You are on page 1of 12

BUILDING EVACUATION: TWO DIFFERENT APPROACHES

Elvezia M. Cepolina
Dipartimento di Ingegneria Civile, Università di Pisa

Introduction
According to the Engineering Guide to Human Behaviour in Fire [BCF02], a
building evacuation is successful if the Available Safe Egress Time results longer
than the Required Safe Evacuation Time.
The Available Safe Egress Time is defined as the time period between the
occurrence of a catastrophic event and the onset of untenable conditions for one
or more building areas. Estimations of the available safe egress time 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.
The Required Safe Evacuation Time is the time needed for the last pedestrian to
leave the building. We will call the Required Safe Evacuation Time simply
evacuation time. Predictions of the evacuation time typically involve estimating
the time that it would take for people to be notified that there might be a danger
(typically 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.

The possibility of having a successful evacuation will increase if the evacuation


time decreases or if the Available Safe Egress Time increases. Since the
evacuation time mainly depends on the building design and on emergency
procedures whilst the Available Safe Egress Time depends on the strength of the
building to the catastrophic event, actions aimed at decreasing the evacuation
time result “softer”. This paper focuses on the evacuation time.

It is possible to operate in two different but complementary ways for reducing the
evacuation time:
1. adapting the building environment to pedestrian behaviour and making it as
suitable as possible for the spontaneous evacuation process: in the following
this approach is called person approach;
2. controlling pedestrian behaviour in such a way to make the best possible use
of a given building: in the following this approach is called system approach.

The person approach requires a deep understanding of pedestrian behaviour in


panic situations and of the way in which the environment affects it. Simulation of
a spontaneous evacuation in a given environment allows to identify critical points
where poor design of architectural spaces and/or signage system could
potentially cause disorientation and confusion. This approach is oriented towards
the definition of prescriptive regulations about the design of architectural spaces.

© Association for European Transport 2004


Given an environment, the system approach assesses which is the pedestrian
behaviour that guarantees the minimum evacuation time; the underpinning
assumption is that this “perfect” behaviour could be imposed to the building
occupants through education and training. This approach is oriented towards the
definition of prescriptive regulations about the occupants behaviour.

The paper is organised as follows: the first section presents the different phases
of an evacuation process. Evacuation time has several components, each one
referring to a specific phase; the relative weight of these components on the
evacuation time changes depending on the circumstances.
Section two concerns the person approach and section three concerns the
system approach. Each section analyses the actions that could be taken for
reducing each specific component of the evacuation time. Different models and
simulations aimed at designing these actions or assessing their impact on the
evacuation process will be reviewed.
The paper concludes with some general indications for the application fields of
the system approach and of the person approach.

1. Phases of an evacuation process


One of the key features of occupant response to catastrophic events illustrated
by some real incidents is that some time may elapse before occupants become
fully aware and convinced of the need to change their behaviour: people are not
in building with emergency uppermost in their mind, but for other purposes to
which they have a considerable commitment. In order to attend to some other
demand they have to recognise its importance and cease their normal activities
[PuB01]. Moreover real fire incidents illustrated to some extend the ”friendly fire
syndrome” in which people may not feel threatened initially and often
misunderstand the rapid rate at which fires, particularly fire inside enclosures, can
grow.

The “Human Behaviour in Fires” report [SCK92] underlines that the time to
escape should take into account:
1. pre-movement time: pre-movement processes begin at an alarm or cue and
end when travel to an exit begins. There are two components:
i. recognition: it begins at an alarm or cue and ends with the first
response;
ii. response: it begins at the first response and ends when the travel to an
exit begins.
2. time to travel to and through exits: it strictly depends on the chaos in the
building that is a direct consequence of the human irrational behaviour that
typically emerges in panic situations. Irrationality in human behaviour could
affect either
 the dynamic of pedestrian flow and
 the exit choice behaviour.
The relative weight of pre-movement time and time to travel to and through exits
on the overall evacuation time is a function of several items.
Pre-movement time results the major determinant of evacuation time:

