Skip to main content



Experiment and Evacuation


The UK WTC9/11 Evacuation Study: An Overview of the Methodologies Employed and Some Preliminary Analysis

This paper briefly describes the methodologies employed in the collection and storage of first-hand accounts of evacuation experiences derived from face-to-face interviews with evacuees from the World Trade Center (WTC) Twin Towers complex on 11 September 2001 and the development of the High-rise Evacuation Evaluation Database (HEED). The main focus of the paper is to present a preliminary analysis of data derived from the evacuation of the North Tower.

Edwin R. Galea, Lynn Hulse, Rachel Day, Asim Siddiqui, Gary Sharp, Karen Boyce, Louise Summerfield, David Canter, Melisa Marselle, Paul V. Greenall

Evacuation Movement in Photoluminescent Stairwells

An experiment was conducted in a 13-storey office building to assess the effectiveness of different installations of photoluminescent material (PLM) in stairwells. For the experiment four identical stairwells were used: three had different PLM installations and one had reduced lighting of an average of 37 lux. Video cameras and a questionnaire were used to gather data on the movement time and behavior of evacuees. Results from the study show that between 65 to 75% of the respondents felt comfortable going down the stairwells with PLM markings, with the visibility assessed as “good or excellent” in the two stairwells with PLM marking across each step. After the sound of the fire alarm, the average time taken by the first occupants to arrive at each stairwell was 1 min 7 s. Overall, the full evacuation lasted about 12 min. The results also indicate that during the busiest moment of the evacuation the mean speed of movement in the stairwells ranged between 0.40 and 0.66 m/s, while the density ranged between 1.56 and 1.60 p/m


. Occupants’ judgment of the installations showed the importance of marking across each step of the stairwell.

Guylène Proulx, Noureddine Bénichou

Automatic Extraction of Pedestrian Trajectories from Video Recordings

To understand and model pedestrian dynamics, reliable empirical data of pedestrian movement are necessary for analysis and verification, but the existing database is small, inaccurate and highly contradictory. For collecting trajectories from extensive experimental series with a large number of persons we are developing a software named


which automatically extracts these trajectories from normal video recordings with high accuracy in space and time.

Maik Boltes, Armin Seyfried, Bernhard Steffen, Andreas Schadschneider

Stairwell Evacuation from Buildings: What We Know We Don’t Know

Occupant descent down stairwells during building evacuations is typically described by measurable engineering variables such as stairwell geometry, speed, density, and pre-evacuation delay. In turn, predictive models of building evacuation use these variables to predict the performance of egress systems for building design, emergency planning, or event reconstruction. This paper provides a summary of literature values for movement speeds and compares these to several new fire drill evacuations. Movement speeds in the current study are observed to be quite similar to the range of literature values. Perhaps most importantly though, the typical engineering parameters are seen to explain only a small fraction of the observed variance in occupant movement speeds. This suggests that traditional measures form an incomplete theory of people movement in stairs. Additional research to better understand the physiological and behavioral aspects of the evacuation process and the difference between fire drill evacuations and real fire emergencies are needed.

Richard D. Peacock, Jason D. Averill, Erica D. Kuligowski

Evacuation of a High Floor Metro Train in a Tunnel Situation: Experimental Findings

The paper investigates the evacuation of passengers from a metro train in a simulated tunnel situation with a focus on the close geometry of the passage between the metro cars and the tunnel walls. Furthermore trains for underground transportation usually have a high floor which means that people have to overcome a height of 1.0 to 1.2 m to the surrounding ground adopting different types of exiting strategies. In addition people, who try to leave the train through one of the doors onto the escape route, and people who pass this door outside the train at the same time, have to interfere and react to each other in some way. Due to those risks two full-scale evacuation experiments were conducted in a newly released metro train presently used by the Vienna Transport Authorities, in which participants were subjected to the close geometry of a tunnel situation and a high floor evacuation.

Monika Oswald, Hubert Kirchberger, Christian Lebeda

Using Laser Scanner Data to Calibrate Certain Aspects of Microscopic Pedestrian Motion Models

In this paper an automatic procedure to obtain trajectory data sets for controlled walking experiments based on laser scanner measurements is investigated. The laser range scanners provide raw data consisting of snapshots of scattered points with a frequency of 10 Hertz. A tracking algorithm is applied in order to convert the laser scanner measurements into trajectory data sets. Suitability of the method is demonstrated via the application to walking experiments performed in the London based walking laboratory PAMELA. Beside evaluating the accuracy of the obtained trajectory data the experiments are also used in order to enable data driven modelling of stopping and turning movements within the social force model paradigm. It is shown that via the modelling of a ‘desired velocity’ term inside the models the observed behavior can be modelled with reasonable accuracy.

