SFPE Human Behavior in Fire by MS23V

The SFPE Task Group on Human Behavior Chairman Daniel Oâ&#x20AC;&#x2122;Connor, P.E. Schirmer Engineering

Members John Bryan, Ed.D. Eugene Cable, P.E. Department of Veterans Affairs Rita Fahy, Ph.D. National Fire Protection Association Michael Ferreira, P.E. Hughes Associates Edwin Galea, Ph.D. The University of Greenwich Norman Groner, Ph.D.

Gordon Hartzell, Ph.D. Hartzell Consulting, Inc.

Guylene Proulx, Ph.D. National Research Council of Canada

Frank Hsu, Ph.D., P.E. Thermal Science and Technology

Steve Smith NSW Fire Brigades

Irving Mande, P.E. EST

Chen-Hsiang Su, P.E. Schirmer Engineering

Harold E. Nelson, P.E. Hughes Associates

Bruce Wallace Royal & Sun Alliance

Jake Pauls Consulting Services Bldg. Use & Safety

Thomas Wright, P.E. FPE Forensics, PSC

Staff Morgan J. Hurley, P.E. Society of Fire Protection Engineers

Printed in the U.S.A.

Foreword

This engineering guide to Human Behavior in Fire is the fifth engineering practice guide published by the Society of Fire Protection Engineers. Its publication is a milestone for the fire protection engineering profession in that this represents the first concise document that summarizes the state of the art in understanding how people might behave in fire. Publication of this guide will help engineers improve fire protection designs in which life safety is a goal by facilitating better predictions of how people may respond in a fire.

The Society of Fire Protection Engineers wishes to acknowledge and thank the NFPA for its generous support of this project.

Contents

Foreword ............................................................................................................................................................. ii Introduction ........................................................................................................................................................ Basic Evacuation Concepts and Goals ............................................................................................................ People–Building–Environment Interactions .................................................................................................... Organization and Use of This Guide ...............................................................................................................

1 1 2 4

Occupant Characteristics .................................................................................................................................. 7 Population Numbers and Density .................................................................................................................... 7 Alone or with Others ....................................................................................................................................... 8 Familiarity with the Building........................................................................................................................... 8 Distribution and Activities ............................................................................................................................... 8 Alertness........................................................................................................................................................... 9 Physical and Cognitive Ability ........................................................................................................................ 9 Social Affiliation .............................................................................................................................................. 9 Role and Responsibility ................................................................................................................................... 9 Location ........................................................................................................................................................... 9 Commitment..................................................................................................................................................... 9 Focal Point ....................................................................................................................................................... 9 Occupant Condition .........................................................................................................................................10 Gender ..............................................................................................................................................................10 Culture..............................................................................................................................................................10 Age ...................................................................................................................................................................10 Other Factors....................................................................................................................................................11 Human Response to Cues ..................................................................................................................................11 Introduction ......................................................................................................................................................11 Receiving Cues ................................................................................................................................................13 Recognizing Cues ............................................................................................................................................14 Interpreting Cues..............................................................................................................................................15 Impact on Fire Protection Engineering Design and Analysis..........................................................................17 Conclusion .......................................................................................................................................................18 Decision Making of People Facing a Fire ........................................................................................................18 “Panic” Behavior .............................................................................................................................................19 Occupant Characteristics .................................................................................................................................19 Information Processing ....................................................................................................................................21 Decision Making During a Fire .......................................................................................................................23 Impact on Fire Protection Engineering Design and Analysis..........................................................................25

iii

Movement............................................................................................................................................................26 Factors That Impact Movement Time..............................................................................................................27 Methods for Calculating Movement Time.......................................................................................................28 Alternatives to Evacuation...............................................................................................................................37 Impact of Smoke on Movement ......................................................................................................................38 Conclusion .......................................................................................................................................................38 Appendix Questions a Potential Model User Should Ask About an Evacuation Model .................................................41 References ...........................................................................................................................................................43

Illustrations FIGURE 1 2 3 4 5 6 7 8 9 10 11

Timeline of the Evacuation Process .............................................................................................................. 1 Interrelationship of Human Factors .............................................................................................................. 5 Cue Validation Process..................................................................................................................................11 Decision-Making Period................................................................................................................................18 Movement/Refuge Time................................................................................................................................26 Comparison of Correlations and Data...........................................................................................................27 Movement Speed as a Function of Density ..................................................................................................29 Specific Flow as a Function of Density ........................................................................................................31 Merging Egress Flows...................................................................................................................................35 Transition in Egress Component ...................................................................................................................35 Walking Speed as a Function of Smoke Extinction Coefficient...................................................................38

Tables TABLE 1 2 3 4 5 6 7 8

Type of Human Behavior Guidance Provided .............................................................................................. 4 Compilation of Visibility Distance for Population Moving Through Smoke...............................................24 Compilation of Distance Moved Through Smoke ........................................................................................24 Compilation of Visibility Distance Relative to Turned Back Behavior .......................................................25 Velocity Factor in Equations 2 and 34 ..........................................................................................................29 Mean Speed for Impaired Individuals...........................................................................................................30 Boundary Layer Width ..................................................................................................................................31 Maximum Specific Flows .............................................................................................................................32

Engineering Guide

Human Behavior in Fire of evacuation response. The information in this guide may also be useful and applicable to postevent analysis.

Introduction To address the fire safety of occupants in a building, it is important to understand and consider the factors that may influence the responses and behaviors of people in threatening fires. The anticipation of human behavior and prediction of human responses is one of the most complex areas of fire protection engineering. Because the understanding of human behavior in fire is limited compared to other areas of fire protection engineering and behavioral study, it is difficult to predict accurately the responses and behaviors of people in fire situations. While it is important to recognize that there is limited understanding and acknowledged uncertainty related to human response and behavior in fire, significant sources of published information on human response and behavior in or related to fires are relevant and useful for addressing human factors in fire situations. This guideâ&#x20AC;&#x2122;s purpose is to identify and review the key factors and considerations that impact the response and behavior of occupants evacuating a building during a fire event. It is also the purpose of this guide to identify both quantitative and qualitative information and resources in the literature. Such information should be considered prior to developing safety factors or exercising engineering judgment in the practical design of buildings, the development of evacuation scenarios for performance-based designs, and the estimation

BASIC EVACUATION CONCEPTS AND GOALS Traditionally, human factors during fire evacuations have not been fully considered in engineered fire protection design. When human factors were considered, they were often limited to simple assumptions that may or may not have been appropriate. The information on addressing human response and behavior in fire in this guide is intended to facilitate a better and more complete understanding and consideration of the human factors as they may impact evacuation of people during a fire incident. Evacuation is the process wherein occupants become aware of a building fire-related emergency and experience a variety of mental processes/actions before and/or while they travel to reach a place of safety within or outside the building. Time is the basic measure of the evacuation process, and Figure 1 depicts conceptually the timeline of the evacuation process and the relative time relationships of key human responses or behaviors that occur during the evacuation process. The initial cues may be direct signs of the fire such as visible smoke or flames or events resulting

<START TIME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . END TIME> FIRE / CUE INITIATION OR DEVELOPMENT

Cue Validation Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . and Continuing Process Receiving Cues

Recognizing Cues

Interpreting Cues

Receiving, Recognizing, Interpreting (RRI)... RRI...RRI...

Decision-Making Period Pre-movement Decisions

Transmovement Decisions Movement / Refuge Time

FIGURE 1. Timeline of the Evacuation Process

from a fire such as fire alarm activation or power outage. When these cues or indicators of the fire reach human senses, the cue(s) validation process begins, which continues for the duration of the event. For example, at some time after cue(s) are received there is recognition of the cue(s) (e.g., smell of smoke) and finally interpretation of the cues (smoke means something is burning). Decision making results from the cue validation process, and specific pre-movement actions may result such as gathering belongings or calling for emergency assistance. As time transpires and additional cues are evaluated, decisions to evacuate or remain in place are expected. Often the decision to move or relocate to an area of refuge will have occupants moving through a building encountering and evaluating new cues along the way, which results in repeated (transmovement) decisions until a place of safety is achieved or exposure to the fire environment precludes evacuation. Arriving at a place of safety is embodied in the concept of Available Safe Egress Time or ASET. ASET is the time period between the ignition of a fire and the onset of untenable conditions for one or more building areas. ASET must be properly evaluated and compared to the Required Safe Evacuation Time or RSET. For a successful evacuation, ASET > RSET. Estimations of ASET typically involve tenability analyses, e.g., the time before a smoke layer reaches a certain height or the time before the instantaneous or cumulative effects to people are predicted to result in incapacitation. Predictions of RSET typically involve estimating the time that it would take for people to be notified that there might be a fire, the time that people would take for “pre-movement” activities such as alerting others, checking on family members, etc., and the time that it would take for people to egress to a safe place. During the course of evacuation there may be many activities that extend the time for evacuation. Time is obviously the primary aspect of an evacuation, and incomplete or cursory attention to the human factors in an evacuation scenario can result in misleading, overestimated, or underestimated evacuation times, and potentially failure to achieve the goal of safe evacuation, where ASET > RSET.

PEOPLE–BUILDING–ENVIRONMENT INTERACTIONS The focus of this guide is human factors. This includes consideration of building characteristics, emergency plans, and the fire environment. Relevant building characteristics that should be identified and reviewed relative to the impact on human behavior and responses include the following: • • • • • • • •

Building type and use Physical dimensions Geometry of enclosures Number and arrangement of means of egress Architectural characteristics and complexity Lighting and signage Emergency information systems Fire safety systems

The responses and behavior of occupants during fire emergencies can be influenced to a small or great extent by the various building characteristics. Certain characteristics, such as number and location of exits, should be quantified because they will directly impact the movement of occupants to a place of safety. Other building characteristics may not be readily quantified but may need to be considered in qualitative terms such as whether exit facilities are direct and obvious or complex and unfamiliar. With regard to fire emergency plans, several facts should be established relative to the building’s occupants: • Does a fire emergency plan exist? • Is the fire emergency plan for a full or partial evacuation? • What is the nature of evacuations (sequenced, zoned, staged)? • Which occupants are familiar with the evacuation plan? • Who participates and how often do evacuation drills occur? • What evacuation times are associated with drills? • Are there injured or disabled occupants in the building? • Are there trained individuals to assist during evacuation (e.g., schoolteachers, staff in a hospital, fire wardens)?

Answers to the above questions should assist with both qualitative and quantitative decisions and judgments relative to assumptions about human response and behavior in a fire. For example, a fire emergency plan may exist but may not be routinely practiced, or is familiar only to a limited number of occupants in a building. If occupants have had no practice with the emergency plan, it should not be expected that they would adhere to the plan or benefit from the plan in an actual emergency. Conversely, if large numbers of employees are routinely trained and drilled on evacuation procedures, then the evacuation data associated with drills can be useful and appropriate for understanding movement in evacuation analysis. The analysis and prediction of human behavior during fire emergencies requires a systems view of the people–building–fire environment. People will respond to fire cues differently depending on their familiarity or experience with the building and their perception of the fire. People will also respond differently because of their awareness of past emergencies. People may only be aware of certain aspects of the building–fire environment. As people assess a situation as dangerous, they are likely to focus their attention on only a few salient pieces of information.1 People’s perceptions and interpretations are important determinants of how they may behave during a fire emergency. Ambiguous information and perception can cause people to misinterpret a situation: ignore signs of fire danger, evacuate through smoke when they would be better off taking refuge, or select inappropriate egress routes. Conversely, the provision of complete and accurate information will enhance human perception and interpretation allowing for appropriate decision making: recognizing quickly that there is fire danger, evacuating or taking refuge as necessary, or selecting the safest and most direct egress route. Perceptions are based on both physical and social factors during a fire emergency. Physical perception involves information provided to people by everything in the environment except other humans. Social perception involves information provided by other people. The actions that people take are influenced by other people’s actions. Sometimes physical and social perception overlap. Voice alarm systems are an excellent example;

both the social attributes of the message (voice, tone, wording) and the physical attributes (audibility and intelligibility) are important determinants of the system’s effectiveness.2 Occupant characteristics can similarly be divided into two categories: physical capabilities and cognitive capabilities. The interaction of these attributes with the building–fire environment can be classified into three categories depending on whether physical, social, or both types of capabilities predominate in influencing human behavior during fire emergencies. 1. Physical capabilities of people interact with the building–fire environment. For example, a person’s mobility limitations can complicate or prevent the use of stairs to leave a burning building. Sensory limitations can interfere with people’s abilities to become aware of a fire situation and to navigate within a building to find a safe route of egress. Apart from variations in their sensory and physical capabilities, people also differ in their susceptibility to products of combustion.3 2. Cognitive capabilities of people interact with the building–fire environment. For example, a person’s role as a waitress could lead her to decide to help her customers as she evacuates during a fire. The manner in which people pursue goals, make decisions, and perceive risks, along with the knowledge that they bring to an emergency, are all important cognitive capabilities that help determine performance. Various attributes of the built environment, including signage and alarm signals, interact with cognitive capabilities in important ways to affect behavior. 3. Cognitive and physical capabilities together interact with the building–fire environment. For example, the cognitive abilities of an intoxicated person can be so impaired that he or she fails to respond appropriately to contextual cues that there is a fire emergency. Cognitive performance limitations related to both age (the very young and very old) and disabilities can be important determinants of performance, depending on the situation. Also, the stress associated with an undefined situation or a fire emergency can affect both peoples’ cognitive and physical capabilities.

ORGANIZATION AND USE OF THIS GUIDE

smoke and gases. Use of any quantitative method or data requires consideration of the conservatism or lack of conservatism of the method or data and the context in which the data applies. The use of quantitative methods or data is appropriate to the extent that the method or data is applied with reasonable assumptions and/or engineering judgment. Table 1 illustrates which sections provide quantitative guidance and which provide qualitative guidance. The following sections of this guide each address a specific aspect of human factors important to the analysis and evaluation of evacuation time during a fire incident. Prior to using the information of any one section, it is important to understand the content of every other section and how to use the information in those sections. The user of this guide should read the following overview and guidance on use of this guide prior to applying the specific elements and methods in it. The information in the following sections is interrelated, and it is generally intended that all sections be used in concert.

This guide has been organized into four human factors subject areas that are key to understanding and estimating the responses and general behavior of people evacuating during a fire incident: 1. 2. 3. 4.

Occupant Characteristics Human Response to Cues Decision Making Movement

The approach to analysis and estimation of occupant evacuation in this guide acknowledges that no comprehensive, validated evacuation models that address behavior and movement exist. However, there is important and useful qualitative and quantitative information that can facilitate and guide the analysis and estimation of occupant evacuation. Qualitative information in this guide will be useful for defining the parameters of an evacuation scenario, identifying critical human factors, developing assumptions, and supporting engineering judgments. For some of the sections, information is available that can be used to develop quantitative estimates. Quantitative information is provided in the form of equations and calculation methods, but also in the form of data from the literature, case studies, and simulated or documented evacuations. The focus of most of the quantitative information relates to time aspects of an evacuation such as time involved in deciding to evacuate, the time related to moving through a building, and the time of exposure to

Occupant Characteristics

The occupant characteristics section identifies factors that would be necessary for considering human behavior in fire. These factors include occupant characteristics such as gender, age, physical capabilities, sensory capabilities, familiarity with the building, past experience and knowledge of fire emergencies, social and cultural roles, presence of others, and commitment to activities. To predict human reactions and behaviors during a fire emergency, the occupant characteristics of a buildingâ&#x20AC;&#x2122;s TABLE 1: Type of Human Behavior Guidance Provided population need to be reviewed to Subject Treatment identify the occupant group or groups that should be considered Occupant Characteristics Qualitative in the analysis. From the list of Human Response to Cues Qualitative discussion of occupant characteristics, a group factors; quantitative information or groups can be distinguished by provided via case studies attributing key characteristics to a Decision Making Qualitative discussion of specific group or groups. Not all factors; quantitative information characteristics need to be essential provided via case studies factors, but those that are critical and expected to influence the Movement Quantitative information provided, reaction and behavior of a group and references provided to other or groups should be noted. methods

In performing the analysis, it may be possible to rely on a single defined occupant group that is recognized as the most critical and is conservatively characterized. However, it may also be necessary to perform additional analysis when several identified occupant groups in a given building are distinguished by their varying characteristics.

