You are on page 1of 10

Introduction

Integrating human
Fire safety management plans are often
behavior and response developed separately from fire safety designs,
issues into fire safety in many cases after the design is complete and
by someone other than the fire safety design
management of facilities engineer. If this occurs, and the design
engineer did not have a good appreciation for
Brian J. Meacham the operation of the facility, and the developer
The author of the fire safety management plan does not
Brian J. Meacham is the Research Director for the
have a good understanding of the fire safety
Society of Fire Protection Engineers. He is currently at systems, the resulting level of safety may be
Clark University, Worcester, Massachusetts, USA. less than expected. This is especially true
where human behavior and response factors
are not adequately addressed in the fire safety
Keywords
design, the fire safety management plan, or
Fire safety, Risk management, Behaviour, Alarms, both.
Fire services
If, for example, the fire safety systems for a
facility are designed in accordance with pre-
Abstract scriptive regulations, there are likely to be
Although there is a growing international movement some built-in assumptions regarding fuel load,
toward the use of engineered or performance-based fire fire spread potential, fire detection, fire alarm
safety design, current practice is dominated by prescrip- signaling, and occupant evacuation. These
tive-based design. In prescriptive-based fire safety assumptions could include the following: the
design, only those requirements prescribed by appropriate fuel load will not change significantly over
building regulations, installation standards, or approved time, fire and smoke doors will operate as
documents tend to be applied. Because these require- expected, fire detection and signaling systems
ments typically include fire protection measures, such as will provide sufficient early warning of a fire,
fire detection and signaling systems, automatic sprinkler and people will begin to evacuate immediately
systems, fire compartmentation, and emergency egress
on activation of the alarm. Unfortunately, all of
systems, there is often an assumption that occupants,
these assumptions could be wrong. As part of
employees, and users of a facility will be safe should a fire
living and working, people will change the
occur. However, there are a variety of factors that could
nature of a fire hazard. Combustibles of all
affect the actual fire safety of a facility that comply with
types accumulate in residences, hotels some-
the appropriate regulations. Fuel type, loading, config-
uration, and location can change, leading to an increase
times use corridors for temporary storage,
in fire risk. Occupants may not see, hear or understand businesses reconfigure open-plan office spaces,
fire alarm signals as fire alarm signals. Fire detection and the types of material stored in a warehouse
signaling systems, fire suppression systems, or smoke may change over time, and production equip-
management systems may not be 100 percent functional ment may change in industry.
at all times. Fortunately, many of these factors can be If the hazards change significantly from
controlled for, if they are understood and addressed, what the design engineer or the regulation
within a fire safety management plan. To assist with such
The author is indebted to Richard Custer and
planning, this paper discusses various human behavior
Pamela Powell for their contributions and
and response issues that may affect life safety during a assistance in preparing the original text of the
fire or emergency, and provides suggestions for inte- sections on human response to alarm signals,
grating these issues into a fire safety management plan. behavioral response to fire effects from which this
paper is derived (Custer and Meacham, 1997).
Electronic access The author would also like to acknowledge the
First International Symposium on Human
The current issue and full text archive of this journal is
Behaviour in Fire, where the above material
available at
serviced from Custer and Meacham (1997) was
http://www.emerald-library.com presented (Shields, 1998). The material presented
in this paper reflects the view of the author and
Facilities does not reflect the views, opinions, or consensus
Volume 17 . Number 9/10 . 1999 . pp. 303±312 of the Society of Fire Protection Engineers, its
# MCB University Press . ISSN 0263-2772 members, or its directors.
303
Integrating human behavior and response issues Facilities
Brian J. Meacham Volume 17 . Number 9/10 . 1999 . 303±312

