the space and not exceeding an airflow at openings into the space. These flows and velocities should be measured. Upper layer temperature of the space should be measured to ensure that considerations about smoke stratification in the atrium are appropriate.

Caution about Smoke Bombs
A caution needs to be given concerning the use of smoke bombs. For zoned snloke control systems, a major problem with most snloke bomb tests of smoke control systems is that they are intended to test some improvement of smoke conditions in the zone where the fire is located. This is based on the mistaken belief that smoke control is capable of producing a significant improvement in tenable conditions within the zone where the fire is located. Smoke bomb tests for zoned smoke control are described here in general terms so that the reader car. recognize this type of test and understand the problems with them. The smoke control system is put in operation. In the zone that is being exhausted, a number of smoke bombs are ignited. The smoke bombs produce all of their smoke in a few minutes, and the zone rapidly fills with smoke. Because the smoke control system is exhausting air and chemical smoke from this zone, the concentration of chemical smoke decreases with time. If

Zoned Smoke Control
For zoned smoke control systems, one zolie should be put into the smoke control mode, and the pressure differences at the boundaries of that zone should be measured. After snloke control operation in that zone has becn deactivated, another zone should be tcsted in the same manner. This should be repeated until all smoke zones have been tested. Systems with automatic

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 system 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 smoke. At present, no information is available relating the smoke obscuration of chemical smoke to that of smoke from building fires. These problems can be overcome by modifications to the test method. However, this would not yield a test relevant for a smoke control system. Because a smoke control system is intended to maintain pressure differences at the boundaries of the smoke zone, the system should be tested by measuring pressure differences. A very serious probleni with this type of smoke bonib test is that it can give building occupants and fire service officials a false sense of security. The test can lead people to wrongly think that zoned smoke control is capable of achie\.ing a significant improvement in tenable conditions \vitliin tlie fire space. Testing the performance of zoned snioke control systems with chemical smoke from smoke bonibs is not realistic for flaming fires in unsprinklered buildings. Possibly the flow of unheated chernical snioke is similar to that of smoke from a sprinklered fire or a smoldering fire. However, the gases produced by a large flaming fire in a building are in the range of 1200°F to 1 800°F (650°C to 1000°C). For chemical smoke to produce tlie same buoyant pressure differences as these gases, the chemical smoke would have to be heated to tlie same temperatures. This is impractical because of the associated danger to life and property. Smoke bonib tests of atriuni smoke management are similar to those for zoned smoke control, except that the smoke bombs are set off in tlie atrium with the atriuni system actwated. While atriuni smoke management systems are intended to iniprove snioke conditions in tlie atrium, cliemical snioke from smoke bombs is so different from hot snioke from a flaming tire tliat these tests are not realistic. As witli zoned smoke control, the concern witli this type of snioke bomb tcst is that it can give building occupants and firs scrvice otlicials a false sense of security. Chemical smokc or a traccr gas (sucli as sulhr liexatlouridc) can bc used to test for snioke fecdback into supply air. Thc gcncral proccdurc for testing witli

Figure 17.1

Setup for measzu.ing pressure difference across a door.

chemical smoke is described here. A number of smoke bombs are placed in a metal container, and all bombs are simultaneously ignited. The container is located near an exhaust inlet in the smoke zone being tested so that all of the chemical smoke produced by the bombs is drawn directly into the exhaust airstream. If chemical smoke is detected in the supply air, its path should be determined, the path should be blocked, and then tlie smoke feedback test should be conducted again. Smoke bombs or other tracers can be useful in locating the leakage paths tliat sometimes defeat a smoke control system. For example, if the construction of a stainvell is unusually leaky, pressurization of that stairwell may not be possible with fans sized for construction of average tightness. Chemical smoke generated within the stairwell will flow through tlie leakage paths and indicate their location so that they can be caulked or sealed. General testing procedures are presented in appendices H and I. As with Appendix G, these are intended as a guide for tlie development of specific procedures for individual smoke control systems. DIFFERENTIAL PRESSURE INSTRUMENTATION The setup for measuring pressure difference across a door is illustrated in Figure 17.1. The convention of this sctup is that tlie instrument is on thc low-pressure side of tlie door. Experience has shown that adherence to a particular convention reduces conlision and, thus, the potential for human error. A hose connected to the highpressure port of tlie instrument goes through a gap underneath and is terminated with a tee on tlie liiglipressure side of the door. The tec is used to minimize any pressure errors duc to air velocity. Alternatively, tlie tube can end \vitIiout a tee, provided it is located so that 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 diameter works well for most cases. A narrow gap may result in a pinched tube, invalidating any measurement. Small diameter metal tubing can sometimes be used in such cases, particularly through the gaps of some gasketed doors. The differential pressure instrument should have a sensitivity of at least 0.01 in. H20(2.5 Pa), and generally a range from 0 to 0.25 in. H20(0 to 62 Pa) is sufficient. Occasionally an instrument with a range of 0 to 0.50 in. H20 (0 to 124 Pa) is needed.

smoke control systems. The gage should have a stand so that it can be set on the floor or other flat surface. The instrument has a zero adjustment that can correct for minor deviations in surface level. Thus, an instrument level adjustment is unnecessary. A differential gage should be calibrated.

Electronic Pressure Transducers
Most electronic differential pressure transducers are of the diaphragm type. Changes in pressure across a diaphragm cause diaphragm displacement, which can be measured by strain gages, piezoelectric elements, inductance pickups, capacitance pickups, etc. These transducers require electrical power and should be calibrated periodically. Many instruments are commercially available with the necessary sensitivity and in appropriate ranges. For many applications, a major advantage of these instruments is that they have analog voltage output suitable for monitoring by computer data acquisition systems. For field tests conducted with hand-held instruments, analog output seems to have little advantage. For this reason and because of the expense of these instruments, they are not generally the instrument of choice for smoke control testing.

Inclined Liquid Manometer
An inclined manometer with a liquid reservoir is illustrated in Figure 17.2. This device indicates pressure by the height of a column of liquid. Before any measurements, the instrument must be ~djusted that it is level. so Generally, the scales of inclined manometers are compensated for the liquid rise in the reservoir so that the pressure difference can be read directly. The zero level of these instruments can be adjusted by adding or removing liquid from the reservoir or by changing the position of the scale. Because the measurement principle of these devices is so fundamental, it is believed that com~nercially available inclined manometers are of sufficient accuracy for smoke control testing \vithout independent calibration.

FLOW INDICATION AND MEASUREMENT
During acceptance and routine testing, there are many situations for which the knowledge of flow direction is desirable. Such cases abound during the initial checkout of a smoke control system. A piece of paper placed in front of an air grille provides an immediate and simple indication of flow and flow direction. Airflow \\.ill cause a hanging strip of tissue paper to noticeably detlect diagonally at flow velocities as low as 15 fpm (0.08 mls). Smoke flow from a punk stick or a cigarette can also be used to detect such low airflows. This section discusses flow measurement appropriate for smoke management applications, but more
Note: The absence of liquid makes his type of gage more convenient than an inclined manometer.

Differential Pressure Gages
A gage without liquid has the advantage of convenience over the inclined manometer. Bourdon-tube gages are the most common type of pressure gages, but the friction of the mecha~~ical linkages of these instruments limits sensitivity. No Bourdon-tube gage is known wit11 sullicient sensitivity for smoke control application. However, a magnetically coupled gage, as illustrated in Figure 17.3, is sufficiently sensitive, and these gages have been used extensively for field tests of

High

Low -Pressure

,Liquid Reservoir

Gage /Stand

1

Level Adjustment

'zero Adjustment

Figu re 1 7.3

~Magr?c~icallv coupled d$ereritial p/-aswe gage.

H

Figure 17.4

Flow hood being used to measure volzrmel~*icfloiv ceiling-tnounted supply. of

detailed information on- this topic is available from For ASHRAE (1997~). field measurement of. fan. flows, .. see AMCA (1 990b).

Figure 17.5

rL;;
= = = =

Principles of Smoke Management

rw1
Equal Rectangular Areas

Centers of Areas Where Velocity is Measured

Open Doorway or Section of Corridor

Flow measurement t t a ~ e r s e for cot-ridots and open doonvaj~s.

Volumetric Flow Rate
Airflow velocity through an open doonvay or across a section of a corridor is generally far from uniform. Such flow is frequently characterized by the presence of large stationary vortices, especially flow through open stairwell doorways. This makes accurate determination of volun~etricairflow difficult unless extreme care is taken. Fortunately, airflow through large openings is not the major principle of smoke control for most building systems. It follows that for the majority of smoke control systems for buildings, flow measurements in doorways and corridors are not necessary. However, flow measurements of the supply and exhaust of a smoke control system are often desired, and sometimes information about the flows through doorways is also needed. Flow can be measured directly by using a flow hood or determined indirectly from a set of velocity are available measurements. Flow h o ~ d s comn~ercially 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. Figure 17.4 illustrates a flow hood being used to measure flow from a ceiling supply. The device can also be used to measure exhaust flows, and it can be oriented for use with wall-mounted inlets and outlets. Provided that the pressure loss through the hood is small compared to the duct losses, the accuracy of flow hoods is believed to be in the range of 10% to 15%. When volun~etricflow is obtained from velocity measurements, a traverse should be made. Traversing open doonvays or sections of corridor can be done in a manner similar to that for rectangular ducts, as illustrated in Figure 17.5. Velocity readings should be taken in the center of equal areas over the cross section. For flow in ducts, the cross section should be divided into 16 to 64 equals spaces. Because of the likely variations of velocity in doonvays and corridors, these openings should be divided into 30 to 64 equals spaces. Flows through doorways in particular should be checked for stationary vortices by use of smoke from a punk stick or cigarette. If stationary vortices exist, care should be taken that flows against the main flow direction should be assigned negati\ L values when calculating the average velocity. The volumetric flow rate for a rectangular duct or other opening is calculated from the formula

where

p H
W

volumetric flow rate, cfin (m3/s); height of opening, ft (m); \vidth of opening, It (m); average velocity, rpm (rills).

U

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 stationary 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 velocity is 300 fpm (1.5 mls). Using Equation (17. I), the flow is 6300 cfm (3.0 m3/s). e

I
Note: The aidlow causes vane to deflect diagonally, and the

Velocity Measurement
Pitot tubes, deflecting-vane anemometers, and thermal anemometers are commonly used to measure airflow in building. These instruments are discussed in the following sections.

/Pin

Suppo"ng the Vane

/ /

Deflecting Vane Anemometer
The deflecting vane anemometer consists of a vane hung from a pin such that air velocity ivill cause a diagonal 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.

these instruments are low in mst and compact. they are useful for soot checks and rough kstima~es.

(a) Principle of operation of deflecting vane anemometer.

Pitot Tube
The stagnation pressure, is the pressure that results when moving gas is brought to rest. An expression for this pressure can be obtained from Bernoulli's equation,
/

(b) Deflecting vane anemometer in use.

i

where
Ps,trs -

Figure 17.6

Deflecting vatle anernometex

- stagnation pressure of the gas, in. HzO (mm

/?~IOI =

HzO): static pressure of the moving gas. in. H20 (mm

For an idealized frictionless fluid, the coefficient

C has a value of one, and the value differs for real tlui: Pitot tubes measure the stagnation pressure of a d . moving gas, and some pitot tubes incorporate static pressure taps, as illustrated in Figure 17.7. Manufacturers of pitot-static tubes frequently supply information about the correction factor as a function of flow velocity or of Reynolds number. The velocity from Equation (1 7.2) can be expressed as

H2

0 ):

c,,, =
P
= = =

correction factor (di~nensionless); densit? of gas, lb/ft3(kdm3); gas \.eloci~y, ( d s ) ; rpm 1097 (4.427).

U

K,,,

Principles of Smoke Manageme*
. .

!

i 1
f

1

where

U

!

velocity, fpm (&S); = pressure difference from manometer, in. H20 (mm
=

H2O);

j 1

p

=

density of air, lb/ft3 (kg/m3);

Tzps Evenly Spaced Around Circunference

Cp, = correction factor (dimensionless);
A pitot-static tube can be used to measure velocities in the range of 400 to 2000 fpm (2 to 10 m/s) when connected to an inclined manometer. With an electronic differential pressure transducer, a pitot tube can be used in the range of 200 to 3000 fpm (l to 15 d s ) .

U
Pstq

Hoses to Differential Pressure Instrument

Figure 17.7

Pitot-static tube.

I

I

11 E x a m ~ l 17.2 e

11The menometer connectki to a pitot-static tube reads 0.08
in. H 2 0 2 . 0 3 mm H20), the air density is 0.075 lb/ft3 (1.2
kg/m3), and the pitot tube correction factor is 1 .OS. The velocity calculated from Equation (17.3) is l l l 0 f p n ~ (5.62 m/s).

Velocitv from Pitot-StaticTube Reading

Thermal Anemometer
Thennal anemon~eters(also called hot-wire anenlometers and hot-film anemometers) are available in two types: constant-current and constant-temperature. Both types have a velocity probe with a filum (fine

wire). For the constant-current type, a filum is subjected to a constant electrical current and the temperature of the filum depends upon the convective cooling of air flowing past it. Thus, temperature is a measure of velocity. The constant-temperature type uses the same principle in a different way. The electrical current through a filum is adjusted so that its temperature remains constant. For this instrument, current is a measurement of velocity. Hand-held, battery-powered, temperaturecompensated thermal anenlonleters are commercially available fbr air temperatures normally encountered in building heating and cooling systems. Such instruments have ranges of approximately 10 to 5000 fpm (0.05 to 25 m/s) with accuracies of about 5%.

Nomenclature
area of opening, leakage path, shaft, test sample, atrium, or fire, ft2 (m2) wind exponent (dimensionless); ga'p thickness"; dilution rate* effective flow area, ft2 (m2) distributed effective flo\v area per unit height, ft (m) vent area, ft2 (m2) temperature factor, in. H20/ft(Pa/m);distance from the opening to the balcony edge, ft (m); or constant on N-gas model 127,000 for CO2 < 5% and -38,600 for CO2 > 5% flow or discharge coefticient (dimensionless); gas or contaminant concentration; or specific heat, Btu/lb°F (J/kg°C) flow coefficient for elevator car (dimensionless) flow coefficient for exponential flow equation, ft3 min-I (in. H20)-" (m3 S-' Pa-"); constant pressure specific heat, Btu/lb°F (kJ/ kg°C) pitot tube correction factor, (dimensionless) conductivity factor, ft"' /s1I2(m''' / s " ~ ) flow coefficient of the vent (dimensionless) wind pressure coefficient (dinlensionless) depth ofsmoke layer below botto~nofexhaust inlet, ft, (m); or distance from the doorknob to the knob side of the door. ft (m) density oroccupant tlon.. pers/ft' (pers/ni'), or dilli~sion coellicient
l. :

depth ofsmoke layer below the smoke vent, ft (m) equivalent diameter of flow path* diameter of fire, ft (m) hydraulic diameter, in. (m) minimum smoke layer depth to prevent plugholing, ft (m) maximum specific flow, pers/min.fi (pers/s.m) diameter of visible axisymmetric plume, ft (m) effect of exposure (ppmmin) total door opening force, Ib (N) friction factor of shaft or duct (dimensionless) flow rate perdmin, (pers/s) fractional effective dose (dimensionless) force to overcome the door closer and other friction, Ib (N),or Froude number (dimensionless) specific flow, pers/min.ft (pers/sm) the flow factor, fpm ( m k j acceleration of gravity height ofatrium, shaft, opening, ceiling above the fire, upwind wall, balcony ceiling above top of hel, fi (m) distanceabove neutral plane, ft (m); depth of smoke layer, ft (m) conv&tive heat transfer coefticient, ~ t u / l i 's "F (w/m2 "C) height o r fuel (m) height limit, l (m) i

E

=

F

=
= = = =

f
F,
FED

F,

Units dcpc~itl [tic spccilic equation on

height of wind measurement, ft (m) velocity factor (dimensionless) thermal inertia ofa material (product ofk, p, and C), ~ t in. h-' ~f' OF^ (kw2 m4 K-~s) u t' friction factor of stairwell (dimensionless) length of gap*; length of shaft or duct, ft (m); height of section of stairwell, ft (m) lethal concentration, Ib ft-3 min (g m-3 min) niass of the sprinkler, Ib (kg) mass flow rate, Ib/s (kgls) niass concentration of fuel burned, lb/@ (g/m3) mass of fuel burned or consumed, Ib (g) masitnum mass rate of exhaust without plugholing, Ibis (kgls) inass concentration of particulate lb/ft3 (g/m3) mass of particdates produced, Ib (g) moment of the door closer and other friction, Ib ft (N m) mass of smoke, Ib (kg) fire location factor (diniensionless);,flow esponent (din~ensionless) N-Gas model indicator (dimensionless) number of exhaust inlets (dimensionless) pressure ditTercnce (dimensionless) flo\~- (dimensionless) rate perimeter of duct or shaft, ft (m); or population pressure' absolute atniosplieric ambient pressilre* Prandtl number (diniensionless) stagnation pressure of the gas, in. H20 (mm static pressure of the moving gas, in. H20 (mm
H201

RTI RTI,

=

S Smi, T
I
lac1

response time index, ftlR S" (m1' sIn) = virtual RTI, ft" S ( ' m S'') ' ' = visibility, ft(m)
I

1,

tQ

Tcp

T/
f/

minimum separation between inlets, ft, (m) temperature14; transmittance (dimensionless); or emc~ztion (minutes) time = time* = time of sprinkler actuation, S = time for population to pass through constraint = transport time lag of ceiling jet, S = absolute centerline axisymrnetric plume R temperature, O (K) = absolute temperatureof the fire space, O (K), or R temperature of gas in full scale facility, "F ("C) = time in full scale facility S
= =
= = =

I

;

T
'g

temperature of the gases in the exhaust fan* growth time, s time to incapacitation du2 to thermal exposure, min modeled evacuation time for an egress route* or time in scale model, s temperature in scale model, "F ("C) temperature of ambient or outside ai; plume temperaturet transport time lag of plume, S absolute reference temperature, "R ("K) velocity, fprn (mls) critical air velocity to prevent smoke backflow, fpm measured wind velocity. fpni (mls) volumetric flow rate, cfm (m3/s) volunietric flow in scale model, ft3/s (m3/s) volume of smoke in a space or test chamber, f$ (m3 factor for CO2-induced hyperventilation

[m

=
=

T,,

T,
Tp
f

= = =
=

T,.

U Uk U,,,,,
V

= =
= =

c,,,
VC

= = =

~vind pressure, in. H20 (Pa) heat release densily. Btu/s ft2 (kw/mz) heat release of thc tire, Btds (kW) H R R at sprinkler actuation, kW (3tds) convective heat release rate, Btuk (kW) heat ge~ieratcd i t h i n tlic control volume, Btu's w (kW) radiant heat release of the fire. Bttds (kW) gas constant (J/kg K) radius or horizontal distance from centerline of plun~e. (m) fi Reynolds number (di~nensionless) ~cparation distancc fiom the centcr of the fire to a target. l1 (m)

''C

V,

W
I

S

)'P

z
z,

limiting average air velocity, fpni ( 4 s ) volume of smoke, ft3 (m') = width ofdoor, corridor, opening, or plume. ft (m) = effective width of stair, in. (m); width of the opening from the fire rooni, ft (m); or spray density, gpndft2 (nids) = depth of gap in flow direction, in. (m), or distance of light travel or the path length. ft (m). = particulate yield (dimensionless) = elevation. ft (m) = mean flarne height, ft (m)
=
=

''Units depend on the specilic equation.

Principles of Smoke Management

= height above balcony, ft (m)

=

time interval15 ambient, "F ("C)

maximum height at which plume is considered buoyant, ft (m) = virtual origin correction of the axisyrnmetric plume, ft (m) = extinction coefficient ft-I (m''); fire growth coefficient, kw/s2 ( ~ t u l s ~or;plugholing ) exponent (dimensionless) = specific extinction coefficient, f t 2 ~ b (m21g) = exhaust location factor (dimensionless) = wind boundary layer height in the vicinity ofthe building, .fi(m), or optical density per unit distance, ft-' (m-') specific optical density (dimensionless) = change in energy of the smoke layer, kJ = chemical heat of combustion, Btullb (kJkg)
=

A mn = minimum temperature rise of plume above T i
E

= =

turbulence dissipation rate; or roughness of the inside surface of the duct, fi (m) wall heat transfer fraction (dimensionless) percentage obscuration (dimensionless) or failure rate absolute viscosity kinematic viscosity, fi2/s (m2/s) dimensionless group of variables density, lb/@ (kg/m3)

/1

=

p v

=

=
=

l l
p

-

= =

density of gases in exhaust fan, lb/@ (kg/m3) pfan = outside or ambient density, lb/fi3 (kg/m3) p .
T
= =

Ap A

mass loss of test sample, Ib (g) = mass optical density, @/lb (m21g) = boundary layer height in the vicinity of the wind anemometer, f (m) t = pressure difl'erence, in. H 2 0 (Pa) = overall pressure difference from one side of a building to another due to wind effect
=

X,

= =

time constant, s (S) allo\vable fraction reduction in mass flow rate through fan convective fraction radiatiue fraction

z,.

l 5 Units depend on the specific equation.

References
Abbott, M.B., and D.R. Basco. 1989. Co~iip~ttational fluid dynamics: AI? introd~rction, e~igi~leers. for New York: Wiley. Achakji, G.Y., and G.T. Tamura. 1988. Pressure drop characteristics of typical stairshafts in high-rise buildings. ASHRAE fi-ansactiom (I): 1223- 1236. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Ahonen, A., M. Kokkala, and H. Weckman. 1984. Burning characteristics of potential ignition sources of room fires, Research Report 285. Technical Research Center of Finland. Alpert, R.L. 1972. Calculation of response time of ceiling-mounted fire detectors. Fire Technology 8(3): 181-195. AMCA. 1987. Air Sj~tetns, AMCA Pub. 200. Arlington Heights, Ill.: Air Movement and Control Association. AMCA. 1990a. Fans and Sjate~iis, AMCA Pub. 20 1-90. Arlington Heights, 111.: Air Moiwnent and Control Association. AMCA. 1990b. Field pe1fofor7nance of ~ n e a s ~ o n , r e f~ i t o fat7 sjate~ns, AMCA Pub. 703-90. Arlington Heights, Ill.: Air Movement and Control Association. Anderson, D.A., J.C. Tarmehill, and R.H. Pletchec 1984. Con7pr/atio17al flitid nreci1m7ic.s and /lea/ t/-ansfer.. New York: Hemisphere. Aris, R. 1962. Vecto~*.s, so^.^, a d //re l)o.sic e q m / i o ~ n ten off7rcid 1nec/7atiic~. New York: Dover. Arpaci. V.S., and A. Agarwal. 1999. Scaling laivs of turbulent ceiling fires. C o n ~ I ~ r ~ . s mu1 iF1c111re 1 16: /io~ 84-93. ASHRAE. 1994. Guideline 5, Coni1nissio17ing sliioke managenlent systems. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers. AS HUA E. 1997a. 1997 ASHRAE handbook-F~rndamentals, Chapter 15, Airflow around buildings. Atlanta: American Society of Heating, Refrigerating and Air-conditioning En,'o~neers. ASHRAE. 1997b. 1997 ASHRAE handbook-F~r17dan~entals,Chapters 25 and 26. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. AS HRAE. 1 9 9 7 ~ .1997 ASHRAE handbooX--F~o7damentals, Chapter 14, Measurement and instruments. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. ASHRAE. 1999. 1999 ASHRAE handbook-Applicalions, Chapter 36, Testing, adjusting and balancing. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. AS HRAE. 2000a. ASHRAE Handbook-*ems and Applications. Atlanta: American Society of Heating, R~frigerating Air-conditioning Engineers. and AS H U E . 2000b. ANSIIASHRA E Standa~d149-2000,
f Laboratory Methods o Testing Falls Used to Exhaust Smoke in Smoke Manageme~u Sjs/e/ns.

Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers. ASH RAE. 2000c. ASHRAE H a n d b o o k - E ~ L ~ ~ ~ I I I E I I ~ , Chapter 18-Fans. Atlanta: American Society of Heating, ~ e f r k p a t i n g Air-conditioning Engiand neers. ASME. 1987. A~iierican / a n d a ~ sqf+ codeJor- elevas d
t o n . escaluto~s,drc~nb~t~ni/enls /no\.i17g 11nlkS. trlrd

References

AI 7. I. New York: American Society of Mechanical . Engineers. Aynsley, R.M. 1989. The estimation of wind pressures at ventilation inlets and outlets on buildings. ASHRAE Transactions 95(2): 707-72 1. Babrauskas, V. 1990. The cone calorimeter Heat Release in Fires, Chapter 3. V. Babrauskas and S.J. Grayson eds. Elsevier, New York. Babrauskas, V., et al. 1991. Toxic measurement for fire hazard analysis, NIST Special Publications 827. Gaithersburg, Md.: National Institute of Standards and Technology. Babrauskas, V. 2002. Burning rates. SFPE Handbook of Fire Protection Engineering, Chapter 3-1. Bethesda, Md.: Society of Fire Protection Engineers. Barrett, R.E., and D.W. Locklin. 1969. A computer technique for predicting smoke movement in tall buildings. Symposiuni on'movement of smoke on escape routes in buildings, pp. 78-87. Watford, Herts, U.K.: Watford College of Technology. Baum, H.R., K.B. McGrattan, and R.G. Rehm. 1996. Large eddy simulation of smoke movement in three dimensions. Interflam 1996, 7th International Interflam Conference, March 26-28, 1996, Cambridge, England. Baum, H.R., K.B. McGrattan, and R.G. Rehm. 1997. Three dimensional simulations of fire plume dynamics. Fire Safety Science Proceedings, 5th International Symposium, March 3-7, 1997, Melbourne, Australia. Bazjanac, V. 1977. Simulation of elevator performance in high-rise buildings under conditions of ernergency. Human response to tall buildings, ed. by D.J. Conway. Stroudsburg, Pa.: Dowden, Hutchinson & Ross, pp. 3 16-328. Bennetts, I.D., et al. 1997. Simulated shopping centre fire tests. Fire code Reform Centre, Broken Hill Proprietary Company Ltd., Melbourne, Australia. Berl, W.G., and B.M. Halpin. 1980. Hunian fatalities from unwanted fires. Johns Hopkins APL Technical Digest l(2): 129-134. Best, R., and D.P. Demers. 1982. Investigation report on the MGM Grand Hotel fire, Las Vegas, Nevada, November 2 1, 1980. Quincy, Mass.: National Fire Protection Association. Beyler, C. 1986. Fire plumes and ceiling jets. FireSafey J., l l , JulyISeptember, p. 53-75. Block, J.A. 197 1. A theoretical and experimental study of nonpropagating free-burning fires, 13th Symposium (International) on Combustion, 1970 August 23-29, Salt Lake City, Utah, Combustion Institute, pp. 97 1-978.

Blockley, W.V. 1973. Biology Data Book. Bethesda, Md.: Federation of American Societies of Experimental Biology. BOCA. 1999. National Building Code. Country Club Hills, Ill.: Building Oficials and Code Administrator International, Inc. Borisenko, A.I., and I.E. Tarapov. 1968. Vector and tensor analysis with applications, transulated by R. A. Silverman. New York: Dover. Bryan, J.L. 2002. Behavioral response to fire and smoke. SFPE Handbook of Fire Protection Engineering. Quincy, Mass.: National Fire Protection Association. Buckingham, E. 1915. Model experiments and the form of empirical equations. Transactions ASME 37: 263-296. Bukowski, R. W., et al. 1989. ~echnical Reference Guide for- HAZARD I Fire Hazard Assessment Method. NIST Handbook 146, Vol. Il. Gaithersburg, Md.: National lnstitute of Standards and Technology. Bukowslc~,R.W., and R.C. Spetzler. 1992. Analysis of the Happyland Social Club fire with HAZARD I. Joltrnal of Fire Protection Engineering 4(4): 1 1717 1
1Jl.

,

,

Burden, R.L., J.D. Faires, and A.C. Reynolds. 1981. N1m7erical analpis, 2d ed. Boston, Mass.: Prindle, Weber & Schmidt. Butcher, E.G., P.J. Fardell, and P.J. Jackman. 1969. Prediction of the behavior of smoke in a building using a computer. Syn~posium movement of smoke in on escape routes in buildings, pp. 70-75. \Vatford, Herts, England: Watford College of Technology. Butcher, E.G., T.H. Cottle, and T.A. Baily. 1971. Smoke tests in the pressurized stairs and lobbies of a 26story ofice building. Building Setvice Engineet- 39: 206-2 10. Butcher, E.G., et al. 1976. Smoke control by pressurization. Fire Engineers Journal 36(103): 16-19. Carhanan, B., H.A. Luther, and J.O. Wilkes. 1969. Applied n~tmerical methods. New York: John Wiley & Sons. Cetegan, B.M., E.E. Zukoski, and T. Kubota. 1982. Entrainment and flame geometry of fire plumes, Ph.D. thesis of Cetegan, California Institute of Technology, Pasadena. Chow, W.K., and A.C.W. Lo. 1995. Scale modelling studies on atrium smoke movement and the smoke filling process. Journnl o f Fire Prolecrion Et@neering 7(2): 56-64. Chow, W.K., and W.M. Siu. 1993. Visualization of smoke movement in scale models of atriums. JOLWnal of Appliecl Science 3(2): 93- 1 1 1.

Principles of Smoke Management'

.-

.-.-.

CIBSE. 1995. Relationships for smoke control calculations, Technical Memoranda TM19: 1995. London: Chartered Institute of Building Services Engineers. Cooper, L.Y. 198 1. Estimating safe available egress time fiom fires. National Bureau of Standards (U.S.), NBSIR 80-2 172. Cooper, L.Y. i982. A mathematical model for estimating available safe egress time in fires. Fire Mater. 6(3/4). Cooper, L.Y. 1985. ASET- computer program for A calculating available safe egress time. Fire Safe& Journal 9: 29-45. Cooper, L.Y., et al. 1981. An experimental study of upper hot layer stratification in full scale multiroom fire scenarios. Paper 8 I-HT-9, American Society of Mechanical En,'olneers. Cooper, L.Y., and G.P. Forney. 1990. The consolidated compartment fire model (CCFM) computer code application CCFM. VENTS, Part I: Physical basis. Cooper, L.Y., and D.W. Stroup. 1982. Calculating safe egress time (ASEV-A cornputer program and user's guide. National Bureau of Standards (U.S.), NBSIR 82-2578. 66th CRC. 1985. Handbook of Chernislty and Pl~ysics, ed. Boca Raton, Fla.: CRC Press, Boca Raton. Cresci, R.J. 1973. Smoke and fire control in high-rise ofice buildings-Part 11: Analysis of stair pressurization systems. Symposium on experience and applications on smoke and fire control at the ASHRAE Annual Meeting, Louisville, Ky., pp. 1623. Atlanta: An~erican Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Davis, W.D. 1999. The zone model JET: A model for the prediction of detector activation and gas temperature in the presence of a smoke layer, NISTIR 6324. Gaithersburg, Md.: National lnstitute of Standards and Technolo~y. Davis, W.D., and L.Y. Cooper. 1989. Estimating the environment and the sprinkler links in compartment fires with draft curtains and fusible-link-actuated ceiling vents, Part 11: User guide for the computer code LAVENT, NISTIR 89-4122. Gaithersburg, Md.: National lnstitute of Standards and Technology. Davis, W.D., G.P. Forney, and J.H. Klote. 1991. Field modeling of room fires. Gaithersburg, Md.: National lnstitute of Standards and Technology, NISTlR 4673. Davis, W.D., K.A. Notarianni, and K.B. McGrattan. 1996. Comparison of fire modcl predictions with experiments conducted in a hangar with a 15 meter ceiling. Gaithersburg, Md.: Nabonal Institute of Standards and Technology, NlSTlR 5927.