Human Response to Cues Occupant Characteristics Decision Making Fire Environment Impact

Human Response to Cues

Movement

The section addressing human response to cues provides guidance on estimating the time from when fire cues FIGURE 2: Interrelationship of Human Factors are perceptible to when the cues are interpreted as requiring a response. The psychonot. A determination should be made if peoplelogical process of cue validation involves receiving alerting cues will be used in the assessment or if cues, recognizing cues, and interpreting cues. such cues are expected to occur and enhance the After occupant group(s) and the associated occucue validation process. pant characteristics are identified, that portion of Building services disruptions, such as power the evacuation or response timeline (see Figure 1) failure, can create cues for occupant groups in areas referred to as the cue validation time can be evalunear or remote from a fire location. For occupants ated. Cues that should be considered and assessed remote from the fire, such cues may be disregarded relative to the occupant group(s) include fire cues, but may also begin the cue validation process. Such building signaling or public address systems, cues cues may be important in this process, but may not from people alerting others, and cues from building always be an expected cue that readily initiates services disruptions. evacuation. Building services disruption cues need Fire cues are generally present in the area of fire to be considered in terms of their expected freorigin, but may also be present in other areas of a quency, magnitude, and location relative to occubuilding due to the spread of smoke and/or flame pants in a building. spread. Fire cues may be a first cue to some occuIn the assessment of cues, one or several cues pant groups but not others and may trigger the initimay be applicable to an occupant group. The ation of other cues such as people alerting others. assessment should determine if one or several cues Automatic cues may result from building signalare applicable for any given occupant group. Given ing or public address systems. In the case of fire the selection of appropriate cues, the reaction to the alarm systems, the presence of automatic detection cues and time for validation of the cues must be and notification equipment needs to be established. established using available research data, case histoThe capabilities, extent of coverage, and efficacy of ries, decision models, or engineering judgment. any notification system need to be considered in the context of the occupant group(s). If detection/notification equipment is present in a building, a decision Decision Making needs to be reached on how to consider the operaThe decision-making section gives guidance on tion of the equipment in the evacuation assessment. determining some of the types of actions that people The effectiveness of visual and verbal cues might take after they have validated the cue or cues from other people varies with occupant characas being a fire incident. The types of actions include teristics. Some types of occupants may be readily seeking additional information, searching for others, influenced by alerting actions while others may

notifying others, fire fighting, beginning to evacuate, or continuing with activities in which they were previously engaged. After the cue validation time period, it is possible that occupants will begin to proceed directly to building exits; however, such movement may often be delayed as the result of occupant decisions to perform other actions deemed important. The decision-making section reviews the types of factors that may result in additional time before definitive movement to safe areas and those factors that may influence delays in movement during an evacuation. This section provides information regarding various case studies; other case studies other than those identified in this guide may also be relevant to evaluating the decision-making time period. The decision to use information from case studies must be based on the relevance of the case study to the building, occupant group(s), and evacuation plans under consideration. For example, where time data from a case study addresses a particular building, such data might or might not be appropriate and directly applicable to a similar building. In deciding on the use of case studies, the analyst should review the context of the case study for relevance to the scenario being addressed.

In the more likely event that no directly relevant movement time data is found, then a suitable calculation method or model needs to be selected and used to estimate movement time. Assumptions of the model and the analysis need to be identified. Where engineering judgment is applied to aspects of the analysis or calculation factors, justification or basis for the engineering judgment should be provided.

Fire Environment Impact When considering human response to fire, it is frequently necessary to consider the manners in which human behavior would be modified by exposure to fire products. The types of effects that the fire environment can have include reduced visibility, breathing difficulty, fatigue, or incapacitation. Guidance in this area is provided in the SFPE Handbook of Fire Protection Engineering,4 an ISO Technical Specification,5 and an NIST Technical Note.3

Context in Design and Analysis Process The SFPE Engineering Guide to PerformanceBased Fire Protection Analysis and Design of Buildings 6 identifies a process for performancebased design. Human factors should be considered in several parts of the design and analysis process. During the development of the project scope, the occupant characteristics should be determined. The occupant characteristics should be further evaluated during the development and analysis of design fire scenarios. Occupant characteristics include the number of people who might be present, the activities in which they might be engaged, their relationships with others, etc. For more information on the types of occupant characteristics that should be considered, see the section of this guide on Occupant Characteristics. As the project goals are developed, a determination will be made as to whether or not protection of people will be contemplated by the design. Designs where the goals only relate to property protection, mission continuity, or environmental protection, such as might be the case in an unattended storage or warehouse facility, will typically not require the types of analyses identified in this guide.

Movement The movement section addresses the period of time after the decision is made to evacuate or relocate. It provides guidance and quantitative methods for estimating the time for occupants to move to a place of safety or refuge. Additionally, this section addresses factors (smoke in egress paths, familiarity with the building, etc.) that might affect the route chosen. This section describes the quantitative factors, calculation procedures, and modeling approaches that can be utilized to estimate the travel or movement time of occupants. In addition to performing mathematical calculations, the engineer should determine if there is available evacuation drill or case study data and determine if the data are relevant to the context of the building and occupants during a fire incident. Such data may be appropriate as the definitive bases for occupant movement time or may be useful in validating mathematical calculations of movement time.

As trial designs are developed, the section on Human Response to Cues can be used to develop designs that have the intended effects of notifying people of the presence of a possible fire or in facilitating the intended actions. Similarly, the information in the section on Decision Making can be used to predict the types of decisions that people will make in the event of a fire and ensure that the design strategy used will incorporate elements that will provide adequate protection. As designs are evaluated, guidance in the section on Movement can be used to estimate the time that it will take people to move through egress paths. Coupled with the information in the section on Decision Making, which can be used to estimate the time that people will spend on pre-evacuation activities, realistic evacuation time predictions can be made. The classical human behavior analysis compares the “required safe egress time,” or RSET, to the “available safe egress time” or ASET. For a design to be acceptable, ASET has to be greater than RSET, after accounting for uncertainty. This guide provides information for predicting RSET. While the predictions of fire phenomena are outside the scope of this guide, other SFPE documents provide guidance in the area of ASET analysis.

Occupant Characteristics The factors that must be considered when predicting human behavior in fire include occupant characteristics such as gender, age, physical capabilities, sensory capabilities, familiarity with the building, past experience and knowledge of fire emergencies, social and cultural roles, presence of others, and commitment to activities. To predict human reactions and behaviors during a fire, the occupant characteristics of a building population need to be reviewed to identify the occupant group or groups that are important in the analysis. Using the list of occupant characteristics, a group or groups can be distinguished by their key characteristics. Not all characteristics are essential factors, but those that are critical and expected to influence the reaction and behavior of a group or groups should be noted. In performing the evacuation analysis, it may be possible to rely on a single defined occupant group that is recognized as the most critical and is conservatively characterized. However, it may also be necessary to perform additional analysis when several identified occupant groups in a given building are distinguished by their varying characteristics.

POPULATION NUMBERS AND DENSITY The code occupant load of a room is the maximum number of persons anticipated to be present for a given configuration or use. Where there is no other information available, the number should be estimated according to use, i.e., by dividing the area of the room or story by an appropriate occupant load factor. However, where actual occupancy load data are available for similar occupancies, these may be used. Potential changes in occupancy or use need to be considered. Conservative design requires use of the maximum potential occupant load. Situations exist where codified information is not sufficiently accurate for the particular design under consideration.7 In such circumstances, the designers should access other data sources or generate the data by carrying out surveys of similar premises. Designers should be mindful that the numbers and distribution of occupants in a building will change with time and activity. In a hotel, during the night

Cautions and Limitations Users of this guide must exercise caution when applying the research cited in this guide to fire protection design in buildings. Caution must be exercised when the results of research are used outside the bounds of the original research. In addition, the assumptions on which the research is based must be evaluated using good engineering judgment prior to applying the research in a building design. Several examples are provided within this guide to illustrate the concepts presented. The data that are used in these examples may be specific to the example and may not be applicable to all situations. It is the responsibility of the engineer performing an analysis to ensure that any data used in calculations are appropriate for the scenario being considered.

every sleeping room in the hotel might be occupied while during the day the sleeping rooms may be occupied by only a handful of housekeeping staff as they do their dayâ&#x20AC;&#x2122;s work. This occurred in the Vista Hotel during the bombing and fire at the World Trade Center in 1993.8,9 Similarly, an arena may have its stands filled with families during a performance, but have its stands empty and its floor space filled during an exhibition. The number of people using a building or space and their distribution or density in that space will affect travel speeds (see the section on Movement beginning on page 26).

Since people are continually exposed to exit signs but are rarely, if ever, required to use non-familiar exits, they may have learned to filter out this information. During an emergency, it is unlikely that occupants will be prepared to try a route they have never used before to leave the building. Occupants are more likely to attempt to leave by their familiar route for which they know the location, conditions, and location of discharge. If their familiar route out of the building is judged impassable due to conditions such as smoke or crowd, then occupants might rely on exit signs to find an alternative way out. In such conditions, the exit signs will need to be highly visible and conspicuous to be distinguished from surrounding information and be easily noticed by occupants. Generally, it cannot be expected that occupants of buildings are familiar with all the emergency exits of the building unless they have been trained as to their locations and use.

ALONE OR WITH OTHERS The presence of other people will influence behavior and decision making. Response to alarms or fire cues is affected by whether people are alone or with others. The presence of other people can have an inhibiting effect on the definition and initiation of action from initial ambiguous cues.10 However, the presence of other people may increase the chances of a person being notified of an emergency and allow for group decisions of what actions to take. People who are alone tend to respond more rapidly to ambiguous cues.

DISTRIBUTION AND ACTIVITIES The evolution of an evacuation event will depend on the extent to which occupants are evenly distributed throughout the occupied spaces or concentrated in particular locations. The initial response may be affected to some extent by the activities the occupant is engaged in immediately before the fire. It is important to obtain pre-movement time data for occupants engaged in different activities (such as eating in a restaurant, shopping, watching a film or entertainment, sleeping, or working).13 The activities that people are engaged in affect their response. (See the subsections below on Commitment and Focal Point.) People who are sleeping or showering, for example, when they are notified of an emergency will need additional time to waken and dress for the outdoors. The distribution of occupants throughout a space will impact movement speeds (the more people, the denser the movement flow and the slower the walking speeds). Occupant density can also impact the ability to communicate instructions. Highly dispersed occupancies may create difficulties in communication; however, densely occupied or noisy occupancies may also hinder the ability to communicate.

FAMILIARITY WITH THE BUILDING Occupant response may be influenced by familiarity with a building and its systems. In some conditions, frequent users of a building may have a complete knowledge of the nearest and alternative egress routes and warning systems. They may be expected to make an effective evacuation, particularly if subjected to regular emergency training and evacuation drills. Infrequent users of a building, such as members of the public, will usually depend more upon signs and staff. These users may be less familiar with, and less responsive to, warning systems. Members of the public may also be more likely to attempt to leave by the route they entered the building once they have decided to evacuate unless they are directed otherwise by signs or systems. Providing exit signs to indicate egress routes does not ensure that occupants will notice the signs or will use these exits to leave the building.11,12

Groups have been found to affect the smooth merging of flows in corridors as they attempt to remain intact. For example, rather than each individual in the group merging singly into the flow, the group will try to stay together.15

ALERTNESS The involvement of people in and commitment to the activities being carried out within the building or their interaction with the other occupants of the building can affect their awareness of other circumstances. When people are in bed or asleep, their response times to a fire alarm can be expected to be considerably delayed. Drug or alcohol use also affects alertness.

ROLE AND RESPONSIBILITY The roles and responsibilities of occupants during the normal use of the building will, in an emergency, influence their behavior and the behavior of others. Sufficient, well-trained, and authoritative staff will shorten the ambiguous, information-gathering phase of pre-movement time. In the Beverly Hills Supper Club fire, for example, the wait staff remained in their roles, assisting in the evacuation of the patrons at “their” tables.16

PHYSICAL AND COGNITIVE ABILITY In many buildings, a proportion of the population may be impaired (cognitively and/or physically) or will present some level of limitation related to injury, illness, poor health, or other medical conditions. Occupants’ preexisting medical conditions may influence tolerance to fire effluents. Some of the occupants may have to rely entirely on assistance of others or may not be capable of being moved. The initial response of disabled people may involve a considerable preparation time prior to movement. The movement of disabled occupants is significantly influenced by the nature of their disability and building elements such as doors, ramps, and stairs. People with a hearing disability may require special means of notification of a fire, although their evacuation movement may not be different than mobile occupants. People with a visual disability may perceive audible information such as a fire alarm or a voice communication message but might need assistance to find a suitable evacuation route. However, once in a stairwell, they might move independently at the speed of the group.

LOCATION Individual occupant responses are influenced by their specific location in relation to the fire, the warning system, and the escape routes. Location can influence the time to notification, comprehension of the alert signal or message, and actual travel distance.

COMMITMENT People are action- and goal-oriented. They have reasons for being in a particular place. Those reasons may continue to guide their behavior even when an emergency occurs. People who have committed a significant amount of time to their activity (for example, waiting in a queue for service or waiting for a meal in a restaurant) will be reluctant to respond to an alarm signal if that means that they must lose their place in line or walk away from a meal.14,17

SOCIAL AFFILIATION The behavior of occupants will be significantly influenced by whether they are alone or with a group. Sometimes this contributes to people starting to move more quickly in response to fire cues. This does not necessarily result in direct movement by separated group members toward the nearest exit route since they are likely first to attempt to reestablish the group.14 In addition, the speed of movement will often be dictated by that of the slowest member of the group.

FOCAL POINT If the setting has a particular focal point, such as a stage in a theater, the population of the building will normally look to that point for guidance in the first moments of an alarm and evacuation. The focal point can be used positively in an emergency, recognizing that people will often look to the stage, lecturer, etc. for “permission” to leave or indication of appropriate exit routes to use.

OCCUPANT CONDITION

of gender and the effects of role can be muddied in some older studies. Further studies should be done to help differentiate the influence of gender and the influence of role as society changes and genderassigned roles are less prevalent.

Throughout the evolution of the fire, occupantsâ&#x20AC;&#x2122; ability to respond will depend on their cumulative and instantaneous exposure to fire effluent.