assumed, the fire safety systems may be less immediately respond positively to the fire
effective than expected. In addition, people alarm signal (i.e. begin to escape as planned in
can inadvertently change the intended func- the fire safety design and written emergency
tion of a system. Fire and smoke doors only procedures)[1]. However, when developing a
work properly when they are able to close fire safety management plan, one should
during a fire. This cannot happen if they are question the validity of such assumptions,
wedged open by wood blocks or tied open by particularly the likelihood that all of these
wire or string. Sprinklers are most effective if actions will occur instantaneously and as
the water can reach the base of the fire. This is expected.
difficult if materials are stored up to the The first issue to consider is that many new
ceiling, effectively blocking sprinkler water and existing buildings only utilize bells, horns,
from reaching a potential fire. The number of or piezo-electric signaling devices to alert
egress paths from a building are selected so people, with voice alarm communication
that there is typically more than one way out. systems typically required or used only in
If materials are stored in an egress corridor, high-rise buildings and large assembly spaces.
an exit is effectively lost. If the materials are Unfortunately, people do not respond well to
combustible, they could make a fire situation non-voice signals. Bryan has cited one study
worse. that found ``the response to fire alarm bells
The above are all factors that most people and sounders tends to be less than optimum''
responsible for fire safety management plans (Bryan, 1995). Similarly, within United
understand and address in one way or States Department of Veterans' Affairs hos-
another. Good housekeeping, maintenance of pitals, the largest delay by staff in responding
fire safety systems, and good fire prevention to alarm signals was observed when coded-
practices (material control, ignition control) bell type signals were used (Cable, 1994).
are hallmarks of a thorough fire safety Part of the problem may be that people are
management plan. The more difficult issues, uncertain about what the signal actually
however, are those related to human behavior indicates. It has been reported that there is
and response once a fire occurs. If the fire usually skepticism as to whether the noise
safety design or the fire safety management indicated is a fire alarm and, if so, whether the
plan assume people will hear a fire alarm, alarm is merely a system test or a drill (Bryan,
recognize it as an alarm, and begin moving 1995). There is also the concern about
immediately to an exit, the time needed for whether or not the signal is a true alarm or
safe egress could be significantly underesti- false alarm. According to Moore, the decision
mated. If the time needed to safely egress is to leave a building when the fire alarm sounds
underestimated, the exposure to untenable will depend on occupant experience with the
conditions could also be underestimated. As a fire alarm system in the building, and the
result, there may be a higher likelihood of decision will often be no (Moore, 1988). In
injury than expected. such a case, delayed response results from the
However, if those responsible for the fire occupants' waning belief that the signal is a
safety management plan understand human true alarm. If the occupants do not acknowl-
behavior and response factors, adjustments edge the signal as a true alarm, they will delay
can be made if necessary. To help those with identifying the signal as indicating a fire or life
this responsibility, the following overview of safety threat. Various references on the
human behavior and response factors is nuisance alarm problem are available in the
provided. literature (e.g. Cable, 1994; Moore, 1988;
Meacham, 1994a; 1994b; Breen, 1985; Bu-
kowski and Jason, 1991).
Human response to alarm signals The next issue to consider is whether
people will be able to hear or see an alarm
Many regulation-prescribed fire safety designs signal. The problem of signal audibility and
assume that building occupants will be alerted visibility in buildings is not new, and is
by a non-voice signal (either from a smoke discussed in detail in the SFPE Handbook of
detector or as part of a building alarm Fire Protection Engineering (Schifiliti et al.,
system), will recognize that signal as a fire 1995) and elsewhere (Butler et al., 1981). For
alarm, will accept the signal as a true alarm this discussion, the focus will be on audibility
(i.e. not a false or nuisance alarm), and will issues other than attenuation though building
304
Integrating human behavior and response issues Facilities
Brian J. Meacham Volume 17 . Number 9/10 . 1999 . 303±312

materials (visibility issues will not be ad- noise ratio of less than 10dBA at the ears
dressed). An important factor in signal failed to awaken sleeping subjects (Kahn,
audibility analysis is the frequency of the 1984). In the case of a 60-65dBA ambient
signal. First, as the frequency of an alarm sound power level (noise) and a 70dBA alarm
signal increases, the sound pressure level of sound power level (signal), the signal to noise
the signal decreases (Schifiliti et al., 1995). ratio would be 10dBA or less. Would the
Second, people generally hear low frequencies occupant awaken? In Kahn's study, even
better than high frequencies. These factors when the signal-to-noise ratio was increased,
would seem to indicate that lower frequency and the subjects awoke, they failed to identify
alarm signals would be best for alerting the smoke detector signal, even though
people. This is supported by a recent study radiant heat and smoke odor cues were
which indicates that people, especially the present as additional indications of a fire. It
young and the elderly, respond better to should also be noted that, to meet code
alarm signals at 500Hz than to alarm signals requirements, apartment building alarm
at 4,000Hz (Huey et al., 1994). However, the sounders located in common corridors (i.e.
signal frequency in many residential smoke not inside units) may be set to sound at more
detectors is in the 4,000Hz range. than 100dBA to reach a level of 70dBA inside
The variability in ambient sound levels the individual units. In this case, the resultant
should also be considered. Ambient sound sound power level within units may be
levels vary by geographical location, season of insufficient to awaken some occupants (see
the year, function of a facility, and several above), and such a loud alarm in common
other factors. However, it is not uncommon corridors or staircases may prevent essential
for a regulation to indicate that the required communication among occupants (Proulx et
sound power level of an alarm signal must be al., 1995a).
at least XdBA or YdBA above ambient, with a Another important factor in alarm signal
minimum of supporting information describ- audibility analysis is the physical condition of
ing ``ambient'' conditions (BOCA, 1993; the occupant. Hearing an alarm or otherwise
NFPA, 1993). In many cases, a minimum sensing the presence of fire is only part of how
sound power level of 70-85dBA or 15dBA occupants respond to fire. Basic physiological
above ambient is specified, with a level of at factors greatly influence how people respond.
least 70-75dBA being required at ``pillow For example, two people could react very
level'' in sleeping rooms. Although basing a differently to the same fire; the same person
design on such minimum values may be could have a different physical reaction to a
adequate in most cases, these values may not fire today or a month from now. The
consider internal or external changes in a occupant's physiological ability to withstand
facility. Such changes may include increasing the physical assault of fire's heat, smoke, and
traffic noise levels over time, the installation gases depends on a combination of age, size,
of a room air-conditioner in summer months, weight, pre-existing physical or medical con-
reconfiguration of office or factory partitions, dition, and the presence of medication, drugs,
or the volume of a television set or piece of or alcohol. Although such inter- and intra-
equipment in an adjacent room, apartment, individual variability is seldom accounted for
or building. in regulations or by design professionals, it
For example, a ``typical'' residential air- should be. According to some estimates
conditioner has an ambient sound level of (Conley and Fahy, 1994), at least 10 percent
55dBA (Nober et al., 1981). With this level of of those who die in residential fires were
ambient noise, a regulation compliant alarm impaired by alcohol or other drugs when they
signal would need to be 70dBA (15dBA died, with 20-64 year-old fire victims being
above ambient). Although this complies with impaired twice as often as the general
the referenced regulations (BOCA, 1993; population. In addition, a 1970s study con-
NFPA, 1993), what would happen if the air- cluded that more than 70 percent of fire
conditioner were noisier, say 60-65dBA? The victims between the ages of 30 and 60 in
required sound power level of the alarm Maryland were legally intoxicated (Berl and
would then be 75-80dBA. Would this be Halpin, 1979). Given the indications that
assumed at the design phase, and if not, people with hearing impairments or who are
would the 70dBA level be sufficient to wake taking sleeping pills or other medication may
someone? Kahn indicates that a signal-to- require alarm signals with sound power levels
305
Integrating human behavior and response issues Facilities
Brian J. Meacham Volume 17 . Number 9/10 . 1999 . 303±312