© Association for European Transport 2004


1. when the time to travel to and through the exits is limited i.e. when congestion
doesn’t occur during the evacuation and when the length of the escape routes
is limited: a series of monitored evacuation studies and investigations of fire
incidents in a range of different building [PuB01] underlined that.
2. in multi-enclosure buildings where the occupants are widely distributed (such
as hotels and apartment blocks). The dominance of the pre-movement time in
this situation could be due to two main reasons:
 Very often occupants could not hear the fire alarm: in a building such
as an hotel, when sounders are used as alarms, it is quite common for
occupants to ignore them completely, particularly those remote from
fire.
 If occupants are isolated each person has to react independently and
there is likely to be a wide variation in times to start evacuating.
Conversely in group situations it was found [PuB01] that response time
is very dependent upon a few key group members who facilitate the
evacuation and the range of times to start an evacuation tend to be
similar.
Experiments [Pro95] showed that in buildings where the alarm was audible to
most occupants, the mean time to start was 2 min 49 s while in other
buildings where 25% of the occupants could not hear the fire alarm in their
apartment, the occupants started their evacuation only when the arriving fire-
fighters knocked on their doors, providing a mean time to start the evacuation
of 9 min 2 s.
3. in building where people are not familiar with evacuation plans and escape
procedures: in the monitored evacuations from two supermarkets [PuR95],
[PuB99] showed that shoppers took very little notice of the alarm sounder, the
majority continuing to shop. This emphasises the ineffectiveness of sounders
in building where the occupants are not trained to respond. In general
supermarkets empty quickly only when a large number of staff are available
to put into practice an emergency plan: alerting the occupants and ushering
them to the exits.

The two approaches on which this paper is focused allow to reduce either the
pre-movement time and the time to travel to and through exits; in the following we
will focus on how the two approaches involve each component of the evacuation
time.

2. Person approach
The configuration of the environment acts as both a constraint and as a facilitator
of building evacuation.
The environment is defined by its morphology which includes aspects related to
the geometry of built space and to alarm and signage systems.
Understanding how the environment affects pedestrian behaviour allows to act on
the environment design in such a way to facilitate the evacuation.
The understanding of human behaviour during emergency situations comes from
observation of real incidents life evacuation processes and field experiments. The
first source of information is quite rare; the second source is characterised by
weak reliability because of the difficulty of reproducing in vitro panic situations.

© Association for European Transport 2004


Researches according to this approach are mainly focused on modelling deeply
the spontaneous human behaviour in a given environment during an evacuation.

2.1. Pre-movement time


According to the person approach, pre-movement time can be reduced mainly
acting on the alarm system.
The alarm system acts on the recognition time. Experimental results confirmed
that a voice alarm is more effective and produced a quicker response than a
sounder and that short messages produce the quickest response [PuB01]. Two
stage alarm systems are often used. The initial alarm alerts security or
management, who are then able to investigate the incident before deciding
whether or not to trigger the general alarm. This prevents evacuations from
occurring for trivial reasons but increases the evacuation time.

2.2 Time to travel to and through exits


Time to travel to and through exits is determined by pedestrian movement
behaviour and wayfinding behaviour.
While pedestrian movement is strongly affected by the geometry of built space,
pedestrian wayfinding is influenced by the signage system.
Hereinafter these two behaviours are separately dealt with.
2.2.1 Pedestrian movement
Many researches have been carried out trying to model realistically the
spontaneous dynamic of pedestrian flow in panic situations. The target is to
understand to which extend environment details affect pedestrian flow and
therefore impact on the evacuation time.

Different theoretical approaches for pedestrian dynamics show to be suitable for


assessing the space geometry effects on pedestrian dynamic in panic situations.
Microscopic and cellular automaton models have the advantage to take into
account pedestrian flow as a collective motion of individuals and differences in
characteristic proprieties of individual pedestrians. Such models are able to
represent important features of flow dynamic and allow to assess the effects of
the environment geometry on pedestrian flow: these models are able to
reproduce among others:
 the capacity drop due to clogging effects. An increase in the desire to leave
the facility may reduce the capacity of the bottleneck due to the formation of
arcs in front of the bottleneck: this is the so-called faster-is-slower effect.
 The lanes formation. In uni-directional flows through bottlenecks, capacity is a
step-function of the width of the bottleneck, depending on the number of lanes
that can be formed within the bottleneck [Hoo04]. Under saturated conditions
but not in panic situations, overlapping lanes form dynamically. The lateral
distance between the lanes is smaller than lateral space required by the
pedestrians, implying that the lanes are in fact merged together (zipped) and
move at the same speed [Hoo04].
 Unseen blockages. In presence of a wide distribution of pedestrian speeds,
people pushing causes some persons to fall and blockages arise.
Cellular automaton models and microscopic models are mainly focused on
pedestrian collision avoidance behaviour: the flow patterns result from the
impulsive reactions of each single pedestrian to other pedestrians or to the