GENDER

CULTURE

In general, females are more likely to alert or warn others and evacuate in response to fire cues than males.18,19 Similarly, in the residential environment, it has been found that men were more likely to fight the fire while women were more likely to gather family members and call the fire department.19,20 A study of occupant behavior in health care facilities also found a higher likelihood of males attempting to fight the fire while female staff were more likely to take protective actions and rescue patients, consistent with their training and assigned responsibilities.21 The re-entry behavior that was observed among men in residential fires was noted to frequently agree with the social role of the male as protector of the family.22 Some of these studies were undertaken more than 20 years ago, and to some degree the findings may have been influenced by a more rigid assignment of roles by gender than exists today. For example, the Project People II final report21 notes that the gender distribution of the participants in the health care study reflected the distribution of occupations among the participants: the study group was largely made up of women, and the predominant occupation among the participants was nursing. The protective actions taken by the women in the study may have been a reflection of their role as nurses (and caregivers) rather than their gender. Today, more and more nurses are men, many women work as security officers and medical doctors, and a health care study done today might show that the protective actions are taken by nursing staff while the firefighting roles are taken by security staff, and gender may be less relevant. Similarly, in residential fires, the protective actions that have long been assigned to women may actually be shared more equally in households where parental responsibilities are shared. The above discussion, however, serves only as a reminder that users should be aware that the effects

There is limited research available on the influence of culture on evacuation performance (one such resource is provided by Ozkaya23). However, culture can be assumed to influence factors such as social affiliation and role and responsibility, given research from the field of organizational behavior.

AGE Variations in evacuation performance as a function of the age of the individual can be expected and should be accounted for. With the aging population trend, this factor bears stronger consideration, especially considering that data for developing evacuation times was typically developed using collegeage students. Kose 24 categorizes the expected performance differences using three categories: 1. Sensory skills 2. Decision making 3. Action (mobility, swiftness, etc.) These three categories can then be expected to affect all evacuation phases, from recognition through actual movement throughout the structure and decision making along the way. Furthermore, this influence can be expected not to be monotonic, but rather with evacuation performance degradation for the very young and the very old. Unfortunately, while the effect of limitations in each of the three categories on evacuation performance can be somewhat readily quantified (for example, in factors such as physical and cognitive ability), the linkage between age and these degradations is less established. Age can also be expected to influence the ability of an individual to withstand exposure to byproducts of fires. Specifically, the very young and the elderly in poor health may be less able to withstand the debilitating effects of smoke and heat.3,4,5

OTHER FACTORS

• Cues from people alerting others • Cues from building services disruptions

Other factors influence people’s behavior that are more environmental than occupant-based also should be considered. These include external factors such as weather, which can inhibit evacuation (e.g., reluctance to exit a building during a thunderstorm or heavy rain, causing a bottleneck at the exit door) and/or increase pre-movement delay time (e.g., additional time to dress for cold winter weather). Other factors to consider are occupant reaction to uneven floor or wall surfaces, complex evacuation routes, lighting levels and noise levels of alarms, or use of strobe lights in corridors.

Fire cues are generally present in the area of fire origin, but may also be present in other areas of a building as a result of the spread of smoke and/or flame spread. Fire cues may be a first cue to some occupant groups and may trigger the initiation of other cues such as people alerting others. Automatic cues may result from building signaling or public address systems. In the case of fire alarm systems, the presence of automatic detection and notification equipment needs to be established. The capabilities, extent of coverage, and efficacy of any notification system need to be considered in the context of the occupant group(s). If detection/notification equipment is present in a building, a decision needs to be reached on how to consider the operation of the equipment in the evacuation assessment. The effectiveness of visual and verbal cues from other people varies with occupant characteristics. Some types of occupants may be readily influenced by alerting actions while others may not. A determination should be made if alerting cues will be used in the assessment or if such cues are expected to occur and enhance the cue validation process. Building services disruptions, such as power failure, can create cues for occupant groups in areas near or remote from a fire location. For occupants remote from the fire, such cues may be disregarded but may also begin the cue validation process. Such cues may be important in this process, but may not always be an expected cue that readily initiates evacuation. Building services disruption cues

Human Response to Cues INTRODUCTION This section provides guidance on estimating the time from when fire cues are perceptible to when the cues are interpreted as requiring a response and evacuation or refuge-seeking decision making begins. The psychological process of cue validation involves receiving cues, recognizing cues, and interpreting cues (RRI). After occupant group(s) and their associated occupant characteristics are identified, the portion of the evacuation or response timeline referred to as the cue validation time can be evaluated. Cues that should be considered and assessed relative to the occupant group(s) include: • Fire cues • Building signaling or public address systems

<START TIME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . END TIME> FIRE / CUE INITIATION OR DEVELOPMENT

Cue Validation Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . and Continuing Process Receiving Cues

Recognizing Cues

Interpreting Cues

Receiving, Recognizing, Interpreting (RRI)... RRI...RRI...

Decision-Making Period Pre-movement Decisions

Transmovement Decisions Movement / Refuge Time

FIGURE 3. Cue Validation Process

need to be considered in terms of their expected frequency, magnitude, and location relative to occupants in a building. In the assessment of cues, one or several cues may be applicable to an occupant group. The assessment should include a determination of whether one or several cues are applicable for any given occupant group. Given the selection of appropriate cues, the reaction to the cues and time for validation of the cues must be established using available research data, case histories, decision models, or engineering judgment. During a fire, factors such as form of ignition, type of fuel ignited, location of fire, architectural features, room geometry, and fire suppression systems will all have their effects, and these effects will take place in time regardless of human awareness. At some point during the fire event, people will become aware of cues that indicate a fire is in progress. The fire will develop as a function of the fuel, ventilation, and enclosure characteristics. In many cases, time elapses between ignition and the first possibility of a person receiving a cue. Human response to cues of fire must be considered in place of the assumption that people will respond immediately. There is considerable evidence in the literature that occupants do not respond within seconds to the initial cues of a fire.13,25,26,27 The time people will spend receiving the cue, recognizing the cue, and interpreting the cue is called â&#x20AC;&#x153;cue validation time.â&#x20AC;? Estimating or quantifying the cue validation time is one component of the human factors to be considered. The cue validation time involves a delay before response that needs to be included in any fire safety analysis and design. This delay time is a component of the decision-making process influenced by notification circumstances. Human behavior and decision take place once cues are recognized and interpreted, including the decision to take no action. The types of cues that could be available include one or more of the following:

3. Cues from others such as verbal staff instructions, observable occupant actions, or fire department arrival 4. Cues from building services disruption such as elevators not working and other utility failures For life safety performance analysis, the following cognitive processes should be considered. All of these apply to the four cue types: 1. Receiving the cue (sense the cue) 2. Recognizing the cue (identify the cue) 3. Interpreting the cue (give meaning to the cue) The intent of describing human response to cues is to estimate the time from when fire cues are available to be perceived by occupants to when the cues are interpreted as requiring a response and decision making begins. The dark shaded area of Figure 3 is a depiction of the process of human response to cues. The initial cues may be direct signs of the fire such as visible smoke or flames or events resulting from a fire such as a warning from other occupants, fire alarm, or power outage. When these cues or indicators of the fire reach human senses, the cue(s) validation process begins, which continues for the duration of the event. For any given cue(s), the cue validation time includes the time to receive the cue(s), recognize the cue(s), and interpret the cue(s). At some time after cue(s) are received, there is recognition of the cue(s) (e.g., the smell is smoke) and finally interpretation of the cues (e.g., smoke means something is burning). The cue validation point or points are those moments in time when a cue or cues are interpreted as requiring a response (e.g., investigate the cue, find family members, or evacuate or relocate). To estimate the occupant response time, it is essential to include the delay time related to onset of cues up to the time the occupant will decide to begin evacuation or relocation action. One important determinant of a personâ&#x20AC;&#x2122;s response in an emergency is what the person was doing at the time of occurrence of available cues. For example, if a person is asleep or watching a movie, the person may not be aware of or pay attention to initial cues.

1. Fire cues such as heat, smoke, or crackling sounds 2. Cues from building signaling systems such as a fire alarm system

Notable in reviewing fire and human behavior studies is the scarcity of data on occupant delay time to start evacuation or relocation. For instance, there is some information available concerning human behavior sequence of actions and probabilistic models to predict likely next action. There is also some information relative to the overall time involved from receiving the first cues to the time occupants reached safety after actual fires. However, there is limited information available concerning the time people spend to make individual decisions. To estimate the cue validation time, it is necessary to consider the type of cue received and the cognitive processes the person will have to engage in to validate the cue(s) in order to decide that the situation is an emergency that requires a specific response.

fire can also emit auditory cues such as crackling sounds. The heat from the fire is also another potential cue that could be felt through the skin. In a study looking at the waking effectiveness of smoke odor, it was found that very few people are actually awakened from sleep by an odor stimulus. Further, only 20% of the subjects tested during their sleep (at stage 2, which is marked by muscle tension and gradual decline in vital signs) were awakened by a smoke odor.28 In a study of 17 young adult volunteers with ages ranging from 18 to 26 (mean = 21.4) with self-described “normal” smell and sleep patterns, 29% of males and 80% of females woke when exposed to a smoke odor. The smoke odor concentration at the pillow was approximately 1 ppm at 1 minute and 6 ppm at 10 minutes. The subjects who awoke in response to the odor did so between 45 and 205 seconds after emission of the odor began, with a mean time of 101 seconds (SD = 56 seconds.)29

RECEIVING CUES The building characteristics will have an impact on the delivery of cues to occupants. Factors such as ambient noise, odors and light, degree of compartmentation, ceiling heights, ventilation conditions, etc., can enhance or deter cue delivery. Occupant characteristics will be a major determinant of the individual’s potential for receiving a fire cue. A person asleep or under the influence of alcohol or other drugs might be oblivious to all the fire cues. There is also variation in sensitivity to smells and heat as well as differences in ability to see smoke. An important factor is a person’s commitment, which can inhibit the ability to perceive these cues. For example, a person working intently at the computer might not notice smoke filling up a room while a person entering that room could readily perceive the unusual cues.

Waking patterns to light and auditory cues similar to those that might be emitted by a fire were tested in a sleep study involving 33 adult volunteers with normal hearing and sleep patterns aged 25 to 55 (mean = 43 years). Ninetyone percent woke to a crackling sound, and 83% awoke to a shuffling sound. The sound was in the range of 42 to 48 dBA at the pillow, and 83% of the people who woke up in response to the cue did so within 30 seconds. The light source used in the experiment was in the range of 1 to 5 lux. Forty-nine percent of the subjects awoke in response to the light cue, and 50% of those did so within the first 30 seconds.29

1. Fire Cues Fire cues are specifically related to the fire and the combustion process. Fire cues include olfactive cues such as the smells of smoke, burning, strong plastic smells, acrid smells, etc. There are also visual cues, such as the visible smoke that can be any shade from white to black and of different density. Other visual cues can also be sparks or flames. The

2. Building Signaling Systems Cues from the building signaling systems could include audible alarm signals such as bells, chimes, horns, and automatic voice messages, and visual alarm signals such as strobes, flashing signs, messages on information screens, etc.

Receiving the building signaling cues will depend on the occupants’ perceptual capacity and the perceptual characteristics of the cues. The audibility and intelligibility of audible signals, as well as the visibility of the visual signals, will determine if occupants can receive these cues. People do not tend to immediately begin evacuation in response to cues from building signaling systems. “The code alarm provisions generally assume fire alarm systems will enable building occupants to initiate effective emergency egress behaviors with minimal delay. Almost without exception, however, the technical literature contradicts this notion and supports the idea that an alarm signal alone does not generate immediate action—but starts the search for ‘confirmation of a second clue.’”30

A study investigated waking in response to light.33 Thirty normally hearing college students and 48 deaf subjects were used, and the light sources were strobes, industrial strobes, and 100 watt incandescent lights. White flashes were perceived as the brightest, and the study concluded that “clearly, signal intensity is a major factor in arousing sleeping people.” Household Industrial Strobe Strobe

In an experiment conducted with 36 subjects aged between 6 and 59 with normal hearing and sleeping patterns, 100% of adults (aged 30 to 59) awoke to a 3-minute, 60 dBA (measured at the pillow) smoke alarm, while 85% of children aged 6 to 17 years old slept through the alarm.31

Incandescent

Deaf subjects awakened

86%

92%

Hearing subjects awakened

82%

78%

59%

Deaf subject mean waking time

16 sec

15.9 sec

31.2 sec

Hearing subject mean waking time

23.9 sec

13.4 sec

53.5 sec

3. Cues from Others Cues from others include verbal alerts as well as observations of the behavior of other people. Such cues obviously will only be available in spaces where other occupants are located or through which other occupants will pass.

An experiment with subjects aged 19 to 29 years examined sleep–waking performance in response to fire alarm signals and environmental noise.32 Alarm sound pressure levels of 70 to 85 dB (at the head of the bed) awakened 20 of 20 subjects. The average waking time in response to a sound pressure level of 70 dB and 85 dB was 9.5 seconds and 7.4 seconds, respectively. Alarm sound pressure levels of 55 dB awakened 80% of the subjects, and the average waking time was 13.6 seconds.

4. Cues from Building Services Disruption Cues from building services disruption include power failures, signals warning of interruption of power supplies, sudden shutoff of services such as air handling units, etc. Such changes in the environment have to be noticeable to occupants in order to be perceived as cues at all.

RECOGNIZING CUES The process of recognizing cues involves the cognitive activity of comparing the information perceived to knowledge from past experience. A person may recognize a cue for what it is, mistake the cue for something else, or not recognize the cue at all. The audibility and intelligibility of audible signals, as well as the visibility of the visual signals, will determine if occupants can receive and understand these cues.

Occupants or staff who have been trained for fire response through fire drills should readily recognize fire alarm signals.

1. Fire Cues Recognizing fire cues will depend on the person’s knowledge and past experience with similar fire cues. Having been through a fire event or simply having experienced very controlled fires in fireplaces or bonfires forms part of the knowledge people have about fires. A person does not need to have been through an actual fire to recognize fire cues.

3. Cues from Others The ability to recognize cues from others will depend in part on the occupant’s knowledge of the other people and related confidence in information that their words or activities may convey. For example, if a neighboring worker is prone to become agitated over trivial events, another occupant may not heed any cues that might have been apparent under other circumstances and might not recognize a warning for what it is. Unusual activities, such as general movement toward exits, or increased levels of conversation mentioning a perceived threat might also alert an occupant to the cues. Arriving fire trucks may also alert occupants to the possibility of a fire or emergency.

2. Building Alarm Signaling Systems Prior exposure to a specific alarm system or to alarm systems in general leads to learning. Based on prior training and experiences, a person will be more likely to recognize the signal.27,34 In a public building when a fire alarm goes off, occupants may not necessarily recognize the signal as a fire alarm because there is a variety of warning signals in the built environment and it is very difficult for the public to discriminate among these different signals.35

4. Building Services Disruption A study conducted in Canada with 307 visitors to public buildings, who had not received any training regarding the three-pulse temporal (temporal-3) pattern evacuation signal, found that only 6% recognized or associated the temporal-3 pattern with a fire or evacuation signal. Though the temporal-3 pattern signal can be used with any sound for the evacuation signal, including all of the sounds used in the study, only a single electronic tone having a fundamental frequency of 505 Hz was used for purposes of evaluating recognition of the temporal-3 pattern. Other sounds tested included a car horn (98% recognition as a car horn), vehicle back-up alarm (71% recognition as a back-up alarm), bell (50% recognition as a fire alarm), slow whoop (23% recognition as a fire alarm) and a buzzer (2% recognition as an industrial buzzer). On an arbitrary scale where people judged the perceived urgency conveyed by the signals, the temporal-3 pattern signal ranked the lowest. Results show that public education is essential to increase the perceived urgency and recognition of the temporal-3 pattern in order for occupants to readily identify the evacuation signal.36

Building services disruption, such as power outages or breaking glass, may not be recognized as a fire cue, but might make subsequent fire cues more recognizable as such.

INTERPRETING CUES At this stage of the process the person is giving meaning to the cue perceived. This meaning can be accurate or not, and it may change over time as new cues are processed. The interpretation of the cue will determine the response applied by the person.