DeCicco, P.R. 1973. Smoke and fire control in high-rise office buildings-Part I: Full-scale tests for establishing standards. Symposium on experience and applications on smoke and fire control at the ASHRAE Annual Meeting, Louisville, Ky., pp. 915. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Dias, C. 1978. Stairwell pressurization in a high-rise commercial building. ASHRAE Journal 20(7): 2426. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Dols, W.S., K.R. Denton, and GN. Walton. 2000. CONTAMW user manual. Gaithersburg, Md.: National Institute of Standards and Technology. Drysdale, D. 1985. An introduction to five dynamics. New York: Wiley. Dyrbye, C., and S.O. Hansen. 1997. Wind loads on structures. New York: Wiley. Evans, D.D. 1993. Sprinkler fire suppression algorithm for HAZARD. Gaithersburg, Md.: National Institute of Standards and Technology, NISTIR 5254. Evans, D.D., and D.W. Stroup. 1986. Methods to calculate the response time of heat and smoke detectors installed below large unobstructed ceilings. Fire Technology 22(1): 54-65. Evans, D.D., D.W. Stroup, and P. Martin. 1986. Evaluating thernlal fire detection systems (S1 Units), NBSSP 71. Gaithersburg, Md.: National Bureau of Standards. Evans, D.H. 2001. Two Las Vegas hotcl and casinos demonstrate new ways of thinking about smoke control. HPAC Engineering 73(1). Evers, E., and A. Waterhouse. 1978. A computer model for analyzing smoke movement in buildings. Borehamwood, Herts, U.K.: Building Research Est. Fang, J.B. 1980. Static pressures produced by room fires. National Bureau Standards, NBSIR 80-1984. Feng, J.B., and J.N. Breese. 1980. Fire development in residential basement rooms. National Bureau of Standards, NBSIR 80-2 120. Forney, G.P. 1989. Personal communications at the Center for Fire Research. Gaithersburg, Md.: National Institute of Standards and Technology. Forney G.P., W.D. Davis, and J.H. Klote. 1992. Simulating the effect of beamed ceilings on smoke flow, Part I. Comparison of Numerical and Experimental Results. Gaithersburg, Md.: National lnstitute of Standards and Technology, NISTIR 4994. Forney, G.P., and W.D. Davis. 1992. Analyzing strategies for elimination of flame blow-down occurring in the Navy's 19F4 fire fighting trainer. Gaithersburg, Md.: National lnstitute of Standards and Technology, NlSTlR 4825.

References

Forney, GP., and K.B. McGrattan. 2000. user's guide for smokeiiew version 1.0- tool for visualizing A fire dynamics simulation data. Gaithersburg, Md.: National Institute of Standards and Technology, NISTIR 65 13. Forney, G.P., R.W. Bukowski, and W.D. Davis. 1993. Field modeling: Effects offlai beamed ceilings on detector and spr-inkier- response. Quincy, Mass.: National Fire Protection Research Foundation. Frantzich, H. 1996. Study of movement on stairs during evacuation using video analysing techniques. Lund Institute of Technology. Friedman, R. 1992. An international survcy of computer models for fire and smoke. Journal qf Fire Protection Engineering 4(3): 8 1-92. Fruin, J.J. 1987. Pedestrian planning and design, revised ed. Mobile, Ala.: Elevator World Educational Services Division. Gann, R.G 200 l . Toxic hazard of building products and furnishings. Proceedings of the Internatio17al Fire Sajety Conference, March 11-14, 2001, San Francisco, Fire Retardant Chemicals Assoc., Lancaster, Pa., pp. 85-9 1. Gross, D. 1962. Experiments on the Burning of Cross Piles of Wood, Journal of Research of NBS, Vol. 66C, NO. 2, pp. 99- 105. Gross, D., and W.L. Haberman. 1988. Analysis and prediction of air leakage through door assemblies. Fire Safety Science, Proceedings of the 2nd International Symposium, Tokyo, Japan, pp. 169-1 78. Gwynne, S., and E.R. Galea. 1999. A review of the methodologies and critical appraisal of computer models used in the sin~ulationof evacuation from the built- environment. Bethesda, Md.: Society of Fire Protection En,' omeers. Gwynne. S., et al. 1999. Escape as a social response, Research report. Bethesda, Md.: Society of Fire Protection Engineers. Haber, F. 1924. Funf Vortrange aus den jaren 19201923, Verlag v011Julius Spanger, Germany. Hadjisophocleous, G.V., G.D. Lougheed, and S. Cao. 19,99. Numerical study of the effectiveness of atrium smoke exhaust systems. .4SHRAE Tramac1io17.7105(1): 699-71 5. Hagglund, B., R. Jansson, and K. Nireus. 1985. Smoke filling experiments in a 6 x 6 ~ 6meter enclosure. Stockl.~olm:National Defense Research Institute of Sweden. Handbook of' HVAC Desigtl. 1990. Nils Grimm and Robert Rosaler, eds. New York: McGraw. Harland, W.A., and W.D. Woolley. 1979. Fire fatality study-University of Glasgow. Building Research Establishment, Borehamwood Information Paper IP
18/79.

Harlow, EH., and P.I. Nakayama. 1968. Transport of turbulence energy decay rate. Los Alamos, N.M.: Los Alamos Scientific Laboratory of the University of California. Hartzell, GE., A.F. Grand, and W.G Switzer. 1990. Toxicity of smoke containing hydrogen chloride, fire and poiymers-Hazards identification and prevention, Nelson, G.I., ed., ASC Symposium Series 425, American Chemical Society, Washington, D.C., pp. 12-20. Hay, GE. 1953. Vector and tensor analusis. New York: Dover. Heskestad, G. 1972. Similarity relations for the initial convective flow generated by fire. Paper 72-WN HT-1 7, American Society of Mechanical Engineers. Heskestad, G. 1975. Physical modeling of fire. J. Fire &? Flatnnzability 6: 253-272. Heskestad, G. 1983. Virtual origins of fire plumes. Fire Safety Jozrrnal 5(2): 109- 114. Heskestad, G. 1984. Engineering relations for fire plumes. Fire Safe02 Journal 7(1): 25-32. Heskestad, G. 1986. Fire plume air entrainment according to two competing assumptions, Twenty-first Symposium (International) on Combustion, Combustion Institute, Pittsburgh, Pa., pp. 111-120. Heskestad, G. 1989. Inflow of air required at wall and ceiling apertures to prevent escape of fire smoke. FMRC J. I. OQ4E4. RU. Nonvood, Mass: Factory Mutual Research Corporation. Heskestad, G. 2002. Fire plumes. SFPE handbook offit-e protection engineering. Boston, Mass.: Society of Fire Protection Engineers. Heskestad, G., and M.A. Delichatsios 1977. Environments of fire detectors-Phase 1: Effect of fire size, ceiling height and materials. Volumes I and 11-Measurenients (NBS-GCR-77-86; NBS-GCR-77-95). Gaithersburg, Md.: National Bureau ofstandards. Hirsch, C. 1988. Nzrmerical computation ofinternal and external flows, Vol. I: Fztnda~nentals ntrmerical of disct-etization. New York: Wiley. Hirsch, C. 1990. Numerical computation ofinternal a17d external flo~rs. Vol. 2: Con7p1rtationalmethods for inviscid a17d visco~rsjlo~vs. York: Wiley. New Hoffmann, K.A. 1989. Conzplrtational fluid ajwan~ics for engineers. Austin, Tex.: Engineering Education System. Hottel, H.C. 1961. Fire modeling. The use of models in fire research, Publication 786, National Academy of Sciences, Washington DC, pp. 32-47. Huggett, C. 19SO. Estimation of rate of heat release by means of oxygen consumption measurements. Fzre and Materials 4(2): 6 1-65. ICBO. 1997. Uniform building code. Whittier, Calif.:

Principles of Smoke Management,

l

i
ICC. 2000. International building code. Idelchik, I.E. 1986. Handbook of hydraulic resistance, 2d edition. New York: Hemisphere Publishing. Incropera, F.P., and D.P. DeWitt. 1985. Fundamentals of heat and mass transfer, 2nd ed., Wiley, New York. Jin, T. 1974. Visibility through fire smoke, in main reports on production, movement and control in buildings, Japanese Association of Fire Science and Engineering, pp. 100-153. Jin, T. 1975. Visibility thorough fire smoke. Report of the Fire Research Institute of Japan 5(42): 12- 18. Jin, T. 1985. Irritating effects of fire smoke on visibility. Fire Science and Technology 5(1): 79-90. Jones, W.W. 1983. A review of compartment fire models. Gaithersburg, Md.: National Bureau of Standards, NBSIR 83-2684. Jones, W.W., and X. Bodart. 1986. Buoyancy driven flow as the forcing function of smoke transport models. Gaithersburg, Md.: National Bureau of Standards. NBSIR 86-3329. Jones, W.W., et al. 2000. A technical reference for CFAST: An engineering tools for estimating fire growth and smoke transport, NIST TN 143 1. Gaithersburg, Md.: National Institute of Standards and Technology. Jorgensen, R. 1983. Fan engineering. Buffalo, N.Y.: Buffalo Forge Co. Joyeux, D. 1997. Natural fires in closed car parks: Car fire tests, INC - 961294d - DJ/NB. Juillerant, E.E. 1964. Jacksonville Hotel disaster. NFPA Quarterly 57(4): 309-3 19. Kandola, B.S. 1986a. A wind tunnel building model for the investigation of smoke movement problems. Fire Safety Journal 1 O(3): 203-209. Kandola, B.S. 1986b. Comparison of wind tunnel pressure measurements and smoke movement computer predictions inside a five-storey model building. Fire Safety Journal lO(3): 229-238. Kandola, B.S. 1986c. The effects of simulated pressure and outside wind on the internal pressure distribution in a five-storey building. Fire Safep Journal 1 O(3): 2 11-227. Karlsson, B., and J.G. Quintiere. 2000. Encloswe fire dynatnics. New York: CRC Press. Keating, J. 1982. The myth of panic. Fire Jo~rmal76(3): 57-61, 147. Keating, J.P., and E.F. Loftus. 1977. Vocal alarm systems for high-rise buildings-A case study. Mass Enietgeticies 2: 25-34. Kendik, E. 1985. Technology Report 85-4. Bethesda, Md.: Society of Fire Protection En,'wneers. Klote, J.H. 1980. Stairwell pressurization. ASHRAE Tratlsactiot7s 86(1): 604-623. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. Klote, J.H. 1982. A computer program for analysis ,of smoke control systems. Gaithersburg, Md.: National Bureau of Standards, NBSIR 82-2512. Klote, J.H. 1983. Elevators as a means of Fire Escape. ASHRAE Transactions 89( I B): 362-378. Klote, J.H. 1984. Smoke control for elevators. A S m E Journal 26(4): 23-33. Klote, J.H. 1986. Smoke control at Veterans Administration hospitals. National Bureau of Standards (U.S.), NBSIR 85-3297. Klote, J.H. 1987. A computer model of smoke movement by air conditioning systems (SMACS), NBSIR 87-3657. Gaithersburg, Md.: National Bureau of Standards. Klote, J.H. 1988. An analysis of the influence of piston effect on elevator smoke control, NBSIR 88-375 1. Gaithersburg, Md.: National Bureau of Standards. Klote, J.H. 1989. Considerations of stack effect in building fires. Gaithersburg, Md.: National lnstitute of Standards and Technology, NlSTlR 89-4035. Klote, J.H. 1990. Fire experiments of zoned smoke control at the Plaza Hotel in Washington, D.C. ASHRAE E-ansactions 96(2): 397-4 16. Klote, J.H. 1995. Design of smoke control systems for elevator fire evacuation including wind effects. 2nd Symposium on Elevators, Fire, and Accessibility, Baltimore, Md. April 19-21. New York: American Society of Mechanical En,'Olneers. Klote, J.H. 1999a. An engineering approach to tenability systems for atrium smoke management. ASHRAE i5-atisactions 105(1): 716-729. Klote, J.H. 1999b. CFD sirnulations of the effects of HVAC-induced flows on smoke detector response. ASHRAE i5-ansactions 105(1): 395-409. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Klote, J.H., D.M. and Alvord. 1992. Routine for analysis of the people movement time for elevator evacuation. Gaithersburg, Md.: National Institute of Standards and Technology, NlSTlR 4730. Klote, J.H., and X. Bodart. 1985. Validation of network models for smoke control analysis. ASHRAE Tramactiota 91(2b): 1134- 1 145. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. Klote, J.H., and J.W. Fothergill. 1983. Design of smoke control systems for buildings. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Klote, J.H., H.E. Nelson, S. Deal, and B.M. Levm. 1992. Staging areas for persons with mobility limi-

!

-

References

tations. NISTIR 4770. National Institute of Standards and Technology. Klote, J.H., and GT. Tamura. -1986a. Elevator piston effect and the smoke problem. Fire Safefy Journal 1l(3): 227-233. Klote, J.H., and GT. Tamura. 1986b. Smoke control and fire evacuation by elevators. ASHRAE Transactions 92(1A): 23 1-245. Klote, J.H., and GT. Tamura. 1987. Experiments of piston effect on elevator smoke control. ASHRAE Transactioi7s 93(2a): 221 7-2228. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. Klote, J.H., et al. 1992. Feasibility and design considerations of emergency evacuation by elevators. National Institute of Standards and Technology, NISTIR 4870. Koplon, N.A. 1973a. Report of the Henry Grady fire tests. Atlanta: City ofAtlanta Building Department. Koplon, N.A. 1973b. A partial report of the Henry Grady fire tests (Atlanta, July 1972). Symposium on Experience and Applications on Smoke and Fire Control at the ASHRAE Annual Meeting, Louisville, Ky., pp. 3-8. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Kreith, F. 1965. Prit~ciples of heat fran.fer, 2d ed. Scranton, Pa.: International Textbook Co. Kumar, S. 1983. Mathematical modelling of natural convection in fire- state of the art review of the A field modelling of variable density turbulent flow. Fire and ,liaterials 7(1): 1-24. Launder, B.E.. and D.B. Spalding. 1974. The numerical computation of turbulent flows. Computer Methods in Applied ~~eclianics Engineering 3269-289. and Law, M. 1982 Air-supported structures: Fire and smoke hazards. Fire Prevention, Vol. 48, pp. 24-28. Law, M. 1986. A note on smoke plun~es from fires in multi-level shopping malls. Fire Safe& Journal 10: 197. Lawson, J.R., W.D. Walton, and W.H. Twilley. 1984. Fire performance of furnishings as measured in the NBS furniture calorimeter, Part I. Gaithersburg, Md.: National Bureau of Standards. Lees, F.P. 1980. Loss prevention in the process industries, Volume 2, Buttenvorths, London. Levin, B.C. 1996. New research avenues in toxicity: 7gas N-gas model, toxicant suppressants, and genetic toxicology. Toxicology. No. 1 15, pp. 89- 106. Levin, B.C. 7000. Personal communication between b i n and klote about the N-gas model and LCSo values, March 2000. Levin, B.C., et al. 1987. Effects of Exposure to Single of Multiple Conibinations of the predominant toxic

gases and low oxygen atmospheres produced by fires. Fundomental and Applied Toxicology 9: 236250. Levin, B.C., et al. 1988. Toxicological effects of diffkrent time exposures to fire gases: Carbon monoxide or hydrogen cyanide or to carbon monoxide combined with hydrogen cyanide or carbon dioxide. Society Of Plastics Industry, Polyurethanes 88, 31St Annual Technicalhlarketing Conference, Philadelphia, PA, October 18-2 1, pp. 240-248. Levin, BC., et al. 1989. Synergistic effects of nitrogen dioxide and carbon dioxide following acute exposures in rats, NISTIR 89-4105. Gaithersburg, Md.: . National Institute of Standards and Technology. Levin, BiC., et al. 1995. Further development of the Ngas mathematical model: An approach for predicting the toxic potency of complex combustion mixtures, fire and polymers 11: Materials and tests for hazard prevention. ACS Symposium Series. No. 599. August 2 1-26, 1994, Washington, D.C., American Chemical Society, Washington, DC., pp. 293311. Liu, H. 199 1. Wind engineering-A handbook for structural engineers. Englewood, N.J.: Prentice Hall. Lougheed, G.D. et al. 1999. Large-scale physical model studies for an. atrium smoke exhaust system. ,4SHRA E Transactions 105(1): 676-698. Lougheed, G.D., and G.V. Hadjisophocleous. 1997. Investigation of atrium smoke exhaust effectiveness. ASHRAE Transactions 1 03(2): 1- 1 5. Lougheed, GD., and GV. Hadjisophocleous. 2000. The smoke hazard from a fire in high spaces. ASHRAE Transactions 107(1). Lougheed, GD., C. McCartney, and B.C. Taber. 2000. Smoke movement for sprinklered fires. ASHRAE Transactions 106(1): 605-619. Lougheed, G.D., C. McCartney, and B.C. Taber. 2001. Sprinklered mercantile fires. ASHRAE Transactions 107(1). MacDonald, A.J. 1975. Wind loading on buildings. New York: Wiley. Madrzykowski, D. 1998. Office building fire research program: An engineering based approach to fire safety design, fire and material, 1998 International Conference, Interscience Communications, London, UK, pp. 23-33. Madrzykowski, D., and R.L. Vettori. 1992. A sprinkler fire suppression algorithm for the GSA engineering fire assessment system. Gaithersburg, Md.: National Institute of Standards and Technology, NISTlR 4833. Markatos, N.C. 1986. The mathematical modelling of turbulent flows. Applied Mathematical Modeling 1 O(3): 190-220.

.

.

.

Principles of Smoke Management,

1
1 .

L

McCaffrey, B.J. 1983. Momentum implications for buoyant diffusion flames. Combustion and Flame 52(2): 149-167. McGrattan, K.B., H.R. Baum, and R.G Rehm. 1996. Large eddy simulations of smoke movement in three dimensions, NISTIR 6030. Gaithersburg, Md.: National Institute of Standards and Technology. McGrattan, K.B., H.R. Baum, and R.G Rehm. 1999. Large eddy simulations of smoke movement. ASHRAE Pansactions 105(1): 426-436. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. McGrattan, K.B., and GP. Forney. 2000. Fire dynamics simulator-User's manual, N[ST[R 6469. Gaithersburg, Md.: National Institute of Standards and Technology. McGrattan, K.B. et al. 2000. Fire dynamics simulatorTechnical reference guide, N[ST[R 6467. Gaithersburg, Md.: National Institute of Standards and Technology. McGuire, J.H., and G.T. Tamura. 1975. Simple analysis of smoke-flow problems in high buildings. Fire Technology 1 1 (I): 15-22. Milke, J.A., and F.W. Mowrer. 1994. Computer aided desig~ifor smoke management in atria and covered malls. ASHRAE Tt-amac/iotu lOO(2): 448-456. Miller, R.W., and L.S. Han. 1971. Pressure losses for laminar floupin the entrance region of ducts of rectangular and equilateral triangular cross sections. Journal ofApplied Mechanics 38: 1083-1087. Mills, F.A. 2001. Case study of a fire engineering approach to a large, unsprinklered, naturally ve~ltilated atrium building. ASHRAE T,.misac/ions 107(1). Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Mitler, H.E. 1984. Zone modeling of forced ventilation fires. Cotm!~itst. Tech. 39: 83-106. Sci. Mitler, H.E. 1996. input data for fire modeling, 13th Meeting of the UJNR Panel on Fire Research and Safety, March 13-20, 1996, NlSTlR 6030, pp. 187199. Gaithersburg, Md.: National Institute of Standards and Technology. Mitler, H.E. 2000. Notes on burning of several Suel packages, Personal com~iii~nicatio~~ between Henri Mitler and John H. Klote on April 27, 2000. Mitler, H.E., and H.W. Emmons. 198 l . Documentation for CFC V, the Fifth Harvard Computer Code. Home Fire Projcct Tech. Rcp. #45, Harvard University. Mitler, H.E., and J.A. Rockeu. 1986. How accuratc is mathematical lire niodeling? Gaitlicrsburg. Md: National Bureau of Slandards. NBSIR 86-3459.

Moodie, K., et al. 1988. Fire at King's Cross Underground station, 18th November 1987-Part 11: Scale model fire growth tests, Incident Report, Health and Safety Executive, Derbyshire, U.K. Morgan, H.P. 1979. Smoke control methods in enclosed shopping complexes of one or more stores: A design summaqr,Building Reseerch Establishment. Morgan, H.P., and GO. Hansell. 1987. Atrium buildings: Calculating smoke flows in atria for smokecontrol design. Fire Safefy Journal 12(1987): 9-35. Morgan, H.P. et al. 1999. Design methodologies for smoke and heat exhaust ventilation. London: CRC Ltd. Morita, M., and M. Hirota. 1989. Numerical analysis and experiments of convective heat flow in fire compartments. Fire Science and Technology 9(1): 11-24. Morton, B.R., G. Taylor, and J.S. Turner. 1956. Turbulent gravitational convection from maintained and instantaneous sources. Proceedings of tize Royal Sociel)~ London 234: 1-23. of Mowrer, F.W. 1990. Lag times associated with fire detection and suppression. Fire Technology 26(3): 244-265. Mowrer, F.W., and R.B. Williamson. 1987. Estimating room temperatures from fires along walls and in corners. Fire Techt~olog?23(2): 133- 145. Mudan, K.S., and P..4. Croce. 1995. Fire hazard calculations for large open hydrocarbon fires. SFPE Handbook of Fire Ptorectiotl E11gir7eet-i11,o, edition. 2nd Quincy, Mass.: Sational Fire Protection Association. Mulholland, G. 2002. Smoke production and properties. SFPE HaridDook o f Fits Ptorection Engineering, 2d ed. Quincy, Mass.: National Fire Protection Association. Mulholland, G., et al. 1981. Smoke filling in an enclosure, Paper 81 -HT-8. New York: American Society of Mechanical Engineers. Murosaki, Y., H. Hayashi, and T. Nishigaki. 1986. Effects of passage width on choice of egress route at a T-junction in a building. 1st International Symposium on Fire Safety Science. NenTYork: Hemisphere, 593-600. Nam, S., and R.G. Bill. 1993. Numerical simulation of thermal plumes. Fits Sqfery Joitrna/ 2 l(3): 23 1 256. NBFU. 1939. Sniokr hazards of air-conditioning systems. NFPA Quarret-i~~ 33(2): 1 13-1 22. Nelson, H.E. 1987. An engineering analysis of the early stages of fire dr.t.elopment-The fire at the Dupont Plaza Hotel and Casino-December 3 1 , 1986. Gaithersburg. Xld.: National lnstitute of Standards SlSTlR S7-3560. and Techno1og~-.

References

Nelson and MacLennan. 2002. Emergency movement. SFPE Handbook of Fire Protection Engineering. Quincy, Mass.: National Fire Protection Association.. Newman, J.S. 1988. Princioles of fire detection. Fire Technology 24(2): 116-127. NFPA. 1976. Life safety code--Code for safety to life from fire in buildings and stiuctures, NFPA 101. Quincy, Mass.: National Fire Protection Association. NFPA. 1977. Life Safity Code. NFPA 101. Quincy, Mass.: National Fire Protection Association. NFPA. 1988. Reconmended practice for smoke control systems, NFPA 92A. Quincy: Mass.: National Fire Protection Association. NFPA. 1995. Fire safety concept tree, NFPA 550. Quincy, Mass.: National Fire Protection Association. NFPA. 1997. Fire Protection Haildbook. Quincy, Mass.: National Fire Protection Association. NFPA. 1999. National fire alarm code, NFPA 72. Quincy, Mass.: National Fire Protection Association. NFPA. 2000a. Recommended practice- for smoke control systems, NFPA 92A. Quincy, Mass.: National Fire Protection Association. ~ I in NFPA. 2000b. Guide for smoke I I I O I U ~ ~ I IS V S I~~ I I ~ S malls, atl-in, and large areas. NFPA 92B. Quincy, Mass.: National Fire Protection Association. NFPA. 2000c. Llfe safefy code, NFPA 101. Quincy, Mass.: National Fire Protection Association. Nowler, S.P. 1987. Enclosure environment characterization testing for the baseline validatiorl of cornputer fire sinlulation codes. Sandia National Laboratories. O'Neill, J.G., W.D. Hayes, and R.H. Zile. 1980. Full scale tests with automatic sprinklers in a patient room phase 11, NBSIR 80-2097. Gaithersburg, Md.: National Institute of Standards and Technology. Parker, W.J. 1982. Calculation of heat release rate by oxygen consunlption for various applications. Gaithersburg, Md.: National Bureau of Standards, NBSIR 82-2427-1. Pauls, J. 1977. Management and movement of building occupants in emergncies, DBR Paper NO. 788. Ottawa, Canada: National Research Council. Pauls, J. 1980. Building evacuation: Research findings and recomniendations. Fires and Hu~nanBehuviour, D. Canter, ed. New York: John Wiley. Pauls, J.L. 1988. Review of standards & codes plus recommendations for accessible means of egress, Funded by U S . Architectural & Transportatio~i Barriers Compliance Board. Dept. of Education. Washington, D.C.

Pauls, J; 2002. Mcvement of people. SFPE Handbook of Fire Protection Engineering. Quincy, Mass.: National Fire Protection Association. Paul~, J.L., and E. Juillet. 1989. Recent technical and social developments influencing the life safety of people with disabilities. Presentation at Pacific Rim Conference of Building Officials, Honolulu, Hawaii, April 9-13. Peacock, R.D., et al. 1991. Software user's guide for HAZARD I, Version l. l. NIST Handbook 146, Vol. I. Gaithersburg, Md.: National Institute of Standards and Technology. Peacock, R.D., et al. 1993. CFAST, the consdlidated model of fire growth and smoke transport. Gaithersburg, Md.: National Institute of Standards and Technolegy, NIST Technical Note 1299. Peacock, R.D., et al.. 1997. A user's guide for FAST: engineering tools for estimating fire growth and smoke transport. Gaithersburg, Md.: National Institute of Standards and Technology, NIST Special Publication 92 1 . Peacock, R.D., et al. 2000. A user's guide for FAST: engineering tools for estimating fire growth and smoke transport, NIST Special Publication 92 1. Gaithersburg, Md.: National Institute of Standards and Technology. Peacock, R.D., P.A. Reneke, R.W. Bukowski, and V. Babrauskas. 1999. Defining flashover for fire hazard calculations. Firs Safety Journal. Vol. 32, pp. 33 1-345. Pitts, W.M. 1994. The global equivalence ratio concept and production of carbon monoxide in enclosure fires, NIST Monograph 197. Gaithersburg, Md.: National Institute of Standards and Technology. Prasad, B. 1995. Testing fire dampers at varying air flows and temperatures. ASHRAE fin~aactions I 0 l (2). Predtechenskii, V.M., and A.I. Milinskii. 1978. Plannins for foot traffic flow in buildings. New Delhi: Arnerind Publishing Co. Press, W.H., et al. 1986. Monerical recipes. New York Cambridge Univ. Press, pp. 258. Proulx, G. 1995. Evacuarion times and movement times in apartment buildings. Fire Safety JOUI-nal 22924: 246. Proulx, G., and R. Fahy. 1997. The time delay to start evacuation: Review of tive case studies. 5th International Symposiun~ on Fire Safety Science. IAFSS, 783-794. Proulx, G., and^. Sime. 1991. To prevent panic in an underground enlcrgency. W l ~ y tell people the nor timth. 3rd International Symposium on Fire Safety Scicncc. Ncw York: Elsevier. 843-853.

Principles of Smoke Management

Purser, D.A. 2002. Toxicity assessment of combustion products, Chapter 1-13. SFPE Handbook of Fire Protection Engineering, 2d ed. Quincy, Mass.: National Fire Protection Association. Quarantelli, E.L. 1979a. Panic behavior in fire situations: Findings and a model from the English langliagi rescarch literature. Columbus, Ohio: Ohio State University. Quarentelli, E.L. 1979b. Fire papers from the panel session on panic, Second International Seminar on Human Behavior in Fire Emergencies, October 29 November 1, 1979, in Proceedings of Seminar, B. M. Levin and R. L. Paulsen (ed.), NBSIR 80-2970. Washington D.C.: National Bureau of Standards. Quintiere, J.G. 1989a. Fundamentals of enclosure fire "zone" models. Journal of Fire Protection Engineering l(3): 99- 119. Quintiere, J.G. 1989b. Scaling applications in fire research. Fire Safefy Journal 1 5(1): 3-29. Quintiere, J.G. 1998. PI-inciples of fire behavior. New York: Delmar Publishers. Quintiere, J.G., and M.E. Dillon. 1997. Scale model reconstruction of fire i n an atrium. 2nd International Symposium on Scale Modeling, June 23-27, University of Kentucky, Lexington, Kentucky, pp. 7386. Quintiere, J.G., B.J. McCaffrey, and T. Kashiwagi. 1978. ConlBustion Science and Tecl~nology 1- 19. 18: Ramachandran, G. 199 1. Informative fire warning systems. Fire Teclmology 27(1): 66-8 l . Read, R.E.H., and M.P. Shipp. 1979. An investigation of fire door closer forces, BRE Report. Borehamwood, Herts, U.K.: Fire Research Station. Rehm, R.G., and H.R. Baum. 1978. The equations of motion for thermally driven, buoyant flows. Journal of Resea~.cll NBS 83: 297-308. of Rilling, J. 1980. Mechanism and conditions of smoke control through a door opening. Champs Sur Marne, France: Centre Scientifique et Technique du Batiment (CSTB). Rockett, J.A., et al. 1987. Comparison of NBSIHarvard V1 simulations and full scale, multi-room fire test data, NBSlR 87-3567. Gaithersburg, Md.: National Bureau of Standards. Sander, D.M. 1974. FORTRAN IV program to calculate air infiltration in buildings, National Research CounciI Canada, DBR Computer Program No. 37. Sander, D.M., and G.T. Tamura. 1973. FORTRAN IV program to stimulate air movement in multi-story buildings, National Research Council Canada, DBR Conlputer Program No. 35. SBBCI. 1 999. Standard building code, Birmingham, Ala.