exceeding 100dBA to alert them (Berry, heard an ``explosion'' (actually, sounds asso-
1978), the condition of occupants at the time ciated with rapid fire growth and spread)
of alarm is a serious concern. (Klem, 1987). Similarly, for more than an
hour, an alarm technician steadfastly believed
that an alarm was merely a trouble signal at
Behavioral response to fire the remote Illinois Bell Central Office facility
in Hinsdale, Illinois; a passing motorist finally
Assuming that building occupants will hear reported the fire (Custer and Powell, 1989).
the fire alarm signal, one should also consider In 1991, employees at the John F. Kennedy
how people may react to the alarm and other Federal Building in Boston, Massachusetts
fire cues. As this is a complex issue, the watched burning debris fall past their win-
discussion provided here will not attempt to dows, without evacuating the building or
duplicate work published elsewhere, such as pulling the manual alarm (Devir, 1996).
in Canter's comprehensive ``Human behavior Delayed threat recognition may seem inex-
in fire'' (Canter, 1990), but will focus on plicable to fire safety engineers, but it does
broad categories of human behavior in fire occur and must be considered in fire protec-
that are particularly relevant to engineered tion design.
designs. Validation: When people are unsure about a
After an occupant senses and understands threat's severity, they seek reassurance that
an alert or alarm signal, or identifies another the threat is mild. When threats seem more
fire cue (such as heat, flame or smoke), the serious, people will ask others for information.
next step is a process of interpreting this Seeking more information proves or validates
information and deciding if the alarm or fire that there is indeed a serious problem. ``Do
cue is a real threat to him/her. It has been you think we really need to evacuate the
reported that the decision process people use building?'' or ``Do you smell smoke?'' are
to decide what to do in a fire has six basic examples of validation questions. Definition:
components: recognition, validation, defini- Through definition, people structure or even
tion, evaluation, commitment, and try to quantify what they know about a threat.
reassessment (Bryan, 1991). Recognition: For Defining how much smoke people smell, how
lay people, the first cues of a fire can be very much heat they feel, or how many flames they
ambiguous and unclear. In fact, people may see helps people relate a threat to their own
worry that ``something isn't right'', and, at the situation. Evaluation: During evaluation,
same time, optimistically hope that the threat people decide whether to use fight or flight to
is false. In some cases, the transition between reduce their danger. The decision often
awareness of a fire cue and recognizing its happens very quickly, within seconds, and
threat is nearly simultaneous (if people see under great stress and threat. The actions of
flames and thick smoke 3m (10ft) in front of other people greatly influence the evaluation
them, they are quite likely to perceive the process. Commitment: Through the commit-
situation as life threatening). In other cases, ment process, people act on the decision
there are time gaps between becoming aware made during evaluation. Reassessment: If an
of fire cues and recognizing their threat. action does not appear to be working, the
Consider the impact of nuisance alarms and person in the fire becomes more anxious and
the ``cry wolf'' syndrome (Pate-Cornell and more frustrated. Decisions become less ra-
Benito-Claudio, 1987). The ratio of nuisance tional, and the likelihood of injury increases.
or false alarms from smoke detectors to real Another way to understand how people act
fires in college dormitories has been reported in fires is to view actions in the light of four
to be as high as 46:1 (Breen, 1990). If the socio-psychological concepts (Proulx, 1994a).
signal has been a ``false'' alarm for the past 40 These concepts are avoidance, commitment,
alarms, how likely are people to respond to affiliation, and role. Avoidance: People feel
the next alarm? that they can protect themselves, psychologi-
In general, recognizing a threat as real takes cally, by denying unpleasant situations.
time, as many fire incidents illustrate. In the Psychological denial is very common during
1986 fire at the Dupont Hotel and Casino in the first moments of a fire, when people find
San Juan, Puerto Rico, occupants remained in reassuring and benign explanations for the
the casino and ``gambling activities apparently cues they see, smell, and hear. Avoidance
continued'' even after they saw smoke and explains why people delay recognizing the
306
Integrating human behavior and response issues Facilities
Brian J. Meacham Volume 17 . Number 9/10 . 1999 . 303±312