1. Fire Cues Interpreting a fire cue as “it is truly a fire” or “it is threatening to me” are two different interpretations that could lead to different decisions. Acknowledging that the smoke a person sees is actually a fire may result in a firefighting response while the interpretation of a personal threat is more likely to lead to protective action on location or moving away from the danger.

2. Building Signaling Systems

Two hundred thirty-eight unannounced fire drills were conducted in U.S. Veterans Affairs facilities in the United States.34 An experimental procedure for measuring health care staff “response delay” and a 48-question participant survey were used to gather information for multiple regression analysis. Correlations between the independent variable “time to action” and the 48 factors revealed two findings with moderate correlation coefficients:

Recognizing an audible signal as a fire alarm does not necessarily mean that there is indeed a fire. The interpretation of the signal could be that it is a system test or fire drill. Interpreting the sound of a fire alarm as a genuine warning of an actual fire incident is more likely when other cues are perceived by the occupant. The likelihood that a warning system will be interpreted as intended depends on its credibility. In interpreting building signals there is a tendency for people to either ignore the warning or to investigate the reason for the alarm to evaluate whether building evacuation is necessary. These activities mean a substantial loss of time for escaping from danger. Breznitz, in his book Cry Wolf Syndrome,37 states that the detection component of a warning system is electrical or mechanical; the management of information gained, as related to occupant notification, must consider the psychological aspects. Breznitz further states that, to avoid potentially irreversible negative consequences of false alarms, one can design warning systems to better manage the signal.

1. Larger fire alarm zone systems correlated with longer staff response delays. 2. Lower confidence in the fire alarm system correlated with longer staff response delays. The following factors also correlated, in order of importance, but explained a very small percentage of variance in response delay: • Coded signal systems correlated with longer response delays. • More fire protection systems and equipment within the zone correlated with longer response delays. • More alarms over the previous 6 months (all alarms) correlated with longer response delays. • More false alarms over the previous 6 months correlated with longer response delays. • If a staff person’s location involved direct patient care, there were longer response delays.

ANSI S3.41-1990, Audible Emergency Evacuation Signal The National Fire Alarm Code38 (2002 edition) specifies in section 4.4.3.6 that fire alarm signals shall be distinctive in sound from other signals and further specifies in section 6.8.6.4 that such signals used to notify occupants of the need for evacuation shall be in accordance with ANSI S3.41, Audible Emergency Evacuation Signal (also an international standard, ISO 8201). The goal is to have anyone hearing the signal in any building immediately recognize the signal as a fire alarm evacuation signal. The use of the distinctive three-pulse temporal pattern fire alarm evacuation signal required by section 6.8.6.4 became effective July 1, 1996 for new systems installed after that date.

Response delay times were measured for the fastest, or first responding, staff person within the zone alarmed. The staff in the study were highly trained with a high sense of responsibility to react quickly and properly. In 75% of 238 cases, no-one responded within 6 seconds. The average response delay for all cases was 27 seconds. Where a continuous bell was used, the average response delay was 13 seconds. Where automatic voice was used, the average response delay was 18 seconds. Where a coded signal was used, the average response delay was 31.6 seconds. If a coded signal was used without a concurrent “code red” announcement, the average response delay was 51.7 seconds.

Proulx suggests that the messages should contain the following information:

Ramachandran conducted two experiments to determine which types of alarm notification are most likely to be interpreted as a fire. The results indicated that providing more detailed information, such as by voice, graphic display, or text, resulted in as much as a six-fold increase over a bell in interpreting notification as indicative of a fire.39

• Identification of the problem • Location of the problem • Instructions Keating and Loftus41 summarize characteristics that would be desirable in voice alarm systems that were identified at a 1971 conference on high-rise fire safety:

3. Cues from Others Interpreting cues from others will depend in part on the credibility of the people generating the cue. A person who repeatedly raises a false alarm will not be credible. The failure of others to take action can also have an inhibiting effect on building occupants. Studies have shown that people have a reluctance to act (e.g., interpret cues as threats) if others around them do not.10

• The system should provide precise instructions under varying emergency situations. • Instructions should be capable of varying in different portions of the building so that people can be provided with the information that is tailored based upon the actions that it is desired that they take. • Alerting tones should precede voice instructions to capture people’s attention. • Pre-recorded messages can be used for preplanned situations. • Pre-recorded voice messages can be automatically activated in response to signals from manual or automatic initiating devices. • The system should be capable of being manually used to deliver instructions.

4. Cues from Building Services Disruption Building services disruption may not be interpreted as a fire cue, but might make subsequent fire cues more recognizable.

IMPACT ON FIRE PROTECTION ENGINEERING DESIGN AND ANALYSIS

Keating and Loftus suggest that the U.S. Federal Communications Commission warning signal be used as the alerting tone. The then-existing FCC signal was a 1000 Hz pure sine wave tone, and, as Keating and Loftus state, the human ear is most sensitive in the 500 to 3000 Hz range. Following the alerting tone, Loftus and Keating suggest that a female voice provide an introduction, and a male voice provide detailed instructions. Keating and Loftus also note that repetition of instructions will facilitate understanding and recall. In health care occupancies, where staff is trained to assist with notification and evacuation of occupants, Cable34 suggests that, to gain maximum life safety benefit from a fire alarm system, the key is to only alarm those individuals who need to take action. Others can be alerted, but not alarmed. A fire alarm system that maintains its early warning function, and yet by design notifies occupants in a way to prevent unnecessary alarms outside the

Proulx35 suggests a number of strategies to improve the likelihood that building occupants will correctly interpret fire alarm cues: • Tell occupants the truth in as much detail as possible. Canter40 notes that people tend to underestimate the speed of fire growth. Since people will tend to try to obtain additional confirmation when they sense ambiguous information, providing as much detail as possible will minimize the time that it takes before people begin to take the appropriate actions. • Immediately use voice instruction to notify occupants once a fire emergency is confirmed. Proulx suggests that live voice instruction over a public address system is superior to the use of pre-recorded messages since the instructions can be tailored to the situation. These forms of notification will facilitate the provision of the most detailed information.25

perform other actions they deem important.13 This section reviews the types of factors that may result in additional time before definitive movement to safety and those factors that may influence delays in movement during an evacuation. This section provides information from various case studies; other case studies may also be relevant to evaluating the decision-making time period. The decision to use information from case studies must be based on the relevance of the case study to the building, occupant group(s), and evacuation plans under consideration. For example, where time data from a case study address a particular building, such data may or may not be appropriate and directly applicable to a similar building. In deciding on the use of case studies, the context of the case study must be reviewed for relevance to the scenario being addressed. When a fire occurs, occupants have to decide if an evacuation is required or necessary. Some regulations and models are predicated on the expectation that, during a fire, people will move quickly toward a perceived or identified area of safety. Although it is often presumed that upon hearing the alarm signal or perceiving some smoke occupants will leave the building immediately by the nearest exit, the analyses of people’s behavior in some occupancy situations have revealed a fairly different picture.22,42,43,44 Delays before deciding to evacuate, time spent looking for others or gathering personal items, as well as attempts made to move toward the fire and fire fighting, are behaviors that have been observed repeatedly.

danger zone(s), would likely provide the maximum life safety. In occupancies where children are present, reliance on audible alarms to wake children will likely not be sufficient. It will be necessary to use adults, who can be depended upon (if they are not impaired), to notify children so that they can take appropriate actions.

CONCLUSION Once an occupant has received the cue, recognized it for what it is, and interpreted it as a warning of fire, the next stage, “decision,” is reached. An iterative process then begins where additional cues are received and validated, and the occupant begins a process that will result in evacuation or defend-in-place.

Decision Making of People Facing a Fire The types of actions that people have taken after they have validated a cue or cues as indicating a fire incident range from seeking additional information, searching for family members, notifying others, fire fighting, beginning to evacuate, or doing nothing— continuing with activities in which they were previously engaged. After the cue validation time period, it is possible that occupants will begin to proceed directly to building exits. However, such movement may often be delayed as the result of occupant decisions to

<START TIME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . END TIME> FIRE / CUE INITIATION OR DEVELOPMENT

Cue Validation Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . and Continuing Process Receiving Cues

Recognizing Cues

Interpreting Cues

Receiving, Recognizing, Interpreting (RRI)... RRI...RRI...

Decision-Making Period Pre-movement Decisions

Transmovement Decisions Movement / Refuge Time

FIGURE 4. Decision-Making Period

starting evacuation or in taking protective action. Once the situation has been defined as a fire, it does not necessarily imply that people will evacuate right away. Time could be spent gathering family members and pets, getting dressed according to the weather, or finding keys, a wallet, or a purse.

“PANIC” BEHAVIOR The first common expectation about human behavior in fire is the assumption that, during a fire, occupants will panic. The possibility of panic behavior in a fire has been considered a “myth” by social scientists since the 1970s.45,46,47 Although the media are very fond of this concept for its drama and sensational connotation, there is little evidence of panic in actual fire situations. It is a widespread misconception to believe that people caught in a fire will panic and try to flee in a stampede, crushing and fighting others. Such crazed behaviors are extremely rare; in fact, altruistic behavior is the norm. Panic, which supposes irrational behavior for a situation, is atypical of human behavior in fire. On the contrary, people appear to apply rational, altruistic decision making in relation to their understanding of the situation at the time of the fire. In retrospect, it is easy to point to some decisions that were not optimal and played a negative part in the outcome of a fire; however, at the time of the fire these decisions were rational to the occupants when all factors were considered. It is commonly observed during interviews following fire events that victims themselves mention that they had panicked during the event. The public often use the word “panic” as synonymous for being frightened, scared, nervous, or anxious; usually it does not have the implication of irrational behavior. The limited knowledge that people have on fire development and fire dynamics does not prepare them to have the best response during fires. A majority of people who are faced with a fire situation react in a rational fashion considering the ambiguity of the initial cues, their limited knowledge about fires, and the restricted time they have to make a decision and to take action. In the initial moments of a fire, upon smelling smoke or hearing the fire alarm, it is often observed that occupants do not react and deny or ignore the situation. This seems especially true in public buildings where occupants do not want to overreact to a false alarm or a situation that is already under control.43 Such avoidance or acceptance of a dangerous situation often results in delays in

OCCUPANT CHARACTERISTICS Some occupant characteristics will have a major impact on the decision-making process. These characteristics will have an impact on the availability of the person to recognize the cues of the developing situation, and some of these characteristics will limit the number of alternatives considered by the person while making decisions for adaptive action. One of these characteristics is the occupant’s familiarity, through education or experience, with the building, the type of warning system in place, and the designated evacuation procedure. A person who has already experienced the sound of the fire alarm signal will be able to recognize this signal more rapidly than a person who is unfamiliar with this system.34 Also, a person who has participated in an evacuation drill in that building in the past will know what behavior is expected and will be able to adopt a rehearsed plan of action that may enhance the decision-making process and the evacuation behavior. Occupants may very well encounter wayfinding problems if they have to evacuate a building by a different exit than the one they are familiar with or the way they entered the building. Wayfinding is a property of the building itself since built environments can be more or less difficult to understand depending on their architectural components, circulation plan, and interior design. Several technologies being developed are intended to facilitate wayfinding during an emergency, such as photoluminescent materials,48 flashing lights,12 directional sound,49 and tactile wayguidance systems.50 Wayfinding is also related to the capacity of a person to develop a cognitive representation of a space and to use this representation for decision making.51,52

A number of evacuation drills have been conducted with regular, non-informed customers in different IKEA stores in Sweden. It was observed that shoppers attempted to exit the stores either by the entrance or by the regular exit moving along the zigzag path that took them through different departments before reaching the exit. Customers passed by several emergency exit doors without using them as long as they were not open.53

time for this evacuation was 24 seconds. This evacuation demonstrates the importance of staff training and the influence staff can have on decision making and evacuation during an emergency in a public building.54 Commitment is one of the most important determinants of decision making. Commitment has two dimensions. First, people who are deeply involved in an activity might not immediately become aware of fire cues. For example, people watching a movie, eating in a restaurant, or gambling at a casino would be committed to these activities and may not notice the sound of an alarm bell or the smell of smoke (note: smoking is still allowed in many restaurants and casinos which tends to mask the early fire-generated smoke cues). Second, once people have noticed that something is happening, they might decide consciously not to address these cues because they are committed to a specific activity. Their initial decision might be to ignore the situation until sufficient information requiring a decision is imposed on them.

Another occupant characteristic is the role played by a person in the building and the personâ&#x20AC;&#x2122;s feeling of responsibility toward the activities occurring in the building. For example, in a single-family house, occupants tend to respond right away when the smoke alarm activates because they know they are responsible to investigate and initiate adaptive action. In a public building, such as a museum or a shopping center, visitors do not assume the responsibility to initiate adaptive behavior if the fire alarm is activated. They expect that they will receive information from staff or a figure of authority if the alarm is for a valid threat. Consequently, the occupant role played in the building is an important factor to consider when examining the decisionmaking process during a fire incident. Occupants who are visitors might simply make the decision to wait and see while staff may decide to investigate the problem or provide occupants with definitive information.

In November 1987, a fire erupted on an escalator killing 31 people at the Kingâ&#x20AC;&#x2122;s Cross underground station in London, England. The fire and smoke spread in the escalator area, the ticket hall, and toward the different corridors and entrances to the station. It was observed during that fire that passengers continued their routine activity of traveling home: they entered the smoke-filled station and went down escalators, sometimes next to the visible flames, in an attempt to catch a train to go home. These people were committed to use the underground to reach a destination and were unlikely to shift their attention to ambiguous circumstances for which they felt they had no responsibility.55

The FireSERT research group at the University of Ulster conducted unannounced evacuations in Marks and Spencerâ&#x20AC;&#x2122;s department stores with neither staff nor customers aware of the exercise. One, for example, was recorded through 46 video cameras and followed with a questionnaire of approximately 300 customers. When the fire alarm bells sounded, customers looked around at the behavior of others. It is only when staff started to close their cash registers and instructed the customers to evacuate that the occupants complied and moved toward exits. Among the evacuees interviewed, 52% said they were prompt to evacuate when requested by staff. The mean pre-movement

Social affiliation is another characteristic that predicts that occupants are likely to attempt to notify and gather with people with whom they have emotional or social ties, such as family members or social or business groups, before deciding on an adaptive behavioral response. This activity of notifying or gathering members may take time, especially

if members are not together at the initial awareness of the fire incident. These activities may also involve movement toward the fire area and through smoke to gather missing members. Because of social affiliation, the initial decision might be to notify and gather members prior to any specific behavioral action of evacuation or control of the fire.

occupant situation involving other occupants and their adaptive or nonadaptive behavior.

INFORMATION PROCESSING The psychological process faced by a person during a fire can be conceptualized in terms of information processing and problem solving. The person facing an emergency has to make decisions and take actions to solve the problem of reaching safety. These cognitive processes are, however, constrained in such circumstances. In fact, the characteristics of the emergency will create psychic stress.56 Because too much stress can impair cognitive processes, it must be reduced in such a way as to allow the person to interpret the situation accurately and to take on and execute appropriate decisions. It is argued that obtaining precise information about the situation during an emergency will prompt action, reduce stress, and support the problem-solving process.