© Association for European Transport 2004


environment within his local surrounding. The implicit assumption in most of
these models is that occupants know their destination and the path for reaching it
and hence they do not require wayfinding information: a route choice behaviour is
not considered and pedestrian planning abilities are not modelled.
Some of the models able to reproduce said aspects of crowd dynamic are
reported hereinafter.
Among the cellular automaton models, it is interesting the floor field model
[KNN03]: according to it each pedestrian can move to one of its unoccupied
neighbour cells at a discrete time step according to different transition
probabilities. The transition probabilities are determined by the interaction
between two discrete floor fields: the dynamic floor field represents a virtual trace
left by moving pedestrians, the static floor filed describes the shortest distance to
an exit door: the field value increases in the direction of the exit.
Individual based models often have been studied via their analogy with granular
flows. The similarity between granular materials and pedestrians is that both
motions are affected by interaction forces from surrounding pedestrians or
granules. The axiomatic difference lies in the fact that pedestrians are provided
with intelligence and intentions. Among the individual based models it is worth to
mention the social force model [HFV00] and the model developed by Tsuji
Yutaka [Tsu03].
Among the commercial simulators, are worth to be mentioned PedGo [KlM03]
and Simulex [ThM94].

These models are aimed to investigate the mechanisms of panic and jamming by
uncoordinated motion in crowds. Simulations allow to test the effects of changes
in the building design. For instance according to the Helbing model, placing an
additional column in a slightly asymmetric position in front of the exit reduces the
number of conflicts in front of the door, reducing the evacuation times [HFV00].
2.2.2 Wayfinding
Garbrecht studied the difference between path and walk, where path refers to an
initial choice of a complete path from origin to destination and walk describes a
movement where a person makes a choice at each intersection. A path choice is
possible if there is knowledge about alternatives paths: pedestrians know the
path by which they entered the building; alternative paths could be known only if
pedestrians are familiar with the building or information are provided in the origin
room, for instance through an evacuation plan.

Cognitive psychologists study hypothesis about how individuals gather


information and how information is used to select routes. Sadalla et alii [SaS80]
underlined that routes are perceived longer if the person has to process a lot of
information on his way. This observation seems to be consistent with the fact that
routes with many turns are perceived longer than routes with few turns and often
occupants try to leave the building following the way they entered it [Sim94].
Moreover it was observed in a simulated fire emergency in apartment buildings
[Pro95] that people move towards areas they are more familiar with [Sim85].
Areas around the elevators and stairs leading to main exits are seen as familiar:
people may use stairs that are less convenient in proximity of their apartment
simply because they spontaneously go to areas which they are familiar. This

© Association for European Transport 2004


behaviour is known as “the affiliation model”. The tendency to limit information
processing results also in pedestrian predisposition to follow persons with
authority, especially in emergency situations [Bro91].
Pedestrian spontaneous exit route choices is not always efficient in emergency
situations also because in many cases pedestrian knowledge about alternative
paths is poor. Occupants may be unaware of the existence of the most suitable
means of escape in emergency situations: emergency exits not used for normal
circulation are often ignored and in emergency situations, where smoke may also
obscure vision, the problem is often fatally compounded. The presence of
signage in an enclosure is therefore essential for reducing the amount of time
spent in wayfinding particularly in large, complex or unfamiliar structures.
Signage systems allow the occupants to gain knowledge of the location of a
previously unknown exit.

Many simulation models concern the way in which pedestrians gather information
and process them during an evacuation process: research have been carried out
for modelling pedestrian interactions with wayfinding systems. The target is to
assess how the morphology of the environment, and particularly the information
systems, affects the occupants’ wayfinding behaviour and thus how the signage
system should be designed to help confused pedestrian in finding a suitable exit
way. The target is to determine whether there are any problem areas, or
unforeseen reactions to the design during egress.
A review of wayfinding literature could be found in [Lov97]. Among the simulators,
are worth to be mentioned EvacSim [Lov98], EXODUS [FGG03] and Virtual
reality models.
In these simulators, the agents are given movement rules that depend on what
they can see from where they are currently in the environment. Vision allows
pedestrians to interact with both the environment and other pedestrians over long
distances, on a scale much greater than their physical size. Individuals are
assumed to select their paths based on their own cognitive information of the
environment but they are allowed to reselect goal, destination or path if they get
new information. Models could differ for the rule the gathered information plays in
the pedestrian decision making process. In EXODUS when occupants become
aware of an exit they will only evacuate through that exit if it is closer than the exit
they were originally heading for.