threat and spend long minutes ignoring the fire cues very early in a fire, but found them
situation. Commitment: People are often very ambiguous (misinterpreting or ignoring
committed to their current activity (standing strange noises, or perhaps discussing them
in line at the supermarket, watching televi- with others). When the cues persisted, people
sion, or working at a desk, for example) when investigated, including entering a room with
they perceive their first cues of a fire. Because smoke or fire in it.
of commitment, people will continue what Researchers at the National Research
they are doing, despite warnings of danger. Council of Canada's National Fire Labora-
Many people, for example, will notice flashing tory and at the National Fire Protection
fire warning lights on their way into a hotel or Association have also studied human behavior
office building, but will continue into the during evacuation drills and actual emergen-
building. Affiliation: Affiliation is one cause of cies. According to Canadian research on
a delayed start of evacuation. Often no one evacuation drills in seven apartment build-
starts to evacuate until everyone in the group ings, it took people from 30 seconds to 14
(family members or friends) is ready. Once a minutes to start evacuating, with most of the
group has started to evacuate, the slowest people starting approximately three minutes
member determines the speed of movement after the alarm was sounded in buildings
for the entire group. Affiliation is such a where the alarm was readily audible
strong concept that it over-rides threat. throughout (Proulx, 1994b; Proulx et al.,
Regardless of the threat, parents will not leave 1994; 1995b). Some of the delays can be
without their children; children are reluctant attributed to not hearing the signal and not
to go without their siblings. People will wait linking the signal to the threat of fire. In all
for their co-workers, friends, or anyone they buildings, some delays were related to people
feel responsible for (such as an elderly deciding what to do (e.g. looking in the
neighbor living down the hall). Role: An corridor, looking for children, getting dressed,
occupant's role or status in a building gathering valuables, etc.). In the evacuation of
determines their response to a fire or other the World Trade Center following a bombing
emergency. For example, visitors to a building (Fahy and Proulx, 1995; 1996), which in-
will be more passive during an emergency volved a much larger and more complex series
than residents or employees. Simply because of buildings, there were remarkable simila-
they are unfamiliar with a building under rities to the Canadian study. In the World
normal conditions, visitors are likely to spend Trade Center evacuation, for example, occu-
more time on threat recognition and valida- pants did not initially consider the situation to
tion. Occupants will turn to supervisors and be extremely serious. Delay between becom-
others in authority for instruction. ing aware of the event and first attempting to
Cataloging what people reported they did leave the building was common; the mean
(and in what sequence) during a fire has been time between awareness of the event and
useful in putting the decision-making process attempting to leaving the building was 8.9
into better perspective. Specific reported minutes in Tower 1 and 39.9 minutes in
actions include ``notifying others'', ``called Tower 2. These two studies highlight some
Fire Department'', ``left building'', and very significant concerns for fire safety en-
``fought fire'' (NFPA, 1991). Note that this gineers to consider: inaudible alarms clearly
type of cataloging captures what occupants delayed the start of evacuation (audibility);
report (which may differ from what they people delayed evacuation further by actions
actually did), lacks a time context, and does such as verifying the signal or gathering
not necessarily reflect actions to prepare for personal effects (commitment); and occu-
evacuation. In recent studies of human pants who did not hear the alarm did not
behavior during emergencies (Fahy and move until the fire department arrived.
Proulx, 1996), occupant actions have been
grouped into the broader categories of ``in-
vestigate'', ``seek information'', ``prepare to Physiological response to fire effects
evacuate'', ``evacuate'', ``alert others or report
incident'', ``assist others'', ``seek refuge'', and As part of the life safety hazard assessment for
``wait''. Survivors of residential fires in the UK design, a critical component is the selection of
engaged in similar patterns, as Canter has tenability or survivability criteria: the trigger
reported (Canter, 1990). Occupants noticed points to assess success or failure of a
307
Integrating human behavior and response issues Facilities
Brian J. Meacham Volume 17 . Number 9/10 . 1999 . 303±312