Dietmar Bauer, Kay Kitazawa

Pedestrian Vision and Collision Avoidance Behavior: Investigation of the Information Process Space of Pedestrians Using an Eye Tracker

This study investigates the Information Process Space (IPS) of pedestrians, which has been widely used in microscopic pedestrian movement simulation models. IPS is a conceptual framework to define the spatial extent within which all objects are considered as potential obstacles for each pedestrian when computing where to move next. Particular foci of our study were on identifying the size and shape of IPS through examining observed gaze patterns of pedestrians. A series of experiments were conducted in a controlled laboratory environment, in which up to 4 participants walked on a platform at their natural speed. Their gaze patterns were recorded by a head-mounted eye tracker and walking paths by laser-range-scanner–based tracking systems at the frequency of 25 Hz. Our findings are three folds: pedestrians pay much more attention to ground surface to detect potential immediate environmental hazards than fixating on obstacles; most their fixations fall within a cone-shape area rather than semicircle; the attention paid to approaching pedestrians is not as high as that to static obstacles. These results led to an insight that the structure of IPS should be re-examined by taking directional characteristics of pedestrians’ vision.

Kay Kitazawa, Taku Fujiyama

FDS+Evac: An Agent Based Fire Evacuation Model

In this paper, an evacuation simulation method is presented, which is embedded in a CFD based fire modelling programme. The evacuation programme allows the modelling of high crowd density situations and the interaction between evacuation simulations and state-of-the-art fire simulations. The evacuation process is modelled as a quasi-2D system, where autonomous agents simulating the escaping humans are moving according to equations of motion and decision making processes. The space and time, where the agents are moving, is taken to be continuous, but the building geometry is discretized using fine meshes. The model follows each agent individually and each agent has its own personal properties, like mass, walking velocity, familiar doors, etc. The fire and evacuation calculations interact via the smoke and gas concentrations. A reaction function model is used to select the exit routes. The model is compared to other evacuation simulation models using some test simulations.

Timo Korhonen, Simo Hostikka, Simo Heliövaara, Harri Ehtamo

Comparisons of Evacuation Efficiency and Pre-travel Activity Times in Response to a Sounder and Two Different Voice Alarm Messages

Previous studies (Bellamy and Geter, 1990, BRE Report BR 172; Proulx and Sime, 1991, Fire Safety Science, 3rd International Symposium, pp. 843–852; Sime, 1998, Human Behaviour in Fire, Proceedings of the First International Symposium, pp. 299–308; Purser and Bensilum, 2001, Safety Science 38, pp. 157–182) have indicated that recorded voice alarm messages can be more efficient than sounders in motivating occupants to evacuate buildings, and that they produce shorter pre-travel activity (


) times (PTAT), sometimes by a wide margin. More recently, with increased use of voice alarms and other voice announcements, people may have become less responsive. For this study, monitored evacuations using voice alarms were conducted in a shopping centre, a theatre and a large office building (Purser and Bensilum, 2001, Safety Science 38, pp. 157–182). An experimental study was then set up in a university teaching room under hidden video surveillance, in order to examine responsiveness to different alarm systems (including a sounder, long and short voice messages) and provide PTAT data. The findings from both the monitored evacuations and the experimental study were that voice alarms provided more reliable and shorter PTAT response times than sounders, especially in “awake and unfamiliar” behavioural scenarios. Occupants tended to listen to the full voice message and sometimes the first repeat before starting to travel. The short voice message produced a shorter but less reliable PTAT response. Group interactions had a major effect on response behaviours and times.

David Purser

Design of Voice Alarms—the Benefit of Mentioning Fire and the Use of a Synthetic Voice

Preliminary results from a study about voice alarms are presented in this paper. The purpose of the study is to explore both how messages should be worded and how they should be presented. The paper focuses on an introductory questionnaire study at an IKEA store and unannounced evacuation experiments at Lund University. The results of these activities suggest that it is preferable to mention the word ‘fire’ in voice alarms since it makes people remember the content of the message more accurately. No difference could be detected between messages that were read by a human and a synthetic (computer generated) voice.

Daniel Nilsson, Håkan Frantzich

Enhanced Empirical Data for the Fundamental Diagram and the Flow Through Bottlenecks

In recent years, several approaches for modeling pedestrian dynamics have been proposed and applied e.g. for design of egress routes. However, so far not much attention has been paid to their


validation. This unsatisfactory situation belongs amongst others on the uncertain and contradictory experimental data base. The fundamental diagram, i.e. the density-dependence of the flow or velocity, is probably the most important relation as it connects the basic parameter to describe the dynamic of crowds. But specifications in different handbooks as well as experimental measurements differ considerably. The same is true for the bottleneck flow. After a comprehensive review of the experimental data base we give an survey of a research project, including experiments with up to 250 persons performed under well controlled laboratory conditions. The trajectories of each person are measured in high precision to analyze the fundamental diagram and the flow through bottlenecks. The trajectories allow to study how the way of measurement influences the resulting relations. Surprisingly we found large deviation amongst the methods. These may be responsible for the deviation in the literature mentioned above. The results are of particular importance for the comparison of experimental data gained in different contexts and for the validation of models.

Armin Seyfried, Maik Boltes, Jens Kähler, Wolfram Klingsch, Andrea Portz, Tobias Rupprecht, Andreas Schadschneider, Bernhard Steffen, Andreas Winkens

Parameters of Pedestrian Flow for Modeling Purposes

Seventy years of foot traffic flow research in Russia provided unique empirical data base of travel speed values at different flow density, ranging from 0 up to 13–14 persons/m


obtained in approximately 35.000 counts in 100 series of observations and experiments in the different buildings, city territories, route types and experimental settings.