It is worth to mention that all the models which reproduce walk choices include a
movement model as the choices are taken step by step as pedestrians proceed
towards the exit.

3. System approach
Evacuation time depends on pedestrian behaviour and spontaneous human
behaviour delays the evacuation process, either as for the pre-movement time
and pedestrian movement and for wayfinding.
The target of the system approach is to define which is the pedestrian behaviour
that guarantees the minimum evacuation time in a given environment.
Whereas the person approach is aimed at adapting the building environment to
pedestrian behaviour and making it as suitable as possible to spontaneous

© Association for European Transport 2004


evacuation process, the system approach aims to control pedestrian behaviour in
such a way to make the best possible usage of a given building.

3.1. Pre-movement time


According to the system approach, the response time can be reduced training the
building population to respond promptly to alarm signs. Guidelines and
recommendations could address a correct behaviour during emergency, telling
the building occupants what they should do and what they shouldn’t do. The need
for this kind of recommendations comes from field experiments: during an
unannounced evacuation [PuB01] from a meeting room occupants responded
quickly after hearing the message but engaged in activities (such as gathering
coats and belongings before leaving the room: one occupant left the room and
then came back to retrieve the jacket) that required 51 s in average, while the
time taken to reach the door was only 5 s.

3.2 Time to travel to and through exits


Time to travel to and through exits strictly depends on the chaos in the building
that is a direct consequence of the irrational movement behaviour and wayfinding
behaviour that typically emerge in panic situations. Increasing people confidence
in, and awareness of, the actions to take is therefore a critical issue.
The actions people have to take should be planned taking into account the
overall context: the given building design and the building population.
3.2.1 Pedestrian movement
As for flow dynamic, the best pedestrian behaviour should be the one that
guarantees no activation of capacity drop. As capacity drop arises when pressure
increases, controlling pedestrian behaviour means do not allow pedestrians to
run and push.
Clogging effects could therefore be avoided either placing a column in front of an
exit according to the person approach or not allowing the building occupants to
push, according to the system approach.

3.2.2 Wayfinding
Controlling pedestrian behaviour in such a way to make the best possible usage
of a given building means “imposing” to the building population the paths people
have to follow. In fact the best evacuation time is achieved when the building
population follows the escape routes that respond to a “system optimum” rather
than the ones spontaneously chosen by each individual on the base of his limited
knowledge.
The information provided to the building population could concern:
1. the overall escape route (path) from each room where pedestrians are when
the evacuation starts. Each route is defined by an exit and by the escape
route leading to the exit the pedestrians in the room have to follow.
2. the turning movements (walk) at each intersection.
In the first case, information are provided once in each origin room. The
information are shared by all pedestrians within the same origin room. The route
guidance is static and people’s awareness on the escape routes could be
achieved through a priori training.

© Association for European Transport 2004


In the second case information are provided via signage systems several times
along the escape path. All the pedestrians passing at a given intersection,
whichever had been their paths from the origin rooms, gather the same
information. The information provided could be static and coherent with the
escape routes specified in the origin rooms but, since this sign system reaches
pedestrians at different times during the evacuation, this signage system should
be suitable also for dynamic route guidance. The route re-planning should
dynamically use flow pattern information.

We will review the models aimed at defining the best set of escape routes
(evacuation plan) for a given building and building population. These models
have:
1. to find, among all the possible evacuation plans, the one that gives the
lower evacuation time: an optimisation algorithm is therefore required; as
the evacuation time we want to minimise could be a multi-peak function
and since its configuration space (given by all the possible evacuation
plans that can be defined in the building) is discrete and extremely large, a
random search algorithm could be useful.
2. to evaluate the evacuation time for each evacuation plan. As evacuations
are characterised by some form of congestion, a dynamic network loading
model is required to keep track of the location of moving queues in the
network and to predict spillbacks and dissipation. As the model has to be
implemented within the optimisation algorithm, it needs to be more
synthetic than the models previously described, i.e. it cannot take into
account the detailed geometry of buildings and obstacles.
As for the second issue, the evacuation time related to a given evacuation plan
could be evaluated:
i. assuming an optimum pedestrian movement which does not activate
capacity drop in any section.
ii. assuming a spontaneous pedestrian movement where impatience might
determine clogging effects.