Schetz, J.A. 1993. Boundary layer analjsis. Englewood, Cliffs, N.J.: Prentice Hall. Schlichting, H. 1960. Boundary layer theory, 4th ed,., J . Kestin, Translato . New York: McGraw. Seader, J., and I. Einhom. 1976. Some physical, chemical, toxicological, and physiological aspects of fire smokes, NSF Report, Utah University. S FP E. 2002. Handbook of fire protection engineering. Boston, Mass.: Society of Fire Protection Engineers. Shaw, C.T., and GT. Tamura. 1977. The calculation of air infiltration rates caused by wind and stack action for tall buildings. ASHME Transactions 83(2): 145- 158. Atlanta: American Society of Heating, r Refrigerating and Air-conditioning Engineers, Inc. Shaw, C.Y., J.T. Reardon, and M.S. Cheung. 1993. Changes in air leakage levels of six canadian office buildings. ASHRAE Journal 35(2): 34-36. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Sherman, F.S. 1990. Escousflow. New York: McGraw. Shields, et al. 1996. Escape of disabled people from fire. A measurement and classification of capability for assessing escape risk, BRE Report 301, Borehamwood, Building Research Establishment. Siegel, R., and J.R. Howell. 1992. Thermal radiation heat transfer, 3rd ed., Hemisphere, Washington, D.C. Simcox, S., N.S. Wilkes, and LP. Jones. 1988. Fire at King's Cross Underground Station, 18th November 1987: Numerical simulation of the buoyant flow and heat transfer. Hanvell Laboratory, UK. Simcox, S., N.S. Wilkes, and I.P. Jones. 1989. Computer simulation of the flows of hot gases from the fires at King's Cross Underground Station. Institution of Mechanical Engineers (IMechE) King's Cross Underground Fire: Fire Dynamics and the Organization of Safety, London, UK, pp. 19-25. Simiu, E., and R.H. Scanlan. 1996. Wind effects on structures: Fundamentals and application to design, 3rd ed. New York: Wiley. Sinclair, R. 200 I . CFD simulation in atrium smoke management system design. ASHRA E Transactions 107(1). Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. SMACNA. 1987. HVAC sysfems applications. Vienna, Va.: Sheet Metal and Air Conditioning Contractors' Association, Inc. s f r n s SMACNA. 1990. ~ ~ ~ ~ - s ~ ducf edesign, 3d ed. Chantilly, Va.: Sheet Metal and Air Conditioning ~ontractor~"~ssociation, Inc. SMACNA. 1993. HVAC Systems Esfing, Adjlrsring and Balancing, 2d ed. Vienna, Va.: Sheet Metal and Air Conditioning Contractors' Association, Inc.

-

~eferinces

Steckler, K.D., H.R. Baum, and J.G Quintiere. 1986. Transactions 96(2): 373-383. Atlanta: American Society of Heating, Refiigerating and Air-CondiSalt water modeling of fire induced flows in multicompartment enclosures. 2 1st Symposium (Internationing Engineers, Inc. tional) on Combustion, Combustion Institute, pp. Tamura, GT. 1990~. tower'tests of stair pressuriiaFire 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, RefrigeratResearch. Gaithersburg, Md.: National Institute of ing and Air-conditioning Engineers, Inc. Standards and Technology. Tamura, GT. 1991. Determination of critical air velociStewart, et al. 1973. Experimental human response to ties to prevent smoke backflow at a stair door openhigh concentrations of carbon monoxide. Archifecing 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-oTamura, G.T. 1994. Fire tower tests on vestibule pressurspace Medicine 40: 1232-1238, November. ization for protection of stairshafts. ASHRAE TransStrakosch, G.R. 1983. Verficaltransporfafion:Elevafors actions lOO(2). Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engiand escnlafot-S, ed. New York: Wiley & Sons. 2d neers, Inc. Streeter, V.L., and E.B. Wylie. 1979. Fluid Mechalzics, Tamura,G.T., and J.H. Klote. 1987a. Experimental Fire 7th Ed. New York: McGraw. Tower Studies on Elevator Pressurization Systems Stioup, W.D. 1999. Scotch pine christmas tree fire tests, for Smoke Control. ASHRAE Transactions 93(2): Report of Test FR 4010. Gaithersburg, Md.: 2235-2257. National Institute of Standards and Technology. Tamura, G.T., and J.H. Klote. 1987b. Experimental fire Stroup, W.D., et al. 2000. Large fire research facility tower studies on mechanical pressurization to con(Building 205) exhaust hood heat release rate meatrol smoke movement caused by fire pressures. Prosurement system. Gaithersburg, Md.: National ceedings of the 2nd Infernational Sytnposilrnl 017 Institute of Standards and Technology, NISTIR Fire Safe01 Scietice, ToXyo, Japan. 6509. Tarnura, G.T., and J.H. Klote. 1988. Experimental fire Sumka, E.H. 19% Presently, elevators are not safe in tower studies on adverse pressures caused by stack fire emergencies. Elevafol- World 36(11): 34-40. and wind action: Studies on smoke movement and S~~ppIemen~ nafinnal building code o f Cannda. 10 /he control, ASTM International Symposiuni on Char1985. Chapter 3, Measures for fire safety in high acterization and Toxicity of Smoke, December 5, buildings. Associate Committee on the National 1988, Phoenix, Ariz. Building Code, National Research Council of CanTamura, G.T., and C.Y Shaw. 1973. Basis for the design ada, Ottawa, NRCC 23178. of smoke shafts. Fire Tecl7nol0,qy 9(3): 209-222. Tamanini, F. 1976. The application of water sprays to Tamura, G.T., and C.Y. Shaw. 1976a. Studies on exterior the extinguishnient of crib fires. Combusfion Sciwall air tightness and air infiltration of tall buildence atid Technolog~ Vol. 14, pp. 17-23. ings. ASHRAE Transactions 82(1): 122-134. Tamura, G.T. 1978a. Exterior wall venting for smoke control in tall office buildings. ASHRAE J O Z I I - I ~ ~ Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. 20(8): 43-48. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Tamura, G.T., and C.Y. Shaw. 1976b. Air leakage data for the design of elevator and stair shaft pressurizaInc. tion systems. ASHRAE T?ansactions 82(2): 179Tamura, G.T. 197%. Experimental studies on exterior 190. Atlanta: American Society of Heating, Refrigwall venting for smoke control in tall buildings. 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! studAir-conditioning Engineers, Inc. ies of mechanical venting for smoke control in tall Ta~nura, G.T. 1990a. Field tests of stairshaft pressurizaoffice buildings. ASHRAE Tramactions 86(1): 54tion systems with overpressure- relief. ASHRAE 7 1. Atlanta: American Society of Heating, RefrigerTrn1l.soc/io17s 96(1): 95 1-958. Atlanta: American ating and Air-conditioning Engineers, Inc. Socicty of Healing, Refrigerating and Air-CondiTamura, G.T., and A.G. Wilson. 1966. Pressure differ~ioni~ig Engineers, Inc. cnces for a nine-story building as a result of chimney effcct and ventilation system operation. Tamura, G.T. 1990b. Fire tower tests of stair pressurizaASHRAE Tt.tn~snc/ions72(1): 180-189. Atlanta: tion systems \vitli overprcssure relief. ASHRAE

.,

,

-

Principles of Smoke Management

i

l
I

'

American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Tamura, GT., and A.G Wilson. 1967a. Building pressures caused by chimney action and mechanical ventilation. ASHRAE Transactions 73(2). Timura, GT., and A.G Wilson. 1967b. Pressure differences caused by chimney effect in'three high buildings. ASHRA E Transactions 73(2). Tanaka, T. 1983. A model of multiroom fire spread. Gaithersburg, Md.: National Bureau of Standards, NBSIR 83-27 18. Tewarson, A. 2002. Generation of Heat and chemical compounds in fires, Chapter 3-4. SFPE Handbook of Fire Protection Engineering, 2d ed. Quincy, Mass.: National Fire Protection Association. Thomas, P.H. 1970. Movement of smoke in horizontal corridors against an air flow. Institute of Fire E@neers Quarterly 30(77): 45-53. Thomas, P.H., et al. 1963; Investigation into the flow of hot gases in roof venting, Tech. Paper No. 7. Boreham Woods, Herts, UK: Fire Research Station. Tsujirnoto, M., T. Takenouchi, and S. Uehara. 1990. A scaling law of smoke movement in atrium. I l th Joint Panel Meeting of the UJNR Panel on Fire Research and Safety. Gaithersburg, Md.: National Institute of Standards and Technology. UL. 1999a. Standard forfire darupers, UL 555. Northbrook, Ill.: Underwriters Laboratories, Inc. UL. 1999b. Standard for smoke dampers, UL 555s. Northbrook, Ill.: Underwriters Laboratories, Inc. Underwriters Laboratories. 1990. Standard for air leakage tests of door assemblies-UL 1784, Northbrook, 111. Wakamatsu, T. 1977. Calculation methods for predicting smoke movement in building tires and desiging smoke control systems, Fire Standards and Safety, ASTM STP-6 14, A. F. Robertson, ed., pp. 16s-193. Philadelphia, Pa.: American Society for Testing and Materials.

Walton, GN. 1989. AIRNET-A computer program for building airflow network modeling. Gaithersburg, Md.: National Institute of Standards and Technology. Walton, GN. 1997. CONTAM96 user manual, NISTIR 6056. Gaithersburg, Md.: National Institute of Standards and Technology. Walton, W.D. 1985. ASET-B: A room fire program for personal computers. Gaithersburg, Md.: National Bureau of Standards, NBSIR 85-3144-1. Walton, W.D. 1988. Suppression of wood crib fires with sprinkler sprays: Test results. Gaithersburg, Md.: National Bureau of Standards, NBSIR 88-3696. Walton, W.D., and P.H. Thornas. 1995. Estimating temperatures in compartment fires. SFPE Handbook of Fire Protection Engineer-irlg, Chapter 3-6. Bethesda, Md.: Society of Fire Protection Engineers. Waters, R.A. 1989. Stansted terminal building and early atrium studies. low-nal of Fire P)-otcction Engirieeririg 1(2): 63-76. White, F.M. 1974. fiscozrs ,fluid floiv. New York: McGran.. Wood, Peter G. 1971. The behaviour of people in fires. Fire Research Note 953. Borehamwood, U.K.: Fire Research Station. Wrap, C.P., and G.K. Yuill. 1993. An evaluation of algorithms for analyzing smoke control systems. ASHRAE Transactions 99( 1 ): 160- 1 74. Yii, E.H. 1998. Exploratory salt \vater experiments of balcony spill plume using laser induced fluorescent technique. Fire Engineering Research Report 9517, University of Canterbury, Neu. Zealand. Yoshida, H. et al. 1979. A FORTRAN IV pro,oram to calculate smoke concentrations in a multi-story building. Ottawa, Canada: National Research Council.Best, R., and D.P. Demers. 1982. Investigation report on the MGM Grand Hotel fire-Las Vegas. Nevada, November 21, 1980. Quincy, Mass.: National Fire Protection Association.

Appendix A

Units of Measurement and Physical Data
hysical quantities such as length, weight, and time are expressed in temls of standard units of measurement. In this book, both English units and international system (SI) units are used. Newton's second law of motion states that the force, F, on a body of fixed mass, m , is proportional to the product of the mass and the acceleration, a: become base units and the other is a derived unit. Theoretically, any three can be selected as base units. However, the only two combinations to be used extensively are:
Base Units Derived Unit

mass, length, and time force, length, and time

force mass

There are three c0111111011 English systems with regard to mass and force: the pound mass and pound force system, the slug and pound system, and the pound mass and poundal system. Introduction of the proportionality constant 11% into the abo1.e relation yields

Table A-I lists the units for these systeins and the S1 system along with the values of g, for each. Generally, a pound is thought of as a unit of force. However, in some engineering applications, the pound also has been used as a unit of mass. One pound mass (Ibm) is the mass of a body that weighs one pound (Ib) at sca level. One slug equals 32.174 Ibni, and one poundal is a force of 0.03108 pounds. For the systems listed in Table A-l for which the value of ,q. is one, Ne\vton's second law can be written as
F =
I110

.

-

This formulation of Neuton's law simplifies derived equations and calculations. I t is accomplished by defining one of the four units (len_rth,mass, time, and force) in terms o f ~ l i other three. Thus, three of h e units e

Because force is a derived unit in the S1 system, that convention is used in the following discussion for the inch pound (I-P) system. For convenience, the unit of mass in the I-P system \\.ill be taken to be the slug. A 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 newton, N, which is the force required to accelerate a mass of one kilogram at a rate of one meter per second squared. The base units and derived unit discussed above relate force and mass, but many more units are needed for engineering calculations. The base units and derived units needed for smoke control applications are listed in Tables A-2 and A-3. In the S1 system, prefixes are used to form decimal multiples and submultiples of the S1 units. The S1 prefixes are listed in Table A-4. Unit systems with g, of 1 can be referred to as I~omogeneousunit systems (Table A-l), and in this text homogeneous unit systems are referred to as being either the S1 system or the slug and pound system with the base units and derived units as listed in Tables A-? 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 Pound Force System foot (ft) second (S) pound mass (Ibm) pound force (Ibt) 32.174 Ibm ft Slug a n d Pound System foot (ft) second (S) slug pound (Ib) 1 slug ft lbf s Pound Mass a n d Poundal System foot (ft) second (S) pound mass (Ibm) poundal 1 Ibm ft 2 poundal s International System (SI) meter (m) second.(s) kilogram (kg) Newton (N) I kgm

Quantity length time mass force
gc

.

I lbf s

2

N s2

Table A-2: Base Units
S1 System Quantity len,oth mass time thennodynamic (absolute) temperature Unit meter kilogram second kelvin Symbol m kg
S

English System Unit Symbol foot ft slug second
O R

slu~
S
O R

K

Table A-3: Derived Units
Quantity force pressure energy, work or heat power, enersy release rate mass flow rate Unit newton pascal joule watt S1 System Symbol N Pa
J W

Formula

Unit pound

English System Synlbol Formula Ib slug ft/s2 lblft2 Ib ft Ib ftls slugls

kg I
~lm' Nm Jls kgls

Table A-4: SI Prefixes
Prefis giga niega kilo Sxmbol G Multiplication Factor
I o9 = I 000 000 000

nano
I.

11

I

= 0.000 000

00 1

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 figures. 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 acceleration of gravity, gas constant of air, and standard atmospheric 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 discussions in the text and figures. However, caution should be exercised to ensure that absolute temperatures are used in calculations where necessary. The following equations can be used to convert between temperature scales:

where TF
=

temperature, OF,

TC = temperature, OC,

TR
TK

= =

temperature, OR, temperature, K.

Tables A-8 and A-9 list density, specific heat, viscosity, and thermal conductivity of air. For further information 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, Gaithersburg, Md.

-Appendix A - Units of Measurement a n d Physical Data

Table A-5: Factors for Conversion to S Units I
Multiply atmosphere, standard (atm) atmosphere, standard (atm) British thermal unit (Btu) British thermal unit (Btu) British thermal unit per hour (Btulh) British thermal unit per pound (Btdlb) British thennal unit per pound degree Fahrenheit [Btu/(lb°F)] British thermal unit per second (Btuls) British thermal unit per second (Btu/s) British thermal unit foot per hour square foot degree Fahrenheit [Btu ft/(h f t Z O ~ ) ] British thermal unit inch per hour square foot degree Fahrenheit [Btu in./(h f t Z O ~ ) ] calorie (cal) centimeter of mercury (cm Hg) centimeter of mercury (cni Hg) centimeter o r water (cm H@) centipoise (cP) centistokes (cSt) cubic foot (ti ) cubic foot (ft3) (li-'Irnin or cfin) cubic foot per ~ninutc cubic I'oot per ~ninute (li3in1in or clin) cubic tbot per second (ct3/s) cubic inch (in.3) cubic inch per minutc (in.'imin) cubic yard (yd3) cubic yard pcr minute (yd'/n~in) dyne (dyn) dyne centinleter ( d y n a n ) dyne per square centillleter (dyn/cm2) erg (erg) erg per second (ergls) foot (li) Toot of mcrcury. conven~ional tig) (It foot of mercury. conventional (It Hg) root ofwater ( f 1 2 ) H0 h o t ol'water (li 1i20) foot pcr hour (lilh) h o t per minutc (lilmi~i Ipnl) or h o t per sccond (lils)
3
-

BY 101325 101.325 1055

To Obtain
pascal (Pa) kilopascal (kPa) joule (J) kilojoule (U) watt (W) joule per kilogram (Jkg) joule per kilogram kelvin [J/(kg+K)] watt (W) kilowatt (kW) watt per meter kelvin [W/(m.K)] watt per meter kelvin [Wl(m-K)] joule (J) pascal (Pa) kilopascal (kPa) pascal (Pa) pascal second (Pa-S) meter squared per second (m2/*) cubic meter (m3) liter (L) cubic meter per sccond (rn3/s) liter per second (Lls) cubic meter per second (m'/s) cubic meter (m3) cubic meter per second (m3/s) cubic meter (m3) cubic meter per second (1n3/s) newton (N) netvton meter (N.ni) pascal (Pa) joule (J) watt (W) meter (m) pascal (Pa) kilopascal (kPa) pascal (Pa) kilopascal (kPa) nieter per second (mls) meter per sccond (m!s) metcr pcr sccond (mls)

1 .055 0.293 2330 4187

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 S Units I
foot pound-force (ft-lbf) foot pound-force per hour (ft.lbt%) foot pound-force per minute (ft-lbflmin) foot pound-force per second (ftdbfls) gallon (Imperial) (gal) gallon (Imperial) (gal) gallon (US.) (gal) gallon (US.) (gal) gallon (US.) per minute (gpm) (gall niin) gallon (US.) per minute (gpm) (gallmin) horsepower (550 ft.lbf/s) horsepower (boiler) horsepower (electric) hour (hr or h) inch (in.) inch (in.) inch of mercury (in. Hg) inch of mercury (in. Hg) inch o f water (in. H 2 0 ) inch per second (in.1~) kilogram-force (kg0 kilogram-force meter (kgfm) kilogram-force per square centimeter (kgf/cm2) kilogram-force per square centimeter (kgf/cm2) kilogram-force per square meter (kgf/m2) kilometer per hour (kmlh) kilowatt hour (kWh) kilowatt hour (kWh) kilowatt-hour (kWh) knot (nautical mile per hour) liter (L) mil (0.00 1 in.) mil (0.001 in.) mile (mi) mile (mi) mile per hour ( m i h or mph) mile per hour (milh o r mph) mile per minute (milmin) mile per second (mils) mile, nautical millimeter of mercury (mmHg) niillimeter of water (1111iiH~O) minute (min) ounce (avoirdupois) (oz) 0.02835 1.356 joule (J) watt (W) watt (W) watt (W) cubic meter (m3) liter (L) cubic meter (m3) liter (L) cubic meter per second (m3/s) liter per second (Lls) watt (W) watt (W) watt (W) second (S) nleter (m) centimeter (cm) pascal (Pa) kilopascal (kPa) pascal (Pa) meter per second (mls) newton (N) newton meter ( N m ) pascal (Pa) kilopascal (kPa) pascal (Pa) meter per second (mls) joule (J) ~negajoule (MJ) joule (J) meter per second (mls) cubic meter (m3) meter (m) millimeter (mm) meter (m) kilometer (km) meter per second (mls) kilome~er hour (kmlh) per n~eter second (mls) per meler-persecond (mls) nleler (m) pascal-(Pa) pascal (Pa) second ( S ) kilogram (kg)

263

Appendix A-Units of Measurement and Physical Data

Table AS: (Continued) Factors for Conversion to S Units I
Multiply
ounce (avoirdupois) (02) ounce (troy or apothecary) (02) ounce (boy or apothecary) (02) poise (P) . pound (avoirdupois) (Ib) pound (troy or apothecary) (Ib) pound force (Ibf) pound per cubic foot (lblft3) pound per hour (lblh) pound per minute (Iblmin) pound per second (Ibls) poundal pound-force (Ibf) pounbforce foot (Ibf.ft) pound-force inch (Ibfh.) pound-force per square foot (lbf/ft2) pound-force per square inch (psi) ( l b ~ i n . ~ ) pound-force per square inch (psi) (1bflin.l) revolution per minute (rpm) (rimin) slug (slug) slug per cubic foot (slug/ft3) slug per foot second [slug/(ft.s)] square foot (ft2) square foot per second (ft2/s) square inch (in2) square inch (in2) square yard (yd2) standard cubic feet per minute (scfm)' standard cubic feet per minute (scfm)' standard cubic feet per minute (scfm)' stokes (St) ton of refrigeration ton, long (2240 Ib) ton, metric (t) ton, short (2000 Ib) ton-force (2000 1bf) ton-force (2000 Ibf) watt hour (W.h) yard (yd)
I.

BY
28.35

To Obtain

gram (g)
kilogram (kg)

gram (g)
pascal second (Pas) kilogram (kg) kilogram (kg) newton (N) kilogram per cubic meter (kg/m3) kilogram per second (kg/s) kilogram per second (kgk.) kilogram per second (kgls) newton (N) newton (N) newton meter (N-m) newton meter ( N m ) pascal (Pa) pascal ( W kilopascal (kPa) radian per second (radls) kilogram (kg) kilogram per cubic meter (kg/m3) pascal second (Pas) square meter (m2) meter squared per second (m2/s) square meter (m2) square centimeter (cm') square meter (m2) standard cubic meter per second (sm3/s) standard lltre per second (sL/s) kilogram per second (kgls) meter squared per second (m2/s) watt (W) kilogram (kg) kilogram (kg) kilogram (kg) newton (N) kilonewton (kN) joule (J) 0.9 144 meter (m)

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 atmosphere, standard (atm)
atmosphere, standard (atm) atmosphere, standard (atm) British thermal unit (Btu) British thermal unit per hour (BtuAi) British thermal unit per hour (Btu/h) British thermal unit per minute (Btulmin) British British British British thermal unit per minute thermal unit per second thermal unit per second thermal unit per second (Btulmin) (Btds) (Btu!s) (Btuk)

BY 2116.2

To Obtain
pound-force per square foot (lbflft2) pound-force per square inch (psi) (~bflin.~) inch of water (in. H20) foot pound-force (ft-lbf) foot pound-force per second (ft4bfls) British thermal unit per second (Btu/s) foot pound-force per second (ft-lbf/s) British thermal unit per second (Btu/s) British thermal unit per minute (Btdmin) British thermal unit per hour (Btdh) foot pound-force per second (ft.lbf/s) foot pound-force per slug (ft4bfIslug) pound-force per square foot (1bflft2) pound-force per square inch (psi) ( ~ b f / i n . ~ ) inch of water (in. H20) pound-force per square foot (lbf/Ft2) pound-force per square inch (psi) (lb~in.') inch of water (in H20) cubic inch ( h 3 ) gallon (US.) (gal) cubic foot per second (ft3/s) cubic foot per minute (cfm) (ft3/min) cubic foot (ft3) cubic foot per minute (cfrn) (ft3/min) cubic foot per second (Ft3/s) cubic foot (ft3) cubic inch (in.)) gallon (U.S.) (gal) cubic foot per minute (cfm) (ft3/min) cubic foot per second (ft3/s) pound-force per square foot (lbf/ft2) pound-force per square inch (psi) (lbf/in2) inch of water (in. H20) pound-force per square foot (lbf/ft2) pound-force per square inch (psi) (lbf/in2) inch of water (in. H20) foot per second (ft/s) foot per niinute.(ft/min) foot per second (Ws) foot per minute (Wmin) British thermal unit (Btu)

Btdlbm centimeter of mercuq (cm Hg) centimeter of mercup (cm Hg) centimeter of mercun (cm Hg) centimeter of water, (cm H20) centimeter of water, (cm HzO) centimeter of water, (cm H,O) cubic foot (ft3) cubic foot (ft3) cubic foot per minute (sfin) (ft3/lnin) cubic foot per second (ft3/s) cubic inch ( h 3 ) cubic inch per minute (in.3/n~in) cubic inch per minute (in.'lmin) cubic yard (yd') cubic yard 0.d') cubic yard (yd3) cubic yard per n~inute (vd3/n~in) cubic yard per minute (yd3/min) foot of mercury, con\.entional (ft Hg) foot of mercury, conventional (ft Hg)

: ) foot of mercury, conventional (ft H
foot of water (ft HIOl foot of water (ft H,O) foot of water (ft H 2 0 ) foot foot foot foot per hour (ft/h) per hour (ftlh) per minute (ftlmin) per second (fi/s)

foot ~ound-force (li.lht'l

Appendix A -Units of Measurement and Physical Data

Table A-6: (Continued) Factors for Conversion to the English Units
Multiply
gallon,(Imperial) (gal) gallon (Imperial) (gal) gallon (Imperial) (gal) gallon (U.S.) (gal) gallon (US.) (gal) gallon (US.) (gal) gallon (U.S.) per minute (gpm) (gal/ min) gallon (U.S.) per minute (gpm) (gal/ min) horsepower hour (h) inch (in.) inch (in) inch o f mercury (in. Hg) inch o f mercury (in. Hg) inch o f mercury (in. Hg) inch of water (in. H 2 0 ) inch of water (in. HzO) inch per second (ink) inch per second (ink) kilogram (kg) kilometer per Iiour (kmlh) kilowatt ( k W ) kilowatt (kW) kilowatt hours (kW h) knor (nautical milc pcr hour) knot (nautical mile per hour) liter per sccond (Lls) liter per second (Lls) meter (m) mil mile (mi) mile (mi) mile (nii) milc per I~our (nii/h) milc per hour (milli) milc per hour (milh) mile, nautical millirncter of mercury (mniHs) niilli~ncter nlcrcury (nitiiHg) of millimctcr of mercury (mmHg) millirncter of water (mmI1,O) niillinietcr of\vatcr (1ii1n14~0) millimctcr ofwatcr (mrnl~l,O)

BY
0.1605

To Obtain cubic foot (ft3)
cubic inch ( h 3 ) gallon (US.) (gal) cubic foot (8') cubic inch (im3) gallon (Imperial) (gal) cubic foot per minute (cfm) (ft!/min) cubic foot per second (ft3/s) foot pound-force per second (ft.lbf/s) second (S) foot (ft) yard (yd) pound-force per-square foot (lbf/ft2)' pound-force per square inch (psi) ( l b ~ i n . ~ ) inch of water (in H 2 0 ) pound-force per square foot (lbf/ft2) pound-force per square inch (psi) (lbf/im2) foot per second ( f ~ k ) foot per minute (fvmin) pound (Ib) foot per second (~L's) foot pound-force per second (ft-lbf/s) British thermal unit per second (Btuls) rt ~ b f foot per second ((L'S) milc per hour ( m i h ) cubic foot per second (ft3/s) cubic foot per minute (rt3/rnin) foot (ft) inch (in.) foot (11) yard (yd) mile. nautical foot per second ( ~ L ' s ) foot per minute (f~'min) knot (nautical milc per hour) mile (mi) pound-forcc per square root (lbf/ft2) pound-forcc pcr squarc inch (psi) (lbf~ir,.~) i~~cli ofwatcr.(iti H2O) pound-forcc per squarc l h t (lbflft') pound-li~rccper squarc inch (psi) (lbflin.') inch oI'w;~tcr(in H?O)

Principles of Smoke Management

Table AB: (Continued) Factors for Conversion to the English Units
Multiply minute (min) ounce (avoirdupois) (02) ounce (avoirdupois) (02) ounce (troy or apothecary) (02) pascal (Pa) pound (Ib) pound (Ib) pound per cubic inch ( ~ b l i n . ~ ) pound per cubic yard (lblyd3) poundal pound-force per square foot (lbf/ft2) pound-force per square foot (lbf/ft2) pound-force per square inch (psi) (1bf/im2) pound-force per square inch (psi) (lbf/im2) slug (slug) square foot (ft ) square foot (ft2) square inch (in2) square inch (in ) square yard (yd2) square yard (yd2) standard cubic feet per minute (scfm)' standard cubic feet per minute (scfm)' ton (refrigeration) ton (refrigeration) ton, long (2240 1b) ton, metric (t) ton, short watt (W) watt (W) watt (W) yard (yd)
I.
scfm
IS a

BY 60 0.9115 0.0625 1 .097 0.0209 16 0.03 108 1728 0.037037 0.03 109 0.006944 0.1922 144 27.68 32.174 0.1111 0.006944 7.7 168-04 144 1296 9 3 . 8 9 10-5 ~ 0.00 125 12,000.00 2594 2240 2205 2000 0.7376 9.4788-04 0.7376 3

To Obtain second (S) ounce (troy or apothecary) (oz) pound (Ib) ounce (avoirdupois) (oz)
pound-force per square foot (lb0ft2) ounce (avoirdupois) (02) slug (slug) pound per cubic foot (lblft3) pound per cubic foot (lblft3) pound-force (IbQ pound-force per square inch (psi) ( l b f h 2 ) inch of water (in. H20) pound-force per square foot (1bf/ft2) inch of water (in H20) pound (Ib) square yard (yd2) square inch ( h 2 ) square yard (yd2) square foot (ft2) square inch ( h 2 ) square foot (ft2) slug per second (slugls) pound per second (Ibls) British thermal unit per hour (Btuh) foot-pound per second (ft Ibfls) pound (Ib) pound (Ib) pound (Ib) foot pound-force per second (ft.lbf/s) British thermal unit per second (Btuls) foot-pound per second (ft Ibfls) foot (ft)
IS ar

2

2

form o f mass flo\\. rate used for air movemenr and, for this text. ir

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 O R 1716. ft IbWslug O R 0.06858 Btullbm O R 101325 Pa 14.696 psi

Standard atmospheric pressure, P,,,

Table A-8: Properties of Air in English units1
T (OF) 0

P
(lbmlft') 0.086

G
(Btullbm "F) 0.239

P
(Ibn~lft S) 1.110~10‘~

v
(ft2/s)
0. l 30x 1 0-j

k (Btulhr ft OF) 0.0133

Principles of Smoke Management

Table A-9: Properties of Air in SI units1

(K)
200

(kglm3) 1.7684

(Jlkg 'C) 1.0061~10~

(Wm 1.3289x 10"

(m21s) 7.5 14x10-~

(Wlrn 'C) 0.0 1809

I.

Note: Noration listed ar bottom ofTable A-S.

Table A-10: Thermal Properties of Materials in IP Units
Density P lblf? 1 69 190 l50 162 169 119 59.9 33.7 15.0 19.9 31.2 59.3 13.7

specific kjeat
C Btullb F 0.2 1 0.1 1 0.18 0.19 0.19 0.20 0.26 0.60 0.30 0.30 0.23 0.20 0.27

Thermal Conductivity
k

Material Aluminum (pure) Steel (0.5% Carbon) Concrete Brick Glass, Plate BrickIConcrete Block Gypsum Board Plywood Fiber Insulation Board Chipboard Aerated Concrete Plasterboard Calcium Silicate Board (Marinits XL)' Alurnina Silicate Block ( ~ a o \ v o o l ) ' Glass Fiber Insulation Expanded Polystyrenc
I.