The laws for crossing of boundaries of adjacent sectors of paths, crowding of people, merging, reforming and diffusion of flows were developed. The flow traffic through door openings, cross and contra-flows, movement on a router with “unlimited” width and other special cases were also investigated.

Accumulated empirical database was used to establish valid psychophysiological law describes relation between flow travel speed and flow density considering emotional state of people. This law was expressed as random function.

Established relations were validated against actual observations, experiments and unannounced evacuations which involved able-body and disabled people. Due to high reliability of established laws they were used in all related Russian building codes and also used for flow/evacuation modeling aimed to provide safety of building occupants. Obtained results are presented in the given paper.

Valerii V. Kholshevnikov, Dmitrii A. Samoshin

Emergency Preparedness in the Case of a Tsunami—Evacuation Analysis and Traffic Optimization for the Indonesian City of Padang

The “Last-Mile Evacuation” research project develops a numerical last mile tsunami early warning and evacuation information system on the basis of detailed earth observation data and techniques as well as unsteady, hydraulic numerical modeling of small-scale flooding and inundation dynamics of the tsunami including evacuation simulations in the urban coastal hinterland for the city of Padang, West Sumatra, Indonesia. It is well documented that Sumatra’s third largest city with almost one million inhabitants is located directly on the coast and partially sited beneath the sea level, and thus, is located in a zone of extreme risk due to severe earthquakes and potential triggered tsunamis. “Last-Mile” takes the inundation dynamics into account and additionally assesses the physical-technical susceptibility and the socio-economic vulnerability of the population with the objective to mitigate human and material losses due to possible tsunamis. By means of discrete multi-agent techniques risk-based, time- and site-dependent forecasts of the evacuation behavior of the population and the flow of traffic in large parts of the road system in the urban coastal strip are simulated and concurrently linked with the other components.

Gregor Lämmel, Marcel Rieser, Kai Nagel, Hannes Taubenböck, Günter Strunz, Nils Goseberg, Thorsten Schlurmann, Hubert Klüpfel, Neysa Setiadi, Jörn Birkmann

Case Studies on Evacuation Behaviour in a Hotel Building in BART and in Real Life

BART is the Behavioural Assessment and Research Tool. BART makes use of an interactive, real-time, physics-based virtual environment with realistic 3D visuals and audio. In BART it is possible to face people with the phenomenon fire in a safe way, without being exposed to the extreme health risk of a real fire. The primary objective of the tool in virtual reality (VR) is to generate the information that fire safety engineers have need of for the design of a safe building that complies with actual human behaviour in fires. For the development of the innovative tool first a trial version of BART is developed, which consists of the visualisation of an existing Dutch hotel building. In this paper some case studies on evacuation behaviour in a real hotel building, as well as in a virtual hotel building in


, are described. Furthermore some key findings of the earliest trials of behavioural assessment in VR are given.

Margrethe Kobes, Nancy Oberijé, Martina Duyvis

Analysis of Empirical Trajectory Data of Pedestrians

We investigate how the characteristics and dynamics of a crowd is changing when the crowd density is increased from a few pedestrians only, up to extremely high crowd densities. Video analysis of the crowd disaster in Mina, Kingdom of Saudi-Arabia, in 2006 gives an empirical base for further analysis which reveals two transitions of the flow; one transition from laminar flow to stop-and-go flow and a second transition to turbulent flow. Finally, an improved specification of the social-force model is suggested in order to explain some of the phenomena occurring in dense crowds.

Anders Johansson, Dirk Helbing

Model-Based Real-Time Estimation of Building Occupancy During Emergency Egress

This paper provides a viable and practical solution to the challenge of real-time estimation of the number of people in areas of a building, during an emergency egress situation. Such estimates would be extremely valuable to first responders to aid in egress management, search-and-rescue, and other emergency response tactics. The approach of this paper uses an extended Kalman filter, which combines sensor readings and a dynamic stochastic model of people movement. The approach is demonstrated using two types of sensors: video with real-time signal processing to detect number of people moving in each direction across a threshold such as an entrance/exit, and passive infra-red motion sensors that detect people occupancy within its field of view. The people movement model uses the key idea that each room has a “high-density” and “low-density” area, where high-density corresponds to a queue of people at a bottleneck exit doorway, and low-density represents unconstrained flow of people. Another key feature of the approach is that constraints on occupancy levels and people flow rates are used to improve the estimation accuracy. The approach is tested using a stochastic discrete-time simulation model of a 1500 square meter office building with occupancy up to 100 people, having a video camera at each of the three exits, and motion sensors in each of the 42 office rooms. The simulation includes stochastic models of video sensors having a probability of detection of 98%, and motion sensors with probability of detection of 80%. Averaged over 100 simulation runs and averaged over the evacuation time, the sensor-only approach produced a mean estimation error per room of 0.35 people, the Kalman filter with cameras only had a mean error of 0.14 people, and the Kalman filter with all sensors produced a mean error of 0.09 people. These results show that an effective combination of models and sensors greatly improves estimation accuracy compared to the state-of-the-art practice of using sensors only.

Robert Tomastik, Satish Narayanan, Andrzej Banaszuk, Sean Meyn

Experiments on Evacuation Dynamics for Different Classes of Situations

The article presents experiments on pedestrian evacuation in different situations. The presented experiments involved evacuation of a lecture room. A group of 31 students took part in four evacuation experiments. The experiments are devoted for three classes of situations: normal conditions, controlled evacuation (noncompetitive evacuation) and panic (competitive evacuation).