proposed design. If a regulation-based design Heat Flux: Heat flux describes how much
is used, these criteria may not be known. In heat is available for transfer to any surface,
this case, it is important for the fire safety including human skin. Generally, the upper
management plan to consider the levels of limit of heat flux without severe pain is
hazard that may be present and adjust as 2.5kW/m2 for three minutes (Babrauskas,
appropriate. Common metrics selected for 1979). To put this into perspective, holding a
tenability and survivability criteria are related hand about 100mm (a few inches) from a
to conditions that cause physiological impacts 100-watt light bulb for three minutes would
on people. Such conditions include tempera- produce a heat flux of about 2.5kW/m2. A
ture, heat flux, smoke obscuration, oxygen recent report prepared for the SFPE provides
depletion, and exposure to fire gases. A an excellent discussion on skin burn issues
detailed discussion of physiological response due to radiant energy, and provides various
is well beyond the scope of this paper. engineering approaches for predicting first
However, this section will briefly review and second degree burns that can be a
physiological response to conditions that are valuable tool for life safety hazard assessment
particularly germane to life safety hazard (Wieczorek and Dembsey, 1997). Smoke
assessments used as part of fire safety obscuration: Visible smoke has two kinds of
management plans. All data provided are for negative physical effects during a fire: reduced
the purpose of illustration only. The refer- visibility and the irritation and toxicity caused
ences should be consulted for selection of by inhaling smoke particles. Smoke may also
appropriate data based on specific design have a marked emotional effect, by causing
parameters. fear. Although, recent studies have shown
Temperature: Effects of temperature vary that smoke often does not stop occupant
with the length of time exposed to heat, the movement, even in worsening conditions
amount of relative humidity (moisture in the (Fahy and Proulx, 1995), reduced visibility
air increases the effects of heat), and the does reduce the ability to escape (NIST,
``breathability'' of clothing worn. Tempera- 1984). Oxygen depletion: Oxygen normally
tures as low as 448C (1128F) can begin to makes up 21 percent of air, and the human
cause skin burns (Wieczorek and Dembsey, respiratory and nervous systems have evolved
1997) with temperatures as low as 508C to function at that concentration. When the
(1228F) producing severe discomfort in the percentage of oxygen drops even slightly,
mouth, nose, and esophagus (NFPA, 1959). physiological effects occur. The precise effects
School fire tests conducted in Los Angeles on an individual, of course, may vary widely
and known as Operation School Burning and are influenced by factors such as age and
concluded that children and teachers could overall condition. Table I summarizes some
effects of oxygen depletion on humans
not be expected to enter a corridor in which
(Custer and Bright, 1974).
the temperature was 658C (1508F) (NFPA,
Exposure to fire gases: Carbon monoxide
1959). It has also been reported that death
poisoning alone reportedly accounts for about
from hyperthermia occurs at temperatures
half of all fire deaths, with carbon monoxide
above 1008C (2128F) (Powell, 1997). Addi-
combined with other factors being responsible
tional information on other physiological
for another 30 percent of all fire fatalities
effects of heat are available in the literature
(Hartzell, 1986). Carbon monoxide easily
(e.g. Pryor and Yuill, 1966).
attaches itself to hemoglobin (in red blood
In estimating the potential for burn injuries,
cells), decreasing the blood's ability to carry
the combination of temperature and exposure
oxygen. It is so dangerous that even relatively
time required to cause a burn must be
considered (Hartzell, 1991). The higher the Table I Effects of oxygen depletion on humans
temperature, the more quickly a burn injury
Percent O2 Time Effect
will occur. For example, one reference reports
that the time required for a second degree 17-21 Indefinite Decreased respiration volume; loss of
burn to occur is 60 seconds with a skin coordination; difficulty thinking
surface temperature of 718C, 30 seconds with 14-17 Two hours Rapid pulse; dizziness
a skin surface temperature of 828C, and 15 11-14 30 minutes Nausea; vomiting; paralysis
9 Five minutes Loss of consciousness
seconds with a skin surface temperature of
6 1-2 minutes Death
1008C (Hartzell, 1991).
308
Integrating human behavior and response issues Facilities
Brian J. Meacham Volume 17 . Number 9/10 . 1999 . 303±312

small amounts may be incapacitating or one should address these factors depends on
lethal. One measure of toxicity is the exposure the facility and the characteristics of people
concentration (C) in parts per million (ppm) expected to live at, work in, visit, or otherwise
times the time (T) of exposure (in minutes) use the facility.
and is expressed as the CT product or ppm- One approach would be to consider the
minutes. Any concentration above 35,000 human factors as part of a life safety risk
ppm-minute is hazardous. Very high concen- assessment, wherein the variables are in-
trations for even very short times can be cluded in probability distributions, and a
incapacitating or lethal. Table II outlines quantitative risk-to-life value is determined
some of the effects of carbon monoxide using stochastic methods (Beck and Yung,
poisoning (Claudy, 1957). 1995; Beck, 1997; Magnusson et al., 1995;
Carbon monoxide levels are frequently Magnusson, 1997). Another approach would
expressed as the percentage of carboxyhemo- be to modify various performance or accep-
globin (COHb) saturation in the blood. In tance criteria by using subjectively or
general, carboxyhemoglobin levels below 30 objectively developed uncertainty and varia-
percent will not harm a person's ability to bility factors (Magnusson et al., 1995;
escape a fire. By 40 percent saturation, Meacham, 1997). This is common practice in
however, escape becomes much more diffi- deterministically developed designs.
cult. At saturation levels of 50-60 percent, Those responsible for the fire safety man-
severe symptoms and even death occur agement plan, however, may not need to
(Hartzell, 1989). It is important to note that perform any detailed risk or hazard assess-
carboxyhemoglobin levels, while a valuable ments. Rather, it may be possible to simply
tool in measuring toxicity, do not consider the determine the design assumptions and con-
length of exposure. The CT measurement, on ditions, verify whether those assumptions and
the other hand, does account for how long a conditions are valid for the facility, make
victim is exposed to the hazard. adjustments to fire safety systems and prac-
tices where necessary, and maintain the
facility fire safety management plan according
Integrating human behavior and to the assumptions and conditions. The
response issues into fire safety following suggestions are aimed at accounting
management of facilities for and integrating various human behavior
and response factors into the fire safety
For many fire safety engineers, integrating management plan.
human factors issues into engineered fire
safety design consists of incorporating fire Human response to alarm signals
detector activation times and travel times in a As noted previously, people have difficulty
computer model, sometimes with safety fac- recognizing fire alarm signals as such, espe-
tors, and developing an available safe egress cially non-voice signals. The degree to which
time (Proulx, 1995). However, as discussed people will have problems will be related, in
throughout this paper, delays are inherent in part, to the facility. In an apartment building,
human response to a fire. Therefore, designs bells or electronic signals may have a better
should consider delays prior to occupants chance of being associated with a fire alarm as
beginning to evacuate, delays during evacua- such sounds are not normal in that environ-
tion, and the potential for some people not to ment. This assumes that the building-wide
evacuate on their own. The extent to which alarm can be distinguished from single-station