The social affiliation behavior was well documented by Jonathan Sime, who studied, among other cases, the behavior of the occupants of the Marquee Show Bar, part of the Summerland holiday leisure complex in England, where a fire killed 50 people in 1973. From interviews, it was determined that occupants in the bar who had left their children in another area of the complex started to move rapidly after the initial cues of a fire. Moving at counter flow to the people evacuating, they attempted to find their children. Groups who were intact in the bar took a longer time to react and to start leaving the premises, putting the whole group in jeopardy. This was because, where the group was not intact, members of the group decided to get the missing member(s).11

During unannounced evacuation drills in an underground station, passengers started their evacuation within 1 minute when provided with precise voice communication instructions regarding the fire event, its location, and action to be undertaken. With limited information such as the fire alarm bells only, many occupants were still in the station after 15 minutes.25

Mental alertness and limitations are other characteristics that, with age, may limit a personâ&#x20AC;&#x2122;s capacity to process information and to react in a given situation. For example, if a fire starts in the middle of the night, occupants who are not alert because they are asleep will require a longer time to respond, or may not respond at all, if not alerted by others. Another dimension is the possibility that occupants may have some limitations that will extend their decision and response time. These limitations could be perceptual, physical, or intellectual or might be due to the consumption of medication, drugs, or alcohol. Finally, very old and very young occupants may have limited physical or cognitive capacity to respond to the ambiguous cues of a fire. These occupant characteristics should be examined when considering occupant decision making. In some cases, occupants may believe that no decision is required as a result of their location or the observable fire cues available at the time. Also, the decision-making process might be biased by the

According to the classic model developed by Polya,57 problem solving involves four cognitive stages. The first stage is understanding the problem. This involves defining the situation and determining the problem to be solved. Devising a plan is the second stage. This requires looking for information, making decisions, and structuring actions. The third stage is carrying out the plan, the execution of the decision taken in the previous stage. Finally, the fourth stage is looking back, that is, assessing if the action taken worked toward solving the problem. The relative importance of each of these problem-solving stages depends on the problem to be solved.58 In the case of a fire, the first stage, â&#x20AC;&#x153;to define the situation,â&#x20AC;? appears to be the most decisive step. Often undervalued in the literature,

this first step is paramount. The time spent in gathering information in order to interpret the situation represents precious seconds, often minutes, invested in non-evacuation behavior which, by the time the situation has been assessed, could leave very little time to reach safety.59,60 Furthermore, the interpretation of the situation is important in determining further decisions. Erroneous decisions can result from a situation that has been inadequately defined.61 The time taken from the perception of the initial cues of a fire to the moment a person starts to evacuate can last from a few seconds to several minutes.13 This first stage of defining the situation is based on the available information. The information used to interpret the situation can come from different sources. It can be provided by the fire itself, taking the form of smoke, heat, noise, or flames. The building can provide information through the activation of an alarm signal or messages from a voice communication system. Other people, either staff or occupants, could also be a source of information by their communications and behavior. Along with this environmental information, the user will rely on information already known about the building and similar environments visited previously. Finally, to assess the information, the person can refer to information gathered from past experiences in dealing with emergencies. Fires vary enormously. Thus, the information available to occupants is generally characterized by its ambiguity. Smelling smoke, recognizing the sound of an alarm, or observing behavior of staff may not indicate to occupants that “there is an actual fire that requires evacuation.” Generally, the initial information available to define the situation as perceived is ambiguous because it is often contradictory, unusual, and unexpected.

A false alarm study was conducted in a college dormitory. In addition to recording numbers of false alarms, the overall effect these had on “people response” was observed. The study concluded that smoke detection systems, together with prevailing resident attitudes, may offer no life safety improvement over simple manual pull stations and local room residential smoke detectors and may actually render the building’s environment less safe because the more numerous alarms are more likely to be ignored with systems smoke detectors present. A false alarm rate of approximately one per week caused students to believe that there was no “credibility status” for the alarm system.63 On devising and carrying out a plan of action, different types of reactions are observed. When people notice ambiguous information or cues, particularly in public buildings, they usually ignore the situation or they investigate.43 Ignoring the situation and pursuing normal activities is a common reaction to ambiguous cues because it corresponds to the users’ role in a public building of not having an assumed responsibility for taking action. Consequently, people tend to maintain their role of customers or visitors, assuming that it is the staff’s responsibility for defining and reacting to the situation. Also with the presence of others, due to social inhibition and the diffusion of responsibility, occupants do not want to overreact to an incident that may be non-threatening or is already under control.10 Only when the person perceives a valid threat from the fire cues will he or she interpret the situation as requiring a decision and adaptive action. Another reaction to ambiguous information is to investigate the situation. This plan of action may involve consulting others, when others are accessible, or asking staff about the situation. The person may also initiate behavior to define the nature of the incident. This behavior often implies moving in the direction of the potential danger to gather information. These reactions to ambiguous information imply that a certain amount of time is spent either ignoring the cues or investigating to define the situation.

More than 36% of survivors of a highrise residential fire, which killed six people, mentioned that their initial interpretation of the situation after hearing the alarm bell and perceiving smoke was that the situation was probably “not at all serious.” 62

Decision making in a risk situation, such as an emergency, will be complicated with uncertainty because the information about the situation is incomplete.68 When a decision has to be made in order to solve a problem, not all possible information and alternatives are considered.69 In dealing with uncertainty, subjects tend to focus their attention on a few information units while disregarding others.70 This could explain the observation that sometimes people report that the fire alarm did not sound although there is evidence that it did sound. The use of a few units of information accords with heuristic models that explain the person’s need to reduce the space of the problem to an easily manageable dimension.71 The decision is made according to Simon’s concepts of “bounded rationality” and “satisficing”: when a good-enough solution for the problem is found.72 According to the literature on decision making in risk situations, it might be expected that, in a fire situation, a person will not refer to all possible information and alternatives as perceived, known, or inferred from the situation. Only a few options, which the subject considers as being more likely to solve the problem, will be retained. A rapid and easy strategy to make decisions and solve a problem is to apply a well-run decision plan.73 Usually, for the person, this translates into evacuation by a familiar route.11 Using a familiar route to leave a building is a less demanding task then developing a new decision plan for a new route.51 In public buildings, this usually means leaving by the main entrance.11 Only when the familiar route is blocked or judged dangerous will the person try to develop an alternative decision plan. Another characteristic of an emergency is the time pressure under which the person is processing information, making decisions, and executing action. Although research on decision making under time pressure has usually dealt with the incentive of gain and loss of money,74 which differs from a life threat, some results are relevant to decision making in fire. According to Wright,75 decision makers are inclined to use less information and to accentuate negative evidence when the environment provides high time pressure or distracting conditions. In fire situations, following failure of initial adaptive actions, alternative actions involving greater risk of

While investigating the situation or after an unsuccessful attempt at evacuation, the formation of clusters of occupants has been observed in numerous fires.64,65 Convergence clusters can be defined as the formation of a group of people in a room or area of refuge who are not previous acquaintances, but who will show altruistic behavior to one another, admit others to the room, and assist until the danger is cleared. During the course of a fire event, the rapidly changing situation will constantly and repeatedly be redefined in relation to new information gathered and feedback obtained from previous decisions and actions.66 The type of action taken will depend on the interpretation the person has made of the situation and on what he or she assumes to be an adaptive reaction to such a situation. To notify or gather others, fight the fire, leave the area or building, or wait to be rescued are all likely behaviors.

DECISION MAKING DURING A FIRE Janis and Mann1 have explained that decision making during an emergency is different from day-to-day decision making in two main ways: “One is that there is much more at stake in emergency decisions—often the personal survival of the decision maker and of the people he (or she) values the most. A second important difference is in the amount of time available to make a choice before crucial options are lost.” To those two main differences, a third one is added for a fire emergency: the ambiguous, incomplete, and unusual nature of the information on which to base the decisionmaking process and usually the impossibility of obtaining more valid information due to the lack of both the time and the means to get information. The problem of defining the situation was very well expressed by people interviewed after the explosion and fire at the World Trade Center in 1993. Occupants did not know what had happened and had difficulty deciding on the best course of action.67

personal injury are conducted. Under time pressure, the person tends to discredit options after looking at only a few salient dimensions. Transposed to a fire emergency, these findings assume that people will give priority to familiar options and disregard unfamiliar options, which they tend to define as likely to increase danger. For example, escaping through an emergency exit is often eliminated as an egress alternative. Indeed, the person may believe that the emergency exit door could be locked or blocked at the bottom or could lead to an underground parking garage or to an unsafe area. Users have limited experience with such means of circulation because use of emergency exits is often prohibited during normal occupancy of the building. These findings help explain why users tend to prioritize evacuation through a familiar route.

The interaction of the perceived threat, the limited time to react, and the limitation of means to gather information will create psychic stress in the user. Knowing that a decision has to be made in an emergency and that this decision could be irrevocable contributes to the stress created. According to Janis and Mann,1 knowing that one has to make a decision is, in itself, stressful. It has been commonly assumed that during a fire occupants would avoid making the decision to move through smoke. This expectation is, however, contrary to accounts provided by people involved in fire events. Studies conducted with fire survivors demonstrate that a large number of occupants moved through smoke to escape.19

In the same study,19 the estimated distance moved through smoke ranged from a few feet to more than 60 feet (a few meters to more than 18 meters) as presented in Table 3.

In the United Kingdom, interviews with 2193 people involved in fire incidents show that 60% moved through smoke; a similar study in the United States with 584 people involved in fire incidents shows that 62.7% said they moved through smoke.19 Table 2 summarizes data compiled in the United Kingdom and the United States and the percentage of building occupants who moved through smoke and their estimation of the visibility distance at the time of their movement.

TABLE 3. Compilation of Distance Moved Through Smoke Distance Moved (m) [ft]

0–0.6 [0-2] 0.9–1.8 [3-6]

TABLE 2. Compilation of Visibility Distance for Population Moving Through Smoke U.K. Sample Population (%)

U.S. Sample Population (%)

0–0.6 [0-2]

12.0

10.2

0.9–1.8 [3-6]

25.0

17.2

2.1–3.7 [7-12]

27.0

20.2

4.0–9.1 [13-30]

11.0

31.7

9.4–11 [31-36]

3.0

2.2

11–14 [37-45]

3.0

3.7

14–18 [46-60]

3.0

7.4

17.0

7.4

>18 [>60]

U.S. Sample Population (%)

3.0

2.3

18.0

8.4

2.1–3.7 [7-12]

30.0

17.1

4.0–9.1 [13-30]

19.0

45.5

9.4–11 [31-36] Visibility Distance (m) [ft]

U.K. Sample Population (%)

11–14 [37-45] 14–18 [46-60] >18 [>60]

5.0

2.0

4.0

4.1

5.0

11.0

15.0

9.6

Although occupants were prepared to move through smoke, many eventually decided to turn back because of the smoke, heat, or a combination of both. For the U.K. population, 26% made the decision to turn back at some point during their evacuation, and the same

the following section, Movement, and an analysis of the tenability of the fire environment are used to track the position of the occupants and determine the exposure to dangerous conditions, given an initial estimated decision time. Some iterations will usually be needed to estimate the time actually available for decisions. With this in hand, a judgment can be made of the ability of the occupants to make successful decisions in the time available. Appropriate safety factors need to be included to compensate for unknowns, uncertainties, and individual variations involved. Means to improve the decision-making process can include a variety of elements:

decision was made by 18.3% of the U.S. sample. Table 4 presents the estimated visibility distance of the two populations when they decided to turn back. TABLE 4. Compilation of Visibility Distance Relative to Turned Back Behavior Visibility Distance (m) [ft]

U.K. Sample Population (%)

U.S. Sample Population (%)

0–0.6 [0-2]

29.0

31.8

0.9–1.8 [3-6]

37.0

22.3

2.1–3.7 [7-12]

25.0

22.3

4.0–9.1 [13-30]

6.0

17.6

0.5

1.2

11–14 [37-45]

9.4–11 [31-36]

1.0

14–18 [46-60]

0.5

4.7

1.0

>18 [>60]

• Providing the maximum amount of key information on the threat so that each individual can assess his or her own risk and weigh the options available. Some of the desirable features include:

IMPACT ON FIRE PROTECTION ENGINEERING DESIGN AND ANALYSIS

1. A communications system that reaches all occupants promptly, clearly, and with sufficient intensity and intelligibility to be understood by all 2. Messages or cues that convey sufficient unambiguous accurate information for individuals to evaluate the threat and the options open to them 3. A limited number of false or inaccurate alarms so that occupants have confidence in the emergency message and interpret the alarm as a warning of a real fire

In view of the scarcity of research and field data related to occupants’ decisions during fire emergencies, the current state of knowledge allows only the most preliminary attempts to quantify the hazard impact of decision making under fire conditions. At this time, the prediction of what emergency decisions each individual in a building will make at the time of a fire and when each decision will be made are among the most difficult elements to take into account when predicting the consequences of any given fire scenario. The impact of the decisions made and the time involved, however, are real and important factors in the ultimate safety of the building occupants. The decision process and impact must be considered in planning or evaluating the fire safety of a building or other facility. It is possible to estimate the expected time that is safely available for occupant decision making. Typically, this exercise is conducted using a series of fire scenarios. Considering uncertainties and capabilities, various fire models are then used to predict the expected time at which life threats could occur in the locations of occupants or the routes they are liable to take to seek refuge or escape the building. In this analysis, the methods described in

• Increasing the familiarity of the occupants with the means of egress. The more familiar the occupants are with the building and the ways out, the more likely they are to use the most efficient routes during an evacuation. Some considerations for designing egress systems are: 1. The exit routes most likely to be used during an emergency are those that are regularly used by the occupants. In a public building, the most familiar routes of customers are the ways they have entered the building; these routes are judged safe and desirable by individuals even if they are more distant than

nearby emergency exits. The familiar ways in and out of a building are the routes most likely to be used in an emergency. Routes that are not obviously safe such as those exiting through back corridors of hotels or the stock rooms of stores are not likely to be used unless there is no other choice or staff who appear knowledgeable instruct occupants to direct people to exits. Exits with big warning signs that read â&#x20AC;&#x153;emergency exit onlyâ&#x20AC;? or that do not appear to lead directly to the outside are unlikely to be used unless occupants have been trained in using these exits or are explicitly instructed to use them. 2. Evacuation drills, even pre-announced drills, are beneficial to allow occupants to experience the exit system. 3. In family dwellings where windows and other abnormal escape routes might need to be used, familiarization exercises such as those spelled out in the NFPA Operation EDITH can avoid confusion and speed up the escape process.

evacuation drill or case study data and determine if the data are relevant to the context of the building and occupants during a fire incident. Such data may be appropriate as the definitive basis for occupant movement time or may be useful in validating mathematical calculations of movement time. In the more likely event that no directly relevant movement time data are found, then a suitable calculation method or model needs to be selected and used to estimate movement time. Assumptions of the model and the analysis need to be identified. Where engineering judgment is applied to aspects of the analysis or calculation factors, justification or basis for the engineering judgment should be provided. Traditionally, the capacity of egress components was measured as a function of the number of units of exit width. (A unit of exit width is defined as 0.56 m [22 inches].) However, the work of Pauls,76 Fruin,77 and Habicht and Braaksma78 has demonstrated that the capacity of egress components approximates a linear function of the clear width of the egress component, less an edge component along each side of the flow path. The resultant width, referred to as effective egress width, is normally about 0.3 m (12 inches) narrower than the actual clear width measurement.79 The full flow capacity of egress routes with width above a minimum is now approximated as a direct function of width.80 The abandonment of the exit unit concept was a major step in recognizing that people do not move in regimented side-by-side lanes down stairs or through doors, but rather move in a staggered arrangement that makes more efficient use of space and permits lateral body sway.81 Measurement and analysis of crowd movement have provided the

Movement This section addresses the period of time after the decision is made to evacuate or relocate. It provides guidance and quantitative methods for estimating the time for occupants to move to a place of safety or refuge. Quantitative factors and calculation procedures and modeling approaches can be utilized to estimate the travel or movement time of occupants. In addition to performing mathematical calculations, the engineer should determine if there is available

<START TIME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . END TIME> FIRE / CUE INITIATION OR DEVELOPMENT

Cue Validation Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . and Continuing Process Receiving Cues

Recognizing Cues

Interpreting Cues

Receiving, Recognizing, Interpreting (RRI)... RRI...RRI...