Jarosław Was

Prediction and Mitigation of Crush Conditions in Emergency Evacuations

Several simulation environments exist for the simulation of large-scale evacuations of buildings, ships, or other enclosed spaces. These offer sophisticated tools for the study of human behaviour, the recreation of environmental factors such as fire or smoke, and the inclusion of architectural or structural features, such as elevators, pillars and exits. Although such simulation environments can provide insights into crowd behaviour, they lack the ability to examine potentially dangerous forces building up


a crowd. These are commonly referred to as

crush conditions

, and are a common cause of death in emergency evacuations.

In this paper, we describe a methodology for the prediction and mitigation of crush conditions. The paper is organised as follows. We first establish the need for such a model, defining the main factors that lead to crush conditions, and describing several exemplar case studies. We then examine current methods for studying crush, and describe their limitations. From this, we develop a three-stage hybrid approach, using a combination of techniques. We conclude with a brief discussion of the potential benefits of our approach.

Peter J. Harding, Martyn Amos, Steve Gwynne

Start Waves and Pedestrian Movement— An Experimental Study

Pedestrian movement has different quantities, e.g. densities and walking velocities. But these quantities can only be measured during a dynamic process of movement, thus they are depending on each other. But an important point is: How do people start walking? In this contribution it will be shown how fast people start walking and how this process is influenced by the initial density. Furthermore the velocity of the start wave moving through the group of pedestrians is determined.

Christian Rogsch

Clearance Time for Pedestrian Crossing

Analysis of free flow pedestrian dynamics is not sufficient when considering pedestrian crossings. In addition, there is no current evidence of how pedestrian flows are affected by the inclusion of additional items like prams or trolley style luggage. This paper presents work that starts to consider the complex dynamics related to pedestrian crossings.

Craig R. Childs, Taku Fujiyama, Nick Tyler

Ship Evacuation—Guidelines, Simulation, Validation, and Acceptance Criteria

This paper deals with various aspects of ship evacuation: Guidelines, Simulation, Validation, and Acceptance Criteria. The first section adresses general aspects of ship evacuation, like environmental influences, ship motion, the special evacuation procedure on board ships, and guidelines. The second section covers the simulation of ship evacuation. Section 3 focuses on calibration, validation, and acceptance criteria. Finally, some information resources are presented together with a summary and an outlook towards the concept of “safe return to port”.

Hubert Klüpfel

Empirical Study of Pedestrians’ Characteristics at Bottlenecks

The design procedures of Nelson and Mowrer in the SFPE Handbook of Fire Protection Engineering (Society of Fire Protection Engineers, Bethesda, MD,


) and Predtetchenskii and Milinskii (PM) in Planning for Foot Traffic Flow in Buildings (Amerind, New Dehli,


) are frequently used for capacity analysis of pedestrian facilities, e.g. egress routes. Both agree that congestion occurs in front of a bottleneck, if the incoming flow exceeds the capacity. However, in case of a present congestion in front of a bottleneck, their approaches differ considerably. Nelson assumes that in this case the flow inside a bottleneck is determined by the bottleneck capacity. PM instead expect that the density in front of the entrance to the bottleneck is significant higher than the density, which is attributed to the capacity, and thus the flow inside is lower than the capacity. Furthermore PM assume that the density inside is generally significant lower than the density in front of the bottleneck.

To resolve these discrepancies we studied the pedestrian flow through a bottleneck (Rupprecht, Diploma thesis, University of Wuppertal,


) as well as the density and the jam occurrence in front of a bottleneck (Winkens, Diploma thesis, University of Wuppertal,


) by an experiment performed under laboratory conditions. The aim was to get reliable data concerning the density directly in front of and inside the bottleneck and thereby to check the assumptions of Nelson and Mowrer and Predtetchenskii and Milinskii.

We found that during the stationary state the density in front of the bottleneck does not depend on the bottleneck width


for 0.8 m≤


≤1.2 m. In conformance with PM the density inside is significantly lower than in front of the bottleneck. In reference to the continuity equation these results cast doubts on the assumption of Nelson that the flow through a bottleneck maintains the value of the capacity if a congestion appears in front of the bottleneck.

Andreas Winkens, Tobias Rupprecht, Armin Seyfried, Wolfram Klingsch

RFID Technology Applied for Validation of an Office Simulation Model

This paper presents the validation of an office utilisation model for the research project called “User Simulation of Space Utilisation (USSU)”. The result of this research is a system that can be used for analysing and evaluating the space utilisation of a building for any given organisation. A system for building usage simulation that produces data about activities of members of an organisation can substantially improve the relevance and performance of building simulation tools. This is relevant for engineering domains as well as for architects to evaluate the performance of a building design. For a thorough evaluation of the system an experiment was executed for assessing its predictive quality in the context of a real building, organisation and actual human behaviour; this experiment was executed using RFID technology. The result of the experiment was observed data about the space utilisation of the selected organisation. This data set was compared with the space utilisation predicted by the USSU system to evaluate the simulation model. The validation of USSU showed that there were no significant differences between the predicated and observed activity behaviour. As a consequence, the output of USSU is considered to be valid.