We know of several methodologies aimed at evaluating the evacuation time for a


given evacuation plan [DRV03], [Smi91], [SMP82]; small is instead the literature
we know which is aimed at finding the evacuation plan which gives the lower
evacuation time [CFS82], [Cep03].
The models aimed at defining the best evacuation plan could be further classified
in the ones which optimise only the escape routes and the models which optimise
both the escape routes and the starting evacuation times.
In fact if evacuation is assumed to start simultaneously from each room in the
building, the resulting loading pattern naturally causes congestion, queues and
spillbacks which might be exaggerated due to the assumption; different
evacuation starting times could be therefore assigned to different areas in the
building. In this case, both the escape routes and the starting evacuation times
affect the building evacuation time and the optimisation process needs to include
starting evacuation times, as well as the escape routes, as control parameters. A
model like this has been developed by Cepolina [Cep04].

© Association for European Transport 2004


We used the object-oriented simulator developed by Cepolina for evaluating how
optimal evacuation plans change either assuming optimal and spontaneous
pedestrian movement, changing the number of building occupants. A building
population evenly spread in the building was assumed [Cep04].
If the building population is small, all the evacuation plans which minimise the
longest travelled distance result optimal. Flow dynamic has not any impact on
evacuation time since congestion doesn’t occur. The number of optimal
evacuation plans could be high.
Increasing the population within the building, the impact of congestion increases.
The number of optimal solutions decreases since the optimal evacuation plans
have to minimise either:
– the longest travelled distance and
– the congestion that occurs in the building: evacuation plans which load evenly
the exits result optimal.
The weight of the two criteria changes, changing the hypothesis on pedestrian
behaviour. If a spontaneous movement is assumed and a significant capacity
drop occurs, the second criteria becomes more relevant.
The simulator allows also to check the effect of differentiating the starting
evacuation times at the different floors. When the building is highly populated, a
proper differentiation of the staring evacuation times can reduce capacity drop
activation; however this effect is not mandatory [Cep04].

4. Conclusions
Two different approaches aimed at reducing the evacuation time and at
facilitating the evacuation process have been reviewed.
The two approaches act on:
– building morphology: through prescriptive regulations it is possible to address
the correct design of spaces either as it concerns the geometry of spaces and
the position of alarm and signage systems;
– the training of the building population: this mainly concerns addressing the
occupants’ route choices through a proper evacuation plan and imposing
behavioural rules about the motion and the actions to take: these
prescriptions should limit the panic effects and the risk of disaster due to
them.
Both the approaches require a correct understanding at a microscopic level of the
dynamic of pedestrian flow and pedestrian psychological behaviour during an
evacuation.
The former approach is always pertinent, the latter could be more incisive but its
application filed is more restricted since it requires the building occupants’
training.

In buildings where people could be trained, evacuation plans are a suitable


method for a successful escape: evacuation plans are useful in schools and
office buildings, i.e. where the building population has enough time to learn the
recommended escape routes.
In this kind of building, if the building population is small and people are spread
distributed within the building, the best evacuation plans result the ones which
minimise the maximum distance travelled by the pedestrians. In fact in this

© Association for European Transport 2004


situation congestion doesn’t affect the evacuation time; a significant role instead
could be played by the pre-movement time.
As the building population become consistent, congestion plays a more important
role and the evacuation plan optimisation could not neglect the dynamic of
pedestrian flow. Moreover the panic effects on the evacuation process have to be
taken into account. Neglecting phenomena like flow break down due to arching at
bottlenecks and merging points leads to a building evacuation time
underestimation with critical impacts on the risk of disaster in the building. In
these circumstances a proper starting evacuation time differentiation could lead
to an evacuation time reduction.

As for buildings visited occasionally by people – e.g. shopping centres –


awareness of evacuation plans is more difficult to achieve. In these cases,
evacuation should be studied as a spontaneous phenomenon and improving the
design of the environment in terms of geometry and alarm and signage systems
becomes the main issue, as suggested by the person approach. In these cases
the staff’s skill in alerting the occupants and ushering them to the exits is decisive
for a successful evacuation.