Decision-Making Period Pre-movement Decisions

Transmovement Decisions Movement / Refuge Time

FIGURE 5. Movement/Refuge Time

Evacuees’ Speed Down Stairs m/sec

FACTORS THAT IMPACT MOVEMENT TIME The time it takes an occupant to move from his or her starting point to a location of safety is simply a function of travel speed and distance:

1.0 Spot Measurement Case Average

0.5

Time (s) = Distance (m) / Speed (m/s)

Fruin (1971) S = 1.08–0.29d

Density, Persons per Square Meter

FIGURE 6. Comparison of Correlations and Data

technical foundation for timed exiting approaches that are used to estimate people movement. Such timed egress approaches are recognized and explained in detail in the SFPE Handbook of Fire Protection Engineering.79 The method and calculations outlined in the Handbook abandon some of the traditional egress assumptions of “units of exit width” and the associated optimistic flow assumptions. Instead, the timed egress analysis is oriented toward a sound and detailed evaluation of people movement and considers parameters that include boundary conditions at exits such as walls and handrails, stair geometry, travel speed, and people density. This section identifies and reviews how these parameters affect the time of movement of occupants once an evacuation is in progress. It is important, however, that the users of these calculation methods be aware that the flow parameters are obtained from carefully measured flows of persons in egress drills and other crowd movement situations that involve a considerable range of variability. Figure 6 shows a typical relationship between the source data and the derived equation. The solid line indicates the correlation representing the equation for stairs presented in this guide. While the flow equation expresses relationships, actual flows must be expected to vary considerably from the calculated flow.

However, many factors impact both travel speed and distance. Distance is a function of exit choice, and exit choice is affected by an occupant’s familiarity 5 with the structure, the availability of exits, and the degree of difficulty of an exit path. Travel speed is quite a bit more complicated. Travel speed is impacted by the occupant’s mobility and the mobility of any accompanying occupants; crowding; light levels; presence of smoke; quality of the floor and wall surfaces (roughness, unevenness, etc.); width, tread width, and riser height of stairs; width of doorways and corridors, etc. Whether or not the occupants have been trained and whether or not there is trained staff guiding evacuation can also impact the speed of travel. Some of these factors are occupant characteristics; others are building characteristics. The designer or engineer must either deal with each factor explicitly or be able to justify why a factor is not relevant to the analysis. A variety of values can be calculated to predict the movement component of total evacuation time: • • • •

Time to clear the building Floor clearing times Stairwell clearing times Time for a person to travel the longest (most remote) path • Location of individuals so that exposure levels can be calculated • Etc. The calculation method chosen will depend to a great degree on the values needed for the evaluation of a design.

There are two principal approaches for estimating the evacuation time for a building. One approach applies a hydraulic analogy, simulating people as fluid particles. Another approach considers the behavioral aspects of the people. A prerequisite for either of these approaches is information on the following people movement characteristics:

mations for crowd density, speed, and flow for various conditions of crowdedness on stairs, along corridors, and in doorways. Other approximations can be found under the heading “Movement Assumptions for Simple, First-Approximation Calculations.” The values are based on work by Fruin and Pauls; however, they are simplified and optimistic, with no reductions for edge effects. Pauls81 reports that using them will result in rough estimates of minimum movement time within 25%. The resultant errors when using these values in calculations of movement times will be acceptable as long as calculated times are considered minimum times for escape movement only. Other behavior, not involving simple movement to the exit, will often be a larger factor in determining total evacuation time. (Pauls reports that the simple, first-approximation calculations presented in Section 3-13 of the SFPE Handbook of Fire Protection Engineering will result in rougher estimates, to within ±33%.) Nelson and Mowrer79 use the equations described above to obtain a first order approximation of the movement time in buildings. The method involves determining the maximum flow rate for each of the egress components in the egress system. The total movement time is estimated as:

• Speed: rate of travel along a corridor, ramp, stair. (The speed on stairs refers to the rate of travel along a diagonal path obtained by connecting the nosings of the stairs.) • Flow: number of persons passing a particular segment of the egress system per unit time (e.g., persons/sec passing through a doorway or over an imaginary line drawn across a corridor) • Specific flow: flow per unit width of the egress component (e.g., persons/sec-m of doorway width) Most of the information on people movement has been collected in fire drills and for normal movement. The parameters have been investigated for people movement on stairs, in corridors, and through doorways.

METHODS FOR CALCULATING MOVEMENT TIME

t = t1 + t2 + t3

There are two principal ways to calculate movement time: hand calculations (or closed form equations) and computer models.

(Eq. 1)

Where: t1 = Time for first person to reach controlling component t2 = Time for population to move through controlling component t3 = Time for last person leaving controlling component to reach place of safety, i.e., exterior of building, area of refuge

Hand Calculations The SFPE Handbook provides a discussion of flow calculations to be used to estimate movement time, and the reader is referred to that discussion.79,82 It is essential that the engineer keep in mind that these calculations deal only with movement time and often will provide optimal results. For example, they do not account for scattered evacuation starting times or delays that occur after travel to exits has begun (e.g., the impact of fatigue). Those need to be factored in separately by the engineer. The Handbook presents two methods of approximation; both estimate minimum movement time.82 Table 3-13.5 of the SFPE Handbook lists approxi-

Speed The speed has been shown to be a function of the density of the occupant flow, type of egress component, and mobility capabilities of the individual.79 For a density greater than 0.55 pers/m2 (0.051 pers/ft2): v = k − akD

(Eq. 2)

For densities less than 0.55 pers/m2 (0.051 pers/ft2), too few other people are present to impede the walking speed of an individual. Maximum walking velocities for level walkways and stairways are: v = 0.85k Where: v = a = k = D =

movement of significant numbers of persons. This density is not comfortable, and it should be expected that, given the opportunity, most persons will increase the space around them and actually operate at a lower density. Higher densities not only slow the flow but also can reduce movement to a shuffling gait and, in the extreme, a crushing condition. People movement is faster for level components than stairs, and faster for stairs with lower risers and deeper treads than stairs with higher risers and shallower treads. Figure 7 graphically illustrates the relationship between people movement speed and population density for various stair and level travel configurations.

(Eq. 3)

Speed, m/s (ft/min) Constant, 0.266 m2/pers (2.86 ft2/pers) Velocity factor, see Table 5, m/s (ft/min) Density of occupant flow, pers/m2 (pers/ft2)

At lower densities, people have a greater freedom to move at their own pace. As the crowd density increases, they become more controlled by others in the moving stream. At about 1.88 persons per square meter (0.175 person per square foot),79 the combination of closeness of individuals and speed of movement are indicated to be the maximum. At this density, exiting individuals would normally be able to see about two treads ahead on stairs or two steps ahead on a flat surface. Such exiting densities are most likely to be encountered in highly populated places of assembly or similar situations involving the

TABLE 5. Velocity Factor in Equations 2 and 34 Egress Component

k (m/s)

k (ft/min)

Corridor, aisle, ramp, doorway

1.40

275

Stair Riser, mm (in.)

Stair Tread, mm (in.)

190 (7.5)

254 (10)

1.00

196

272 (7.0)

279 (11)

1.08

212

165 (6.5)

305 (12)

1.16

229

165 (6.5)

330 (13)

1.23

242

250

Movement Speed (ft/min)

Level Travel

Conversion Factors:

6.5/13 stair 200 6.5 /12 stair 7/11 stair

1 ft = 0.005 --m ------min s

7.5/10 stair

1 1 ------= 10.8 -----ft2 m2

150

100

0 0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

Density (persons/sq ft)

FIGURE 7. Movement Speed as a Function of Density (Plot of Tables 6 and 7)

The speed correlations presented in Equations 2 and 3 principally relate to adult, mobile individuals. Proulx81 indicates that the mean speed on stairs for children under 6 and the elderly was approximately 0.45 m/s in unannounced drills in multi-story apartment buildings. The speed for an “encumbered” adult is 0.22 to 0.79 m/s, also appreciably less than the maximum speed noted in Equation 3. (An encumbered adult is an individual carrying packages, luggage, or a child.) Mean velocities for impaired individuals summarized by Shields et al.84 are presented in Table 6. Typical densities of people movement range from 1.0 to 2.0 persons per square meter (0.093 to 0.19 persons per square foot).77,81,85,86 The density of a flow can be determined in several ways:

moving past a point in the egress route per unit time per unit distance of effective width can be determined. This flow is termed “specific flow” and is calculated as the speed multiplied by the density of the population. The specific flow is analogous to the mass flux in hydraulic systems. As such, the specific flow is the product of the density and the speed: Fs = Dv

Expressions for the specific flow as a function of density only can be obtained by substituting for the speed from Equations 2 and 3: • For a density greater than 0.55 pers/m2 (0.051 pers/ft2):

• 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). For the equations in this guide, the density is defined in terms of this approach. • The ratio of the floor area occupied by each individual person in the group divided by the total floor area occupied by the group (including the area between individuals).

Fs = Dv = (1 − aD)kD

(Eq. 5)

• For densities less than 0.55 pers/m2 (0.051 pers/ft2): Fs = 0.85kD

(Eq. 6)

Where: Fs = Specific flow, pers/s-m (pers/min-ft) The width referenced in the units for the specific flow equations relates to the “effective width” as defined by Pauls.81 The concept of effective width is based on the observation that people do not

Specific Flow By combining the concept of people movement speed with effective width, the number of persons

TABLE 6. Mean Speed for Impaired Individuals Impairment

(Eq. 4)

Level Walkway (m/s) [ft/s]

Stairs—Down (m/s) [ft/s]

Stairs—Up (m/s) [ft/s]

Electric wheelchair

0.89 [2.9]

Manual wheelchair

0.69 [2.2]

Crutches

0.94 [3.1]

0.22 [0.72]

Walking stick

0.81 [2.7]

0.32 [1.0]

0.34 [1.1]

Walking frame/walker

0.51 [1.7]

Rollator

0.61 [2.0]

No aid

0.93 [3.1]

0.33 [1.1]

0.41 [1.3]

No disability

1.24 [4.1]

0.70 [2.3]

generally occupy the entire width of an egress component, staying a small distance away from the walls or edge of the component. Nelson and Mowrer79 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 7. Considering the quadratic function of the specific flow, a maximum specific flow is achieved at a density of: 1 Dmax = — 2a

(Eq. 7)

TABLE 7. Boundary Layer Width

Because a is independent 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 Milinskii85 provide results from their data indicating differences in the density where the specific flow is maximized for different types of egress components. In understanding the concept of specific flow, it is important to realize that the flow of people can reach only a specific maximum value as illustrated in Figure 8. By reviewing Figure 8, starting at a density of 0 and speed of 0, it is noted that the specific flow of occupants is shown to increase as the density increases. Although the speed of moving occupants is decreasing, it is more than compensated for by the greater density of moving people.

Boundary Layer mm (inch)

Component

Theater chairs, stadium benches

0 (0)

Railings, handrails*

89 (3.5)

Obstacles

100 (4)

Stairways, doors,† archways

150 (6)

Corridors and ramp walls

200 (8)

* Where handrails are present, use the value that results in the lesser effective width. †There are no data to substantiate the effective width of doors; the value is an extrapolation from the value used for stairs and is believed to result in estimates that do not underestimate the movement time through doors.

Specific Flow (persons/min/ft)

Level Travel

6.5/13 Stair 6.5 /12 Stair 7/11 Stair 7.5/10 Stair

Conversion Factors: 10

1 = 0.055 ------1 ---------ft-min m-s 5

0 0.00

1 1 ------= 10.8 -----m2 ft2

0.02

0.04

0.06

0.08

0.10

0.12

0.14

Density (persons/sq ft)

FIGURE 8. Specific Flow as a Function of Density

0.16

0.18

0.20

0.22

0.24

However, the specific flow begins to diminish and an optimum flow condition is reached at a density of approximately 1.94 persons per square meter (0.18 persons per square foot or 1 person per 5.6 square feet). At densities greater than 1.94 persons per square meter, their speed continues to decrease, but the increasing crowd density rather than compensating for the continuing decreasing speed now makes occupant flow more difficult until the specific flow drops to zero at a jam-point density of 3.77 persons per square meter (0.35 persons per square foot or 1 person per 2.8 square feet). Recognizing that flows may vary significantly between optimal and less than optimal conditions, the timed exiting evaluation of a building or portion of a building may need to consider a range of conditions rather than being based solely on optimal flow conditions. The maximum flow rate occurs when the specific flow is maximized, i.e., where Dmax occurs (see Equation 7). Maximum specific flows for a variety of egress components are provided in Table 8. The controlling egress component is the component with the smallest maximum flow rate, relating to where a queue is expected to form if Dmax occurs in an upstream component. It is typically assumed that, if there are enough people present, the maximum

specific flow will be attained. However, such an assumption is an optimization, and analyses should account for uncertainty in this assumption.

Total Flow Capacity The specific flow mentioned above provides a measurement of the flow capability of an egress component on a per unit width basis (e.g., per meter). In the evaluation of an egress component or multiple egress components, the total flow can be calculated and related to the affected population. Multiplying the specific flow by the total effective width of all exits permits the calculation of a predicted flow rate of persons passing through an exit route or routes. Hence, Flow Capacity = Fc = Fs We Substituting Fs = (1 â&#x20AC;&#x201C; aD)kD Gives Fc = (1 â&#x20AC;&#x201C; aD)kD We The movement time for a populated area through one exit element is the population, P, divided by the flow capacity of the exit element, plus the travel time through the exit element. The following examples demonstrate the use of this method.

TABLE 8. Maximum Specific Flows Egress Component

Fs pers/sec-m of Effective Width (pers/min-ft of Effective Width)

Corridor, aisle, ramp, doorway

1.32 (24.0)

Stair Riser, mm (in.)

Stair Tread, mm (in.)

190 (7.5)

254 (10)

0.94 (17.1)

272 (7.0)

279 (11)

1.01 (18.5)

165 (6.5)

305 (12)

1.09 (20.0)

165 (6.5)

330 (13)

1.16 (21.2)

Examples Example 1 The total movement time for a room containing 300 people will be determined. The room has a travel distance of up to 200 feet to egress through two 32-inch doors that lead to two enclosed 44-inch stairs (height and depth of tread of 7 inches and 11 inches, respectively) and down 50 feet of stairs to a wide discharge at grade. T

d v

Assuming that neither of the stair entrance doors is blocked by the fire and that the occupants in the room are equally distributed (low density; 0.05 person/sq ft), one would first consider if the time to travel to the stairs is greater than the time for occupants to move through the doors into the stairs or move down the stairs.