Vincent Tabak, Bauke de Vries, Jan Dijkstra

Study on Crowd Flow Outside a Hall via Considering Velocity Distribution of Pedestrians

The dynamical behaviors of pedestrians outside a hall is investigated by using the extended lattice-gas model with different maximum velocities. A numerical simulation illustrates that fast and slow pedestrians will affect some dynamical characteristic from the choking flow to the decaying flow. In the choking-flow region, the scaling relation of crowd flow rate and the evacuation time are obtained, respectively. The slow pedestrians, initial density of walkers and the location of the exit influence the evacuation time.

Xiang Shu Liu, Jia Xiu Pan, Liang Yujuan, Yu Xue

Analysis on the Propagation Speed of Pedestrian Reaction: Velocity of Starting Wave and Stopping Wave

We investigate the reaction speed of a pedestrian to the predecessor in a queue which significantly affects the flow inside the queue. This is one of the most essential points for studying the dynamics of pedestrians. We have performed experiments of pedestrians’ walk in a queue on a flat road, upslope and downslope to measure the propagation speed of the starting wave and stopping wave. We have found that the propagation speed of stopping wave is faster than starting wave on a flat road, whereas the speed of stopping wave is slower than starting wave on the other cases. This reverse result has been studied by using stochastic cellular automaton model in which a perspective of predecessors and the inertial effect are introduced. Moreover, in the case of starting wave we can estimate the elapsed time of the last pedestrian in a queue to move after the leader starts by using mean field analysis.

Akiyasu Tomoeda, Daichi Yanagisawa, Katsuhiro Nishinari

Simulation and Modelling

Toward Smooth Movement of Crowds

“Jamology” is an interdisciplinary research of all sorts of jams, e.g. those of vehicles, pedestrians, ants, etc. Our model of pedestrians, called the floor field model, is based on this study, and it is a two-dimensional generalization of an ant trail model. It is a rule-based cellular automaton model, and efficient in computations since the long-range interaction between pedestrians is imitated by the memory of the floor of only neighboring cells. Recently several generalizations of this model are proposed to make the model more realistic. We use an extended model to study how to make crowd movement smooth. Not only computer simulations but also experiments are shown in this paper. Introduction of pedestrians’ anticipation into the model affects the crowd movement significantly, and leads the counterflow smooth. Moreover it is clearly shown experimentally that evacuation dynamics near a bottleneck becomes smooth if we put an obstacle at a suitable place.

Katsuhiro Nishinari, Yushi Suma, Daichi Yanagisawa, Akiyasu Tomoeda, Ayako Kimura, Ryousuke Nishi

Modeling Evacuees’ Exit Selection with Best Response Dynamics

We present a model for occupants’ exit selection in emergency evacuations. The model is based on the game theoretic concept of best response dynamics, where each player updates his strategy periodically according to other players’ strategies. A fixed point of the system of all players’ best response functions defines a Nash equilibrium of the game. In the model the players are the occupants and the strategies are the possible target exits. We present a mathematical formulation for the model and analyze its properties with simple test simulations.

Harri Ehtamo, Simo Heliövaara, Simo Hostikka, Timo Korhonen

Front-to-Back Communication in a Microscopic Crowd Model

Failures in front-to-back communication (F2BC) in crowd disasters are commonly cited, but mechanisms and effects of F2BC have not been studied. We develop a plausible characterization and model of F2BC and evaluate it in a simple scenario. To study F2BC in a naturalistic context we then reconstruct a consistent geometry for the Who concert disaster, explore the mechanisms for that disaster, and introduce F2BC. Our qualitative analysis suggests that F2BC can reduce injuries at the cost of lower exit rates.

Colin Marc Henein, Tony White

Comparison of Various Methods for the Calculation of the Distance Potential Field

The distance from a given position toward one or more destinations, exits, and way points is an important input variable in most models of pedestrian dynamics. Except for special cases without obstacles in a concave scenario—i.e. each position is visible from any other—the calculation of these distances is a non-trivial task. This is not a big problem as long as the model only demands the distances to be stored in a

Static Floor Field


Potential Field

), which never changes throughout the whole simulation. Then a pre-calculation once before the simulation starts is sufficient. But if one wants to allow changes of the geometry during a simulation run—imagine doors or the blocking of a corridor due to some hazard—in the

Distance Potential Field

, calculation time matters strongly. We give an overview over existing and new exact and approximate methods to calculate a potential field, analytical investigations for their exactness, and tests of their computation speed. The advantages and drawbacks of the methods are discussed.

Tobias Kretz, Cornelia Bönisch, Peter Vortisch

Agent-Based Simulation of Evacuation: An Office Building Case Study

Understanding people behavior and movement characteristics during building evacuation is valuable in evaluating building designs and the effectiveness of evacuation policies. Simulation can be a powerful tool since real data on building evacuation are rarely available and costly to obtain. On the other hand, state-of-the-art in evacuation modeling is the use of agent-based simulations, which are computationally expensive for simulating evacuation in very large buildings. In this paper, we present an agent-based simulation model developed for a 2-story office building verified using the evacuation data collected using video cameras during fire drills in the building. Following model parameter calibration to reflect the actual building traffic characteristics, it is shown that the agent-based simulation model is able to match the real data with high accuracy in terms of the cumulative number of people exiting the building during the evacuation. The paper also presents a graph-based complexity reduction approach that can reduce computation requirements and thus be used for large-scale applications.