REFERENCES
[BCF02] Bryan J., Cable E., Fahy R. et al.(2002), “Engineering Guide to
Human Behaviour in Fire”, Society of Fire Protection Engineers.
[Bro91] Brown R. (1991) “Group processes – Dynamic within and between
Groups”, Blackwell, London.
[Cep03] Cepolina E. (2003) “Un modello per la definizione di percorsi
pedonali di evacuazione” Paper presented at Annual Convention
Metodi e Tecnologie dell'Ingegnaria dei Trasporti, Reggio Calabria,
Italy. (Full version in press).
[Cep04] Cepolina E. (2004) “Starting evacuation times and evacuation
plans” Working Paper, University of Pisa.
[CFS82] Chalmet L.G., Francis R. L., Saunders P. B. (1982) “Network
models for building evacuation”, Management science, vol.28, N°1,
printed in USA.
[DRV03] Di Gangi M., Russo F., Vitetta A.(2003) “A Mesoscopic Method for
Evacuation Simulation on Passenger Ships”, Proc. of the
“Pedestrian and Evacuation Dynamics”, Greenwich, London, pp
197-208.
[FGG03] Filippidis L, Gwynne S., Galea E.R., Lawrence P. (2003) “Simulating
the interaction of pedestrians with wayfinding systems” Proceedings
of the 2nd International Conference “Pedestrian and Evacuation
Dynamic”.
[HFV00] Helbing, D., I. Farkas, and T. Vicsek. (2000) “Simulating Dynamical
Features of Escape Panic”, Nature 407, 487-490.

© Association for European Transport 2004


[Hoo04] Hoogendoorn, S.P. (2004) “Pedestrian flow modeling by adaptive
control”, TRB 2004 Annual Meeting CD-ROM
[KlM03] Klupfel H., Meyer-Konig T. (2003) “Characteristics of the PedGo
software for crowd movement and egress simulation”, Proc. of the
“Pedestrian and Evacuation Dynamics”, Greenwich, London.
[KNN03] Kirchner A., Namazi A., Nishinari K., Schadschneider A. (2003)
“Role of conflicts in the floor field cellular automaton model for
pedestrian dynamics”, Proc. of the “Pedestrian and Evacuation
Dynamics”, Greenwich, London pp 51-62.
[Lov97] Lovas G. G. (1997) “Models of wayfinding in emergency
evacuations”, European Journal of Operational Research 105, 371-
389.
[Pro95] Proulx G. (1995) “Evacuation time and Movement in Apartment
Building”, Fire Safety Journal 24 229-246.
[PuB01] Purser D. A., Bensilum M. (2001) “Quantification of behaviour for
engineering design standards and escape time calculations”, Safety
Science, Pergamon.
[PuB99] Purser D. A., Bensilum M. (1999) “Quantification of escape during
experimental evacuations”, Building research Establishment Report
CR 30/99. Building Research Establishment Ltd, Garston, Watford,
UK.
[PuR95] Purser D.A., Raggio A.J.T. (1995) “Behaviour of crowds when
subject to fire intelligence” Building research Establishment Report
CR 143/95. Building Research Establishment Ltd, Garston, Watford,
UK.
[SaS80] Sadalla E. K., Staplin L. J. (1980) “An information storage model for
distance cognition”, Environment and Behaviour 12 (2), 183-193.
[SCK92] Sime J. D., Creed C., Kimura M., Powell J. (1992) “Human
Behaviour in Fires – summary report” Research Report Number 45,
Joint Committee on Fire Research by Jonathan D. Sime.
[Sim85] Sime J. D. (1985) “Movement towards the familiar: person and
place affiliation in a fire entrapment setting” Environment and
behaviour 17 (b) pp 697-724.
[Sim94] Sime J.D. (1994) “Human Behaviour in domestic and hospital fires”,
Building Research Establishments Occasional paper OP59, Building
Research Establishment Ltd, Garston, Watford, UK.
[Smi91] Smith J.MG. (1991) “State-dependent queueing models in
emergency evacuation networks”, Transportation Research-B,
vol.25B,N°6, pp373-389.
[SMP82] Sheffi Y., Mahmassani H., Powell W. (1982) “A Transportation
Network Evacuation Model”, Transportation Research – A, Vol. 16A,
N°3, pp 209-218.

© Association for European Transport 2004


[ThM94] Thompson P.A., Marchant, E.W. (1994) “Simulex: Developing New
Techniques for Modelling Evacuation”, Proc. of the Fourth
International Symposium on Fire Safety Science, National Congress
Centre, Ottawa, Canada, 13-17 June 1994, Elsevier Science
Publishers, London.
[Tsu03] Tsuji Yutaka (2003) “Numerical simulation of pedestrian flow at high
densities”, Proc. of the “Pedestrian and Evacuation Dynamics
2003”, Greenwich, London, pp 27-38.

© Association for European Transport 2004

You might also like