Table II Human response to carbon monoxide


CCO (ppm) Time (min) CT (ppm-min) Effects
200 120-180 24,000-36,000 Mild headache
800 45 36,000 Mild headache
3,200 10-15 32,000-48,000 Dizziness
3,200 30 96,000 Death possible
6,900 1-2 6,900-13,800 Dizziness
12,800 0.1 (2-3 breaths) 1,280 Unconsciousness
12,800 1-3 12,800-38,400 Death

309
Integrating human behavior and response issues Facilities
Brian J. Meacham Volume 17 . Number 9/10 . 1999 . 303±312

smoke detectors (in individual apartments), back on line when work is completed, and
microwave oven alarms, alarm clocks, and that some type of fire watch is in place while
similar ``normal'' alarm signals. In medical or the system is off line.
industrial facilities, where bells or electronic People also delay because they do not
signals may be used for other purposes, the believe the information they are given by
distinction will be more difficult. If multiple alarm systems. In such cases, it is useful and
non-voice signals are used in a facility, helpful to have fire wardens (in appropriate
periodic reminders (memos) or drills could be occupancies) and to drill at least once a year.
used to help occupants (employees) maintain If fire wardens are used, they should be
a needed familiarity with the signal and with people in positions of respect, especially in
the action to be taken on hearing it. The best business and industrial occupancies. This is
approach, where feasible, would be to sup- necessary so that if word is given to evacuate
plement the non-voice alarm with a voice the building, the orders are followed by all
alarm system. In residential occupancies, employees without the need for the employee
such as apartment buildings and hotels, to gain the approval of his or her supervisor.
installation of a voice alarm system speaker in Fire safety management plans should be
each apartment or guest suite would be ideal. developed with an understanding of depen-
People may also have difficulty hearing or dencies and affiliations that may exist
seeing an alarm signal. From a signaling between occupants of a facility. Familial
perspective, this problem is simple to test for affiliations are clear in residential facilities, as
using a sound power meter or a light meter are medical dependencies in health care
respectively. For audible signals, the first step occupancies. There are also teacher-student
is to determine the ambient and peak sound affiliations in schools. However, there may be
power levels throughout the facility. The alarm familial affiliations in such facilities as stores,
should then be sounded and verified as being malls, restaurants, and places of assembly,
at least 15dBA above the ambient levels (or where one member of a family may not leave
peak levels if sustained for long periods of without another, even under worsening con-
time). For visual alarms, testing can begin by ditions. If these affiliations are recognized,
visually evaluating the signals. If the device training of fire wardens can be improved.
cannot be seen directly, the light should still be Overall, one of the most important factors
visible in all parts of a room. If the light cannot in successfully addressing human behavior
be seen, or is dim, additional visual alarm and response in the fire safety management
devices are likely required. Calculations for plan is to provide people with information.
determining audible and visual alarm signal Written instructions for emergencies, well-
requirements can be found in the SFPE trained fire wardens, and well-conducted fire
Handbook (Schifiliti et al., 1995). As noted drills are useful mechanisms for accomplish-
earlier in this paper, factors such as occupant ing this. The more relevant information that
age, physical condition, and medical condition people have at their disposal, the better
should be considered when selecting appro- decisions they will normally make. This
priate alarm signals and signal strengths. applies not only during emergency situations,
when occupants, employees, and visitors need
Behavioral response to fire to be notified of the emergency, but during
One of the reasons that people delay in normal operations as well. If maintenance
responding to an alarm signal, even after they staff, for example, understand the importance
have identified it as such, is that they are of maintaining fire safety systems, and of not
uncertain if the alarm is ``real''. To avoid blocking egress paths, the likelihood of unin-
delays of this type, fire safety management tended consequences from a fire emergency
plans should include efforts to reduce false diminishes.
and nuisance alarms. This can be done by
utilizing new technology, by systematically
cleaning and maintaining detection devices, Summary
and by carefully taking parts of a detection
system off line during times of construction, Fire safety engineers may be unaccustomed to
heavy dust, or production of other non-fire considering human behavior and response
contaminants. When the latter is done, steps factors in their analyses and designs due to
should be taken to ensure the system is placed routine application of prescriptive-based
310
Integrating human behavior and response issues Facilities
Brian J. Meacham Volume 17 . Number 9/10 . 1999 . 303±312