Yiqing Lin, Igor Fedchenia, Bob LaBarre, Robert Tomastik

A Genetic Algorithm Module for Spatial Optimization in Pedestrian Simulation

Regarding pedestrian simulation applications, technologies to optimize the built-up environment apart from pure analysis of pedestrian flows, and based on simulation results, are of crucial importance for the wider acceptance of pedestrian simulation. Apart from conventional pedestrian analysis measures such as density maps, flow rates and travel times, optimization of spatial configurations, leading to congestion or travel time reduction, promises an additional benefit for users of the simulation. Spatial optimization therefore delivers specific solutions for the application of pedestrian simulation in general.

Here we present a genetic algorithm optimizer module, a prototype created for the pedestrian simulation software SimWalk. Based on CAD plans, the module allows optimizing plans and objects (walls, obstacles, etc.) automatically. The user defines and marks a plan section for optimization where, for example, pedestrian density problems occur. Additionally, the user defines which changes of the built-up environment are allowed, based on boundary conditions predefined by his or her architectural or engineering knowledge. After having defined these boundary conditions, the evolutionary process performed by the genetic algorithm gets started, and a first generation of plans and predefined populations is generated. Every succeeding plan shows random variations of the selected obstacles. To evaluate the fitness of each generation, density maps and travel times generated by the software are used to optimize the selected environment. The ultimate goal consists in finding plan configurations with low densities and shorter travel times. If the first generation is established, the best plans can be identified. Based on “elite selection” (“survival of the fittest”), the next generation then gets started, using various GA operators like random generators, selectors, recombination and mutation to generate new plan variations. Every generation, in optimal cases, results in a better plan configuration. A main topic of the research project consisted in mapping the scalability of plan obstacles to the chromosomes of an already existing GA framework of the research institute. To get trend information during the software development, it was necessary to develop a graphical user interface (GUI). It made it possible to edit and prepare plans for optimization, and additionally to select interim solutions for simulation with different parameters, boundary values, population sizes and operators. Statistical tests have shown that with the existing operator set and favorably chosen parameters, after a few generations a significantly improved plan can be achieved. With this prototype, a first result for the optimization of spatial environments in pedestrian simulation regarding congestion and travel times has been accomplished.

Further research will include an extended operator framework to find better results in a shorter time. Additionally, the application workflow will be improved for more intuitive work.

Lukas Kellenberger, Ruedi Müller

Opinion Formation and Propagation Induced by Pedestrian Flow

In this paper, we extend a rule on two-way and four-way traffic to study the formation and propagation of opinions in virtue of the mobility of pedestrian. The results show that formation and propagation of opinion is generated by occurrence of phase transition. Collective behavior of pedestrians is in favor of opinion formation and propagation. When pedestrian density and initial density of opinion are below their critical value, there is no formation of the consistent opinion. Even though pedestrian traffic has congested, if initial density of opinion is below its critical value, it is very difficult to give rise to consensus, else, it must undergo great fluctuation to generate a stable opinion.

Yu Xue, Yan-fang Wei, Huan-huan Tian, Li-juan Liang

Passenger Dynamics at Airport Terminal Environment

We report on a specific calibration for an individual-based simulation environment. For this purpose field data of traveling people inside an airport terminal was recorded. The advantage of using video surveillance system is the granting unbiased gathering of person behavior. The presented results are derived from an observed area of 10×30 m area between the check-in facilities and the security control at the Dresden International Airport. By means of statistical analyses a significant difference in behavior between business and leisure related passenger groups was resolved. The influence of carry-on baggage after the check-in procedure is very small and remaining trolley bags do not affect the maximum speed of passengers. In contrast, the size of a passenger group has a significant influence on walking speed, whereas large groups tend to diverge into smaller groups with 2–3 members.

Michael Schultz, Christian Schulz, Hartmut Fricke

Application Modes of Egress Simulation

Egress models are being used more frequently to simulate people movement; i.e. how people enter, use and leave a building. However, little has been written on the different aspects of people movement that can be examined and how these models may achieve this. This paper outlines six modes in which an egress model can be applied:










; and


. The paper outlines what is needed to enable these application modes, in terms of data, expertise and model functionality, and the benefits of doing so. This is intended to highlight the challenges faced by egress models and the complexities of the subject matter being examined: people movement under emergency and non-emergency scenarios. Currently, no model includes all of the six modes identified. The authors hope that this discussion will identify the importance of these modes, the need for them to be addressed within the same model and the clear benefits of doing so.

Steve M. V. Gwynne, Erica D. Kuligowski

Investigating the Impact of Aircraft Exit Availability on Egress Time Using Computer Simulation

This paper examines the influence of exit availability on evacuation time for narrow body aircraft under certification trial conditions using computer simulation. A narrow body aircraft which has previously passed the certification trial is used as the test configuration. While maintaining the certification requirement of 50% of the available exits, six different exit configurations are examined. These include the standard certification configuration and five other exit configurations based on commonly occurring exit combinations found in accidents. These configurations are based on data derived from the AASK database and the evacuation simulations are performed using the airEXODUS evacuation software. The results show that the certification practise of using half of the available exits predominately down one side of the aircraft is neither statistically relevant nor challenging. For the aircraft cabin layout examined, the exit configuration used in certification trial produces the shortest egress times. Furthermore, three of the six exit combinations investigated result in predicted egress times in excess of 90 seconds, suggesting that the aircraft would not satisfy the certification requirement under these conditions.