Btulh ft OF 119 31 0.92 0.46 0.44 0.42 0.10 0.07 0.3 1 O.OP7 0.15 0.092 0.064-0.08 I

kpC t3trr2 ft4 OF-' h-' 4300 1700 25 14 14 10 1.5 1.4 1.4 I .3 1.1 1. I 0.74-0.95

5.75 1.25

0.19 0.36

0.02 1 0.020

0.0 15 0.0088

Trade namcs imply no ~ r d u r s c ~ i i ~h!.t iIIIC ;lulhors or tlic ~ U ~ I I J I I C ~ S . l

Appendix A-Units o Measurement and Physical Data f

Table A-ll: Thermal Properties of Materials in SI Units
Density
specific

P
Material
Aluminum (pure) 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 Expanded Polystyrene 60 20 (kglm3 2710

C
(kJ/kg K) 0.895

Thermal Conductivity k 103 (kWIm K) 206

PC
k w 2 m4 K% 500

0.8 1.5

0.037 0.034

0.0018 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. Investigation of hydraulic drag coefficients in stairInformation Bulletin No. I I, wells. Fire Pt-eve~tion, Moscow, USSR, pp. 63-66. Appleton, I.C. 1976. A model of smoke movement in of buildings. CIB Symposium on Co17t1.01 Smoke Movetnent it1 Bzrilding Fires, Vol. I . Fire Research Station, Garston Watford, U.K., pp. 127-138. Atkinson, B. 1992. Fire safety engineering: The development of an Australian standard hot smoke test for large compartments, 2nd CIB Workshop at the Fire Research Station, Jan 30-3 1, 1992, Borehamwood, U.K. Atkinson, B., and T. Atkinson, T. 1992. Case study NFPA 92B and the extra dimension. Proceedings of Fire the I11tert~afiot7al Safety Engineering Coffere w e , 18-20 October 1992, Sydney Australia. Atkinson, B., and R. Marchant. 1992. Case study-The Deeds Road experiments. Proceedings of the Infernafional Fire Safefy Engineering Conference, 18-20 October 1992, Sydney Australia. Barrett, R.E., and D.W. Locklin. 1969. A computer technique for predicting smoke movement in tall buildings. $n7posi~m on kIo1~eti1etif Smoke on Escape of Roufes in Buildings, Watford College of Technology, Watford, Herts, U.K., pp. 78-87. BenjaminIClarke Assoc. 1984. Operation San Francisco-SmokeISprinkler Test, Operation Life Safety. Beyler, C. 1993. Revicw of design of smoke management systems. Jo~r~nc~l PtafecfionEngit1eer($Fire itlg 5(1): 33-34. Billings, L. 198 1. Mt. Sinai Medical Center: A new concept in smoke control. ASHRAE Journal 23(2): 3437. Bryner, N.P., E.L. Johnsson, and W.M. Pitts. 1994. Carbon monoxide production in compartment firesReduced-scale enclosure test facility. National Institute of Standards and Technology, NISTIR 5568. Buchman, A.H. 1998. Modelling post-flashover fires with FASTLite. Journal of Fire protectiotl Engineering 9(3): 1- 1 1. Bukowski, R.W. 1978. Smoke measurements in large and small scale fire testing. National Bureau of Standards, NBSIR 78- 1502. Bukowski, R.W. 1991. Fire models, the future is now! NFPA Jownal2(85): 60-69. Bukowski, R.\V. 1986. Quantitative determination of smoke tosicity hazard-Practical approach for current use. Proceedings of the First International Symposium on Fire Safety Science, October 7-11, 1985, C. Grant and P. Pagni, eds., pp. 1089-1 100. New York: Hemisphere Publications. Butcher, E.G., and A. C. Parnell. 1979. Smoke control in fire safety design. London: E. & F. N. Spon. Butcher, E.G., T.H. Cottle, and T.A. Baily. 197 1. Smoke tests in the pressurized stairs and lobbies of a 26story ofice building. Building Service Engineer 39: 206-210. Butcher, E.G., P.J. Fardell, and P.J. Jackman. 1969. Prediction of the behavior of smoke in a building using a computer. Symposi~tm017 movemenf of smoke in escape mires in buildings, pp. 70-75. Watford, Herts, England: Watford College of Technology.

Appendix B - Bibliography

Chow, W.K. 1990. Field tests on a staircase pressurization system installed in a commercial building. ASHRAE Transactions 96(2). Chow, WK. 1993. Simulation of.the atrium fire environment in Hong Kong. ASHRAE Transactions 99(2). chow, W.K. 1995. Simulation of car park fires using zone models. Journal of Fire Protection Engineers 7(2): 65-74. Chow, W.K. 1999. Atrium smoke filling process in shopping malls of Hong Kong. Journal of Fire Protection Engineering 9(4): 8-30. Chow, W.K., and L.W. Lam. 1993. Evaluation of a staircase pressurization system. ASHRAE Transacrio~a 99(2). Chow, W.K., and K.W. Lau. 1995. Field tests on atrium smoke control systems. ASHRAE Transactions lOl(1). Chow, W.K., and C.W. Lo. 1995. Scale modeling studies on atrium smoke movement and the smoke filling process. Jorwnal of Fire ProfecfionEngitieers 7(2): 55-64. Clark, F.R.S. 1963. Visibility in smoke. 3rclSynposizrtr7 0 1 1 Cotnbus~ibili~~, Plastics, Toronto Oct 24-25, atid 1983, National Research Council of Canada. Clark, F.R.S. 196s. Strategies for improving visibility in fires. Canadian Brrilcfir~gDigesr, CBD 246, January. Clark, J.A. and J.W. Harris. 1989. Stairwell pressurization in a cold climate. ASHRAE Ti-amactions95(l). Clarke, F.B. 1997. Physiological effects of smoke: managing escape. ASHRAE Jorrrnal39(3):47-56. Collins, B.L., M.S. Dahir, and D. Madrzykowski. 1992, Visibility of exit signs in clear and smoky conditions. Jout-nal o f /lhrminating Engineering Socie!~:. Cooper, L.Y. 1988. Ceiling jet-driven wall flows in compartment fires. Co117busf. Sci. aiid Tech. 62: 285-296. Cooper, L.Y., G.P. Fomey, and J.H. Klote. 1990. FANRES-Calculate the flow through a fadduct forced ventilation system. CCFM Computer Code Application CCFM. VENTS-Part Ill: Catalog of algorithms and subprogranis, Cooper L.Y. and Forney, G.P., editors, National Institute of Standards and Technology, NISTIR 4344. Cuzzillo, B.R., and P.J. Pagni. 1998. Thermal breakage o of double-pane glazing by fire. Jour~nl f f i r e Prorecfiou Etigi17eering9( l ): l - l l . DeCicco, P.R., and R.J. Cresci. 1975.Smoke and tire control in large atrium spaces. ASHRAE fimunct i o n ~ l(2): 3 19-335. 8 Degenkolb, J. 1991. Elevator usage during a building on tire. ASME sy~liposiuln tire and elevators. Baltiniorc. IM~.,February 19-20. 1991. New York:

American Society of Mechanical Engineers, pp. 76-79. Degenkolb, J.G 1975. Firesafety for atrium type buildings. Building Standard 44(2): 16-18. Degenkolb, J.G. 1983. Atrium. Building Standat-& 52(1): 7-14. Dias, C. 1978. Stairwell pressurization in a high-rise commercial building. ASHRAE Journal 20: 7: 2426. Ditzel, P. 1982. Theater of death-December 30, 1903: Chicago's Iroquois Theater becomes a stage for an inferno that.kills 602. Firehouse, pp. 52-58. Drysdale, D. 1985. At7 ill/mduction IO fire dyna~nics. New York: Wiley. Earp, D. 1998. Natural and mechanical smoke control systems. Seminar Fire Safety-Smoke Control: Standards and Practice, CIBSE-ASHRAE Group and .. Colt International. Edwards, J. 1987. Preventing fire risk in the theatre-A lengthy process ofevolution. Fire 77(980): 30. Edwards, J. 1987. Theatre fires-The importance of smoke ventilation. Fire Prevetition (198): 37-38. Egilsrud, P.E. 1984. Pre\:ention and control of highway tunnel fires. U.S. Department of Transportation, Report FHWA/RD-83!032. Evans, D.H. 1999. Guide to the 1997 UBC smoke-control provisions. International Conference of Building Officials, Whittier. CA. Evers, E. and A. Waterhouse. 1978. A computer model for analyzing smoke movement in buildings. Borehamwood, Herts, U. K.: Building Research Est. Gablin, K.A. 1997. The memorial tunnel fire ventilation test program. ASHRAE Jo~irnal39(2): 26-30. Galbreath, M., and G.W. Shorter. 1965. Air pressures contribute to a smoke-free corridor, National Research Council of Canada Fire Research Yote No. 3 1966, Reprinted from Fire Fighting in Canada. Hadjisophocleous, G.V., and M. Cacambouras. 1993. Computer modeling of comparhnent fires. Joro-nal of Fire Protecfio17Et7,oineet-s 39-52. Hagglund, B., R. Jansson. and K. Nireus. 1985. Smoke filling experiments in a 6 x 6 ~ 6meter enclosure. Stockholni, Sweden: National Defense Research institute of Sweden. Hansell, GO., and H.P. Morzan. 1988. Smoke control in atrium buildings using depressurization, PD 66'88, Borehamwood. Fire Research Station. Hansell, G.O., and H.P. Morgan. 1994. Design approaches for smoke control in atrium buildings, BR 258. Borchamwood. Hem, U.K.: Fire ResearchStation. Hasemi, Y. 1985. Analysis of failures of automatic sprinklers in actual fires. UJNR Panel on Fire

Principles of Smoke Management

-

Research and Safety, Joint Meeting May 13-21, 1985. Tsukuba, Japan, pp. 794-807. He, Y. V. and Beck. 1996. Estimation of neutral plane in high rise buildings. Journal of Fire Sciences 14: 235-248. Heating and Ventilating Research kssociation. Heselden, A.J.M. 1970. Smoke travel in shopping malls experiments in co-operation with Glasgow fire brigade-Part 2, F. R. Note No. 854. Borehamwood, Herts, U.K.: Fire Research Station. Heselden, A.J.M. 1976. Studies of smoke movement and control at the fire research station. CIB Symposium on cont1-01 of smoke movement in building fires, Vol. I, Fire Research Station, Garston Watford, U. K., pp. 185- 198. Heselden, A.J.M. 1978. studies of fire and smoke behavior relevant to tunnels. BRECP, Borehamwood, Herts, U.K.: Building Research Establishment, Heselden, A.J.M., and R. Baldwin. 1976. The movement and control of smoke and escape routes in buildings, BRECP, Building Research Establishment, Borehamwood, Hertfordshire, U.K. Heselden, A.J.M., and P.L. Hinkley. 1970. Smoke travel in shopping malls experiments in co-operation with glasgow fire brigade-Part I, F. R. Note No. 832, Fire Research Station, Borehamwood, Herts, U.K. Heskestad, G. 1972. Detem~ination of gas venting geometry and capacity of air pollution control system at factory mutual research center, FMRC Serial No. 2058 1, Factory Mutual Research, Nonvood, Mass. Heskestad, G., and J.P. Hill. 1987. Propagation of fire smoke in a corridor. P,-oceeditigs of the 1987 ASME-JSME Thermal Engineering Joint Cotference, 1 March 22-27, 1987, Honolulu, H.I. Arnerican Society of Mechanical Engineers, Martro, P. J. and Tanasawa, I., Editors, pp. 37 1-3 79. Hinkley, P.L. 1970. A preliminary note on the movement of smoke in an enclosed shopping mall, Fire Research Note 806. Borehamwood, Herts, U.K.: Fire Research Station. Hinkley, P.L. 1970. The flow of hot gases along an enclosed shopping mall: A tentative theory, F.R. Note No. 807. Borehamwood, Herts, U.K.:Fire Research Station. Hinkley, P.L. 197 1. Some notes on the control of smoke in enclosed shopping centres, Fire Research Note 875. Borehamwood, U.K.: Fire Research Station. Hinkley, P.L. 1975. Work by thc firc research station on the control of smoke in covcrcd shopping centres. Borehamwood. U.K.: Fire Rescarch Station.

Hobson, PJ., and L.J. Stewart. 1973. Pressurization of escape routes in buildings, heating and ventilating research association. Berkshire, U.K.: Brachnell. Joshi, A.A., and P.J. Pagni. 1991. User's guide to BREAKI, The Berkeley algorithm for br~aking window glass in a compartment fire. Gaithersburg, Md.: National Institute of Standards and Technology, NIST-GCR-91-596. Kennedy, W.D. 1997. The influence of the memorial tunnel fire tests on transit tunnel fire emergency ventilation analysis, Draft of unpublished paper. Klem, T.J. 1987. Ninety-seven die in arson fire at dupont plaza hotel. Fire Journal 81(3): 77, 79, 83, 104, 105. Klote, J.H. 1980. Stairwell pressurization. ASHRAE Transactions 86(1): 604-623. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. Klote, J.H. 1982. A Computer program for analysis of smoke control systems. National Bureau of Standards, NBSIR 82-2512. Klote, J.H. 1982. Smoke movement through a suspended ceiling system. National Bureau of Standards, NBSIR 8 1-2444. Klote, J.H. 1983. Designing effective zoned smoke control systems. Building Design and Constructior7 24( 1 l ): 90-93. Klote, J.H. 1984. Field tests of the smoke control system at the San Diego VA Hospital. National Bureau of Standards, NBSIR 84-2948. Klote, J.H. 1984. Smoke control for elevators. ASHRAE Journal 26(4): 23-33. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Klote, J.H. 1985. Field tests of the smoke control system at the Bay Pines VA Hospital. ASHRAE Transactions 91 (I B): 802-8 19. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Klote, J.H. 1985. Computer modeling for smoke control design. Fire Safeg Jout.nal9: 181- 188. Klote, J.H. 1985. Smoke control in VA Hospitals. ASHRAE Journal 27(4): 42-45. Klote, J.H. 1988. Inspecting and testing air moving systems for fire safety, heatinglpipinglair conditioning, April. Klote, J.H. 1989. Considerations of stack effect in building fires. Gaithersburg, Md.: National Institute of Standards and Technology, NISTIR 89-4035. Klote, J.H. 1991. A general routine for analysis of stack effect. Gaithersburg, Md.: National Institute of Standards and Technology, NISTIR 4588. Klote, J.H. 1993. Design of smoke control systems for areas of refuge. ASHRAE Transactions 99(2): 793-

Appendix B- Bibliography

807. Atlanta; American Society of Heating, Refj-igerating and Air-conditioning Engineers. Klote, J.H. 1994. Fire and smoke control: An historical perspective. ASHRAE Journal 36(7): 46-50. Klote, J.H. 1994. Method of predicting smoke movement in atria with application to smoke management. Gaithersbig, Md.: National Institute of Standards and Technology, NISTIR 5516. Klote, J.H. 1998. North American approaches to smoke management, Seminar Fire Safety-Smoke Control: Standards and Practice, CIBSE-ASHRAE Group and Colt International. Klote, J.H., and X. Bodart. 1985. Validation of network models for smoke control analysis. ASHRAE Pansactions 9 l(2B): 1134-1 145. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. Klote, J.H., and L.Y. Cooper. 1989. Model of a simple fan-resistance ventilation systenl and its application to fire modeling. Gaithersburg, Md.: National Institute of Standards and Technology, NISTIR 894141. Klote, J.H., and G.P. Fomey. 1993. Zone fire modeling with natural building flows and a zero order shaft model. Gaithersburg, Md.; National Institute of Standards and Technology, NISTIR 525 1. Klote, S.H., and R.IH. Zile. 1981. Smoke movement and smoke control on merchant ships. Gaithersburg, Md.: National Bureau of Standards, NBSIR S12433. Klote, J.H., D.M. Alvord, B.M. Levin, and N.E. Groner. 1992. Feasibility and design considerations of emergency evacuation by elevators. Gaithersburg, Md.: National Institute of Standards and Technology, NlSTlR 4870. Klote, J.H., and R.H. Zile. l98 1. Smoke movement and smoke control on merchant ships. NBSIR 81-2433. Gailhel-sburs, Md.: National Bureau of Standards. Klote, J.H., H.E. Nelson, S. Deal, and B.M. Levin. 1992. Staging areas for persolis with mobility liiiiitations. Gaithersburg, Md.: National Institute of Standards and Technology, NISTIR 4770. Lamming, S.D., and J. Salmon. 1994. Wind data for design of smoke control systems, Final Report 816TRP. Atlanta: American Society of Heating, Refrigerating and Air-Conditioni~g Engineers, Inc. Lougheed, G.D. 1997. Expected size of shielded fires in sprin klcred office buildings. ASHRA E T,-nn.suctiot~s 103j l ). -Atlanta: American Society of Heating. Refrigerating and Air-conditioning Engineers, Inc. Lougheed, G.D., and D.W. Carpenter. 1995. Probability of occurrence and espectcd size of shielded fires in sprinklcrcd buildings: Phase 2 full scale tire tests. Final l1spo1-l ASIIRAE RP-S38. Atlanta: Amcri01'

can Society of Heating, Reffigerating and Air-Conditioning Engineers, Inc. Marshall, N.R 1985. The behavior of hot gases flowing within a staircase. Fire Safety Journal 9(3): 245,255. Marshall, N.R 1986. Air entrainment into smoke and hot gases in open shafts. Fire Safe07 Journal lO(1): 37-46. Matsushita, T., and I.H. Klote. 1992. Smoke movement in a corridor-Hybrid model, simple model and comparison with experiments. Gaithersburg, Md.: National Institute of Standards and Technology, NISTIR 4982. Mawhinney, J.R., end G.T. Tamura. 1994. Effect of autoniatic sprinkler protection on smoke control systems. ASHRAE Transactions lOO(1). McGuire, J.H. 1967. Control of smoke in buildings. Fire Technology 3(4): 281-290. McGuire, J.H. 1967. Smoke movement in buildings. Fire Techr?ology3(3): 122-133. McGuire, J.H., and GT. Tamura. 1971. Smoke control in high-rise buildings, CBC 134. National Research Council, Ottawa, Canada. Milke, J.A., and J.H. Klote. 1998. Smoke management in large space buildings. Melbourne, Australia: Broken Hill Proprietary Company Limited. Morgan, H.P. 1977. The flow of buoyant fire gases beneath corridor ceilings: A theory, FR Note 1076. Borehamwood, Herts, U.K.: Fire Research Station. Mogan, H.P. 1979. Smoke control methods in enclosed shopping complexes of one or more stores: 4 design summary. Building Research Establishment. Morgan, H.P. 1991. Smoke control in shopping malls and atria, fire & safety in buildings. Symposium in Hong Kong, Hong Kong Polytechnic. Morgan, H.P. 1993. Fire safety-Smoke control: developments in European Standards, Seminar fir5 safety--Smoke control: Standards and practice. ClBSE-ASHRAEGroup and Colt International. Morgan, H.P. 1998. Sprinklers and fire safety design. Fire Safety Engirleer-irlg 5(1): 16 - 20. Morgan, H.P., and J.P. Gardner. 1990. Dcsign principles for smoke ventilation in enclosed shopping centres, BR 186. Borehamwood, Herts, U.K.: Fire Research Station. Morgan, H.P., and (3.0.Hansell. 1987. Atrium buildings: Calculating smoke flows in arria for smokecontrol design. Fire Safely Jozo-nal 12(1987): 9-35. Morgan, H.P., and N.R.~arshall. 1975. Smoke hazards in covered, ,multi-level shopping malls: An experimentally-based theory for smoke production. Borehamwood, U.K.: Building Research Establishment. Morgan, H.P., and N.11. Marshall. 1978. Smoke hazards in covercd, mulli-lcvcl shopping malls: A method

Principles of Smoke Management

of extracting smoke from each level separately. Borehamwood, U.K.: Building Research Establishment. Morgan, H.P., and GO. Hansell. 1987. Atrium buildings: Calculahg smoke flows in atria for smoke control design. Fire Safety Journal 12: 9-35. Morgan, H.P., and N.R. Marshall. 1975. Smoke hazards in covered, multi-level shopping malls: An experimentally based theory for smoke production. CP 48/75. Borehamwood, Hertfordshire, U.K.: Building Research Establishment, Fire Research Station. Morgan, H.P., and N.R. Marshall. 1978. Smoke hazards in covered, multi-level shopping malls: A method of extracting smoke from each level separately. Borehamwood, Hertfordshire, U.K.: Building Research Establishment. Morgan, H.P., N.K. Marshall, and B.M. Gladstone. 1976. Smoke hazards in covered multi-level shopping malls: Smoke studies using a model 2-story mall. Borehamwood, U.K.: Building Research Establishment. Morgan, H.P., N.R. Marshall, and B.M. Gladstone. 1976. Smoke hazards in covered, multi-level shopping malls: Smoke studies using a model 2-story mall. CP 45/76. Boreham\vood, Hertfordshire, U.K.: Building Research Establishment. Moulen, A.W., and S.G. Grubits. 1975. Stairwell pressurization in a 26-story building. North Ryde, N.S.W. Australia: Experimental Building Station. Moulen, A.W., and S.G. Grubits. 1975. Stainvell Pressurization in a 22-story building. North Ryde, N.S.W. Australia: Experimental Building Station. Quintiere, J.G. 1989. Fundamentals of enclosure fire "zone" n~odels. Journa/ of Fire Protectioti Engineet-itig l(3): 99- 1 19. Quiter, J.R. 1996. An application of performance based concepts at the stratosphere tower Las Vargas, Nevada. Presented at NFPA Research Foundation Syn~posiun~, Francisco, Calif.. June 26, 1996. San Said, M.N.A. 1988. A review of smoke control models. ASHRAE Journal 30(4): 36-40. Sander, D.M. 1974. FORTRAN IV program to calculate air infiltration in buildings. National Research Council Canada, DBR Computer Program No. 37. Sander, D.M., and G.T. Tamura. 1973. FORTRAN IV program to stimulate air movement in multi-story buildings. National Research Council Cauada, DBR Computer Program No. 35. Schmidt, W.A., and J.H. Klote. 1952. In case of fireUse the stainvclls, Elevators aren't safe. Spec$i~ing Engitieer 47(5). Shannon, J.M.A. 1976. Compulcr analysis of [he mo\/ement and conttol of smoke in buildings with mccha~iical and natural vcntilat~on CIB Sympo-

sium on Control of Smoke Movement in Building Fires, Vol. I, Fire Research Station, Garston Watford, U.K., pp. 99- 126. Steckler, K.D. 1989. Fire induced flows in corridors-A review of efforts to model key features. Gaithersburg, Md.: National Institute of Standards and Technology, NIST3R 89-4050. Tamura, G.T. 1969. Computer analysis of smoke movement in tall buildings. ASHRAE Transactions 75(2): 8 1-93. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Tamura, G.T. 1972. Pressure difference for a nine-story building as a result of chimney effect and ventilation system operation. ASHRAE Transactiom 72(1): 180- 189. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers. Inc. . Tamura, G.T. 1974. Experimental studies on pressurized escape routes. ASHRAE Tratuactiom 80(2): 224237. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Tamura, G.T. 1978. Experimental studies on exterior wall venting for smoke control in tall buildings. ASHRAE Transactiotis 84(2): 204-2 15. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Tamura, G.T. 1978. Exterior wall venting for smoke control in tall office buildings. ASHRAE Jotit-tin/ 20(8): 43-48. Atlanla: American Society of Heating, Refrigerating and Air-conditioning Engineers. Inc. Tamura, G.T. 1980. The performance of a vestibule pressurization system for the protection of escape routes of a 17-story hotel. ASHRAE Transactiotis 86(1): 593-603. Atlanta: American Society of Heating. Refrigerating and Air-conditioning Engineers, Inc. Tamura, G.T. 1982. A smoke control system for highrise office buildings. ASHRAE Jourtial 24(5): 2932. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. Tamura, G.T., and A.G. Wilson. 1967. Pressure differences caused by chimney effect in three high buildings. ASHRAE Transac~iom73(2): 111. 1-111. 10. Atlanta: American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. Wakamatsu, T. 1968. Calculation of smoke movement in buildings-1st Report, Building Research Institute, Japan. Wakamatsu, T. 197 1 . Calculation of smoke movement in buildings-2nd Report, Building Research Institule, Japan. Wakamatsu, T. 1976. Unsleady-state calculation of s~noke movement in an actually tired building. CIB Syn~posiumon Control of Smoke Movement in

Appendix B-Bibliography

Building Fires, 1: 81-98. Garston Watford, U.K.: Fire Research Station. Wakamatsu, T 1977. Calculation methods for predicting . smoke movement in building fires and designing smoke control systems, fire standards and safety, ASTM STP-614, A. F. Robertsan, ed. Philadelphia,

Pa.: American Society for Testing and Materials, pp. 168-193. Yoshida, H., C.Y. Shaw, and GT. Tamura. 1979. A FORTRAN IV program to calculate smoke concentrations in a multi-story building. Ottawa, Canada: National Research Council.

Appendix C

Calculation of Elevator Evacuation Time
NOMENCLATURE acceleration, Ws (nl/s2) = number of elevators = number of round trips = number of people entering or leaving the elevator N Nd,,: = number of people entering or leaving the elevator during the dwell time S = distance, ft (m) ST = total floor to floor travel distance for trip, fi (m) t = time, s (S) = elevator evacuation start-up time, s (S) I, ~d = time for elevator doors to open and close, s (S)
=

Subscripts
2

a J m

T = end of leveling car motion (also end of travel) 1 = end of constant acceleration motion 2 = end of transitional acceleration motion 3 = end of constant velocity motion 4 = end of transitional deceleration motion 5 = end of constant deceleration motion During building fires, elevators are almost always taken out of service, and vertical evacuation is by the use of stairs. Elevators have been used for emergency evacuation in a few unique situations (subway stations several stories underground, luxury apartments, and towers). The Life Safety Code (NFPA 1977) recognizes elevators as a second means of egress for towers. This appendix presents a detailed method of analysis of people movement by elevators during emergency building evacuation, based on principles of elevator engineering (Strakosch 1983). Bazjanac (1977) and Pauls (1977) have developed methods of calculation of evacuation time by elevator, but the method presented here incorporates more detail about elevator motion and elevator loading and unloading. The ELVAC computer program by Klote and Alvord (1992) is based on the analysis of this appendix, and an example ELVAC analysis is included in this appendix. The sequence of elevator operation for emergency evacuation is complicated and has many possible variations. The following general sequence is presented to provide a framework for the method of analysis presented in this paper. Upon activation of emergency evacuation, elevators in normal service will go to a discharge floor where any passengers on the elevators will exit. This discharge floor may either lead to the outside

tdL, = dwell time for elevator doors, s (S) t, t,, t, t,,
I,
= = = = = = = = = =

t,. t, t,,

!, ,
V

e\.acuation time, s (S) time for leveling of elevator car, s (S) time for A' people to enter elevator car, s (S) time for one person to enter elevator car, s (S) travel time from ele\.ator lobby to outside or to other safe location, s (S) time for elevator car to make a round trip, s (S) standing time, s (S) time for A1people to leave elevator car. s (S) time for one person to leave elevator car. s (S) \.eloci t: ftk (mk) ) basic transfer inefficiency total transfer inefficiency. , = a + E + ; U / door transfer inefXcicncy other transfer inefticiency trip inefficiency

V,,, = normal operating velocity. Ws ( d s )

a
p
E

= =
= = =

y

11

Appendix C-Calculation of Elevator Evacuation Time

or lead to an area of relative safety where people~may stay during the fire. The elevators .will make a number of round trips to transfer occupants from other floors to the discharge floor. During evacuation, the elevators may be under a special emergency evacuation mode of automatic control or under manual control. The evacuation time addressed in this paper is an idealized time for people movement that does not account, for the complex human behavior that often occurs during emergencies. It is believed that the analysis of this paper is about as accurate as that for evacuation by stairs. EVACUATION TIME Analysis of people movement during elevator evacuation must take into account the number and arrangement of elevators in a building. Genkrally, elevators are 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 elevator groups are discussed later. The method of analysis and the computer program of this paper are for the calculation of the evacuation time for one group of elevators. For buildings with multiple groups of elevators, the approach presented in this paper can be applied separately to each group of elevators. Ideally, the time to evacuate a number of people using one group of'elevators consists of the sum of all the round trip times divided by the number of elevators plus the time needed to start up the elevator evacuation and the travel time from the elevator lobby to the outs ~ d (or to another safe location). Accounting for ineffic ciencies of elevator operation, this e\acuation lime can be expressed as

elevators being out-of-service depends on a rlumber of factors, including the age of the elevators and the quality of maintenance. Because the out of service condition can significantly increase elevator evacuation time, any analysis of elevator evacuation should take this condition into account. START-UP TIME The elevator evacuation start-up time is the time from activation to the start of the round trips that evacuate people. For automatic elevator operation during evacuation, a simple approach is to start elevator evacuation after aH of the elevators have been moved to the discharge floor. For this approach, the start up time, I, consists of the time for elevators to go to the discharge floor plus the time for the passengers to leave the elevators. This can be expressed as

where rr is the travel time for the elevator car to go from the farthest floor to the discharge floor. l,, is the time for passengers to leave the elevator, td is the time for the doors to open and close once, and p is the total transfer inefficiency. These terms are discussed in detail later. An alternative to the simple approach discussed above consists of starting the evacuation operation individually for each elevator when it reaches the discharge floor. This alternative could result in slightly reduced evacuation time. However, this alternative is not discussed further here because of its limited benefit and added complexity. For manual elevator operation, the tinie for ele\.ator operators to be alerted and then get to the elevators must be included in the estimate of start-up time. This additional time may be considerably greater than that calculated 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 round trips, J is the number of elevators. 11 is the trip inefficiency, t,, is elevator evacuation start-up time, and 1 , is the travel ti~ilefrom the elevator lobby to the outside or to another safe location. The round trip time depends on the travel timc of the elevator and on the nulnbcr ol'pcople carried by the elevator, as discussed latcr. The travcl timc from the elevator lobby to a safe location can be evaluated by conventional~n~ethods people niwen~cnt(Chapter 4). of The trip inefticiency accounts tbr trips to empty floors and trips to pick only a few stragglers. The elevator evacuation start-up time is discussed in the next section. The nunibcr of elevators, .J, used in Equation ( C l ) may bc less than the number ofclcvators in the group to account h r out-of-service elevators. Thc probability of'

The round trip starts at the discharge floor and consists of the following sequence: elevator doors close. car travels to another floor, elevator doors open, passengers enter the car, doors close, car travels to discharge floor, doors open, and passengers leave the car. The round trip time, t,, can be written as

where t, is the standing tinie and f7-isthe travel time for one way of the round trip. This equation is based on the elevator only stopping at one floor to pick up passengers. It is expected that most elevators will fill up on one lloor and procccd to the discliary floor. Whnt coristitutcs a full elevator is disci~sscd later. If an clcvator stops to pick up pas-

Principles of Smoke Management

sengers at more than one floor during a round trip, Equation (C3) can be modified accordingly. However, the trip hiefficiency accounts for such multiple stops.

chosen to be 0.05 for hospital elevators. Generally, for o f i c e buildings, y is taken as zero. The time, fd,for the doors to open and close depends on the width and type of the doors, as listed in Table C-l. The kinetic energy of closing doors is limited by elevator safety codes and is usually not more than 0.29 J (7 ft poundal'6). This is why doors from different manufacturers take about the same time to open and close. Types of elevator doors are shown in Figure C3. Door 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 closed and locked, and passengers cannot leave an elevator until the doors are fully opened or nearly fully opened. Generally, elevator doors do not open until the car has stopped and is level with the floor. However, some center opening doors start opening while the car is leveling, and the times listed in Table C- l should be reduced by one second for these preopening doors. The time, fi, for pcople to enter an elevator depends on the number, N,of people entering and on the door operation. As previously stated, it is expected that most elevators will fill up on one floor and proceed to the discharge floor. However, elevators will be less than full when there are not enough people waiting in the lobby to fill an elevator or elevators. Thus, the analysis must include partially filled elevators. Strakosch (1983) has observed elevator loadings in which passengers do not board an elevator and choose to wait for the next one. These observed values are based on 0.22 m2 (2.3 ft') of floor space in the elevator car per person. It should be noted that the ASME A17.1 (1987) elevator standard allows a maximum loading at 0.14 m2 (I .S ft2) per person, but this high density is not achieved in normal practice. The observed values of Strakosch are suggested as the number of persons in a full elevator car, and these loadings are listed in Tables C-2 and C-3. When elevator doors open, the doors remain open for a least fixed time, referred to as the dwell-time, rd,,The time that the door is open can be extended beyond the dwell-time by blocking of the light beam across the door opening or by pushing the door safety edge. The time, ri, for N people to enter an elevator car can be expressed as
I6.The poundal is the unit of force in the pound mass-poundal system of units, and one poundal equals 0.03 1 1 pounds force.