regulations. However, these factors are im- Breen, D.E. (1990), ``Toward more reliable residential
portant, as they impact the time required for smoke detection systems'', Journal of Fire Protection
Engineering, Vol. 2 No. 1, Society of Fire Protection
people to recognize a fire alarm as an
Engineers, Boston, MA, pp. 1-10.
indication of a fire threat, for people to decide Bryan, J.L. (1991), unless otherwise noted, the discussion
to evacuate, for emergency service to rescue in this section is based on Bryan, J.L. ``Human
people, and for fire effects to impact people. behavior and fire'', Fire Protection Handbook, 17th
Recognizing these factors in a fire safety edition, National Fire Protection Association,
management plan is also important, as the Quincy, MA.
Bryan, J.L. (1995), ``Behavioral response to fire and
building's fire safety systems can be tested
smoke'', FPE Handbook of Fire Protection Engi-
and evaluated with these factors in mind. In neering, 2nd ed., National Fire Protection
addition, information and training can be Association, Quincy, MA.
provided to help facilitate fire prevention and Bukowski, R. and Jason, N.H. (1991), International Fire
timely and orderly emergency egress. Human Detection Literature Review and Technical Analysis,
National Fire Protection Research Foundation,
behavior and response factors pertinent to life
Quincy, MA.
safety hazard assessments and fire safety Butler, H., Bowyer, A. and Kew, J. (1981), ``Locating fire
management plans have been provided, as alarm sounders for audibility'', Building Services
well as references for various response criteria, Research and Information Association.
and suggested methods for addressing these Cable, E.A. (1994), ``Cry wolf syndrome: radical changes
factors. Although significant research and solve the false alarm problem'', Department of
Veterans' Affairs, Albany, NY, January.
development is yet required to readily quan-
Canter, D. (1990), ``Human behavior in fire'', Fires and
tify and integrate human factors issues into Human Behavior, 2nd ed., Fulton Publishers Ltd,
engineered fire safety design on a widespread London.
basis, the present state of knowledge provides Claudy, W.D. (1957), adapted from Claudy, W.D.
useful information for enhancing current fire Respiratory Hazard of the Fire Services, National
Fire Protection Association, Boston, MA.
safety management plans.
Conley, C.J. and Fahy, R.F. (1994), ``Who dies in fires in
the United States?'', NFPA Journal, May/June,
p. 103.
Note Custer, R.L.P. and Bright, R.G. (1974), Fire Detection:
State of the Art, National Bureau of Standards
1 Material in this section, as well as in the sections Technical Note 839, Gaithersburg, MD.
Behavioral Response to Fire and Physiological Custer, R.L.P. and Meacham, B.J. (1997), Introduction to
Response to Fire Effects, is excerpted from Chapter Performance-Based Fire Safety, ISBN 0-87765-421-
8 of Custer and Meacham (1997), and is used with 2, Society of Fire Protection Engineers (published by
permission. NFPA, Quincy, MA).
Custer, R.L.P. and Powell, P. (1989), Firesafety at NYNEX:
Your Piece of the Action.
Devir, J. (1997), Private communication between P.A.
References Powell and J. Devir, GSA.
Fahy, R.F. and Proulx, G. (1995), ``Collective common
Babrauskas, V. (1979), Full-scale Burning of Upholstered sense: a study of human behavior during the World
Chairs, Technical Note 1103, National Bureau of Trade Center evacuation'', NFPA Journal, March/
Standards. April, pp. 59-67.
Beck, V.R. (1997), ``Performance-based fire engineering Fahy, R. and Proulx, G. (1996), ``A study of occupant
design and its application in Australia'', Proceedings behavior during the World Trade Center Evacua-
of the 5th International Symposium on Fire Safety tion'', Proceedings of Interflam `96, the 7th
Science, ISBN 4-9900625-5-5, IAFSS, pp. 23-40. International Fire Science and Engineering Confer-
Beck, V.R. and Yung, D. (1995), ``Building fire safety risk ence, Cambridge, Interscience Communications,
analysis'', SFPE Handbook of Fire Protection March 26-28, pp. 793-802.
Engineering, Section 5, Chapter 11, SFPE/National Hartzell, G.E. (1986), ``Combustion products and their
Fire Protection Association, Quincy, MA. effects on life safety'', Fire Protection Handbook,
Berl, W.G. and Halpin, B.M. (1979), ``Human fatalities in 16th ed., National Fire Protection Association,
unwanted fires'', Fire Journal, September. Quincy, MA.
Berry, C.H. (1978), ``Will your smoke detector wake you?'', Hartzell, G.E. (Ed.) (1989), Advances in Combustion
Fire Journal, Vol. 72, p. 105. Toxicology, Vol. 1, Technomic Publishing Company
BOCA (1993), The BOCA National Building Code 1993, Inc., p. 23.
Building Officials and Code Administrators Interna- Hartzell, G.E. (1991), ``Combustion products and their
tional. effects on life safety'', Fire Protection Handbook,
Breen, D.E. (1985), ``False fire alarms in college 17th ed., National Fire Protection Association,
dormitories: the problem revisited'', Journal of Fire Quincy, MA.
Protection Engineering, Vol. 3 No. 3, Society of Fire Huey, R.W., Buckley, D.S. and Lerner, N.D. (1994),
Protection Engineers, Boston, MA. ``Audible performance of smoke alarm sounds'',
311
Integrating human behavior and response issues Facilities
Brian J. Meacham Volume 17 . Number 9/10 . 1999 . 303±312