Edwin R. Galea, Madeleine Togher, Peter Lawrence

Bounded Rationality Choice Model Incorporating Attribute Threshold, Mental Effort, and Risk Attitude: Illustration to Pedestrian Walking Direction Choice Decision in Shopping Streets

Discrete choice models have been widely used in modeling individual choice behavior. By assuming that people are rational and utility-maximizers, these models are effective in capturing underlying preferences and choice outcomes. However, these models are not explicitly developed to represent decision processes. On the other hand, models of bounded rationality, for the majority based on decision heuristics, are meant to be more realistic representations of decision processes based on the assumption that people have limited mental capacity for making a decision.

People may apply different heuristics for a decision. To reflect this idea, we first introduce a conceptual framework which articulates (i) that individuals use attribute thresholds to map external continuous factors into discrete internal representations, (ii) information search costs effort which is considered important for individual selecting heuristics, and (iii) individual’s risk attitude towards heuristics, reflecting expected outcome diversity implied in a heuristic decision. The framework is general enough to identify several typical heuristics, including conjunctive, disjunctive and lexicographic rules. A decision can be modeled as a two-stage process, first choosing a heuristic, and then applied it to reach a decision.

To illustrate this approach, it is applied to a dataset about pedestrians choosing walking directions in a shopping street in Shanghai, China. Results about parameter estimates and estimated probabilities of applying different heuristics are presented. Implications of the model results are discussed.

Wei Zhu, Harry Timmermans

A SCA-Based Model for Open Crowd Aggregation

This paper proposes a SCA-based model for crowd dynamics phenomena. SCA (Situated Cellular Agents) is a modeling and simulation approach based on Multi Agent Systems principles that is characterized by the representation of an explicit spatial structure. This paper is focused on the crowd aggregation phenomenon described by Elias Canetti. This work will provide a methodology example of translation of a social theory into a SCA-based computational model.

Stefania Bandini, Mizar Luca Federici, Sara Manzoni, Stefano Redaelli

Hardware Implementation of a Crowd Evacuation Model Based on Cellular Automata

Cellular Automata (CA) can sufficiently represent phenomena of arbitrary complexity and at the same time they can be precisely simulated by digital computers, because of their intrinsic discreteness. A two-dimensional (2-D) CA dynamic system has been proposed to efficiently model crowd behaviour inside bounded areas to contribute to the upgrade of public facilities. This paper examines the on-chip realisation of the proposed model. The hardware implementation of the CA model is based on FPGA logic. CA cells obtain discrete values, thus indicating their status; either free or occupied. Significant parameters of the local CA rule, such as the number and the allocation of the exits or the obstacles are inputs of the dedicated processor. Initial data is loaded to the dedicated processor in a semi-parallel way, i.e. all rows of the CA grid are loaded simultaneously while data propagates in a serial way from one cell of column


to the other cell of its successive column,


+1. The automatic response of the processor provides the signals that guide the crowd in correspondence to its density around exits. In collaboration with smart cameras, the proposed FPGA processor could be incorporated in an efficient, real-time, decision-support system that would be able to guide the crowd in cases of emergency, using sound and optical signals.

Ioakeim G. Georgoudas, Georgios C. Sirakoulis, Ioannis T. Andreadis

Applying a Discrete Event System Approach to Problems of Collective Motion in Emergency Situations

A proper management of evacuation of crowds in panic situations requires the adaption of strategies and escape routes, depending on information collected from the real scenario. To design an effective supervisory controller, a model representing the main aspects of the evacuation dynamics and based on data coming from a distributed sensors network is necessary. To this aim, a Discrete Event System model is developed here to describe the sequence of events occurring in escape procedures. In particular, a systematic approach consisting of few steps allows to define a modular Petri Net representing flows of individuals during egress. A simple case study shows the feasibility of the proposed method.

Paolo Lino, Guido Maione

SIMULEM: Introducing Goal Oriented Behaviours in Crowd Simulation

Simulating human activity is a research field with multiple applications, especially regarding the validation of public buildings. Early work in this area focused on security considerations, especially for evacuation scenarios. However, recent needs have emerged to validate a public building considering its normal operation in order to provide the best quality of service. This problem is addressed here by proposing a simulation tool for crowds of autonomous pedestrians, where individual behaviours take into account medium term goals. We propose an agent’s model with embodied and situated decision abilities which determine its movements inside the environment as well as its interactions with equipments. This model is applied in a tool called


, which supports all the simulation phases from the environment’s specification to the results analysis. Finally, this tool is operated on real cases for the simulation of train stations by our industrial partners (





Sébastien Paris, Delphine Lefebvre, Stéphane Donikian

Conflicts at an Exit in Pedestrian Dynamics

In this paper, we have detailedly studied the effect of conflicts on the pedestrian outflow through an exit. Pedestrians conflict each other at the exit, which is a bottle neck, when they evacuate from a room. The pedestrian outflow decreases when there are many conflicts. In the floor field model, which is a pedestrian model using cellular automata, the conflicts are taken into account by the

friction parameter

. However, the friction parameter is a constant and does not depends on the number of the pedestrians conflicting at the same time. We have extended the friction parameter to the

friction function

, which is a function of the number of the pedestrians involved in the conflict. The results of theoretical analysis using the friction function agree with the experimental results much better than using the friction parameter. We have also found that putting an obstacle in front of the exit increase the pedestrian outflow from our experiments. The friction function clearly explains the mechanism of the effect of the obstacle, i.e., the obstacle blocks a pedestrian moving to the exit and decrease the average number of pedestrians involved in the conflicts.