200 ft d = = 0.9 min 0. 85 × 275 ft / min 0.85k

The time for occupants to move through the door or on the stairs will indicate if occupants are T =

P Fsmax We

queuing at the stair entry doors. Movement time through the doors: 300

= 3. 8 min

 2 × (32 in. − 2 × 6 in.)   24 pers/min − ft ×    12 in./ft

Movement time on the stairs: T =

P Fs max We

300  2 × (4 4 in. − 2 × 6 in.)  1 8.5 pers/min − ft ×   12 in./ft  

d v

d 0 .85 k

3 . 0 min

only takes 0.9 minutes for the occupants to travel to the doors and 3.0 minutes for the occupants to travel past a point on the stairs (i.e., to pass the top of the stairs). Movement time down the stairs:

Since the movement time on the stairs is less than the movement time through the doors, the stair entry doors control the flow of occupants. All the occupants have moved through the stair entry doors in approximately 3.8 minutes (assuming that their density is Dmax), while it T =

50 ft 0 .85 × 212 ft / min

The total movement time, assuming once again that the occupants in the room are equally distributed and the first occupants to egress are initially at the stair entry doors, equals 3.8 minutes for moving through the stair entry doors plus 0.3 minutes for the last person to walk down the stairs. This results in a total movement time of approximately 4.1 minutes. NOTE: It is important for the user to analyze

= 0 .3 min

each individual situation carefully, considering the realm of possible building and occupant arrangements. For instance, in the above example, one could assume that one of the stair entry doors was blocked by the fire or that all occupants were in a meeting in the far corner of the space. Adjustments to the calculations would then be made, resulting in different and increased total people movement times.

Example 2 Determine the total movement time for a 5-story building with the following characteristics:

are 32" wide (0.81 m). The stair design includes 7/11 risers and treads. The floor-to-floor distance is 12 feet, and the landing between floors is 4 x 8 feet (1.22 x 2.44 m). Handrails are provided on both sides of the stairways 2.5 inches (64 mm) from the wall.

• There are 200 people on each floor. Each floor is served by two 44" (1.12 m) wide stairways. The doors leading into and from the stairway Solution:

Component Door into stairway Stairway Landing Door from stairway

Effective Width, m (ft) 0.51 (1.67) 0.81 (2.67) 0.81 (2.67) 0.51 (1.67)

The controlling component is the door leading from the stairway. The time required for the half of the building occupants on the upper floors

P F smaxW e

Specific Flow, pers/m-sec (pers/ft-min ) 1.31 (24.0) 1.01 (18.5) 1.31 (24.0) 1.31 (24.0)

Flow, pers/sec (pers/min) 0.67 (40) 0.82 (49) 1.1 (65) 0.67 (40)

(400 persons) to pass through this doorway is estimated as

400  32 in. − 2 × 6 in.  24 pers/min − ft ×   12 in./ft  

Next, calculate the time required for the first person to travel down the stairs.

10 min

that for the stairway, giving a total length of travel of 30.3 feet (9.2 m). The time required to traverse this distance at the speed achieved on the stairways is 0.29 min (17 sec). Since a queue is expected to form at the door leading from the stair, it is only necessary to consider the time for the first person on the second floor to travel down one flight of stairs to the first floor since people coming from upper floors are expected to encounter a queue. As the time that people would wait in the queue would be greater than the time that would be needed to travel down the stairs if the queue did not exist, we can neglect the time that it would take for people on upper floors to travel down the stairs. This results in a total movement time of 0.3 min + 10 min = 10.3 min, neglecting the time needed to travel to the stairs.

Time to travel down one flight of stairs: The hypotenuse of 7/11 stair is 13 inches (330 mm). Thus, to travel a vertical distance of 12 ft (3.6 m) requires traveling a diagonal distance of 22.3 feet (6.8 m). Using Equation 1, the speed at Dmax of 0.175 pers/ft2 (1.88 pers/m2) is 106 ft/min (0.54 m/s). The length of travel along each landing is 8 feet (2.4 m) (assuming an average length of travel on the middle of the landing). Because the speed on a stairway is less than that for a horizontal component such as a landing, the speed on the landing is limited to that achieved on the stairway. As such, the length of travel on the landing can be added to

The type of analysis shown in the previous examples is most relevant in situations where a queue is expected to form at the controlling egress component. Generally, these situations consist of cases where an appreciable number of people occupy the area of the building being modeled. 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-bycomponent basis. These analyses also may be applied to provide a more accurate assessment in cases where queuing is likely. The component-by-component analysis involves a determination of the time for the population to traverse each egress component. Transitions between different components, changes in component width, and mergers are addressed via the following rules:

â&#x20AC;˘ Where the width of the egress component changes, then the density of the flow is also expected to change. The new density is determined by the following relationship: FC1 = FC 2

Again, if the incoming flow rate leading to the transition point is greater than the capacity of the flow rate for the egress component leading from the transition, a queue is expected to form at the transition.

â&#x20AC;˘ The combined flow rate of people entering an intersection equals the flow rate of people from the intersection (see Figure 9): FC1 + FC2 = FC3

(Eq. 9)

FIGURE 10. Transition in Egress Component

(Eq. 8) The density of a flow of occupants proceeding away from a transition is determined by solving either Equation 5 or 6. Where Equation 5 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, since, if an occupant flow at the maximum density were approaching a widening corridor, the solution of Equation 5 would yield one density greater than the maximum and one less. However, in the case of a widening corridor, it is unreasonable to expect the density to increase (and speed to decrease) 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 proportion to the respective capacities or other characteristics of the available paths.85,87

1 FIGURE 9. Merging Egress Flows

If the combined flow rate of egress components leading to the intersection is greater than the capacity of the flow 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 maximum capacity of the component.

Nelson and Mowrer79 recommend the following method of determining the densities and flow rates following the passage of a transition point:

[Fs(in-1)We(in-1)] + â&#x20AC;Ś + [Fs(in-n)We(in-n)] = [Fc(out-1)We(out-1)] + â&#x20AC;Ś + [Fc(out-n)We(out-n)] (Eq. 10(c))

1. The flow after a transition point is a function, within limits, of the flow(s) entering the transition point. 2. The calculated flow, Fc , following a transition point cannot exceed the maximum specific flow, Fsm , for the route element involved multiplied by the effective width, We , of that element. 3. Within the limits of rule 2, the specific flow, Fs , of the route departing from a transition point is determined by the following equations:

where the letter n in the subscripts (in-n) and (out-n) is a number equal to the total number of routes entering (in-n) or leaving (out-n) the transition point. 4. Where the calculated specific flow, Fs , for the route(s) leaving a transition point, as derived from the equations in rule 3, exceeds the maximum specific flow, Fsm , a queue will form at the incoming side of the transition point. The number of persons in the queue will grow at a rate equal to the calculated flow, Fc , in the arriving route minus the calculated flow leaving the route through the transition point. 5. Where the calculated outgoing specific flow, Fs(out), is less than the maximum specific flow, Fsm , for that route(s), there may be no way to predetermine how the incoming routes will merge. The routes may share access through the transition point equally, or there may be a total dominance of one route over the other. For conservative calculations, assume that the route of interest is dominated by the other route(s). If all routes are of concern, it is necessary to conduct a series of calculations to establish the bounds on each route under each condition of dominance.

(a) For cases involving one flow into and one flow out of a transition point: Fs(out) = Fs(in)We(in) We(out)

(Eq. 10(a))

Where: Fs(out) = Specific flow departing from transition point Fs(in) = Specific flow arriving at transition point We(in) = Effective width prior to transition point We(out) = Effective width after passing transition point (b) For cases involving two incoming flows and one outflow from a transition point, such as that which occurs with the merger of a flow down a stair and the entering flow at a floor:

Computer Models There are two principal types of movement calculation models: hydraulic or network flow models and behavioral models. Hydraulic models treat occupants like water flowing through channels. They result in optimal results since they generally move occupants in the most efficient manner. One major advantage they have is that they are generally inexpensive to purchase and simple to run. However, these models do not track individuals, so they are not able to match occupants to specific locations and thus accumulate

Fs(out) = {[Fc(in-1)We(in-1)] +

[Fc(in-2)We(in-2)]}/ We(out) (Eq. 10(b)) where the subscripts (in-1) and (in-2) indicate the values for the two incoming flows. (c) For cases involving other merger geometries, the following general relationship applies:

toxic exposures. They are best used as a first approximation in the evaluation of a design. If the results indicate that a structure cannot be evacuated before untenable conditions are predicted to occur in occupied areas, then design changes must be made. Behavioral models attempt to realistically predict the actions and decisions made by occupants during an evacuation. These models are attractive because they seem to more accurately simulate evacuations. However, due to the scarcity of behavioral data, they tend to rely heavily on assumptions, and it is not possible to gauge with confidence their predictive accuracy. Users of such models need to establish confidence in the assumptions before relying on the results. To set up the travel options for occupants, evacuation models use either a network of nodes and arcs or a mesh structure. When spaces and travel paths are defined using nodes and arcs, the movement of occupants is restricted to those paths, and the predicted movement is seldom smooth. (Occupants tend to jump from node to node, for example.) A mesh structure lays a grid of “tiles” over entire enclosures on a floor plan, and occupants are able to occupy or move from tile to tile. This allows the more precise location of occupants throughout spaces. Setting up the description of a floor plan using nodes and arcs can be very time-consuming, but, once done, it does not have to be redefined unless the structure is redesigned. CAD packages can often be used to input the floor plan description for a mesh structure, but models that use a mesh are generally more time-consuming to run because of the complexity of the calculations used to move people through such open spaces. The degree to which behavior is simulated varies extensively among available models. Some require the user to estimate and input the pre-movement delays that are appropriate for a particular scenario and structure. Others include behavior “rules” and will predict behavior according to those rules. The engineer must be cautious in choosing a model. The complexity of some models implies a greater predictive capability, but, again, the scarcity of data available on behavior means that a great number of assumptions are embedded in the models, and the

appropriateness of those assumptions is critical in evaluating the validity of a model’s results. The SFPE Handbook79 includes guidance on the selection of evacuation models. These are the questions the user should be able to answer about the model selected for an analysis. (See the Appendix.) Of particular importance are the questions concerning the appropriateness of a model to the task at hand. The user must be clear on the assumptions that are explicitly stated by the model developer. Even more importantly, the user must be aware of the assumptions embedded in a model. For example, if a model uses a constant travel speed for occupant movement, the user must understand the source of that value and its applicability to the scenario being modeled. If all occupants will move at the same speed, the user must be able to justify the absence of differently abled occupants. The user must take into consideration the issue of safety factors: Do the model results incorporate safety factors, and, if so, how is that done? If safety factors must be applied to the results by the user, that must be specifically stated, and appropriate methods for doing so must be described.

Evaluation of Models At this time, a number of models exist. Due to the dynamic nature of model development, no list is presented here. Users are encouraged to answer the questions posed in the Appendix when selecting a model for use in their application.

ALTERNATIVES TO EVACUATION When an emergency occurs, evacuating the building is not an occupant’s only option. In fact, it is often not in the best interests of the occupant to attempt to leave the building. Several studies of high-rise apartment building fires have shown that the victims were the people who left their apartments and were trapped and overcome in stairwells.64 In other cases, such as hospitals, evacuation of all occupants is not possible or advisable. Some buildings are designed today to incorporate defendin-place strategies where disabled occupants can remain safe while awaiting rescue.

In such cases, the engineer needs to model movement for some occupants only to locations of safety. The simplest scenarios to calculate will be those where no evacuation will take place, and the engineer only has to predict the development of hazardous conditions in the areas that will remain occupied.

Other research has shown in experiments travel speeds of 0.2 to 0.4 m/s reported in heavy smoke and that, in actual fires, survivors on average moved only 10 meters in heavy smoke.14 Where evacuation through smoke is involved, the movement speed of evacuation should be no greater than that appropriate for the expected density and irritation properties of the smoke. Pending further research, the adjustment should be made using the values plotted in the Figure 11. The engineer must account for the reduction in travel speed when model results show that occupants will encounter smoke at some point during their evacuation. Some evacuation models might account for this; the model user is responsible for determining if the accounting is adequate.

IMPACT OF SMOKE ON MOVEMENT The presence of smoke will impact movement in two ways: 1. It can decrease the probability that occupants will move into an area or continue their evacuation. 2. It can reduce their walking speed.

Walking Speed (m/s)

The emergency movement speeds reported in most sources were derived from experiments and observations conducted in smoke-free environments. Non-irritant smoke While most emergency egress of popula1.0 tions does occur in environments that are smoke-free or nearly so, some emergency evacuations involve movement 0.5 through smoke conditions. Both the Irritant smoke density and optic-irritating properties of 0 the smoke can impact movement speed. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 88 Jin reports the results and analysis Extinction Coefficient Cs (1/m) of several series of investigations he conducted involving human subjects moving through a smoke-laden environment. FIGURE 11. Walking Speed as a Function of Smoke Jin used two types of smoke, a highly Extinction Coefficient irritating smoke produced by burning wood cribs with narrow spacing between CONCLUSION the sticks and a less irritating smoke produced by burning kerosene. The graph in Figure 11 is typical Although the purpose of this section is to discuss of the results observed. With the less irritating movement calculations, it cannot be stressed enough smoke, movement speed decreased gradually as the that the key component in predicting total evacuaextinction coefficient increased. With the highly tion time is an accurate estimate of pre-movement irritating smoke, the evacuation movement speed delays. Studies have shown that, in the evacuation dropped precipitously once the extinction coeffiof residential buildings, movement time can make cient reached 0.4/m. up as little as 25% of total evacuation time.13 In While Jinâ&#x20AC;&#x2122;s work is a major contribution to contrast, however, pre-movement delays can be a understanding the impact, the user should underfairly small percentage of total evacuation time in stand that the underlying experiments were conoffice buildings. ducted in a controlled setting where all the subjects were aware that their lives were not at risk.

Crowding or the presence of merging flows will slow evacuees along their routes. The presence of smoke, especially irritant smoke, will impede or even stop the movement of affected occupants. The occupant characteristics that are selected for a building will also greatly impact the predicted evacuation time. For example, the mobility of

occupants will clearly influence their travel speed. The engineer must prepare an accurately conservative description of a facilityâ&#x20AC;&#x2122;s expected occupant group. Otherwise, an insupportably optimistic prediction of the performance of a building under emergency conditions may result.

Appendix A

Questions a Potential Model User Should Ask About an Evacuation Model Source: SFPE Handbook, 3d ed., section 3-14

Evacuation Model Type Is the model based on optimization, simulation, or risk assessment? Is the type of model suitable for the application? What are the limitations of the model with respect to the application?

Enclosure Representation Is the model based on a fine network or a coarse network? How are different spaces and areas within spaces represented? How are connections between spaces represented? How are obstructions within a space represented? How do these representations influence the model results? How many nodes, connections, and obstructions can the model handle? How are the data entered to represent spaces, connections, and obstructions?

Population Perspective Does the model use a global or an individual perspective? If the perspective is global, what general characteristics of the population are represented? If the perspective is individual, what individual characteristics of the population are represented? How are the individual or global characteristics of the population entered in the model?

Behavioral Perspective What type of behavioral perspective does the model employ (none, implicit, rule-based, functional analogy-based, or artificial intelligence-based)? How does the model treat peopleâ&#x20AC;&#x201C;people interactions and their effects on behavior? How does the model treat peopleâ&#x20AC;&#x201C;enclosure interactions and their effects on behavior? How does the model treat peopleâ&#x20AC;&#x201C;environment interactions and their effects on behavior? How does the model address physiological factors that influence decision making? How does the model address psychological factors that influence decision making? How does the model address sociological factors that influence decision making?

Model Validation Has the model been validated? If so, how and to what extent? How has the model validation been reported?

Model Implementation Has the model been implemented on a computer? What computer platforms does the model support?

Model Support Is the model currently supported by the author(s)? Is the model supported by another agency? Is the model still being developed? If so, how are users notified of upgrades?

Model Costs What is the initial cost of the model? What are the ongoing costs for upgrades, support, and maintenance?

Appropriateness to Task What inputs does the model require of the user? Are these available? Does the model consider elements needed for the task at hand: • • • • •

Speed of movement, impact of density on speed Queuing or other congestion Merging of flows Pre-movement decisions Decisions/actions during movement

Does the model produce an output meeting the needs of the task at hand?