Proceedings of the Human Factors and Ergonomics Factors and Ergonomics Society 38th Annual
Society 38th Annual Meeting, pp. 147-51. Meeting, pp. 811-15.
Kahn, M.J. (1984), ``Human awakening and subsequent Proulx, G. (1995), ``Human factors in fires and fire safety
identification of fire related cues'', Fire Technology, engineering'', SFPE Bulletin, Society of Fire Protec-
Vol. 20 No. 1, pp. 20-6. tion Engineers, Boston, MA, Winter, p. 15.
Klem, T.J. (1987), ``97 die in arson fire at Dupont Plaza Proulx, G., Laroche, C. and Latour, J.C. (1995a),
Hotel'', Fire Journal, May/June, p. 79. ``Audibility problems with fire alarms in apartment
Magnusson, S.E. (1997), ``Risk assessment'', Proceedings buildings'', Proceedings of the Human Factors and
of the 5th International Symposium on Fire Safety Ergonomics Society 39th Annual Meeting.
Science, ISBN 4-9900625-5-5, IAFSS, pp. 41-58. Proulx, G., Latour, J. and MacLaurin, J. (1994), Housing
Magnusson, S.E., Frantzich, H. and Harada, K. (1995), Evacuation of Mixed Abilities Occupants, National
``Fire safety design based on calculations: uncer- Research Council Internal Report, No. 661, National
tainty analysis and safety verification'', ISSN 1102- Research Council Canada, Ottawa, July.
8246, Report 3078, Lund University, Sweden. Proulx, G., Latour, J., MacLaurin, J., Pineau, J., Hoffman, L.
Meacham, B.J. (1994a), ``International developments in and Laroche, C. (1995b), Housing Evacuation of
fire sensor technology'', Journal of Fire Protection Mixed Abilities Occupants in Highrise Buildings,
Engineering, Society of Fire Protection Engineers, National Research Council Report, No. 706,
Boston, MA, Vol. 6 No. 2. National Research Council Canada, Ottawa,
Meacham, B.J. (1994b), ``The use of artificial intelligence Ontario, August.
techniques for signal discrimination in fire detection Pryor, A.J., and Yuill, C.H. (1966), Mass Fire Life Hazard,
systems'', Journal of Fire Protection Engineering, OCD Work Unit 2537A, Southwest Research
Society of Fire Protection Engineers, Boston, MA, Institute, San Antonio, TX.
Vol. 6 No. 3. Schifiliti, R.P., Meacham, B.J. and Custer, R.L.P. (1995),
Meacham, B.J. (1997), ``Identifying and addressing ``Design of detection systems'', Section 4, Chapter 1,
uncertainty in fire protection engineering'', SFPE Handbook of Fire Protection Engineering, 2nd
Proceedings of the 2nd International Conference on ed., National Fire Protection Association, Quincy,
Fire Research and Engineering, NIST, Gaithersburg, MA, pp. 4-1±4-29.
MD, August. Shields, T.J. (Ed.) (1998), ``Integrating human factors into
Moore, W. (1988), ``When the fire alarm sounds, will engineered fire safety design'', Human Behaviour in
anyone leave the building?'', Fire Journal, Vol. 82 Fire: Proceedings of the 1st International Sympo-
No. 1, pp. 54-6. sium, Fire SERT Centre, University of Ulster.
NFPA (1959), Operation School Burning, National Fire Wieczorek, C.J. and Dembsey, N.A. (1997), Engineering
Protection Association, p. 25. Guide for Predicting First and Second Degree Skin
NFPA (1991), Adapted from Tables 7-1E and 7-1F, Burns, prepared for the SFPE Engineering Task
``Human behavior in fire'', Fire Protection Handbook, Group on Engineering Practices, SFPE, March.
17th ed., National Fire Protection Association,
Quincy, MA. (Brian J. Meacham is the research director
NFPA 72 (1993), National Fire Alarm Code, National Fire for the Society of Fire Protection Engineers.
Protection Association. He holds an MSc degree in Fire Protection
NIST (1991), Technical Reference Guide to HAZARD I, Engineering from Worcester Polytechnic
Version 1.1, National Institute of Standards and Institute, and is currently working towards a
Technology, pp. 87-8.
PhD in Risk and Public Policy at Clark
Nober, E.H., Pierce, H., Well, A, Johnson, C.C. and Clifton,
C. (1981), ``Waking effectiveness of household University, Worcester, Massachusetts. He is
smoke and fire detection devices'', Fire Journal, a licensed professional engineer with over 14
Vol. 75, p. 86. years' experience in the USA and Europe.
Pate-Cornell, M.E. and Benito-Claudio, C.P. (1987), He is the SFPE liaison to CIB W14, CIB
``Warning systems: response models and optimiza- TG39 and ISO TC92, and a member of
tion'', Uncertainties in Risk Assessment, Risk
several national and international commit-
Management, and Decision Making, Plenum Press,
New York, NY, pp. 457-68. tees. He has written over 50 publications in
Powell, P. (1997), ``Learning fire and life safety funda- the area of building fire safety analysis and
mentals'', Fire and Life Safety Educator, 2nd ed., design. He is currently a co-principal in-
International Fire Service Training Association, vestigator for a joint US National Science
Oklahoma State University, Stillwater, OK. foundation/service sector initiative entitled
Proulx, G. (1994a), ``Human response to fires'', Fire
``encouraging the use of risk concepts in
Research News, National Research Council Canada,
Winter. performance-based building and fire regu-
Proulx, G. (1994b), ``The time delay to start evacuating on lation development''. He can be contacted
hearing a fire alarm'', Proceedings of the Human by email: bmeacham@sfpe.org)

312

You might also like