STANDING TIME
The standing time is the sum of the time to open and close the elevator doors twice, the time for people to enter the elevator, and the time for people to leave the elevator. Considering transfer inefficiencies, the standing time for a round trip can be expressed as

where p = a + E t y. The basic transfer inefficiency, 4 allows for rounding off of probable stops, door operating time, door starting and stopping time, and the unpredictability of people. Typically, a value of 0.10 is used for thebasic transfer inefficiency for commonly accepted arrangements of elevator groups, as illustrated in Figure C I . For each of these arrangements, the configuration of the elevator lobby is such that passengers can recognize which elevator has arrived and get on the elevator without excessive delay. Further, these lobbies have sufficient space so that people exiting one elevator will have a minimal impact on the flow of people leaving another elevator. Arrangements of elevator groups other than those commonly accepted can be less efficient and require an increased value of the basic transfer inefficiency. These unusual arrangements include cars separated, too many cars in a line, angular arrangement, and cornered arrangement (Figure C2). Separation of elevators results in increased boarding time for people waiting by one elevator who have to walk to another when it arrives. If the separation is too large, some passengers choose to let elevators go by without boarding. Use of too many elevators in a line has similar inefficiencies. With the angular arrangement [Figure C ~ C ]cars at the narrow , end tend to be too close together while cars at the wide end tend to be too far apart. In the cornered arrangement (Figure C2d), passengers entering or leaving corner cars tend to interfere with each other. The door inefficiency, E. is used to adjust for any increase in transfer time over that of a 1200 mm (48 in.) wide center opening door. Values of E are listed in Table C-1. The inefficiency, y, is used to account for any other inefficiencies in people transfer into or out of elevators, such as increased movement times within an elevator car due to an unusual elevator car shape or limited physical capability of passengers. For example, y often is

~ ~ ~ eC - calculation of Elevator ~vacuation n d k T~A

B l A

U

(a) Two Car Group

1.5Ai; B r 2A

B = 1 S A , but not less thzn 1.8 m (6ft)

(b) TWO Car Group

(c) Three Car Group

B = 1.5A, but not less than 2.4 m (8ft)

(e) Four Car Group

(d) Three Car Group

-r-

A

l
I I

2Open or Closed
B

-

-

_--_J

L

B

I

, --J--

B = 2A

Both Ends of Lobby Open

1.5As B c 2 A

(f) Four Car Group

(g) Six Car Group

-r
B
L

A

1

, Open or
Closed

_1

Both Ends of Lobby Open

1.75As B 52A

B = 2A

(h) Six Car Group
Figure C1
Conzmor71~~ accepted elevafor- a~ar7gemer71s.

(i),Eight Car Group

Principles of Smoke Management

ti =

{

tdw

forNI2.

t,,

for N I 2

tdH,+ tio(N- Ndw) for N > 2

(C51

{ldw+ tuo(N- Ndl) for N > 2.

(c61

where Nd, is the number of people entering the elevator during the dwell time, and tio is the average time for one person to enter the elevator. The number of people entering the elevator during the dwell time is the tern (td,JIio), rounded down to the nearest integer. The time for Npeople to leave an elevator can be expressed in a similar manner.

For the computer program of this paper, the dwelltime is taken to be 4 seconds, the average time for one Passenger to enter an elevator is taken to be 1 second, and the average time for one Passenger to leave an elevator is taken to be 0.6 seconds.

Table C-l: Door Operating Time and Transfer Inefficiency
~ i r n e to Open ' Width Door Type Single-Slide 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: Centeropening 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.

Door Transfer Inefficiency
E

and Close

mm (in.)

900 (36)

.

td (S) 6.6

0.10

(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
(a) Single-Sliding Doa
(d) Center Opening Doon

L

I
(b)

Two-Speed Sliding Doom

I

0 0 0
( l Two-Speed. Center-Opening O c

L
(e) Venical Bipaning Doom

m

Figure C3

Types of elevator doors.

Table C-2: Car Size and Observed Loading in SI Units
Capacity kg (Ib) 1200 (2640) 1400 (3080) 1600 (3520) 1600 (alt.) 1800 (3960) 1800 (alt.) 2000 (4400) 2250 (4950) 2700 (5940)
I.

Car Inside (mm) Wide Deep 2100 1300 2100 l450 2100 1650 2350 1450 2100 1800 2350 1650 2350 1800 2350 1950 2350 2150

Area (m ) 2.73 3.05 3.47 3.4 1 3.78 3.88 4.23 4.58 5.05

2

Observed Loading' (people) 10 12 16 16 18 18 20 22 25

Sec footnote on Table C-3.

Table C-3: Car Size and Observed Loading in English Units
Capacity (Ib) 2000 2500 3000 3500 3500 (alt.) 4000 4000 (alt.) 4500 5000
I.

Car Inside (in.) Wide Deep 68 5l

Area (ft2) 24.1

Observed ~ o a d i n (people) ~' 8

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
Elevator motion is depicted in Figure C4 for most 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 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). Deceleration has the same magnitude as the acceleration, and the total acceleration time equals the total deceleration .. 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, elevators do not always reach the nonnal operating velocity, and methods of analysis for these short trips are presented later.

The distance traveled by the end of transitional acceleration is

The one-way travel time is

The leveling time must be added to the above time to get the total travel time for a one-way trip.

Usually elevators do not stop exactly at the desired floor at the end deceleration, so the elevator must be moved slowly up or down to get it nearly level with the 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 (going to point I on Figure C4) is

Motion Reaching Transitional Acceleration
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 acceleration are given by Equations (C7) and (CS). The velocity at the end of transitional acceleration is

The distance traveled during constant acceleration is

Transitional accsleration is approximated by considering rhe product of velocity and acceleration to be a constanr. The time to reach the end of transitional acceleration (point 2 of Figure C4) is

The time at the end of transitional acceleration is

The one-way travel time is

/
Constant Accelerat~on

Transitional Accelerabon
I

Transittonal Decelerat~on\

Leveltng

\

2

0

t,

t2

Time

t,

t4 .

t 5 t6

Figure C4

Id

o c i of ~ ~ elevarot- reaching normal operating ve1ocif.v. V,,.

Appendix C-Calculation of Elevator Evacuation Time

Transitional Acceleration

Transitional Deceleration

Leveling

Constdnt Acceleration

-l

Constant Deceleration

0

fl

f2

4

5 fr

-

0

t1

5

tr

Time
( a ) Car Reaching

Time
(b) Car Not Reaching

Transitional Acceleration

Transitional Acceleration

Figure C 5

V e l o c i of elevators not reaching normal operating velociv ~

Motion Not Reaching Transitional Acceleration
When the trip does not go beyond constant acceleration, the motion is illustrated in Figure C5b. The oneway travel time is

89.1 s (the same as calculated in Example Cl). In order to move 90 people from floor 21, the elevator trips are considered to consist of five trips with a full car (16 people) plus one trip of a partially filled (10 people) car. 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. On floor 10 of this example, 3% of 90 people are evacuated-this is rounded up to three people. Because this is done by one trip, the round trip time of 45.8 s listed in Table C-5 is for moving three people rather than the full car load of sixteen. The total round trip time of 5395.6 s is the sum of all the round trips to move people from all the floors. The evacuation time of 1258.3 S using five elevators was calculated from Equation (Cl).

COMPUTER EVACUATION ANALYSIS
The computer program ELVAC, written in Quick BASIC, calculates evacuation time for one group of elevators. For buildings with multiple groups of elevators, the program can be used a number of times to calculate the evacuation time for each group. Discussion of Table C-5 provides insight into the computer program. The round-trip time for floor 21 is

Principles of Smoke Management

Table C-4: Parameters for Example C3
Number of stories Number ~f elevator cars Number of people per floor Percent of people evacuating by elevators from floors 2 to 10 Percent of people evacuating by elevators from floors 11 to 21 Height between floors Operating velocity of'elevator car, V , Car acceleration, a Other transfer inefficiency, y Trip inefficiency, g Car full load

21 5 90 3 100 3.2 m (10.5 ft) 3.0 m/s (590 fpm)
1.20 m/s2 (3.94 ft/s2) 0 0.10 16 people

Table C-5: Elevator Trip and Evacuation Time Calculated by ELVAC Computer Program
One-way Elevation Floor 21 (m) 64.0
(ft

Trip Time (S) 24.4

210.0

Round Trip Time (S) 89.1

People on Floor 90

Percent Elevator Evacuation 100

Number of Round Trips 6

Time per ~loor-(S) 524.1

Total round trip time (S) Start-up time (S) Time to get outside after leaving the elevator (S) Evacuation time using five elevators (S)

=
-

=

5395.6 41.3 30.0 1258.3

Appendix C-Calculation of Elevator Evacuation Time

'

I

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 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 elevator 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 From Equation (C4). the standing time is
Is

S.

I, =

4 + 0.6(N- 6) = 4 + 0.6(16 - 6) = 10 S.

= (ti+tU+2td)(l + p ) = ( 1 6 7 10+2(5.3))(1 +O.l) = 40.26s.

lI
1

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 =

2a

--

v:

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) 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

Il
1

11 A 1600 kg elevator in an ofice building

-~

--

-.. - - - - r
~

~

- - - -

~.

lI

a mund trip fmm the ground floor to pick up a fdl load of passengers horn h e 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 elevator shape is not unusual and the passenger capability is normal, so y is 0. The total transfer ineficiency is p = a + ~ + y = O . 1 0 + 0 + = 0.10.' 0 From Equation (CS), the time for 16 people to enter the elevator is ti = N = 16 S . From Equation (C6), the time for 16 people to leave the elevator is

Il
II

11

tU=4+0.6(N-6)=4+0.6(16-6) = 10s.

I
l 11
I
1

From Equation (C4), the standing time is 1, = ( t i + t,, + 2td)(1 + p ) = (16 + 10 + 2(5.3))(1 + 0.1) = 40.26 Consider V is 60% of V,,,, then V, = 0.6Vn, = 0.6(3) = 1.8 d s . 1 From Equation (C7), the time at the end of constant acceleration is t l = Vi/a = 1.8/1.2 = 1.5 From Equation (Cgj, the distance traveled during constant acceleration is
S. S.

Il

II

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
S

I
ll

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

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~floors included in the elevator evacuation. The rest of the people are 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-ofservice; 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 CONTAMW is a computer program developed by George Walton at the National Institute of Standards and Technology. The program is a multizone indoor air quality and ventilation analysis program that is useful in a variety of applications. For smoke management purposes, the program can be used to help calculate roomto-room airflows and pressure differences induced by mechanical and natural forces. 5. Attachment 2 (Floor Layouts)--provides floor layouts with zones and airflow paths identified in this example application. Attachment 3 (Simulation Results)-provides the tabular and schematic results of the simulation for this example application. An example of the shaft report generated by the program is also included.

6.

SECTION 1 DESCRIPTION OF EXAMPLE : The CONTAMW computer program is described in this appendix by way of the following example. Building Description. The building has 12 stories and a roof. The overall dimensions of the building are 246 ft (75 m) by 328 ft (100 m) by 157 ft (48 m) (in height). The building has two stairwells that have the dimensions 7.9 ft (2.4 m) by 33 ft (10 m). Each stair door is 2.9 ft (0.91 m) in width and 7 ft (2.13 m) in height. The center of the doorknob is 3 in. (0.076 m) from the edge of the door, and gaps around the closed doors are 0.125 in. (0.00305 m) (vertical gaps) and 0.25 in. (0.00636 m) (horizontal gaps). The interior doors of both stairs are open on the 1st floor. All other stair doors are closed. Wall and floor construction in this building is "loose." Smoke Management Description. The building has zoned smoke control and two pressurized stairwells. The 6th floor is the smoke zone and is exhausted at a rate of 30,000 scfm (14,158 SUS).The 5th and 7th floors are pressurized at this same flow rate. The stairwells are each pressurized by a fan on the roof. The minimum and maximum allowable pressure differences for the stairwellare0.05 in. W.C. (12.5 Pa)and 0.36in. W.C. (90 Pa). InteriorIExterior Environmental Description. The building temperature is 72°F (22°C). The stairwell

What is the purpose of this appendix?
This appendix describes the use of CONTAMW by way of an example application. The data for this example application are selected as an example of illustrating the input and output form of CONTAMW and are not intended to represent recommended values for this program.

How is this appendix organized?
This appendix is organized as follows: Section I (Description of Example)-provides a brief narrative of the example application. Section 2 (Data Input Process+provides a detailed description of how to input data into the CONTAMW program using the input data from the example application. Section 3 (Data Output Process+provides a description of how to run simulations in CONTAMW and how to read the output data. Attachment I (Input Databprovides input data used in this example application.

Appendix D - Application of CONTAMW

Figure D1 Simple schematic offirstf200~ tempirature is 76°F (24OC) on the 1st floor and increases linearly to 90°F (32OC) on the 12th floor. The outside temperature is 85°F (20°C), representing a summer condition. The wind speed is 10 mph (4.5 d s ) at 30 ft (9.1 m) above ground level. There are two wind coefficients-0.8 and -0.8-that are used to simulate the pressures on windward and leeward walls, respectively. The terrain around the building is considered "urban." Other quantitative and qualitative input data are provided in Attachment 1. refer to the CONTAMW Help Program or the CONTAM96 User's Manual. Note: Since this is a schematic diagram, the drawing does not need to be to scale. However, geometric relationships of relevant building features will facilitate review of model outputs. In addition, the user should refrain from creating additional building levels before identifying all zones and airflow paths on the floor level drawn initially. This will allow all building elements to line up from floor to floor. The creation of new floor levels is discussed further in Step 7. When drauing the roof level of a building, all walls and zones should be deleted from the drawing. In general, the only elements that show on the roof level are airflow elements, such as leaks, stair openings, fans, etc. Further discussion on roof drawings is discussed further under Steps 4 and 8. Example Application: Four exterior walls, two stairs, and one shaft (representing an elevator shaft) are presented. Figure D1 illustrates the program screen for drawing the example schematic.

SECTION 2: DATA INPUT PROCESS
The data input process for the example application is broken into the following seven steps:
1. 2.

Drawing Building Components Identifying Zones Describing Zones Identifying Airflow Paths Describing Airflow Paths Describing Weather Data Creating and Defining New Levels

3.

4. 5. 6. 7.

Step 2: Identifying Zones
Zones indicate a volume of air with uniform temperature and contaminant concentrations. The user can identify a zone as any area of interest (i.e., stair, room, smoke management zone, etc.). Once walls have been drawn, "normal" zone icons are placed within the wall boundaries of the building. In addition to "normal" zones within the building interior, the model represents, by default, an "ambient" zone for the exterior environment surrounding the building. Note: Every physical division in the building, as defined by the schematic, must be identified with a single and uniquely defined zone icon.

Each of these steps is discussed in detail in this section. The purpose of each step is presented, followed by its use in the example application. Examples of the program's screens are provided to assist the user in understandir.2 the narrative. Notes are identified in areas where the user should be cautious during the data input process.

Step 1: Drawing Building Components
Use the SketchPad to draw the desired structural components of the building, including exterior walls. interior walls, stairs, ducts, shafts, etc. For further assistance on drawing building components. the user should

Principles of Smoke Management

'Normal" zone (has

I

(~nddned zone

11
ICo16.~or14

I w l l : of1 le<,l

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) Building interior (Rml) (ignore compartmentation within the building interior) Elevator shaft
volume:

... ...

---

-.........

- ............... .

-p , -

I

3.
4.

Tempedre: pressure:

21 4.

10.
G Variable

PO

_r]

c Constant
196.

In addition, the computer identified the "ambient zone" by default. Figure D2 illustrates the proram screen used to identify zones.

r

n

Il

Step 3: Describing Zones
Each zone must be identified by numeric parameters. The zone data include the name of the zone associated with the zone icon, the zone temperature, pressure (constant, variable, or none), volume/floor area, and initial contaminant concentration if applicable.

)lI .d~ a n i
Figure D3 zone properties for stair I on rhe first jlooy.

il

Note: The zone name must be unique for this level and is limited to four characters. In addition, the user inputs either the floor area or volume information. The program will automatically calculate the field that is not entered by the user. Example Application: Variable pressure is used where pressures are determined by the model for each zone based on an analysis of conservation of mass. Figure D3 illustrates the program screen used to describe

zone properties. Contaminant data were not used in this application.

Step 4: Identifying Airflow Paths
An airflow path indicates some building feature by which air can move from one zone to another. The user should identify any flow element on an exterior wall, interior wall, door, or floor (e.g., leakage paths. other openings, fans or shafts).

Appendix D- Application of CONTAMW

p - - - -

I

t o l S ~ p IS . w

-- ,--

L ~ V O I (1

>.

1 0f1

ZJ
n

Figure D4 Selecting an aitflowpath onjrstjlooc
Note: Floor leaks must be identified at 0 meter elevation for the floor level in question. The propm describes these leaks as airflow between that level and the level below. In addition, airflow paths must be identified on each exterior wall ifthe user is taking into account wind effects. Path elements can be identified for each wall or one path element can be identified and the areas involved described in parallel. When adding a fan on the roof for stair pressurization, the user should make sure that the airflow element is joined up such that it is within the boundary of the stair enclosure. "Large" openings and "small" openings can be used interchangeably at the user's discretion to help facilitate review. Example Application: Horizontal and vertical airflow paths are identified. Horizontal airflo\v paths include "large" openings, such as stair doors, exterior doors, and elevator doors. They also include "small" openings, such as leaks in exterior walls, interior walls, and closed doors. Vertical airflow paths include "large" openings, such as stair and elevator enclosures. They also include "small" openings such as leaks between levels. Since the 1st floor does not have a level below it, there is no floor leak identified from that floor. Figure D4 illustrates the program screen used to identify an airflow path. age characteristics. The height of the flow element should be entered as the midheight elevation of the element. Note: Once a new element is defined, the identified element is stored in the user-defined library. The user can edit an existing element at any time and the changes will apply to all airflow paths identified as that particular element. Example Application: The filter and schedule tab is not used in this example application. New userdefined elements were added for the horizontal and vertical airflow paths in this example. Figure D5 illustrates the program screen used to describe a new airflow element. Airflow path properties are broken into three intermediate steps: defining airflow element characteristics, identifying flow paths, and determining wind pressures (where applicable).

Itzternzediate Step I : Defining Airflow Eleineizt Characteris tics .
Airflow elements describe the mathematical relationship between the flow through an airflow path and the pressure drop across the path. CONTAMW includes choices of several types of flow elements and mathematical models relating the pressure difference, area, and mass flow. A mathematical model must be selected for each new airflow path identified. Note: Fields in the model input screens are either userdefined or default values. Refer to Attachment 1 for details on the values used for this example application. Example Application: The following models are used:
1.

Step 5: Describing Airflow Paths
Once placed on tlie SketchPad, the user can define airflow characteristics and move, copy, and delete them. The user should refer to the CONTAMW Help Program or CONTAM96 User's Manual for additional guidance on describing airflow paths. When describing airflo\v element characteristics, flow paths are specified as either new elements or as an existing element in the CONTAMW library. Input data for each airflow path include information on the zones that the paths connect, their height, and olhcr quantitative information on leak-

One-way flow using powerlaw model, orifice area 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 _
T ~ F l o Models w

C Tes(DataCLpom!J

r T w Opening
-.........

- -

C ,

C I

__ - _ _ _

_-

_

r
...........

One Openmg
-

.....

- ...

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. One-way flow using powerlaw model, stairwell, for all vertical stair openings behveen levels. One-way flow using powerlaw model, shaft, for vertical elevator shaft openings behveen levels. Fan and forced flow model, constant mass flow, for all mechanical fans (exhaust and pressurization fans).
Hydraulic Diorneter. R e y n o l d s Number:

3.
4.

10 9 8 4 3 7 8
130

n

5.

Description: 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. Figures D7 through D1 1 illustrate the inpilt data required for the models used in the example application.

1

C L a r g e oponing

. . . . . . . . . .
0K

0

-1

I

I

I

Cancel

Intermediate Step 2: Ident$j~zg Flow Paths
Once numeric parameters for flow characteristics have been added to the model fields, flow path properties (flow path tab shown in Figure 0 5 ) must be defined. When defining flow path properties, the elevation of

Figure D7 Pon.erlaw model, orifice area dafa.
flow elements is most accurate at midheight of the opening. The default value in the program is the mid-height of the room.

Appendix D - Application o CONTAMW f

..

Figure D8 Powerlaw model, leakage area.

Figure D 11 Powerlaw model, stainoefi.

Perimeter.
Roughness:

(121.d
11 07 .Design (mm) flow rate:
- -

Descaipuon:
elt~cal floor leak In shafl loose mnstruaon. 3 X 3 m m

i

l

1

G Small openmg

o
0

~ ~ I O W that rate

provides desired d e s ~ g n pressure i i l

C Large opening

l
I .

--

-,

..

1 0 1 1 .
1cvlasl
5

-

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 horizontal flow path is the midheight of the staidelevator doors and the elevation of the vertical flow path is zero. In addition, the user must identify the positive flow d i i t i o n r of an airflow path f ~all fans (or other element type where a flow rate is designated). For all other elements, the propm arbitrari!y se!ects the positive flow direction. Example Application: In this example, the positive flow direction from ambient to the building interior is defined for the fans on the roof and the 5th and 7th floors (in this manner, these fans supply rather than exhaust air to the floor). The positive flow direction from the building interior to ambient is defined for the fan on the 6th floor. Elevation of flows was at midheight of level for all walls, mid-height of doors for all doors, or at floor level for all vertical flows. Figure D12 illustrates the program screen used to define flow path elevation and flow direction.

sure option is used for the leaks on ,the roof and all airflow paths inside the building interior. Figure D13 illustrates the screen used to describe wind pressure., For variable wind pressures, three inputs are required: a wind pressure modifier, a wall azimuth angle, and a wind pressure profile. The wind pressure modifier is determined using the equation

Intermediate Step 3: Determining W i n d Pressure
Wind pressure characteristics are included only for elements for which a flow exists between "normal" and "ambient" zones. Three wind pressure options for openings exist in the CONTAMW program: no wind pressure, constant pressure, and variable pressure (dependent on wind speed and direction). Note: Data entry corresponding to the wind pressure option selected is required. Example Application: Variable wind pressure is chosen for all exterior airflow paths. The no wind pres-

where Ch is the wind pressure modifier, A, and a depend on the terrain around the building (ASHRAE 1989, p. 14.3), and H is the height of the roof or wall. The wall azimuth angle is defined as the direction the wall faces with north being 0 degrees, east 90 degrees, south 180 degrees, and west 270 degrees. A default azimuth angle is provided based on the orientation of the wall on the SketchPad with the top of the SketchPad. being north. The wind pressure profile is based on wind coefficients and their respective wind azimuth angles. Example Application: The wind pressure modifier was calculated using the building height (48 meters) and urban terrain factors (A, = 0.35 and a = 0.40). Figure D14 illustrates the weather and wind parameters screen from which weather and wind characteristics are presented. In addition, the default azimuth angle of 0 degrees is used in this example application. The wind coefficients used are 0.8 for the windward wall and -0.8 for all other walls. As mentioned, the windward wall azimuth angle is 0 degrees (north). Figure D15 illustrates the wind pressure display based on these wind coefficients.

Flow Element

1 Falter dnd schedule

-

..

+-U

Path Number

6 -1

h o w Path

. l'.
'<

W n q d

&br&r%

. .

-L.,.

,...-, -.

l

-

.

'

>

7 .

.

R e l a w Elevation:

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 parameters with the exception of ambient temperature, which is 20°C. Figure D16 illustrates the location specific weather parameters used.

Local Terrain Constmt
Velocity Profile Eqonent Wnd Speed Modifier.

v
1 3 5
10 4797

II

Il

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 workI 1 ing on a drawing right away without having to create a Figure D14 Wind chat-acferisfics urban seffing. for new level. Each level should be given a name, an elevation of the level above ground, and the height of the level from floor to ceiling. Whenever a new level is created-whether it is a blank level or a copy of another level-XONTAMW will give it a default name that will Step 6: Describing Weather Data consist of a number enclosed within the "<" and '5" characters. The user can modify the default names as Weather parameters (i.e., ambient temperature, required for the particular building. baroixctric pressure, wind speed, and wind direction) 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 purwind data. CONTAMW uses default values for temperaposes. As previously mentioned, it is recommended that ture and pressure (with no wind): 20°C and 1 atmothe first level be completed first with all "typical" zones and airflow paths, so that the elements line up from floor to sphere (approximately 101 kPa), respectively.The wind floor. Other elements can be added or deleted as deemed speed field is used to address the reference velocity used necessary. In addition, vertical leakage paths (i.e., stairs, for the wind.

ll

Step 7: Creating and Defining New Levels

-

Principles of Smoke Management

Figure D15 Ifindpress~ti-e profile for defined wind coej7ccients.
shafts, floor leaks) are included when new levels are created since leaks via the floor are indicated oh the levels above. The user must be careful to make sure that all airflow paths specific to each level are accounted for. All level names must be unique. In addition, the value for the "distance to the level above" is used by CONTAMW to calculate zone volumes based on the floor area of each zone.

WenIher Wind

l

1 Locaaon

Wind Pressure Display

/

These values are used tor deterrnln~ng mnd pressure jdtsplayfen(ures onlyi .bmbrent~e&erahlre Atlsoluie h s s u r e Wind Speed Wind DseQion.

T I P , 11 y l m p h d

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 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 as the roof, and floor leaks have been adde6 on the 2nd through roof levels. The elements referring to the exterior doors have been deleted from the 2nd through roof levels. Attachment 2 provides the layout of each level (including the zones and airflow paths present).

Figure D16 Bztilding site related weather- data.

Once the results are available, the user can view results in two primary formats: schematic form and tabular form.

SECTION 3: DATA OUTPUT PROCESS
CONTAMW provides the user with several types of simulation results. The results available after a simulation depend on the simulation method and output parameter settings. The simulation is run based on set simulation parameters. For this example application, the simulation parameters used were default values provided by CONTAMW. Figure D17 illustrates the tab that is used to run the simulation.

Schematic Results. The CONTAMW program displays color-coded bars indicating the relative airflow rates and pressure drops associated nith each airflow path on the current level of the Sketchpad. Airtlow rates are shown in blue and pressure dift-crences in red. Downward positioned bars indicate that airflows/pressures are going from that level to the level below. Upward positioned bars indicate that airf~ows/pressures are going from that level to the level above. Schematic results for the example application arc provided in Attachment 3. Note: When a airflow path icon is highlighted, the respective airflow and pressure results (in addition to level and

Appendix D - Application of CONTAMW

aifflow element name) will appear in the lower left-hand corner of the prograin screen.
-

Tabular Results. CONTAMW can export results to comma-separated format file that can later be imported to a spreadsheet program for further analysis. The user can plot aifflow, contaminant, exposure, and pressure results. In addition to comma-delimited format, CONTAMW generates a shaft report- special reporting a feature f i r shafts. The shaft report generated by the program will display pressure drop, airflow rates, and

direction of airflow on every level of the shaft where there are'airflow path icons in the same location on the Sketchpad. As with the comma-delimited results, the shaft report can be saved as a text file. Tabular results (comma-delimited and shaft report format). for the example application are provided in Attachment 3, Figure D18 illustrates the program screen used to export comma-delimited results and generate a shaft report.

NOTE: A simulation must be run first in order to export results/generateshaftreport

Figure D17 Rmnitzg the CONTAMWsimlclation.

!

........

= l
! JenllD0:OO:OO

.

--....-F

Zone

h b t

Lwel cl>: l ol 13

-.

.I

Figure D 18 E.vpo~rir~g ~zslrlfs.

Principles of Smoke Management

Powerlaw Model: Leakage Area Airflow properties Element N:lme Preswre Flow 1,eskage Discharge I)rop Cocfficicnt Exponent c : dclault tlcl>~ult le~ult c\tbuw:llll 0 1 4 ,,,2
c t s r , l

,
0,

,
n12

c l ~ l t

i

ilcf;~ul~ default default default

(~C~~~LII~
default default dcftlul[
I
I

j

II

estwall l estwall2 1loorlc;lk
I

default default tlcSault
I I

0,1051n2 1,275 ,,,2 0,00153 ,,,2

II

Illcakslilil

dePdult tlcli~ult default defadt

defidtrlt dclhult default default

default
I

Illc;~ks~rl 0 , 0 0 4 ~ 8

dd;lult default default

/

intivallrhft intwallstr

0,0042 m 2 0,014 m2

Flow Path Relative Multiplier Positive Flow Direction Elcv:~tio~~ tlrli~ul[ cdculated by 2 Droeraln I I I calculated by dclhult 2 program calculated by default 2 Droeram default calculated by 2 program calculated by 0 default Drocraln l l I calculated by dcfault 0 program . tlcliult calc~~latcd by 0 program calculated by 2 default program calculated by 2 default I program 1
I Y B

Limits Nonc
II

Wind Pressure Wind Pressure Wind Pressure Modifier Option Variable 0.47972 1
I
I

Wall Azimuth Angle 901270 011 80 011 80 901270 None None None None None

None None None None
II

Variable Variable Variitble None
4

0.479721 0.47972 1 0.479721 .None None None None
'

U

. -

None

None

,
None
'

None None None

None None

II

None

Powcrlaw Model: Stairwell Airlluw I'ropcrtics Dist;~ncc Cross-Section:~l Density of' Stair Between arc;^ People Trei~tls Lcwls 4 111 24 1n2 Dcljult Closed
m

llclative Multiplier Positivc Flow Elev:~tio~~ Dircctiou 0 delh~lt
calculated by

Limits

Wind Pressure Option None

Wind I'rcssitre Wind Presst4r-e Modifier None

1
Wall Azimuth

None

program Powcrl:~rvModel: Shaft Airflow Properties Element Name Distance Cross-Sectional Perimeter Roughness Iletwccn Arca Levels default flleakslift2 4111 12 m 9 m2

1:low 1';1tl1 Wind I'ressure Relative Multiplier Positive Flow Limits Wind Pressure Wind Pressure Elevation Direction Option Modifier 0 default calculated by program None None. None

Wall Azimuth Angle None

lI

JOOJ q8no~qi aql JOOU puosas aql uaam~aq yea1 JOOU ps!uaA

I

II
1I

01

P'Z

L 1000'0

P

1

SL

1

Sfooo'o

I

z\le~\lxsl I

Appendix D - Application o CONTAMW f

ATTACHMENT 2 FLOOR LAYOUTS

slairdwrl

,

extslrwall2

6/

-,intwallslr

.

extwalll

i

++ - p'extdwr
.

1

I

i

FIRST FLOOR

flleakslr2

-+
I

-b

edstrwall2

extwalll

p
e~stwaIl1,--.-\.

-- .+st,.
stairdoor2 Elev

'
I

j
Rml

l

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

,
elevdoor openstdoor

.
A. . . .