References 1 2 3

4 5 6 7 8 9 10 11 12

15 16 17 18 19 20 21

Janis, L., and L. Mann, Decision Making, New York, The Free Press, 1977. “Speech Intelligibility,” Fire Protection Engineering (Fall 2002) 23-25. Gann, R., et al., “International Study of the Sublethal Effects of Fire Smoke on Survivability and Health (SEFS): Phase I Final Report,” NIST Technical Note 1439, Gaithersburg, Md., National Institute of Standards and Technology, 2001. Purser, D., “Toxicity Assessment of Combustion Products,” SFPE Handbook of Fire Protection Engineering, 3d ed., Quincy, Mass., National Fire Protection Association, 2002. “Life threat from fires—Guidance on the Estimation of Time Available for Escape Using Fire Data,” ISO TS 13571, Geneva, International Standards Organization, 2002. SFPE Engineering Guide to Performance-Based Fire Protection Analysis and Design of Buildings, Quincy, Mass., National Fire Protection Association, 2000. Milke, J., and T. Caro, “A Survey of Occupant Load Factors in Contemporary Office Buildings,” J. of Fire Protection Engineering 8:4 (1997). “Special Issue: World Trade Center Bombing,” Fire Engineering 96:12 (December 1993). “World Trade Center Response,” WNYF 54:3 3d issue (1993). Latane, B., and J. Darley, “Group Inhibition of Bystander Intervention in Emergencies,” J. of Personality and Social Psychology 10:3 (November 1968) 215-221. Sime, Jonathan, “Movement Toward the Familiar Person and Place Affiliation in a Fire Entrapment Setting,” Environment and Behaviour 17:6 (1985) 697-724. McClintock, T., et al., “A Behavioural Solution to the Learned Irrelevance of Emergency Exit Signage,” 2d International Symposium on Human Behavior in Fire, March 26-28, 2001, Boston, MIT InterScience Communications, 23-33. Fahy, R., and G. Proulx, “Toward Creating a Database on Delay Times to Start Evacuation and Walking Speeds for Use in Evacuation Modeling,” Human Behaviour in Fire, Proceedings of the Second International Symposium, London, InterScience Communications, 2001, 175-183. Sime, J., “The Outcome of Escape Behavior in the Summerland Fire: Panic or Affiliation?” International Conference on Building Use and Safety Technology Proceedings. Los Angeles, March 12-14, 1985, National Institute of Building Sciences. Fineburg, W., “Primary Group Size and Fatality Risk in a Fire Disaster,” Human Behaviour in Fire, Proceedings of the Second International Symposium, London, InterScience Communications, 2001. Schwartz, J., “Human Behavior in the Beverly Hills Fire,” Fire Journal 73:3 (May 1979) 73-74, 108. Purser, D., “Quantification of Behaviour for Engineering Design Standards and Escape Time Calculations,” Proceedings of the First International Symposium on Human Behaviour in Fire, University of Ulster, 1998. Saunders, W., “Gender Differences in Response to Fires,” Human Behaviour in Fire, Proceedings of the Second International Symposium, London, InterScience Communications, 2001. Bryan, J., Smoke as a Determinant of Human Behavior in Fire Situations (Project People), NBS-GCR-77-94, Gaithersburg, Md., National Bureau of Standards, 1977. Wood, P., The Behaviour of People in Fires, Fire Research Note 953, Borehamwood, Fire Research Station, 1972. Bryan, J.L., and J.A. Milke, “The Determination of Behavior Response Patterns in Fire Situations, Project People II. Final Report—Health Care,” NBS-GCR-81-343, Washington, National Bureau of Standards, September 1981. Bryan, J.L., “Implications for Codes and Behavior Models from the Analysis of Behavior Response Patterns in Fire Situations as Selected from the Project People and Project People II Study Programs,” NBS-GCR-83-425, Washington, National Bureau of Standards, Center for Fire Research, March 1983. Ozkaya, Aydn, “A Qualitative Approach to Children of Developing Countries from Human Behaviour in Fire Aspect,” Human Behaviour in Fire, Proceedings of the Second International Symposium, London, InterScience Communications, 2001.

24 25

27 28 29 30 31 32 33 34

35 36 37 38 39 40 41 42

43 44 45 46 46 48 49

Kose, S., “Emergence of Aged Populace: Who Is at Higher Risk in Fires?” Human Behaviour in Fire, Proceedings of the First International Symposium, Jim Shields, ed., University of Ulster, Ulster, 1998. Proulx, G., and J. Sime, “To Prevent Panic in an Underground Emergency: Why Not Tell People the Truth?” Fire Safety Science—Proceedings of the Third International Symposium, London, Elsevier Applied Science, 1991, 843-852. Proulx, G., and R. Fahy, “The Time Delay to Start Evacuation: Review of Five Case Studies,” Fire Safety Science—Proceedings of the Fifth International Symposium, London, International Association for Fire Safety Science, 1997. Cable, E., “An Analysis of Delay in Staff Response to Fire Alarm Signals in Health Care Occupancies,” M.S. Thesis, Worcester, Mass., Worcester Polytechnic Institute, 1993. Lynch, J., “Nocturnal Olfactory Response to Smoke Odor,” Human Behaviour in Fire, Proceedings of the First International Symposium, Belfast, University of Ulster, 1998. Bruck, D., and P. Brennan, “Recognition of Fire Cues During Sleep,” Human Behaviour in Fire, Proceedings of the Second International Symposium, London, InterScience Communications, 2001. Stahl, F., et al., “Time-Based Capabilities of Occupant to Escape Fires in Public Buildings: A Review of Code Provisions and Technical Literature,” NBSIR 82-2480, Gaithersburg, Md., National Bureau of Standards, 1982. Bruck, D., “Non-awakening in children in response to a smoke detector alarm,” Fire Safety Journal 32: 369-376. Nober, E., et al., “Waking Effectiveness of Household Smoke and Fire Detector Devices,” NBS-CGR-83-439, Gaithersburg, Md., National Bureau of Standards, 1983. Nober, E., et al., “Smoke Alarms for the Hearing Impaired,” Fire Journal, Quincy, Mass., National Fire Protection Association (Jan./Feb. 1990). Cable, E., “Cry Wolf Syndrome: Radical Changes Solve the False Alarm Problem,” The 4th National Symposium & Trade Exhibition on Health Care Safety and the Environment, Proceedings, American Hospital Association, 1994. Proulx, G., “Occupant Response to Fire Alarm Signals,” National Fire Alarm Code Handbook, Quincy, Mass., National Fire Protection Association, 1999, 403-412. Proulx, G., et al., “Fire Alarm Signal Recognition,” Internal Report 828; IRC-IR-828, Ottawa, National Research Council of Canada, 2001. Breznitz, S., Cry Wolf: The Psychology of False Alarms, Hillsdale, N.J., Lawrence Erlbaum Associates, 1984. NFPA 72, National Fire Alarm Code, Quincy, Mass., National Fire Protection Association, 2002. Ramachandran, G., “Informative Fire Warning Systems,” Fire Technology 27:1 66–81. Canter, D., Fires and Human Behavior, London, David Fulton Publishers, 1990. Keating, J., and E. Loftus, “People Care in Fire Emergencies—Psychological Aspects,” SFPE Technology Report 75-4, Bethesda, Md., Society of Fire Protection Engineers, 1975. Proulx, G., “The Impact of Voice Communication Messages During a Residential Highrise Fire,” Human Behavior in Fire—Proceedings of the First International Symposium, Belfast, University of Ulster, 1998, 265-274. Canter, D., “Studies of Human Behavior in Fire: Empirical Results and Their Implications for Education and Design,” [Garston], U.K., Building Research Establishment, Department of the Environment, 1985. Wood, G., “A Survey of Behavior in Fires,” Fires and Human Behavior, D. Canter, ed., Chichester, John Wiley and Sons Ltd., 1980, 83-95. Sime, J., “The Concept of Panic,” Fires and Human Behavior, D. Canter, ed., Chichester, John Wiley and Sons Ltd, 1980, 63-81. Keating, P., “The Myth of Panic,” Fire Journal (May 1982) 57-61. Quarantelli, E., “Panic Behavior: Some Empirical Observations,” Human Response to Tall Buildings, D.J. Conway, ed., Stoudsburg, Dowden Hutchinson and Ross, 1977, 336-350. Webber, G., and C. Aizlewood, “Emergency Wayfinding Lighting Systems in Smoke,” Building Research Establishment Information Paper IP17/94, Garston, U.K., Building Research Establishment, 1994. Withington, D., “Life Saving Applications of Directional Sound,” Pedestrian and Evacuation Dynamics, M. Schreckenberg and S. Sharma, eds., Berlin, Springer-Verlag, 2002.

51 52 53 54 55 56 57 58 59 60

61 62

63 64 65 66 67 68 69 70 71 72 73

74 75

Heskestad, A., and K. Pedersen, “Escape Through Smoke: Assessment of Human Behaviour and Performance of Wayguidance Systems,” Human Behaviour in Fire, Proceedings of the First International Symposium, [Belfast], University of Ulster, 1999. Passini, R., Wayfinding in Architecture, New York, Van Nostrand Reinhold, 1984. Arthur, P., and R. Passini, Wayfinding, People, Signs, and Architecture, New York, McGraw-Hill, 1992. Frantzich, H., “Occupant Behavior and Response Time Results From Evacuation Experiments,” Proceedings of the 2nd Symposium on Human Behavior in Fire, London, InterScience Communications, 2001, 159-165. Shields, T.J., K. Boyce, and G. Silcock, “Towards the Characterization of Large Retail Stores,” Proceedings of the First International Symposium on Human Behavior in Fire, [Belfast], University of Ulster, 1998, 277-289. Donald, I., and D. Canter, “Behavioral Aspects of the King’s Cross Disaster,” Fires and Human Behavior, D. Canter, ed., 2d ed., David Fulton Publishers Ltd., 1990, 15-30. Wood, G., “Behavior Under Stress: People in Fires,” Doctoral Thesis, Loughborough University of Technology, 1979. Polya, G., How to Solve It, Garden City, N.Y., Doubleday Anchor, 1957. Howard, D., Cognitive Psychology: Memory, Language, and Thought, New York, MacMillan Pub., 1983. Sime, J., and M. Kimura, “The Timing of Escape: Exit Choice Behavior in Fires and Building Evacuations,” Safety in the Built Environment, J. Sime, ed., London, E & F N. Spon, 1988, 48-61. Sime, J., “Perceived Time Available: The Margin of Safety in Fires,” Fire Safety Science—Proceedings of the First International Symposium, C. Grant and P. Pagni, eds., Washington, Hemisphere Pub. Corp., 1986, 561-570. Canter, D., J. Breaux, and J. Sime, “Domestic, Multiple Occupancy, and Hospital Fires,” Fires and Human Behavior, D. Canter, ed., [London], John Wiley & Sons Ltd., 1980, 117-136. Proulx, G., J.Pineau, J. Latour, and L. Stewart, “Study of the Occupants’ Behavior During the 2 Forest Laneway Fire in North York, Ontario on January 6, 1995,” IRC-IR-705, Internal Report, Ottawa, Institute for Research in Construction, National Research Council Canada, 1995, 74. Breen, D., “Do Smoke Detectors Work in College Dormitories?” SFPE Technology Report 84-8, Bethesda, Md., Society of Fire Protection Engineers, 1984. Proulx, G., “Highrise Evacuation: A Questionable Concept, Proceedings of the Second International Symposium on Human Behavior in Fire, London, InterScience Communications, 2001, 221-230. Bryan J., “Convergence Clusters,” Fire Journal (November 1985) 27-90. Lazarus, R., and S. Folkman, Stress, Appraisal, and Coping, New York, Springer Publishing Company, 1984. Fahy, R., and G. Proulx, “Collective Common Sense: A Study of Human Behavior During the World Trade Center Evacuation,” NFPA Journal 9:2 (March/April 1995) 59-67. Scholz, W., “Decision-Making Under Uncertainty: Biases, Fallacies, and the Development of Decision-Making,” Decision-making Under Uncertainty, R. Scholz, ed., North-Holland, Elsevier Science Pub., 1983, 3-18. Engländer, T., and T. Tyszka, “Information Seeking in Open Decision Situations,” Acta Psychologica 45:3 (1980) 167-176. Kahneman, D., and A. Tversky, “Variants of Uncertainty,” Judgment under Uncertainty; Heuristics and Biases, D. Kahneman, P. Slovic, and A. Tversky, eds., [?], Cambridge University Press, 1982, 509-520. Groner, R., M. Groner, and F. Bischof, “The Role of Heuristics in Models of Decision,” Decision-making Under Uncertainty, R. Scholz, ed., North-Holland, Elsevier Science Pub. 1983, 87-108. Simon, H. Models of Man: Social and Rational, New York, John Wiley & Sons, 1957. Keating, P., “Human Response During Fire Situations: A Role for Social Engineering,” Proceedings of Research and Design 85: Architectural Applications of Design and Technology Research, Los Angeles (1985) 285-288. Eysenck, M., “Incentives,” Stress and Fatigue in Human Performance, G. Hockey, ed., [London], John Wiley & Sons Ltd., 1983, 169-201. Wright, P., “Time Pressure, Distractions and the Use of Evidence,” Journal of Applied Psychology 59:5 (1974) 551-561.

76 77 78 79 80 81 82 83 84

85 86 87

Pauls, J., “Building Evacuation: Research Findings and Recommendations,” Fires and Human Behaviour, D. Canter, ed., New York, John Wiley & Sons, 1980. Fruin, J., Pedestrian Planning and Design, Revised ed., Mobile, Ala., Elevator World Educational Services Division, 1987. Habicht, A., and J. Braaksma, “Effective Width of Pedestrian Corridors,” J. of Transportation Engineering 110:1 (January 1984). Nelson, H., and F. Mowrer, “Emergency Movement,” SFPE Handbook of Fire Protection Engineering, 3d ed., Quincy, Mass., National Fire Protection Association, 2002. NFPA 101, Life Safety Code, Quincy, Mass., National Fire Protection Association, 2003. Pauls, J., “Movement of People,” SFPE Handbook of Fire Protection Engineering, 2d ed., Quincy, Mass., National Fire Protection Association, 1995. Proulx, G., “Movement of People: The Evacuation Timing,” SFPE Handbook of Fire Protection Engineering, 3d ed., Quincy, Mass., National Fire Protection Association, 2002. Proulx, G., “Evacuation Times and Movement Times in Apartment Buildings,” Fire Safety Journal 24 (1995) 229-246. Shields, T.J., K. Dunlop, and G. Silcock, “Escape of Disabled People from Fire: A Measurement and Classification of Capability for Assessing Escape Risk,” BRE Report 301, Borehamwood, Building Research Establishment, 1996. Predtechenskii, V., and A. Milinskii, Planning for Foot Traffic Flow in Buildings, New Delhi, Amerind Publishing Co., 1978. Frantzich, H., “Study of Movement on Stairs During Evacuation Using Video Analysing Techniques,” Lund, Lund Institute of Technology, 1996. Murosaki, Y., H. Hayashi, and T. Nishigaki, “Effects of Passage Width on Choice of Egress Route at a T-Junction in a Building,” 1st International Symposium on Fire Safety Science, New York, Hemisphere, 1986, 593-600. Jin, Y., “Visibility and Human Behavior in Fire Smoke,” SFPE Handbook of Fire Protection Engineering, 3d ed., Quincy, Mass., National Fire Protection Association, 2002.