St2

:

*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 o Smoke Management f

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 L1 -~

1 1 0 d 13 0 ~

-

'..

'.

'

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 description: JanI simulation date : simulation time : 00:00:00 ambient temperature : 20.0 "C barometric pressure : 101325.0 Pa 10.0 mph wind speed : 0.0 deg wind direction : level: < l > elevation: 0.0 m zone Stl Sun D& 03 10:22:54 2000

P 53.5

T 24.0

path flleakstr2 extstnvall2 stairdoor 1 intwallstr extstnvall l openstdoor flleakshft2 inkallshft elevdoor floorleak estwalll estdoor inhvallstr openstdoor inhvallshft esnvall2 esnvall2 elevdoor openstdoor inhvallstr extwall l estdoor flleakstr2 openstdoor intwallstr extstnvall I stairdoor l eststn\~all2

From St 1/<2> Amb t Ambt Rml/<l> Ambt Rml/<l> Elev/<Z> Rml/<l> Rml/<l>

Flow 1 7119.15 -48.70 -268.35 -101.72 -226.93 -6473.46

Flow2

Elev

Rml

Rm 1/<2> Ambt Ambt Stl/<l> Stl/<l> Elev/< l > Ambt Ambt Elev/< l > St2/<1> SW< 1> Ambt Ambt
St2/<2> Rml/<l> Rml/<l> Ambt Ambt Ambt

level: <2> ele\,ation: 4.0 m zone St l
P 7.2

T 24.7

path tlleakstr2 eststnvall2 tlleakstr2 int\\,allstr

from St 1/<3> Ambt Stl/<l> Rm 1/<2>

Flow l 7718.22 -49.28 -7119.15 -1 38.00

Flou.2

312

Principles of Smoke Management

Ambt Rml/G> flleakshfu intwallshft flleakshfu elevdoor Rml -19.6 20.0 floorleak exhvall I inhvallstr stairdoor2 intwallshft exhvall2 exhvall2 elevdoor floorleak stairdoor2 inhvallstr exhvall1 flleakstr2 stairdoor2 inhvallstr extst&alll flleakstr2 extsmvall2 level: <3> elevation: 8.0 m zone Stl
P -38.9

f
I

4..

$

1

I

Rm1/<3> Ambt Stl/G> St 1/<2> Elev/G> Ambt Ambt Elev/G> Rml/<l> St2/<2> St2/<2> Ambt st2/<3> Rm1/<2> Rm 1/<2> Ambt St2/<1> Ambt

T 25.5

path flleakstr2 extstrwall2 flleakstr2 inhvallstr extstrwalll stairdoor2 flleakshft2 inhvallshft flleakshft2 elevdoor floorleak exhvall1 inhvallstr stairdoor2 inhvallshft exhvall2 eshvall2 elevdoor floorleak stairdoor2 inhvallstr extwall l

from St 1 /<4> Ambt St 1/G> Rm 1 /<3> Ambt Rm 1/<3>

Flow I 8335.10 -49.96 -77 18.22 - 145.27 -231.78 - 189.89

Flow2

Elev

-73.1

20.0

Rm 1/<4> Ambt St l / - + St1/<3> Elev/:3> Ambt Ambt Elev/<3> Rm 1/<2> St2/<3> St2/<3> Ambt

Appendix D - Application of CONTAMW

St2

-39.0

25.5

flleakstr2 stairdoor2 intwallstr extstrwalll flleakstr2 extstnvall2

StU<4> Rrn1/-=3> N1/<3> Ambt sue> Ambt

level: <4> elevation: 12.0 m zone St l

P -84.8

path flleakstr2 extstnvall2 flleakstr2 inhvallstr extstnvall 1 stairdoor2 flleakshft2 intwallshft flleakshft2 elevdoor floorleak exhi.all1 intwallstr stairdoor2 intwallshft exhvall2 exhvall2 elevdoor floorlrak stairdoor2 intwallstr extuall I flleakstr2 stairdoor2 intwallstr extstn5,alll flleakstr2 extstn\.all2

from St 1/ G > Ambt St 1/<3> Rm l /<4> Ambt Rrn 1 /<4> Elev/<5> Rm 1/<4> Elev/<3> Rm 1/<4>

Elev

Rm l

Rrn 1 / G > Ambt St 1/<4> St 1/<4> Elev/<4> Arnbt Arnbt Elev/<4> Rm 1/<3> St2/<4> St2/<4> Arnbt
St2/<5> Rm 1/<4> Rn1 1/<4> Arnbt St3/<3> Ambt

level: <5> elevation: 16.0 m zone St l
P
- 1 30.5

T 26.9

path flleakjtr2 cxtstn5-all2 flleakstr2 intwallstr extstnvall l stairdoor2 tlleakshft2 intwallshli Illeakshfi2

from St 1/<6> Ambt St 1 /<4> Rrn 1/<5> Ambt Rrn 1/<5> Elev/<6> Rm 1 /<S> Elev/<4>

Flow 1 9177.54 -51.61 -8945.69 -100.10 -238.22 -141.91 -1249.5 1 34.99 891.96

Flow2

Elcv

Principles of Smoke Management elevdoor floorleak extwalll intwallstr stairdoor2 fan3 intwallshfi extwall2 extwall2 elevdoor floorleak stairdoor2 intwallstr extwall I flleakstr2 stairdoor2 intwallstr extstnvall l flleakstr2 extstnvall2 level: <6> elevation: 20.0 m zone Stl P -176.1

Rm1/<5> Rm l/<6> Ambt St 1 / 4 > st Ambt Elev/<S> Ambt Ambt Elev/<S> Rm l/<& St2/<5> St2/<5> Arnbt St2/<6> Rrn 1 /<S> Rm1/<S> Arnbt St2/<4> Ambt

T 27.6

path flleakstr2 extstnvall2 flleakstr2 intwallstr extstnvall I stairdoor2 flleakshft2 intwallshft flleakshft2 elevdoor tloorleak extwall I intwallstr stairdoor2 fan4 intwallshft extwall2 extwall2 elevdoor tloorleak stairdoor2 intwallstr eshvall l flleakstr2 stairdoor2 inlwallstr

from St l/<7> An1bt St 1 /<S> Rm 1/<6> Ambt Rm 1/<6> Elev/<7> Rrn 1/<6> Elev/<S> Rni l /<G>
Rm 1 /<7>

Flow l 10253.70 -52.59 -9477.54 -220.1 1 -242.02 -26 1.43

Flow2

Elev

-2 14.7

20.0

Rml

-231.8

20.0

Ambt St 1/<6> St 1/<6> Arnbt Elev/<6> Ambt Ambt Elev/<6> Rm I /<5> St2/<6> St2/<6> An1bt St2!<7> Rni l /<G> Rni I /<G>

3 l5

Appendix D- Application i f CONTAMW

extstrwalll flleakstr2 extstrwall;? level: zone Stl

Ambt St2/<5> Ambt

a>elevation: 24.0 m
P -221.5

T
28.4

path flleakstr2 extstrwall2 flleakstr2 intwallstr extstrwalll stairdoor2 flleakshfu intwallshfl flleakshfu elevdoor floorleak extwall l intwallstr stairdoor2 fan3 intwallshft extwall2 extwall2 elevdoor floorleak stairdoor2 intwallstr extwalll flleakstr2 stairdoor2 intwallstr extstrwalll flleakstr2 extstrwall2

from St l/<8> Ambt St l/<6> Rm Ambt Rm l/<7> Elev/<8> Rm 1/ < P Elev/<6> R n1/<7> r

Elev

Rm l/<8> Arnbt St1/<7> St1/<7> Ambt Elevl<7> h b t Ambt Elev/<7> Rm l/<6> StU<7> stU<7> Ambt
StU<8> Rm 1/<l'> Rm1/<7> Ambt StU<6> Ambt

level: <8> elevation: 28.0 m zone St l

P -266.7

T 29.1

path flleakstr2 extstrwall2 flleakstr2 intwallstr extstrwall I stairdood flleakshft2 intwallsh ft flleakshft2 elevdoor

from St 1/<9> Ambt St1/<7> Rm 1/<8> Ambt Rm l/<8> Elev/<9> Rm 1/<8> Elev/<7> Rm 1/<8>

Flow I 11573.08 -54.85 -10851.84 -186.12 -256.91 -229.35

Flow2

Elev

316

Principles of Smoke Management

Rml

-309.6

20.0

floorleak exhvall 1 inhvallstr stairdoor2 intwallshft extwall2 ext;;.all2 elevdoor floorleak stairdoor2 inhvallstr exhvall I flleakstr2 stairdoor2 inhvallstr extstrwalll flleakstr2 extstrwall2

Rm1/<g> Ambt St1/<8> St 1/<8> Elev/<8> Ambt h b t Elev/<8> RmlM> St2/<8> St2k8> Ambt St2/<9> Rni 1/<8> Rm 1/43> Ambt St2/<7> Ambt

level: <9> elevation: 32.0 m zone St l

P -31 1.6

T 29.8

path flleakstr2 extst1kall2 fl leakstr2 intwallstr extstnvall I stairdoor2 flleakshft2 inhi-allshft flleakshft2 elevdoor floorleak exhvalll inhvallstr stairdoor2 inh\.allshft eshvall2 exhvall2 elevdoor floorleak stairdoor2 inhvallstr eshvall I flleakstr2 stairdoor2 inh\-allstr eststnval l l flleakstr2 eststnvall2

from Stl/<lO> hlbt St 1 !<8> Rnl1/<9> Arnbt Rm 1 /<9>

Flow1 12356.22 -56.14 -11573.08 -2 14.84 -255.97 -256.19

Elev

-356.3

20.0

Rrn l

-365.4

20.0

St?.'<lO> Rm 1 /<9> Rrn 1 /<9> Anbt St7/<S> Arnbt

Appendix D - Application of C ~ A M W

level: <10> elevation: 36.0 m zone Stl path flleakstr2 extstrwall2 flleakstr2 intwallstr extstnvall I stairdoor2 flleakshft2 intwallshft flleakshft2 elevdoor floorleak extwa l l l intw?!lstr stairdoor2 intwallshft extwall2 extwall2 elevdoor floorleak stairdoor2 intwallstr exhvall l flleakstr2 stairdoor2 inhvaktr extst~wall l flleakstr2 extstnvall2 fiom Stl/<l l> Ambt St 1/<9> Rml/<lO> Ambt Rm1/<10> Elev/<lI> Rml/<lO> Elev/<9> Rm1/<10> Rml/<l l> Ambt Stl/<lO> Stl/<lO> Elev/<l O > Ambt Ambt Elev/< l O > Rm 1 /<9> St2/< 10> St2/<10> Ambt St2/< I I > Rtnl/<lO> R111 /< 1 0> 1 Ambt St2/<9> Ambt Flowl 13 179.90 -57.54 - 12356.22 -232.47 -261.49 -272.18 Flow2

Elev

Rm l

St2

level: <l l> elevation: 40.0 111 zone St l path flleakstr2 extstnvall2 fl leakstr2 intwallstr extstnvall l stairdoor2 flleakshft2 intwallr;hli tlleakshfi2 elevdoor floorleak extwall l intwallstr stairdool" intwallslili from St1/<12> Ambt Stl/<lO> Rm1/<1 l> Ambt R1111/<1l> Elev/< 12> R~iil/<I I> Elev/< l O > R~nl/<l l> Rm1/<12> Ambt Stl/<l l> Stl/<l l> Elev/< l l > Flow I 14035.98 -59.05 - 13179.90 -245.65 -267.47 -283.9 1 Flow2

Elev

R111I

31s

Principles of Smoke Management

ex,twall2 extwall2 elevdoor floorleak stairdoor2 intwallstr _extwalll flleakstr2 stairdoor2 intwallstr extstrwalll flleakstr2 extstrwall2 level: < l 2> elevation: 44.0 m zone St l P -445.1

Ambt Ambt Elev/<l l > Rml/<lO> St2K 1 l> St2/<1l> Arnbt SW< 12> Rml/<ll> Rml/<l l> Ambt St2k1o> Ambt

T 32.0

path flleakstrl fan l extstrwall2 flleakstr2 intwallstr extstrwall 1 stairdoor2 flleakshft l intwallshft flleakshft2 elevdoor floorleak extwall1 intwallstr stairdoor2 intwallshft extwall2 extwall2 elevdoor floorleak stairdoor2 intwallstr extwall l fan2 flleakstr l btairdoor2 intwallstr extstrwall l flleakstr2 eststrwall2

f?om Ambt Ambt Ambt Stl/<ll> Rm1/<12> Ambt Rm1/<12> Ambt Rm1/<12> Elev/< l l > Rm1/<12> Ambt Ambt St1/<12> St1/<12> Elev/<l2> Ambt Ambt Elev/< l 2> Rml/<l l> St2k 12> St2k12> Ambt Ambt Ambt Rm1/<12> Rm1/<12> Arnbt St2/<1 l> Ambt

dP -76.0 -76.0 -70.4 -1.2 -72.5 -79.6 -72.0

Flow lFlow2 -77.4 1 15000.00 -60.68 -14035.98 -257.60 -273.92 -294.4 1

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
+----------v----------------

[pal

[scfm]

zone

+
> 257.60 ------ l 67.2 > 245.65 ------ I 61.7 > 232.47

< 1 2>Rm I

72.5 < 257.60

Stl Stl Stl

72.5

Rml Rml Rml Rml Rml

I -----<l I>/Rml

< I O>/Rm I

67.2 < 245.65 1 -----61.7 < 232.47

- - p - - -

I

Stl Stl
I -------

24.3
------ I

> 127.32
>

Rml Rml Rml

56.3

220.11

16.7 < 100.10

Stl
- p - - - -

I

Stl St l Stl St l

28.0 > 140.33 ------ I 29.5 > 145.27
-p----

Rml Rml Rml

l

27.2
-p----

> 138.00
l

17.0

> 101.72

Rm l

Appendix E

ASMET Documentation

1

l

NOMENCLATURE
A
a
= = = =

TCP
t,

=

cross-sectional area of the atrium, m2 fire growth coefficient, kw/s2 0.071 0.026
= = =

absolute centerline plume temperature at elevation z, K growth time, s average plume temperature at elevation z, 'C volumetric smoke flow at elevation z, m3/s height above top of fuel, m mean flame height, ni virtual origin of the plume, m convective fraction of heat release density of air or plume gases, kg/m3 density of ambient air, kg/m3 density of plume gases at elevation z, kg/m3

Cl

Tp
V

C2

z
Z J

=
= = = = = =

CS C7
C8

= = = =

9.1 0.235 0.0018 0.166 1.11

I ,

j

p p, pp

C9
Clo

=

Note: The variables above are given in S1 units only, because internal calculations in ASMET are in SI.

Cp

=

specific heat of plume gases, 1.005 kJ/kg-K diameter of fire, m

PART 1: ASMET DESCRIPTION
Below are the equations used in each section of ASMET, except for ASET-C, which is discussed in Appendix F.
Steady Filling Equation (Solve for z)

DJ
g

= =

H
n

acceleration of gravity, 9.807 rn/s2 = ceiling height above the fire, m mass flow in plume at height z, kg/s absolute pressure, Pa heat release rate of the fire, kW convective heat release rate of fire, kW gas constant, 287 J kg.K time, S absolute temperature, K ambient temperature. O C

= = = = = = =
=

P Q
QC

R
t

Steady Filling Equation (Solve for t)

T Ta

-Appendix E- ASMET Documentation

Unsteady Filling Equation (Solve for z)

The density of air and plume gases:

Unsteady Filling Equation (Solve for t)

Plume Centerline Temperature

Plume cecterline temperature:

Simple Plume Equations

Mass flow of plume:
iil

= ClQC z

1/3 5/3

+C8Qc

The virtual origin of the plume and the mean flame height by the equations of the previous section, "Plume with Virtual Origin Correction." Convective portion of the heat release rate:

Mean flame height:
:l-

- - C Q2/5
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:

The density of air and plume gases:

PART 2: ASMET USERS GUIDE ASMET is a collection of tools that can be used for analysis of atria smoke management systems. This program is for a personal computer with a DOS operating system, and the program was \vritten in C. When ASMET is in the active directory, the program is activated 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.
The equations used for each routine are listed in Appendix C, except for ASET-C, which is described in Appendix E. Theequations of Appendix C are also addressed in the body of the text. The first time the program is run, it starts in S1 units, and the user can change units by pressing E for English units or I for S[ units. The program stores a unit indicator in file UNITS so that it M-ill start up with the unit selection from the last time the program was run. The other menu items are selected by pressing the key that is in bold type (or yellow on a color monitor). The first menu item is selected'by pressing S, and the screen for this menu is shown in Table E-2. There are two ways to enter data from this menu. The first is by pressing the key that is in bold for that menu item. The second is by moving the indicator at the right of the menu item with the up and down arrows. This indicator is next to the first menu item (ceilinz height above fire) in Table E-2. Once an item has been selected, the number 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 printer by pressing P. and pressing D returns the user to the main menu. To send results to a file, press f and enter the lile name. Use of the other items in the main menu is similar to that discussed above.

Plume with Virtual Origin Correction

Mass flow of plume:
til

= C l QC (Z- z ~ ) ' / ~ [ c ~ Q ~ / ' ( z - z0 ,-5/3] 1+

I /;

(E10)

This equation can be rearranged to simplify calculation:
tii

= c l Q C (I-z0)

I ;

5/3

+C8Qc

(El 1)

Virtual origin of the plume:
z0 = C;@
715

- 1.02 Of

(E121

Mean flame height:
zf =

c,Q'/' 1.020, -

(E131

Average plume temperature:
Q C TI' = T , + 111 c,

(E14)

The volumetric flow of a plume:
f .

=

c4 ! !
p,

(E15)

Principles o Smoke Management f

-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 cross-sectional area of atrium heat release rate of tire time Print results (to LPTI) Print results to file disabled

H (m):
A (mA2):

Q (kW):
t (S):

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 cross-sectional area of atrium he31 release rate of fire time
X

H (m):

80.00

A (mA2): 20000.00 Q (kW): 10000.00 t (S): 1200.00 +

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 cross-sectional area of atrium heat release rate of fire time H (m):
A (m2):

Q (kW): t (S):

30.00 5000.00 5000.00 300.00

Height of smoke layer above fire, z is 17.4 m or 57.2 f t

------------------------------------------------------------------------Steady Filling Equation (Solve fort) Atrium filling time for steady fire ceiling height above fire cross-sectional area of atrium heat release rate of fire height of smoke layer above fire H (m): 40.00 A ( I ) : 10000.00 Q (kW): 5000.00 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 cross-sectional area of atrium fire growth constant (Menu) time At 800 seconds, the tire is H (m): A( I ) : a (kw/sZ): t (S): 300 l0 kW or 30.00 8000.00 0.04659 800.00 28445 Btds.

Height ofsmoke laycr above tire, z, is 10.7 m or 35.0 ti
-------------------------------------------------------------------------

Unsteady Filli~lg Equation (Solve fort) Atrium tilling time for unsteady tire ceiling height above fire cross-sectional arca ofatrium fire growth constant (Menu) height of smoke layer ab0i.e tire 50.00 12000.00 a ( k ~ l s ' ) : 0.04659 z ( m ) 10.00
A (m Z):

H (m):

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 Heat release rate ol'lirc Ambicnt tcmpcl-atcw
r (m):
Q (k\V): Ta (C):

50.00

25000.00
7 1 .00

Principles of Smoke Management

At elevation z, the plume has: Mass flow of 1282.4 kg/s Volumetric flow of 1117.2 m3/s Average temperature of 35°C Mean flame height of 8.3 m

2827.2 Ib/s 2367016 c h 94°F 27.1 ft
-

.............................................................
Plume with V i a l Origin Correction Mass flow rate and average plume temperature
z (m): Elevation Heat release rate of fire Q (kW): fire diameter Df (m): Ambient temperature Ta ("C):

I

50.00 25000.00 4.00 21.00 2766.1 Ib/s 23 1 8 122 cfm 95°F 2.3 fi 30.9 fi

At elevation z, the plume has: Mass flow of 1254.7 kg/s Volumetric flow of 1094.2 mA3/s Average temperature of 35°C Virtual origin at 0.7 m Mean flame height of 9.4 m

Plume Centerline Temperature Calculate centerline plume temperature Elevation Heat release rate of fire fire diameter Convective fraction of heat release Ambient temperature At elevation z, the plume has: Centerline temperature 46°C Virtual origin at 0.7 m Mean flame height of 9.4 m 50.00 z (m): Q (kW): 25000.00 Df (m): 4.00 (0.6 to 1): 0.70 21.00 Ta ("C): 115°F 2.3 fi 30.9 fi
. .

EXAMPLE OUTPUT (ENGLISH UNITS)

Height of smoke layer during atrium filling From a steady fire ceiling height above fire cross-sectional area of atrium heat release rate of fire time

H (R): 98.40 53800.00 A (g): Q (Btu%): 4740.00 200.00 t (S):

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 cross-sectional area of atrium heat release rate of fire height of smoke layer above tire
H (R): 131.00 lO7OOO.OO A (g): Q (Btuls): 4740.00 z (i: f) 26.20

Appendix E- ASMET Documentation

Unsteady Filling Equation (Solve for z) Atrium filling time for unsteady fire ceiling height above fire cross-sectional area of amum fire growth constant (Menu) time H (R): 98.40

86100.00 A (g): a ( ~ t u l s ~ ) : 0.04444 800.00 t (S):

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 cross-sectional area of atrium fire growth constant (Menu) height of smoke layer above fire H (ft): A (P): a (~tuls'): z (ft) 164.00 129000.00 0.0444 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 Heat release rate of fire Ambient temperature

z (ft): Q (Btds): Ta (F):

164.00 23700.00 70.00

At elevation z, the plume has: 12s1.9 kg/s Mass flow of m3/s

2826.2 Ib/s

Volumetric flow of I 117.2 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 Heat release rate of fire fire diameter Ambient temperature
z (ft): Q (Btds): D f ( ft): Ta (F):

164.00 23700.00 13.10 70.00

At elevation z, the plume has: Mass flow of 1253.9 kg/s Volu~netric flow of Average temperature of Virtual origin at Mean flame height of 1093.9 m3/s 3j3C 0.7 m 9.4 m

7764.4 Ib/s 2317599 cfm 95°F 2.3 ft 30.9 fi

Principles of Smoke Management

Plume Centerline Temperature Calculate centerline plume temperature Elevation Heat release rate of fire Fire diameter Convective fraction of heat release Ambient temperature
At elevation z, the plume has: 46OC Centerline temperature Virtual origin at 0.7 m Mean flame height of 9.4 m
z (R): Q (Btuls): Df (R): (0.6 to l): Ta (F):

.

164.00 23700.00 13.10 0.70 70.00

Appendix F ASET-C: A Room Fire Program for Personal Computers
INTRODUCTION
Cooper (1981) of the Center for Fire Research, National Bureau of Standards, introduced ASET, a mathematical model for estimating available safe egress time in fires. Cooper and Stroup (1982) published a computer program to perform the calculations in the mathematical model; thus, the computer program also became known as ASET. ASET was not specifically written for the personal computer environment because at the time it was being developed, personal computers were just emerging as a tool for use in the engineering office. Since the introduction of ASET, the use of personal computers has become widespread and there has been significant interest in running ASET on personal computers. In response to this interest, Walton (1985) introduced ASET-B, a program for personal computers based on the original ASET mathematics1 model. The B was used to indicate basic, brief, BASIC, and beta. ASET is a 1500-line FORTRAN program that has many features. ASET-B is a 100-line BASIC program that was developed to be as simple and fast as possible. The most significant change in ASET-B is the use of a different mathematical procedure to solve the primary equations. ASET-B employs an equation solver that is at least five times faster than that used in ASET, while retaining mathematical agreement to within a fraction of a percent. ASET-B is an interactive program requiring a minimunl of input. These features make ASET-B easy to learn and apply. In many con\ ersations with practicing fire protection engineers, the author has found that ASET-B has become very popular. This appendix describes the ASET-C routine, which t is p a ~ ofthe ASMET program. ASET-C is a C language version of ASET-B with improved interactive input and a few added features. The interactive input was made to be consistent with the other ASMET routines. The added features consist of allowing fire data input from a file and the use of a t-squared fire. Most of the material in this appendix is adapted from Walton's (1985) paper on ASET-B and, in many places, the adaptation consisted only of changing ASET-B to ASET-C.

DESCRIPTION OF THE MODEL
The mathematical model that is the basis for ASET, ASET-B, and ASET-C has been presented in detail by Cooper (1981, 1982) and will be only summarized here. It is based on a single room or enclosure with all doors, windows, or vents closed except for a small leak at floor level. This leak prevents the pressure from increasing in the room. A fire starts at some point below the ceiling and releases cnergy and produc:~ of combustion. The rate at which energy and products of combustion are released may change with time. The hot products of combustion form a plume, which, due to buoyancy, rises toward the ceiling. As the plume rises, it draws in cool air from the room, which decreases the plume's temperature and increases its volume flow rate. When the plume reaches the ceiling, it spreads out and forms a hot gas layer, which descends with time as the plume's gases continue to flow into it. There is a relatively sharp interface between the hot upper layer and the air in the lower. part of the room, which, in this model, is considered to be at ambient, temperature. The only interchange between the air in the lower part of the room and the hot upper layer is through the plume. ASET could therefore be described as a two-layer or zone model. The basic 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

Air at Approximately ---c Ambient Temperature

Leak at Floor Level

l

"--

I

Figure F1 Schematic offire phenonzena.

The two unknowns in ASET-C are the height of the hot layer interface above the fire, Z, and the average temperature of the upper layer, P. It should be noted that the notation used here to describe the model is consistent with the variable names u;ed in the computer program. The unknowns, Z and P, are often referred to as the (dimensionless) height and temperature of the smoke layer since, consistent with the model formulation, smoke can only be found i n the plume and the hot 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 gravity, G. In addition, the ambient temperature, PA, density, DA, and specific heat, CP, of air must be kno\vn. The final known quantities are the rate at which heat is released by the fire as a function of time, QT, the fraction of the total heat release, which is given off as radiation, LR, and the fraction of total heat release rate, which is lost to the contents and surrounding surfaces of the room, LC. The unknown height and temperature are determined by using conservation of mass and energy in conjunction 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 differential equations. In developing the original equations for ASET (Cooper 198l , 1982), two dimensionless groups of problem parameters, C1 and C2, were introduced. Also introduced were dimensionless forms of the variables: time, height, and temperature of the smoke layer, initial height of the smoke layer, height of the base of the fire, and the rate of heat release. These variables are made dimensionless by dividing them bv a characteristic quantity with the same dimensions or units. Thus, the dimensionless temperature, P, is the actual temperature of the smoke layer. PF (converted to R), divided by the ambient temperature, PA (R). Similarly, rate the din~ensionless of heat release, QT, is the actual rate of heat release, QA (kW), divided by the initial rate of heat release, Q0 (kW). Finally, the dimensionless variables, height of the smoke layer, Z, initial height of

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 as one foot. Thus. the dimensionless lengths Z, ZO, and 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 dimensionless time, T, is therefore numerically equal to the actual time in seconds. Since engineering units are used in ASET, this convention has been continued here for consistency. Conversion to S1 units is provided in the computer pro,oram. The d~fferentialequations for the dimensionless height of the layer above the fire, 2,and average temperature of the layer, P, are given below.
-Cl . Q T - C 2 . Q T ' / ' 2 5 1 3

1

for 0 < Z < zo

0

for Z = -F for o < Z < zo

P [ C I . P T - ( P - 1 ) c 2- Q T 1 / 3 t 5 / 3 1 / ( z 0 + Z )

2 I /3 C2 = (0.21 . C T / A ) [ ( I - L R ) . QO. G.CL / ( D A - C P - P A ) ]

In order to solve the equations for Z and P, the initial conditions must be known. One set of initial conditions, which were derived in Cooper (1981, 1982) and will be used here, assume that the fire starts with a small heat release rate, Q0, at time T = 0. Under such conditions, the initial conditions are

Although dPldT is indeterminate in the above equation at T = 0, its actual value has been found in Cooper (1981, 1981) to be

n_P - ~ 2DQO+ (Cl + C2 - ZO"') . d T - C2 6.zo~/~
where DQO = dQT/dTat time T = 0. SOLUTION OF THE EQUATIONS In general, the differential equations for-Z and P cannot be solved explicitly; that is, an algebraic expression cannot be written that describes Zand P at any time 7. As a result, the equations must be solved numerically. ASET sol\-esthe difl'erential equations using a variation of the fourth-order Runge-Kutta method with variable time step. While this mcthod has a high degree of accu-

Principles of Smoke Management
3

j

1
2

racy, it has been determined that the improved Euler's method has sufficient accuracy for this problem. The improved Euler's method is a simple predictor-corrector type and is described in most books on numerical methods (Carhanan, Luther, and Wilkes 1969). The improved Euler's method used in ASET-C requires substantially fewer calculations than the method used in ASET, resulting in ASET-C running much faster than ASET. The improved Euler's method as applied in ASET-C is basically a technique for stepping the solution forward in time. Given the values of Zand P at a particular time, T, the method is used to determine the values of Z and P at time T + DT, where DT is a small time increment. This process is started at time T = 0 and continued until Z and P are known at all times of interest. In the case of ASET-C, an increment of one second has been found to yield results that agree well with ASET for problems of practical interest. In ASET-C, ZI, and PI are used to indicate the values of Z and P at time i7 For the first step, these are the initial values at time T = 0. 22 and P2 are used to indicate the values of Z and P to be calculated at time T + Di7 To determine 22 and P2 it is observed that the differential equations for Z and P represent the time rate of change of these quantities. The time rate of change multiplied by the time step yields the change that occurs over the time step. This would be an exact result if the equations were linear or the time steps were infinitely small. Since the equations are nonlinear, and it is impractical to make the time step infinitely small, an approximation must be used. In the improved Euler's method, 22 and P2 are first predicted using the derivatives evaluated at time I: Using 22 and P2, the derivatives are then evaluated at time T + DT. Corrected values of Z2 and P2 are then calculated using the average of the derivatives evaluated at times T and T + DT. 22 and P2 are predicted by
2 2 = 2 + DZI - DT , 1 P2 = PI + DRI - DT

time step. If the difference is greater than 0.001, the predicted values become the corrected values and the derivatives at time T + DT are recalculated. New corrected values are then calculated. In ASET-C, this procedure is repeated for a maximum of thirty times. If the differences are still greater than 0.001, a warning is printed, and the program proceeds to the next time step. The evaluation of the derivatives of Z and P requires the dimensionless heat release rate, QT, be known for all times, I: For heat release rates that are not constant with time, ASET-C requires the heat release be specified for each one-second time interval. To simplify this procedure, ASET-C uses point specified heat release rates with linear interpolation. Heat release rates can be specified at as many as 100 different times. Linear interpolation is then performed to determine the heat release rate at each time step. R U N N I N G THE P R O G R A M General Instructions ASET-C is written as an interactive program; that is, the program prompts the user with questions. As previously stated, ASET-C is part of the ASMET package of routines for atrium analysis, and a description of this package is provided in Appendix E. The mechanics of input for ASET-C, are consistent with the other routine in this package. To use ASET-C, data niust be entered for the items discussed below. Program Inputs Heat Loss Fraction. The first input is tlie heat loss fraction. This quantity is the instantaneous faction of the heat release rate of the fire that is lost to the bounding surfaces of the room and its contents. Cooper (1 98 1. 1982) has provided guidelines for selecting this parameter, which is called Lambda C () or ALMAC in ASET. ? , He has detirmined that the approximate range is 0.G 0.9. The lower value corresponds to high aspect ratio spaces (ratio of ceiling span to room height) with smooth ceilings and fires positioned far away from the walls. The intermediate to high values corresponds to low aspect ratio spaces, rooms with irregular surfaces. or rooms in which the fire is within one ceiling height of the wall. The temperature of the upper layer is a function of the heat loss fraction and the heat release rate of the fire. The greater the heat loss fraction, the lower the temperature in the xpper layer. The heat loss fraction for a room mith insulated walls will be lower than the fraction for the same.room with uninsulated walls. Both ASET and ASET-C treat tlie heat loss parameter as a constant. That is, the heat lost from the room is a constant fraction of tlie heat release rate of the fire. As the heat release rate of the tire changes. tlie quantity of

.

where DZ1 = dZldT and DP1 = dPldT are evaluated using Z = ZI and P = PI. The derivatives at time T + DT, D22 = dZlfl: and D M = dPld7; are then evaluated using Z = 22 and P = P2. Corrected values for Z2 and F2 are calculated using the average derivatives
Z2C = Z l + [ ( D Z I + D Z 2 ) / 2 ] . D T , P2C = P1 + [ ( D P I + D P 2 ) / 2 ] . DT .

The predicted values of Z and P are then compared to the corrected values. In ASET-C, if the absolute value of the difference between the predicted and corrected values is less than 0.001, the solution is considered to have converged and the program proceeds to the next

-Appendix F-ASET-C: A Room F i e Program for Personal Computers

heat lost will also change, but in direct proportion to the fire. Therefore, the room will not cool down even though the heat release rate of the fire goes to zero. Height of the Base of the Fire. The second input is the height of the base of the fire above the floor in feet. For fuel items of relatively uniform surface height, such as beds, this is simply the height of the surface. For three dimensional h e 1 items, such as sofas, an average height weighted to reflect the distribution of surfaces should be used. The rate of growth of the upper layer is strongly dependent on the difference between the height of the base of the fire and the height of the smoke layer interface. Room Ceiling Height and Floor Area. The third and fourth inputs are the room ceiling height in feet and the floor area in square feet. According to Cooper (1981, 1982), the calculations may not be valid when applied to room length-to-width aspect ratios greater than 10: 1 or with a ratio of height to minimum horizontal dimension exceeding one. The equations are based on the assumption that the upper layer is well mixed and at a uniform temperature. Therefore, the results for a square room and a rectangular room of equal height and area will be the same. Output Interval. The fifth input is the output interval. This is the time step for results that are sent to the screen or printed. The output interval of ASET-B was set at five seconds, and this is the default interval for ASET-C. Maximum Time. The sixth input is the niaximuni time for the simulation in seconds. The results of the calculations will be printed at five-second intervals until the maximum time or until the end of the heat release data. Fire Growth Constant. The seventh input is the description of heat release rate of the fire. A fire gro\vth constant can be entered to define a t-squared fire, or the Menu can be activated that allows selection of a fire growth constant for typical fires (slow, medium, fast, or ultra-fast). From the menu, the user also can choose to enter data as sets of points, as was done with ASET-B. When the user selects data points, the computer waits for the run command to request the data. However, the following is a discussion of input by data points. As described earlier, the program can accommodate up to 100 pairs of times and comesponding heat release rates. The program performs a linear interpolation between the specified points to determine the heat release rates at the required times during the calculations. The data are entered by typing the time in seconds, follo\ved by a comma, followed by the heat release rate i n kilowatts. A return or enter is then typed to proceed to the nest linc.

Heat release rates entered as less than 0.1 kilowatt will be converted to that value. The program will automatically assume a starting value 0.1 kilowatt at time zero. A heat release rate at time zero does not have to be entered unless a greater initial heat release rate is required. When all of the desired times and heat release rates have been entered, a -9,-9 followed by a return is entered to terminate the data entry and begin the calculations. Actually, any negative time followed by a heat release rate will result in the same action. Optional Upper Limit on Fire. Fire growth may be approximated by the t-squared curve for some time. Because of the action of a suppression system, limitations of fuel, or limitations of combustion air, t-squared fire growth eventually must stop. The optional upper limit on fire growth allows the user to specify a heat release rate at which the fire curve reaches steady burnmg. Send Results to Printer or to File. To sent results to the printer, press P. To send results to a file, press t and enter the file name. Run Simulation. To run ASET-C, press R. If heat release rate by point entry has been selected from the Menu, the data points will be requested after the run starts. Program Outputs. The output of the ASET-B program is a summary of the input data and a table of the conditions in the room as a function of time. The first colunln in the table is the simulation time in seconds. The second and third columns are the temperature in the upper layer in degrees Celsius and Fahrenheit. The fourth and fifth columns are the height above the floor of the interface between the upper and lower layers. The sixth and seventh columns are the heat release rate of the fire in kilowatts and Btu per second. The output has the same number of significant digits as does ASET-B, which allows users to verify that this program produces the same results as ASET-B for the same input. LIMITATIONS OF ASET The use of ASET-C or any design aid requires the design engineer to make the final evaluation as to the appropriateness of the design. The ASET-C programs are based on certain engineering approximations of the fire environment and should be used to supplement rather than replace sound engineering judgment. The program results should be treated as approximate and the user is encourayed to become familiar with how changes in the input variables affect the program results. The temperature of the upper layer and the height of the interface respond differently to changes in the input data. Appropriate factors of safety should be applied to either the input data or ths program results.

1
3 ;j
' I

Principles of Smoke Management

.

,.

Some of the limitations of the program have been presented in conjunction with the input data requirements. There are, however, some additional limitations. The mathematical procedure used in ASET-C is very harrly; that is, the procedure will normally converge and produce results. There are combinations of input data for which the program will either fail to converge or halt due to an illegal mathematical operation. If the procedure for solving the equations fails to converge, a warning will be printed and the solution will continue. The results following this message may be in error and should be treated as such. The failure to converge is usually a result of a heat release value that changes too rapidly. In most cases, this problem can be corrected by minor smoothing of the input heat release curve.

VERIFICATION O F ASET Results of the ASET program have been compared to data from a limited number of actual fire experiments (Cooper 1981, 1982). These comparisons can be extended to the ASET-B and ASET-C programs since they produce results that are within a few percent of those produced by ASET. The fire experiments considered a mockup of a hospital room-corridor building space. Comparisons were found to be generally favorable. This does not necessarily mean that the comparison will be favorable ia all cases. Clearly, additional studies are required in this area and that work is ongoing.

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

SAMPLE RUN (ENGLISH UNITS)
HEAT LOSS FRACTION = FIRE HEIGHT ROOM HEIGHT = ROOM AREA = 0.80 1.OOft 9.00ft 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
sec 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 70.0 75.0 80.0 85.0 90.0 95.0 100.0 105.0 110.0 1 15.0 120.0 125.0 130.0 135.0 140.0 145.0 150.0 155.0 160.0 165.0 170.0 175.0 180.0

TEMP C
21.3 23.4 24.9 26.3 27.7 29.2 30.8 32.6 34.4 36.4 38.6 40.9 43.3 46.0 48.8 51.8 55.0 55.3 61.9 65.8 69.8 74.2 79.0 84.3 90.0 96.2 102.9 110.1 1 17.7 125.9 134.6 143.7 153.3 163.3 173.7 184.5 195.9

TEMP F
70.3 74.2 76.7 79.3 81.8 84.6 87.5 90.6 93.9. 97.5 101.4 105.6 110.0 114.7 119.8 125.2 130.9 137.0 143.5 150.4 157.6 165.5 174.2 183.7 194.0 205.1 217.2 230.1 243.9 258.7 274.3 290.7 307.9 325.9 344.6 364.2 384.7

LAYER

LAY ER
ft 9.0 8.7 8.3 7.8 7.3 6.9 6.5 6.0 5.7 5.3 5.0 4.7
4.4

FIRE

FIRE
Btuls 0.1 9.6 19.0 28.5 37.9 47.4 56.9 66.4 75.9 85.4 94.8 104.3 113.8 123.3 132.8 142.3 151.8 161.2 170.7 180.2 189.7 207.5 225.3 243.1 260.5 278.6 296.4 314.2 332.0 349.8 367.5 385.3 403.1 420.9 438.7 456.5 474.2

kW
0.1 10.1 20.1 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 1 10.0 120.0 130.0 140.0 150.0 160.0 170.0 180.0 190.0 200.0 218.8 237.5 256.2 275.0 293.8 3 12.5 33 1.2 350.0 368.8 387.5 406.2 425.0 443.8 462.5 48 l 2 500.0

1. I 3.9 j.6 3.4 3.2 3.0 2.8 2.6 2.5 2.3 2.1 2.0 I .S 1.7 I .5 1.3 1.2 1 .o 0.8 0.6 0.4 0.2 0.0 0.0

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 sec
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 70.0 75.0 80.0 85.0 90.0 95.0 100.0 105.0 110.0 115.0 120.0 125.0 130.0 135.0 140.0 145.0 150.0 155.0 160.0 165.0 170.0 175.0 180.0

TEMP C
21.2 21.5 21.9 22.6 23.3 24.3 25.4 26.8 28.4 30.3 32.5 35.1 38.1 4 1.5 45.3 49.8 54.8 60.5 66.9 74.1 82.1 9 1.2 101.3 112.5 124.8 138.5 153.8 171.0 190.3 211.9 236.2 263.5 293.3 329.0 368.1 412.5 362.7

TEMP F
70.2 70.6 71.5 72.6 74.0 75.7 77.8 80.2 83.1 86.6 90.5 95.2 100.5 106.6 113.6 121.6 130.6 140.8 152.4 165.3 179.9 196.1 214.3 234.4 256.7 281.3 308.8 339.8 373.5 413.4 457.1 506.3 56 1.7 613.1 694.6 774.4 864.9

LAYER m
3.0 2.9 2.8 2.7 2.5 2.3 2.2 2.0 1.8 1.6 1.5 1.3 1.2 1.1

FIRE kW
0. I 1.2

FIRE Btds
0.1 1.1 4.4 10.0 17.8 27.8 40.0 54.5 71.2 90.1 111.2 134.5 160.1 187.9 217.9 250.2 284.6 321.3 360.2 401.4 444.8 490.3 538.1 588.2 640.4 694.9 751.6 810.6 871.7 935.1 1000.7 1068.5 1138.6 1210.8 1285.3 1362.1 1441.0

I .o
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0. I 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

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 discussion of the results of this example, see Example 10.4 in Chapter 10.

(a) Ground F!oor

(b) Typical Floor

Symbols Single-door Leakage area of dosed single 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 Leakage area of mstruction cracks and gaps in slairwe!l walls SW-wall 81dg-floor Leakage area of construction cracks and gaps in budding floor of the El vent Vent to the outside at the IOD elevator shafl Effectwe area to a m u n t fo; fncl~onlosses m slamell Sj[r_floor Etfecl ve area to a m u n t for fnct~on losses in e1eva:or snaft €1-floor SWI Stairwell 1 SW2 Stawell2 FL Open plan space on the flool EL Elevafor shafl

Notes. l . Values of flow areasare Iksted in Table E2. 2. This door is either opened or closed (Single-door or Open-door)

Figure G 1 CONTAM t-e~~~-~setitatiot~~fot- 10.4: (a) gro~o~d~floot. (1,) ~picalfloot: Exatnple atid

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 Number of stories Height between stories Outside winter design temperature Outside summer design temperature Building design temperature Minimum design pressure difference Maximum design pressure difference 4 15 12.0 ft (3.66 m) 14OF(-1 O°C) 93OF (34OC) 73OF (23OC) 0.05 in. H20 (12.4 Pa) 0.30 in. H20 (87 Pa)

Table 6-2: Flow
CONTAM Path Name Doors: Single - Closed Double -Closed Single - Opened Elevator - Closed Walls (per floor): Exterior Elevator Stairwell to Building Stairwell to Outside Building Floor Elevator Shaft Vent Effective Areas of shafts2 Stairwell Elevator
I. 2.

re as' for Example 10.4
f?

Tight Building mz

Loose Building ft2 m2

Single-door Double-door Open-door Elev-door Ext-wall Elev-wall SW-wall SW-wall Bldg-floor Elev vent SW-floor Elev-floor 41 1290 3.8 120 41 1290 3.8 120

A flow cocfficient, C, of 0.65 is used for all flow areas cscepr for open stairwell doors for which C = 0.35.

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 Run
File

Stair Supply ~ i r ' Stair Doors open2
cfm
rn3/s

Season Summer Summer Winter Winter Summer Winter

Leakage Loose Loose Loose Loose Tight Tight

1 2 3 4 5 6
1.

EX-10-4D EX-10-4A EX-10-4C EX-10-4B EX-10-4E EX-10-4F

G, 2 , 3 , 4 , 5 G G, 12, i3, 14, 15 G G, 2 , 3 , 4 , 5 G

20,500 20,500 20,500 20,500 13,900 13,900

9.67 9.67 9.67 9.67 6.56 6.56

The flow orsupply pressurization air was obtained by running thc computer program several times for runs I and 6 toobrain pressure differences 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 in. H 2 0 Pa Run 3 in. H 2 0 Pa
NA NA

Run 4 in. H 2 0 Pa
NA
0.204 0.214 0.228 0.239 0.248 0.253 0.256 0.257 0.259 0.262 0.267 0.271 0.275 0.276

Run S in. H 2 0 Pa
NA NA

Run 6 in. H 2 0 Pa
NA
0.110 0.1 10 0.110 0.110 0.1 10 0.110 0.109 0.109 0.109 0.109 0.108 0.108 0.108 0.107

N% .
0.171 0.162 0.159 0.152 0.149 0.147 0.145 0.144 0.143 0.141 0.139 0 . 1 0.134 0.133

.NA
42.5 40.3 39.6 37.8 37.1 36.6 36.1 35.8 35.6 35.1 34.6 34.1 33.3 33.1

NA
50.8 53.2 56.7 59.5 61.7 61.9 63.7 63.9 64.4 65.2 66.4 67.4 68.4 68.7

NA
27.4 27.4 27.4 27.4 27.4 27.4 27.1 27.1 27.1 27.1 26.9 26.9 26.9 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~c ground 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
project: EX-10-4D description: E x a m p l e 1 0 . 4 Surrrmer

Run 1

-

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 zone
EL

e l e v a t i o n : 0.0 f t
P 0.103

T . 73.4

path Elev-f l s o r Elev w i l l ~levrdoor Elev-docr

Flow 579.50 -32.15 - 273.68 -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 Open-docr SW-wa l l SW-wall f SW- l o o r Open-doo: SW-wall SW-wa l l

11643.36 -11451.74 - 1 l C . 94 - 80.68 11643.36 -1lrsi.74 -80.68 - 110.94

73.4

level: 2 zone EL

e l e v a t i o n : 12.0 f t

P - 0.069

T 73.4

path Elev-floor Elev-wall Elev-flocr Elev-door Elev-door Bldg-floor Elev-wall Elev-door Eiev-door Open-docr 0per.-door Blcig-floor SW- l l wa S5i-w a l l Ext-wall

Flow - 271.45 47.20 - 579.50 401.87 401.87 - 155.15 - 47.20 - 401.87 -401.87 1712.12 1712.12 -926.11 16.66 16.66 - 1525.34

FL

-0.OC8

73.4

Principles of Smoke Management

SW1

-0.005

73.4 SW-floor Open-door supply SW-wall SW-wall SW-floor 73.4 ~w-floo; Open-door supply SW-wall SW-wall SW-£ loor SW2/3 FL/2 Supply FL/2 Ambt SW2/G

SW2

-0.005

level: 3 zone EL

elevation: 24.0 ft

P -0.241

T 73.4

path Elev-floor Elev-wall Elev-floor Elev-door Elev-door

from EL/4 FL/3 EL/2 FL/3 FL/3 FL/4 EL/3 EL/3 EL/3 SW1/3 SW2/3 FL/2 3W1/3 SW2/3 Ambt

Flow -1104.74 46.22 271.45 393.53 393.53

SW1

-0.181

73.4 SW-floor Open-door SW-wall SW-wa11 SW-floor 73.4 SW-floor Open-door SW-wall SW-wall SW-floor

3W1/4 FL/3 Ambt FL/3 sh'1/2 3W2/4 FL/3 FL/3 Ambt SW2/2

SW2

-0.181

level: 4 zone . EL

elevation: 36.0 ft

P -0.412

T 73.4

path Elev-floor Elev-wall Elev-floor Elev-door Elev-door

r rom EL/5 FL/4 EL/3 FL/4 FL/4

Flow -1917.62 45.09 1104.74 383.89 383.89

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 Open-door SW-wa l i SW-wall SW-floor SW-floor Open-doo r SW-wall SW-wall SW-floor SW1/5 EL/4 Ambt FL/4 SW1/3 SW2/5 FL/4 FL/4 Ambt SW2/3

SW2

-0.355.

73.4

level: 5 zone EL

elevation: 48.0 ft

P -0.584

T 73.4

path Elev-floor Elev-wall Elev-f loor .Elev-door Elev-door

Flow -2711.67 44.05 1917.62 375.00 375.00 -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 OpeR-do cr SW-wa 11 SW-wa1 1 SW-£ loo= SW-floor Open-door SW-wall SW-wall SW-floor

SW1/6 FL/5 Ambt FL/5 SW1/4 SW2/6 FL/5 FL/5 Ambt SW2/4

-3039.04 -1361.45 -102.81 -13.24 4516.54 -3039.04 -1361.45 -13.24 -102.81 4516.54

w 2

-0.529

73.4

level: 6 zone

elevation: 60.0ft

P
-0.756

T

L

73.4

from path Elev-floor EL/7 Elev wall FL/6 ~ l e v ~ f l o c r EL/5 Elev-docr FL/6 FL/6 Elev-door

dP 0.000 -0. G01 0.000 -0.001 -0.001

Flow -2598.ii -6.21 2711.67 -53.35 -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 Single-door SW-wall SW-wall SW-floor SW-floor Single-door SW-wall SW-wall SW-f loor SW1/7 FL/6 Ambt FL/6. SW1/5 SW2/7 FL/6 FL/6 Ambt SW2/5

SW2

-0.701

73.4

level: 7 zone EL

elevation: 72.0 ft

P -0.927

T 73.4

path Elev-floor Elev-wall Elev-floor Elev door ~1ev:door FL/8 EL/: EL/: EL/? SW1/7 SW2/7 FL/6 SW1/7 SW2/7 Ambt

Flow -2302.12 -16.45 2598.71 -140.07 -140.07

SW1

-0.874

73.4

SW-floor Single-door SW-wall SW-wall SW-floor

SWl/% FL/7 Ambt FL/7 SW1/5 SW2/8 FL/7 FL/7 Ambt SW2/6

level: 8 zone EL

elevation: 84.0 ft

P -1.099

T 73.4

path Elev-f loor Elev-wall

Flow -1994.48 -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 Elev-wall Elev-door Elev-dqor Single-door Single-door Bldg-f loo r SW-wall SW-wall Ext-wall -FL/9 EL/8 EL/8 EL/8 SW1/8 SW2/8 FL/7 SW1/8 SW2/8 Ambt

SW2

-1.046

73.4

SW-floor Single-door SW-wall SW-wall SW-floor

level: 9 zone EL

elevation: 96.0 ft
T

P - 1.271

73.4

path Elev-floor Elev-wall Elev-f loor Elev-door Elev-door

from EL/10 FL/9 EL/8 FL/9 €L/9 €L/10 EL/9 EL/9 EL/9 SW1/9 SW2/9 FL/8 SW1/9 SW2/9 Ambt

dP 0.000 -0.008 0.000 -0.008 -0.008

Flow -1688.91 -16.95 1994.48 -144.31 -144.31

SW1

-1.218

73.4

SW-floor Single-door SW-wall SW-wall SW-floor SW-floor Single-dozr SW-wall SW-wall SW-floor

SW1/10 FL/9
Amb t

FL/9 SW1/8

SW2

-1.213

73.4

Levei: 1 0

elsvation: 108.0 f z

Principles of Smoke Management

! ,

zone EL

P -1.442

T 73.4

path Elev-floor Elev-wall Elev-floor Elev-door Elev-door

from EL/ 11 FL/lO EL/9 FL/10 IL/lO "/l1 ZL/ 10 EL/10 ZL/lO SXl/lO S:i2/10 - 3/9 s.11 ;:/0 sx2/10 .=nb t SXl/ll '/0 L1 .'?mbt fL/10 %1/9

Flow -1399.17 -16.07 1688.91 -136.83 -136.83 111.30 16.07 136.83 136.83 151.44 151.44 -87.32 69.41 69.41 -755.43 -1462.62 -151.44 -88.81 -69.41 1772.29 -1462.62 -151.44 -69.41 -88.81 1772.29

SW2

-1.390

73.4

SW-floor Single-door SW wall ll SW-floor

SWIW~

-C;.;2/11 3/10 3/10 .=Xb t S>;2 9 /

level: i l zone EL

elevation: 120.0 ft

P -1.614

T 73.4

patn Elev-floor Elev-wall Elev-floor Elev-door Elev-door Bldg-floor Elev-wall Elev-door Elev-door Singie-door Single-door 91dg-floor S ' wall % ll Ext-wa 11

dP 0.000 -0.006 0.000 -0.006 -0.006

Flow -1137.18 -14.53 1399.17 -123.73 -123.73

FL

- 1.620

73.4

SWIW~

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 SW-floor level: 12 zone EL elevation: 132.0 ft

Ambt SW2/10

P
-1.786

T 73.4
-

path Elev-floor Elev-walf Elev-floor Elev-door Elev-door

from EL/13 FL/12 EL/11 FL/12 FL/12 FL/13 EL/12 EL/12 EL/12 SW1/12 SW2/12 FL/11 SW1/12 SW2/12 Ambt

Flow -911.05 -12.54 1137.18 -106.80 -106.80

SW1

-1.734

73.4

SW-floor SW1/13 Si~.gle.-door FL/12 Ambt SW-wa ll FL/12 SW-wall SW-f loor SW1/11 SW-floor Single-door SW-wall SW-wall SW-floor SW2/13 FL/12 FL/12 Ambt SW2/11

SW2

-1.734

73.4

level: 13 zone EL

elevation: 1'44.0 ft

P -1.958

T 73.4

path Elev-floor Elev-wall Elev-f loor Elev-door Elev-door

from EL/14 FL/13 EL/12 FL/13 FL/13 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

Flow -726.25 -10.25 911.05 -87.24 -87.24

.

.

Principles of Smoke Management

SW2

-1.906

73.4

SW-floor Single-door SW-wall SW.-wa l l SW-floor

SW2/14 =/l3 FL/13 Ambt SW2/12

0.000 -0.055 -0.055 -0.079 0.000

-568.54 -145.53 -66.70 -79.85 860.62

level: 14 zone EL

-

elevation: 156.0 ift

T
73.4

path Elev-f loor Elev-wall Elev-floor Elev-door Elev-door

from EL/15 FW14 EL/13 FL/14 FL/14 FL/lS EL/14 EL/14 EL/14 SW1/14 SW2/14 FL/13 . SW1/14 SW2/14 Ambt SW1/15 FL/ 14 Amb t FL/14 SW1/13 SW2/15 FL/i4 FL/14 Ambt SW2/13

Flow -584.77 -7.85 726.25 -66.85 -66.85 78.72 7.85 66.85 66.85 143.91 143.91 -106.76 65.96 65.96 -533.25 -281.99 -143.91 -76.68 -65.96 568.54 -281.99 -143.91 -65.96 -76.68 568.54

FL

..

SW1

SW2

level: 15 zone EL

elevation: 168.0 ft from EL/16 FL/lS EL/14 FL/15 FL/15 EL/15 EL/l5 EL/15 SW1/15 SW2/15 FL/14 SW1/15 SW2/15 Ambt FL/15 Ambc Flow -473.52 -6.17 584.77 -52.54 -52.54 6.17 52.54 52.54 143.04 143.04 -78.72 65.56 65. 56 -449.73 -143.04 -73.39

P -2.301

Appendix G-Data and Computer Output forStairwell Example

SW wall ~~Ifloor
SW2

FL/15 SW1/14 FL/15 FL/15 Ambt SW2/14

-2.249

73.4

Single door SW wali ~ ~ 3 a l l SW-floor

level: 16 zone EL

elevation: 180.0 ft

P .-2.473

T
73.4

path Elev-Vent Elev-floor

from Ambt EL/15

dP -0.008 0.000

Flow -473.52 473.52

systems: name Exhust supply Note:

air flows: recirc outside 0.00 0.00 0.00 40999.97

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 Eshaust ~loorl cfn' 2800 2800 2800 2800 2SOO 2800 m3/s 1.32 1.37 1.37 1.37 1.32 1.32 Adjacent Floor SUPP~Y cfm 2800 2800 2800 2800 2800 2800 ni3/s 1.32 1.32 1.37 1.32 1.32 1.32 Staircwll Supply cfni l600 l600 l600 l600 1600 l600 n19s 0.761 0.761 0.761 0.761 0.764 0.764

Run
1 2 3 4

File EX-12-5A EX-12-5C EX-12-5B EX-12-5E EX-12-5D EX-12-SF

Season Summer Summer Summer Winter Winter Winter

Building Leakage Loose Loose Loose Loose ~oose Loose

G 2 7 G

5

-I
7

6
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 Run 2 3 4 5 6
1.

Floor Below to Fire ~ l o o r ' in. HzO 0.060 0.053 NA 0.087 0.087 Pa 11.9 13.2 NA 3 1.6 21.6

Floor Above to Fire Floor in. HzO 0.066 0.072 0.069 0.051 0.091 Pa 16.4 17.9 17.7 13.4 23.1

in. HzO 0.065 0.053 0.063 0.054 0.103
N A indicates "not spplic;~hlc."

Pa 16.2 13.2 l 5.7 13.1 75.6

.

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

1 Table H3. CONTAM Flow and Pressure Output for Example i 2 . 5 , ~ u n
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 29.9 in. Hg barometric pressure: wind speed: 0.0 mph wind direction: 0.0 deg Levei: G zone EL elevation: 0.0 ft

D
0.044

T
73.4

path Elev-floor Elev-wall Elev-door Elev-door Bldg-floor Elev-wall Elev-door Elev-door Single-door Single-door return SW-wa ll SW-wall Ext-wall Double-door SW-floor Single-door SW-wall SW-wall SW-floor Single-door SW-wall SW-wall

from EL/2 FL/G FL/G FL/G FL/2 EL/G EL/G EL/G Ambt Ambt Exhust SWl/G SW2/G Ambt Ambt SW1/2 Ambt Ambt FL/G SW2/2 Ambt FL/G Ambt

dP 0.000 -0.061 -0.061 -0.061

Flow1 855.97 -47.48 -404.24 -404.24

€L

-0.018

73.4

S?; l

0.069

73.4

0.000 -0.067 -0.066 -C.086 0.000 -0.067 -0.086 -0.066

317.03 -160.52 -72.87 -83.64 317.03 -160.52 -83.64 -72.87

Sii2

0.069

73.4

level: 2 zcne EL

elevation: 12.0 ft

P -0.128

T
73.4

path Elev-floor Elev-wall Elev-floor Elev-door Elev-door Bldg-floor Elev-wall Elev-door Elev-door Single-door supply

f rom EL/3 FL/2 EL/G FL/2 FL/2 FL/3 EL/2 EL/2 EL/2 SW1/2 Supply

dP 0.000 0.024 0.000 0.024 0.024 -0.G24 -0.024 -0.024 -0.024 0.001 n/a

Flow1 322.91 29.57 -855.97 251.74 251.74 -479.55 -29.57 -251.74 -251.74 19.87 2800.00

€'L

-0.104

73.4

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 %-floor Single-door supply SW-wall SW-wall SW-floor SW2/3 FL/2 Supply FL/2 Ambt SW2/G

SW2

-0.103

73.4

level: 3 zone EL

elevation: 24.0 ft

P -0.300

T 73.4

path Elev-floor Elev-wall Elev-floor Elev-door Elev-door

from EL/4 FL/3 EL/2 FL/3 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 Single-door SW-wall SW-wall SW-floor SW-floor Single-docr SW-wall SW-wall SW-floor

SW1/4 FL/3 Ambt FL/3 SW1/2 SW2/4 FL/3 FL/3 Ambt SW2/2

SW2

-0.275

73.4

level: 4 zone EL

eleva~ion:36.0 ft

P -0.47i

T path 73.4. Elev-floor Elev wail

from EL/5 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 Single-door SW-wall SW-wall SW-floor SW1/5 FL/4 Ambt FL/4 SW1/3 SW2/5 FL/4 FL/4 Ambt SW2/3 level: 5 zone EL elevation: 48.0 ft

P
-0.643

T 73.4

path Elev-f loor Elev-wall Elev-f loor Elev-door Elev-door Bldg-f loor Elev-wall Elev-door Elev door single-door Single-door Bldg-f loor S W-wa ll SW-wall Ext-wall SW-floor Single-door S W-wa ll SW-wall SW-floor

from EL/6 FL/5 EL/4 FL/5 FL/5 FL/6 EL/5 EL/5 EL/5 SW1/5 SW2/5 FL/4 SW1/5 SW2/5 Ambt SW1/6 FL/5 Fmbt FL/5 SW1/4 SW2/6 FL/5 FL/5 Ambt

FL

-0.64 4

73.4

SW1

-0.618

73.4

SW2

Principleso Smoke Management f

SW-floor level: 6 zone EL elevation: 60.0 ft

P -0.8 15

T 73.4

path Elev-f loor Elev-wal l ~lev-f loor Elev-door Elev-door Bldg-f loor Elev-wa 11 Elev-door Elev-door Single-door Single-door Bldg-floor SW-wall SW-wall Ext-wall

from EL/7 FL/6 EL/5 FL/6 FL/6 FL/7 EL/6 EL/6 EL/6 SW1/6 SW2/6 FL/5 SW1/6 SW2/6 Ambt SW1/7 FL/6 Ambt FL/6 SW1/5 SW2/7 FL/6 FL/6 Ambt SW2/5

Flowl 668.01 -4.68 -583.62 -39.86 -39.86

FL

-0.8 15

73.4

SW1

-0.790

73.4 SW-floor Single-docr SW-wall SW-wall SW-floor 73.4 SW-floor Single-door SW-wall SW-wall SW-floor

SW2

-0.790

level: 7 zone EL

elevation: 72.0 ft

P -0.986

T
73.4

path Elev-f loor Elev-wall Elev-floor Elev-door

from EL/8 FL/7 EL/6 'FL/7 FL/7 FL/8 EL/7 EL/7 EL/7 SW1/7 SW2/7 FL/6 SW1/7 SW2i7 Ambt SW1/8 FL/7 Ambt

Flowl 609.22 3.26 -668.01 27.77 27.77

.

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

SW-wall SW-£ loor
S W ~

FL/7 SW1/6 SW2 /8 FL/7 FL/7 Ambt SW2/6

-0.024 0.000 0.000 -0.024 -0.024 -0.029 0.000

-44.35 368.28 -178.95 -96.76 -44.35 -48.22 368.28

-0.962

73.4

SW-floor Single-door SW-wall SW-wall SW-f loor

-

level: 8 zone EL

elevation: 84.0 ft

P
-1.158

T
73.4

path Elev-floor Elev-wall Elev-f loor Elev-door Elev-door

from EL/9 FL/8 EL/7 FL/8 FL/8 EL/8 EL/8 EL/8 SW1/8 SW2/8 FL/7 SW1/8 SW2/8 Ambt FL/8 Ambt FL/8 SW1/7

Flowl 456.28 8.48 -609.22 72-23 72.23

SW2

- 1.134

73.4 Single-door SW-wall SW-wall SW-f loor

level: 9 zone EL

elevation: 96.0 ft

P
-1.330

T
73.4

path Elev-Vent Elev-floor

from Ambt EL/8

Flowl 456.28 -456.28

systems: name Exhust Supply

air flows: recirc outside 0.00 0.00 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
T h e inspection procedures described in this appendix apply to smoke control systems that are dedicated only to controlling smoke in building fires o r that make use o f air-moving equipment with another function, such as heating and air conditioning. These procedures are, o f a general nature, intended as a guide for tlie development of specific procedures for individual smoke control systems. These procedures address tlie major components of smoke control systems but, by their general nature, cannot address all possible coniponents. In this appendix, the phrase "as specified" is used to mean as specified in accordance with a contract documents, a code, or some other standard o r standards that have been agreed upon by the owner, designer, builder, code official, and other involved parties. Check duct installation. Duct installation, including the hangers, must not reduce the fire resistance rating of structural members and of assemblies. Frequently, structural members and asse~iiblieshave fire protective coverings, such as drywall construction or a sprayed-on layer. Check that ducts are installed in such a manner that these protective coverings are not damaged. Check that clearance from ducts to conibustible construction is as specified. In addition, check that where ducts pass through walls, floors, or partitions, the openings in construction around tlie ducts are as specified. Clieck that installation and materials of duct connectors and flexible duct connectors are as specitied. CAUTION: Become 11le cllaracteristics of duct co1it7ectors atid j1e.rible drtc~co~itiecfors are diffe~wir, orie sliorrld not be srrbs~i~rrted 111e for otliet: Check duct coverings and linings to verify that their fire safety requirements are as specified. Check that duct coverings do not conceal any service opening. Check direct access and inspection provisions. Service openings and telescoping or removable duct sections are used for direct access and inspection. Check tliat a service opening or a telescoping or removable duct section is provided in ducts as specified adjacent to fire dampers, smoke dampers. and smoke detectors. Check that these access openings are identified wit11 letters as specified. Check that service openings are

BARRIERS
a. Clieck walls, partitions, floors, and ceilings of barriers of smoke control systems for obvious and unusual openings that could adversely affect smoke control performance. Check tliat gaps around doors do not exceed the limits specified. If gasketing is required, check that it is as specitied. Check that automatic door closers in barriers of smoke control systems are as specified.

b.

c.

AIR-MOVING EQUIPIMENT
a. Check ducts to veriQ that materials ofduct material and construction are as specified.

. Appendix I - Inspection Procedures for Smoke Control Systems

provided in horizontal ducts and plenums where specified.

f
g.

Check air filters to verify that they have the classification specified. Check that the location, fire protection rating, and installation of fire, ceiling, and smoke dampers are as specified. Generally, fire, ceiling, and smoke dampers should be installed in accordance with the conditions of their listing and the manufacturer's installation instructions that are supplied with the damper. Further, check installation by removing hsible link (where applicable) and operate damper to verify that it fUUy closes. It is desirable to operate dampers

with normal air flow to ensure that they are not held open by the airstream. Remember to reinstall all hsible links that have been removed during inspection.

CONTROLS
a. Check manual controls. Check that devices for manual activation and deactivation of the smoke control system are of materials and installation as specified. Check automatic controls. Check that devices for automatic activation and deactivation and control of the smoke control system are of materials and installation as specified.
,

.,

b.

Principles of Smoke Management

Table 1-1: inspection Checklist-Barriers of Pressurized Stairwells
Project: Inspection agent: Date:

YES
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 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:

S

REMARKS

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: DESCRIPTION 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: Date:

YES

REMARKS

PrincipIes of Smoke Management

Table 1-3: Inspection Checklist-Barriers of Zoned Smoke Control Systems
Project: Inspection agent: DESCRIPTION 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: YES Date: NO REMARKS

Appendix I - Inspection Procedures for Smoke Control Systems

Table 14: Inspection Check List-Fire Safety Controls in HVAC Systems
Project: Inspection agent: 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: Date:-

Appendix J Test Procedures for Stairwell Pressurization .Svstems
SCOPE
The test procedures described in this appendix apply to systems for stainvell pressurization.
'

STAIRWELL PRESSURIZATION TEST
Activate the stairwell pressurization systems by a putting a detector in alarm as required by the contract documents. Test each pressurized stairwell by conducting the following steps. a. With all stairwell doors closed (except for the exterior ground floor door if it is required to be opened upon system activation), measure and record pressure differences across each closed stainvell door. Open tlie exterior ground floor stairwell door (except if the exterior ground floor door is required to be opened upon system activation), and measure and record pressure differences across each closed stairwell door. For stainvells without a ground floor exterior door, another highly severe open door condition must be tested. This can be an exterior door not at the ground floor or a large flow path to the outside created by opening the stainvell door and other doors, including an exterior building door. Open an additional stainvell door, and measure and record pressure differences across each closed stainvell door. Repeat this step, ope!ling anotiier door each time, until the required number of d o o k is opened. The required number of doors is that nuniber that must be opened during testing as stipulated i l l tlie applicable codes or contract docu~nents. With the required number of doors opened, clicck flow direction tliroi~gh open door-

EMERGENCY POWER
If standby power or other emergency power has been provided for the operation of the stainvell pressurization control system, acceptance testing shall be conducted with emergency power and normal power. During one test started under normal power conditions, the normal power shall be shut off to determine the ability of the stairwell pressurization systems and all associated systems to properly operate under standby power or other emergency power.

b.

NORMAL OPERATION TEST
With all building HVAC systems in normal operation, any zoned smoke control systems shut off, and the stairwell doors closed, measure and record the pressure differences across each stainvell door. The sign convention for all pressure difference readings in the stairwell tests is: a pressure dityerence resulting from a flow from the stairwell is positive, and a pressure difference resulting from a flow to the stainvell is negative. Evaluate these pressure differences to detennine that they are appropriate for the balanced HVAC system. Generally, this would be about 0.01 inches of water gage, but pressure differences as large as 0.03 inches water gage are not a causc for concern. However, higher pressure differences may occur for special systems such as those intended to control airbornc pollutants. Additionally, greater pressure dillkrcnccs can be caused by stack effect (as explained in Chapter 5).

c.

d.

Appendix 1-Test Procedures for stairwell Pressurizatiori Systems

e.

ways using a 6 ft strip of tissue paper secured at the top of the door h m e . Check that the measured pressure difference is within the acceptable range, as defined in the contract documents. If the pressure difference is not in the acceptable range, double check that the states of fans, dampers, and doors is as required. if any of these were not as required, they should be fixed and the zone retested. After this, if the pressure difference is not acceptable, the flow rate of air to the stairwell in question should be measured and adjusted as appropriate. If the

pressure differences are too low after these actions, excessive air leakage paths in the construction should be filled, caulked, or sealed as appropriate. (Often it is very difficult to locate leakage paths in buildings. Chemical smoke from smoke bombs can be used to find these leakage paths. The stairwell is filled with chemical smoke and pressurized, while the low-pressure side of the stairwell barriers is examined for smoke leakage that indicates the location of a leakage path.) Then the zone should be retested.

Principles of Smoke Management

Table J-1: Test Work Sheet-Pressurized Stairwell
Project Stairwell No. Test Agent: Doors in Pressurized Stairwell Pressure Difference (inches of water gage)
- Flow Direction From Stair To Stair

-

Comments:

Appendix K Test Procedures for Zoned Smoke Control Systems
SCOPE
The test procedures described in this appendix apply to zoned smoke control systems that are either dedicated systems or part of systems for heating, ventilating, and air conditioning (HVAC). gage are not a cause for concern. However, higher pressure differences may occur for special systems, such as those intended to control airborne pollutants. Additionally, greater pressure differences can be caused by stack effect (as explained in Chapter 5).

EMERGENCY POWER
If standby power or other emergency power has been provided for the operation of the zoned smoke control system, acceptance testing shall be conducted with emergency power and normal power. During one test stated under normal power conditions, the normal power shall be shut off to determine the ability of the zoned smoke control systems and all associated systems to properly operate under standby power or other emergency power.

SMOKE MODE TEST
Each smoke zone is to be individually tested by performing the following sequence. Activate smoke control system operation in the zone. This should be accomplished by putting one of the detectors into alarm that are intended to activate the snioke control system in that zone. Check that the operation of fans is as required by the contract documents. Check that the position of smoke dampers is as required by the contract documents. Also, check that any smoke dampers required to be closed are fully and tightly closed. Check to verify that all doors required by the contract documents -:o be closed during smoke control system operation are fully closed and that they operate freely, allowing use during evacuation without becoming jammed in their door frames. This should include doors in the boundary of the smoke zone being tested. Measure and record pressure difTerences across all the closed doors in the boundary of the smoke zone being tested. Pressure differences resulting from air flowing to the snioke zone being tested are to be recorded

SMOKE CONTROL DIAGRAM
Identify the exact location of each smoke control zone. If it is not part of the building plans, make a smoke control zone diagram of the building. This diagram should include the locations of all zone boundaries and of all doors in those boundaries.

NORMAL OPERATION TEST
With all building HVAC systems in normal operation, the zoned smoke control system shut off, and the smoke barrier doors closed, measure and record the pressure differences across each smoke barrier door. Evaluate these pressure differences to determine that they are appropriate for the balanced HVAC system. Generally, this would be about 0.01 inches water gage, but pressure differences as large as 0.03 inches \vater

Appendix K-Test Procedures for Zoned Smoke Control Systems

as positive values, and pressure differences resulting fiom air flowing fiom the -smoke zone being tested are to be recorded as negative values.

f.

Check that the measured pressure difference is within the acceptable range, as defined in the contract documents. If the pressure difference is not in the acceptable range, double check that the state of fans, dampers, and doors is as required. If any of these are not as required, they should be fixed and the zone retested. After this, if the pressure difference is not acceptable, the flow rates of air to and from the smoke zones in question should be measured and adjusted as appropriate. If the pressure differences are too low after these actions, excessive air leakage paths in the construction should be filled, caulked, or sealed as appropriate. (It is often very difficult to locate leakage paths in buildings. Chemical smoke from smoke bombs can be used to find these leakage paths. The high-pressure sides of smoke barriers are exposed to heavy concentrations of chemical smoke, while the low-pressure side of the barrier is examined for smoke leakage that indicates the location of a leak-

age path. Exterior walls, interior partitions, floors, and ceilings, including areas above suspended ceilings, must not be overlooked when hunting for excessive leakage areas.) Then the zone should be retested. g. Test for smoke feedback into supply air. Place six smoke bombs (three-minute duntion size) in a metal container, simultaneously ignite all bombs, and locate container near exhaust inlet in smoke zone being tested so that all of the chemical smoke produced by the bombs is drawn directly into the exhaust airstream. Check that air supplied to other zones of the building has no trace of chemical smoke. If chemical smoke is detected in this supply air, its path should be determined, the path should be blocked, and then the smoke feedback test should be conducted again. (The two most likely causes of smoke feedback are a leaky or party opened return air damper and an outside air inlet located in the vicinity of the exhaust air outlet.) Make sure that this zone has been returned to its normal setting before continuing to test other zones.

h.

Principles of Smoke Management

Tz51e K-l: Test Work Sheet-Zoned Smoke Control System in Normal Operation
Project: Test Agent: Doors of Smoke Control Zone Pressure Difference (inches of water gage) Date: Flow Direction 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: Yes Fans operating appropriately Smoke dampers in required position Pass feedback test

Date: No
-

-

Doors of Smoke Control Zone

Pressure Difference (inches of water gage)

Flow Direction From Zone To Zone

Comments:

Appendix L Inspection Procedures for Atria Smoke Exhaust Systems
SCOPE
The inspection procedures described in this appendix apply to atrium smoke exhaust systems. These procedures are of a general nature, intended as a guide for the development of specific procedures for individual smoke control systems. These procedures address the major components of smoke control systems but, by their general nature, cannot address all possible components. In this appendix, the phrase "as specified" is used to mean as specified in accordance with a contract of documents, a code, or some other standard or standards that have been agreed upon by the owner, designer, builder, code offkial, and other involved parties.

as specified. CAUTION: Because the characteristics oJ duct connectors and flexible duct connectors are d~fferent,one should not be substitutedJor the otheu.
d. Check duct coverings and linings to verify that their fire safety requirements are as specified. Check that duct coverings do not conceal any service opening. Check direct access and inspection provisions. Service openings and telescoping or removable duct sections are used for direct access and inspection. Check that a service opening or a telescoping or removable duct section is provided in ducts, as specified adjacent to fire dampers, smoke dampers, and smoke detectors. Check that these access openings are identified with letters as specified. Check that service openings are provided in horizontal ducts and plenums where specified. Check air filters to verify that they have the classification specified. Check that the location, fire protection rating, and installation of fire, ceiling, and smoke dampers are as specified. Generally, fire, ceiling, and smoke dampers should be installed in accordznce with the conditions of their listing and the manufacturer's installation instructions that are supplied with the damper. Further check installation by removing the fusible link (where applicable) and operate damper to verify that it fully closes. It is desirable to operate dampers with normal airflow to ensure that they are not held open by the airstream. Remember

e.

AIR-MOVING EQUIPMENT
a. b. Check ducts to verify that materials of duct material and construction are as specified. Check duct installation. Duct installation, including the hangers, must not reduce the fire resistance rating of structural members and of assemblies. Frequently, structural members and assemblies have fire protective coverings. such as drywall construction or a sprayed-on layer. Check that ducts are installed in such a manner that these protective coverings are not damaged. Check that clearance from ducts to combustible construction is as specified. In addition, check that where ducts pass through walls. floors, or partitions, the openings in construction around the ducts are as specified. Check that installation and materials of duct connectors and tkxible duct connectors are

f. g.

c.

Appendix L-Inspection Procedures for Atria Smoke Exhaust Systems

to reinstall all hsible links that have bken removed during inspection. b.

CONTROLS
a. Check manual controls. Check that devices for manual activation and deactivation of

the smoke control system are of materials and installation as specified. Check automatic controls. Check that devices for automatic activation and deactivation and control of the smoke control system are of materials and installation as specified.

Test Procedures for Atria Smoke Exhaust Svstems
SCOPE
The test procedures described in this appendix apply to systems for atrium smoke exhaust systems. to properly operate under standby power or other emergency power.

EXHAUST OPERATION TEST
With all building HVAC systems in normal operation and any pressurized stainvells, zoned smoke control systems, and other smoke management systems shut off, activate the atrium smoke exhaust system by a signal from a smoke detector or initiating device. After activation, determine that the smoke exhaust fans are operating as intended. The volun~etricflow of the smoke exhaust fans should be measured before the eshaust operation test.

EMERGENCY POMrER
If standby power or other emergency power has been provided for the operation of the atrium smoke exhaust system, acceptance testing shall be conducted with emergency power and normal power. During one test started under normal power conditions, the normal power shall be shut off to determine the ability of the atrium smoke exhaust system and all associated systems

Principles of Smoke Management

Index
A
Acceleration of gravity 66, 90, 93, 121, 122, 183, 184, 191, 196,197,207,218,219,222,243,261,268,321 Activation 8, 148, 154, 168, 205, 206, 208, 236, 249, 277,278,356,36 1,370,371 Air density 67,74,78,79,81,82,92,93, 143, 153, 158, 191, 195-196,207,241 gas constant 67, 190,261,268 properties 97,220,268,269 specific heat 269- 270 Airborne matter 2,63 Airflow 2-4, 6, 70, 71, 74, 78, 79. 87-95, 97, 109, 112, 113, 115, 117-119, 120-122, 142, 148, 150, 154, 158, 169, 172, 173, 175, 179, 181, 197, 206, 207, 210, 213, 226, 235, 236, 238-240, 247, 257, 289-293, 295-301, 3 12,369 Anemometer 78,240,241,245 ASCOS 119, 120, 122 ASET 120, 122, 123, 126, 199,202,203,249,257,321, 323,329-333 ASMET 120, l26,32 l-323,329,33 1 Atria 4, 8, 120, 131, 181-185, 189, 192, 195, 196, 199, 201, 206, 207, 2 10, 215, 2 17, 221, 223, 225, 253, 254, 274,275,322,323 Atrium mechanical exhaust 199 natural venting 4, 190, 199, 203,207 smoke filling 129, 199,201,205,221,248,272,323 Attenuation coefficient (see extinction coefficient) AZONE 120, 123, 195, 199? 200, 202, 203, 205, 206, 211-215

Carbon dioxide (CO2) 27,36,37, 252 Carbon monoxide (CO) 8, 27, 34, 36, 38, 252, 254, 256, 271 Carboxyhemoglobin (COHb) 38 Chimney effect (see stack effect) Church Street fire tests 4 Clear height 202, 203 Colebrook equation 10 1 Commissioning 3, 7, 9, 105, 146, 152, 161, 167, 175, 235,236,247 Communicating spaces 197,210,224 Compartmentation 2, 3, 5, 6, 32, 87, 129, 172, 180, 199, 291 Computational fluid dynamics (CFD) 3, 197, 247, 250 Confined Flow 190 Conservation of energy 123, 125,219 Conservation of mass 84, 121, 123, 125, 219, 229, 291, 330 Conservation of momentum 219,228,230 CONTAM 119-122, 130, 132, 137, 139, 154, 155, 161, 165, 180, 206, 257, 289, 290, 292, 293, 295-298, 312, 320,337,338,349 Contaminant 3,87, 88, 120-123,243,290,291, 298 123, 125, 158, 188, 189, 225, 244 Control volun~e Convective fraction 24, 182, 184, 202, 204, 205, 223, 245,321,322,325,327 Critical air velocity 89, 90, 244

D
Dampers6,9, 111, 113, 114, 117, 139, 169, 175, 178, 236, 357-359, 362,366, 369 balancing 1 17 barometric 148, 149, 168 bypass 149 chatter 149, 169 control 1 17 curtain 1 17 fire 3, 117. 149,254,257, 355, 357-359 leakage classification 1 18, 178 multi-blade 117, 1 18 return 178, 179, 366 smoke 3, 79, 87, 1 17, 1 18, 178, 179,257, 355-360, 365,368,369 Darcy-Weisbach equation 101 Decision tree 5, 6 DETACT-QS 2 l , 120, 126 DETACT-T2 2 1, 120 Detectors 19, 127, 169, 247, 250, 365 Diameter fire 182, 184,325-327 hydraulic 92, 93, 95, 96, 101, 104, 243, 299 plume 183

B
Barriers 5,87.88,2 10,235,236,254,355,357-359,362, 366 Base fuel package 23- 26 Benioulli's equation 93, 240 Boundary conditions 229, 232, 233 Boundary layer 56, 75, 78. 226.230, 245, 255 Buoyancy 2-4, 66, 71, 73, 74. 79. SO. 87, SS, 92, 107. 129, 150, 175, 176, 179, 181, 189-191, 195, 207, 217, 220,221,251,329

C
Calorimeter 13, 14, 25, 252 cone 14,248 open air 14, 15 oxygen consumption 13. I ? room 14

Index

Differential pressure (see Pressure difference) Differential pressure instruments 237 Dilution 2,3,45-47,87,88, 130, 172, 177,243 Dimensional Analysis 2 17 Dimensionless groups 2 19,330 Door-opening force 105-107, 145 Duct 3, 8, 79, 101-104, 111, 112, 114, 117, 118, 140, 149, 169, 173, 178, 179, 236, 239, 243-245, 253, 255, 272,290,355- 359,369 Duct, access 357- 359

E
Economizer 1 13 Egress 7, 27, 5 1-53, 56, 57, 59, 60, 88, 107, 120, 126, 161,244,249,253,254,277,329 Elevator 1- 4, 49, 63, 80-82, 89, 97-99, 133, 139, 142, 143, 155, 157-159, 161, 165-169, 171, 172, 236, 247, 248, 250-252, 256, 272-274, 277-279, 281- 287, 291293,295,301,338,358 car motion 68,69, 158, 277 evacuation 119, 120. 157, 158, 161, 165, 166, 167, 277,278,285,287 piston effect 66,79, 129, 160,252 ELVAC 119,120,277,284,285,287 Energy conservation 7, 11l , 149, 157 English units (I-P units) 3, 259, 265, 268, 282, 322, 325, 334 Evacuation 1, 3, 6, 7, 27, 29, 37, 48-52, 56-63, 87, 119, 120, 130, 133, 140, 141, 146, 148, 157-159, 161, 165168, 175, 199- 202,205, 207, 244, 250- 252, 254, 274, 277-279,284,285,287,365 component-by-component 57,59,60 constrained flow 57 density 6,52,53, 55, 56, 58-61 empir~cal correlations 5 I hydraulic a~alogy 1. 56 5 velocity 5 1,52,53, 55,58-62 Evacuation 53 Exhaust fan 176 Exhaust inlets, number 193, 194,2 10,213, 244 Exhaust inlets, separation 175, 194 Expansion 66,74,89, 129. 172, 175 Exponential flow equation 94, 96, 97, 243 Extinction (attenuation) coefficient 28, 29,3 1, 32, 245

F
Fan 2-4, 6, 9, 10, 66, 79, 87-89, 92, 109, 111, 112, 114, 121, 129, 139, 140, 146, 138, 149, 152, 158, 161, 167169, 171, 175, 177, 236, 737, 239, 247, 251, 272, 274, 289-295, 297, 301, 315, 316, 319, 357-360, 362, 365, 366,368

airfoil blade 115 axial 115 backward flow 115, 1 17 backward rotation 115, 117 centrifugal 114, 115, 117, 141, 146, 148, 154 exhaust 111,113,120,169,172,175-177,205,206, 244,245,371 forward curved 115 propeller 115, 141, 142 return 113, 179 roof-mounted 11 1, 140, 141 supply 7, 1 13, 154, 179,225 temperature 176, 177 tubeaxial 115, 1 17 vaneaxial 115, 117 variable flow 6, 169 FAST21, 120, 126, 130, 1)2, 137,254,271 Fire building 3,4,5,7, l l , 13,29,36,49,63,7 1, 79, 88, 90, 107, 131, 139, 157, 166, 167, 177,237, 251,252,257,271-273,275- 277.355 design 11,21, 129, 180, 188, 199,203, 205. 207 fighters 8,81, 139, 149, 166, 167 flaming 8, 13,29, 32,33,34,237 fully developed 13, 18, 37, 133, 188 growth coefficient 22.245, 221 growth time 206 research tower 7 1, 96 'scenario 2 , 21,4 1 , 50, 129, 249 size 7, 11,21, 192, 250 smoldering 8, 29, 237 spread 87, 172, 185,257 sprinklered 2, 7, 19, 107, 180, 188, 232, 237, 252 steady 1 1, 21, 192, 200-203, 205, 207, 2 11, 215, 323-325 suppression 5, 6, 7, l 1. 19, 25, 9 1, 129, 199, 249, 252,253, 257.332 test 2,4,23,28,44, 73.21 7,248,251,252,255,256, 271,273, 274 t-squared 18, 21-23, 192, 201, 202, 214, 215, 329, 332 unsteady 21, 192,201,203,205,207,21 l , 324,326 ventilation controlled 13, 18, 188 Fire Dynamics Simulator (FDS) 23 1, 253 Flame height 19, 182-186, 204, 205, 244,321,322,325327 Flashover 1 1, 13, 254 Flexibility 6, 7 Flow area effective 63-66, 70, 79-51, 143-145, 150, 152. 161, 172-174,243 parallel paths 63-65 serles paths 63-65

l

I

l

Principles of Smoke Management,

Flow coefficient 4, 63-66, 70,78,79, 81, 82, 93, 94, 96, 97, 103, 133, 145, 152, 159, 163, 164, 174, 191, 196, 243,338 Fractional effective dose (FED) 36,46, 131,243 Fractional incapacitating dose (Fm) 40,41, 44, 131 Friction losses 66, 93, 143, 150, 161, 163, 164,338 Fuel package 14, 19,23-26, 129,253

G
Gas law (see Ideal gas law) Governing equations 2 19,220,225,230,231

H
Haber's Law 36 Hazard analysis 3, 5, 7, 61,87, 119, 122, 129, 130, 131, 133, 168,207,248 Heat exposure 3,27,44,45,47, 130, 131, 133,207 Heat release density 22, 184,244 Heat release rate (HRR) automobile 17 Christmas tree 14, 15 cribs 17 furniture 13-15, 21 kiosk 14 pallets 17, 22 peak 14, 15, 17,25 sprinklered fires 19 Heat Transfer Scaling 223 Height limit 145-147,243 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, 361,365 Hydrogen bromide (HBr) 34,36 Hydrogen chloride (HCl) 34,36,250 Hydrogen cyanide (HCN) 34,36,252

Manual stations 8 Mass optical density 28,30,32,34,46,47, 131, 133,245 Metric units (see International system units) MGM Grand fire 1,71, 157,248,257 Modeling detector activation 120, 126, 192,226, 227 Froude 21 7,221,222,224 network 104, 180 pressure 221,222 salt water 256 saltwater 221 turbulence 229- 232 zonefire4,5, 180, 181,211,274

K
Navier-Stokes (NS) equation 94 Neutral plane 63,67, 70,71, 73,74, 82- 85,243,273 Newton Raphson method 102 Newton's second law 2 18 N-Gas model 39,40,42,243,244,252 Nomenclature 277,321

0
Objectives, smoke management 5 Open doors 74,87,97, 105, 140, 141, 154,338 Optical density 28,29,3 1,32,34,46,47, 13 1, 133, 245 Orifice equation 70,93, 94, 96,97, 100 8, Oxygen (02) 13,37,38,39,41,42, 74, 79, 89,90-92, 175,250,252,254

P
Panic 49,50,25 1,254,255 Percentage obscuration 28-30,245 Perfect gas law (see Ideal gas law) Physical (scale) modeling 130, 197, 217,219, 22 1 Pirot tube 240,241 P!ugholing 120, 181, 193-195, 210,211,213,244, 245 Plume average temperature 188, 189,208,325,326 axisymmetric 181-186, 188, 189, 199, 204, 21 1, 244,245 balcony 186,187, 197,204,257 centerline temperature 126, 182-184,322,323,325, 327 corner 185, 186, 188 maximum height 189,245 wall 185, 186, 188 window 188,204 Poiseuille Flow 94 Post-flashover fire 13,27 1 Power law 75 Prandtl number 2 17, 2 18, 220

I
Ideal gas law 67, 143 Ignition 2, 5, 11, 19, 22-25, 34,45, 124, 129, 223, 237, 247 Inspection 88, 235, 236,355-360, 369, 370 lnternational system (SI) units 3, 259, 261

J
JET 2 1, 120, 122, 126, 127,249 Johnson City Retirement Center fire I
;

L
LAVENT 2 l , 120, 122, 126 Leakage area (see flow area)

,

M
Manometer 238,24 1

Index

Pressure difference average 142, 145, 162 critical 158, 159 design 107, 109, 162, 168, 172, 175,338 Pressure sandwich 8, 17 1 Pressurization 2-5,7,8,87&9,97-99, 105, 113-115, 119, 120, 122, 129, 139, 140-146, 148-150, 152-155, 157159, 161, 165-169, 173-176, 180, 210, 226, 237, 248, 249,251,256,272,273,275,292,293,338,361 Pull box (see Manual station) Purging 87, 88, 149, 168, 172

R
Radiant fraction 24 Reliability 8, 9, 62, 166 Remote control center 8 Resiliency 6, 7 Response time index (RTI) 20,21, 127,244 Reynolds averaging 229 Reynolds number 78, 92-96, 101, 217, 220-224, 240, 244,299 Roosevelt Hotel fire 1 Roughness 101-103,245,300

Stack effect 66, 70, 71, 73, 79-84, 107, 108, 129, 142, 167,179,251,273,361,365 Stack effect, normal 66,67,175 Stairwell pressurization analysis 147 compartmentation 141 multipleinjection 140, 141, 146, 148, 150 pressure profile 142, 145, 146 single injection 140 vestibules 141 with open doors 146, 148, 150 Stairwell, pressure losses 104 Standard atmospheric pressure 67, 73, 93, 120, 190, 261, 268 Stratification 207,208,236,249 Symmetry 104,105, 146, 152,229,283

T
Temperature, conversion 261 Thermal inertia 223, 224, 244 Thermal radiation (radiant heat flux) l l, 23, 24, 27, 45, 47,48,125, 130, l31,255 Thornas's equation 89,90,92 Time lag ceiling jet 191, 192, 205 plume 192, 205 Toxicity 3, 27, 34,36-39, 42,43,45, 47, 109, 13 1. 133, 135,250,252,255,256,271 Tracer gas 237 Transient fuel 21,9 1 Transmittance 27-29,244

S
Safety factors 7, 146, 152, 161, 167 Scaling relations 217, 221-223 Shopping malls 92, l81,252,272-275 Similitude 217,219,22 1 Smoke backflow 88-92,181,244,256 bombs (see Smoke, chemical) chemical 236,237,362,366 dampers (see Damper, smoke) definition 27 detectors 8,208,209, 224, 249, 355, 369 exhaust 4, 5, 7, 87, 114, 123, 129, 149, 169, 175, 194, 195, 203-206, 210. 214, 250, 252, 369,371 filling 129, 18 1, 199, 200, 201, 205, 248, 272,323 horizontal flow 126, 195, 196,295, 301 layer interface 122, 190, 195,204, 332 shaft 3,87, 142, 149,150, 169, 172, 175,256 venting 3,4, 129, 149, 169, 172, 190,207,210 Specific heat, constant pressure 125,2!8,228,265 Specific heat, constant volume 125, 2 18, 21 9 Specific heat, ratio 20, 46, 89, 125, 177, 189, 204, 2 12, 219,243,261,269,270,321,330 Sprinkler activation 2 time constant 19,20 Stack action (see Stack effect)

U
Units of measurement 259

v
Vector 227,228,247,248,250 Vestibules 141 Virtual origin 182-185, 245,250, 321-327 Viscosity, dynamic (absolute) 93,94,217-2 19,227,228, 23 1,245,268 Viscosity, kinematic 92,95,96, 218,245 Visibility 3,27,29,3 1,32,34,45-48, 130, 13 l, 133. 134, 136,244,251,272 Volumetric flow 64, 74, 92- 94, 96, 103, 1 12, 144, 150, 152, 174, 176, 177, 190, 194, 204, 205, 213, 222: 223, 239,240,244,322,325,326,37 1

W
Weather data 109, 290, 296, 297 Wind 6, 66-69, 74, 75, 78-80, 104, 107-109, 120. 129, 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

Wind data 78, 109,274,296

z
Zero floor leakage idealization 70, 7 1, 142

Zoned smoke control 2, 3,4, 8, 74, 89, 120, 139, 142, 149, 171, 172, 175, 178-180, 236, 237,273, 349, 359, 361,365,367,368,371