Daichi Yanagisawa, Akiyasu Tomoeda, Katsuhiro Nishinari

Improving Pedestrian Dynamics Modeling Using Fuzzy Logic

The complementary nature of MEMS based pedestrian dead-reckoning (PDR) navigation and GNSS (Global Navigation Satellite System) has long been recognized. The advantages are quite clear for those applications requiring indoor positioning and that, for one reason or another, cannot rely on short-range infrastructure-based positioning systems (e.g. WiFi, UWB) to cope with the lack of availability of GNSS indoors. One such example of application is firemen coordination during emergency interventions.

Classification of human displacement using signal pattern recognition techniques often rely on an estimation model or statistical data to compute the step length or horizontal speed information. In general, an initial calibration phase is needed which can constrain the ability to follow the quasi-erratic behavior of a pedestrian in real time. Moreover, existing state-of-the-art PDR solutions enable only the reconstruction of the 2D trajectory.

This paper introduces a different approach to PDR navigation, in which pattern recognition is correlated to biomechanical principles and combined with fuzzy logic for detection and classification of a broader range of walking behaviors in 3D. Furthermore, to avoid the aforementioned limitations of stride length estimation, the step length is effectively computed by a simple inverse segment model during a specific phase of the gait cycle.

Besides a description of the algorithm, this paper includes results of a real-time implementation capable of detecting/classifying four different types of steps: forward walk, stair climbing, stair descent forward and stair descent backward. This development has been conducted in the framework of the European project LIAISON [Renaudin et al., Technical Reports D046 (


) and D077 (


), LIAISON Consortium Deliverable] funded by the Sixth Framework Program to specifically address one of its test case scenarios, the coordination of a fire brigade intervention.

Phillip Tomé, François Bonzon, Bertrand Merminod, Kamiar Aminian

Modeling the Link Volume Counts as a Function of Temporally Dependent OD-Flows

This paper proposes a model for parameterizing time dependent origin-destination (OD) matrices. This model builds the basis for the exploitation of high frequent count data for the estimation of OD-matrices. Such data sets become increasingly available due to recent advances in automatic pedestrian counting technology paired with a business interest to analyze the number as well as the fluctuation of the number of customers for shopping centres or passengers for junctions of mass transport. OD-matrices are needed for simulation, monitoring of the demand and (depending on the location of the counting sensors) for the evaluation of the attractivity of parts of the building. The model contains a full specification of the data set including a stochastic model for the measurement errors which directly suggests a method for the estimation of the parameters.

Dietmar Bauer

Effect of Subconscious Behavior on Pedestrian Counterflow in a Lattice Gas Model Under Open Boundary Conditions

In this paper, human subconscious behavior in pedestrian counterflow is investigated under open boundary conditions by using an extended lattice gas model with different maximum velocities. Four types of walkers are distinguished in the model and their dynamical characteristics are discussed. The simulation results show that the model can capture some essential features of pedestrian counterflows, such as lane formation, segregation effects and phase separation at higher densities. By analyzing the obtained phase diagram it is found that subconscious behavior plays a key role in reducing the occurrence of jam clusters in comparison to the case of un-subconscious behavior.

Kuang Hua, Song Tao, Li Xingli, Dai Shiqiang

Hand-Calculation Methods for Evacuation Calculation—Last Chance for an Old-Fashioned Approach or a Real Alternative to Microscopic Simulation Tools?

Evaluation and optimization of emergency systems can be done by means of several engineering methods, which are entirely different: macroscopic hydraulic models, which can be calculated by hand (the so called “Hand-calculation Methods”), and microscopic computer simulation methods. Both allow forecasting of evacuation-times for various settings. The authors compare results of four commercial software tools (ASERI, buildingEXODUS, PedGo, Simulex) and some macroscopic hydraulic models with real evacuation trials in high-rise buildings. Furthermore a theoretical research of a school building is shown.

Christian Rogsch, Henning Weigel, Wolfram Klingsch

Adding Higher Intelligent Functions to Pedestrian Agent Model

Pedestrian dynamics studies have brought the potential for application to a wide variety of urban planning field, not only crowd accident risk analysis but also simulation-oriented spatial design. Our up-versioned ASPF (Agent Simulator of Pedestrian Flows) ver.4 has newly introduced higher function such as a target maintaining (Helmsman) function, a route-choice function etc. into a pedestrian agent. In an example case of patio-shaped shopping mall, Asunal Kanayama, we have verified that pedestrian agents can walk autonomously along way points to destination points in a complicated space, and that ASPFver.4 can be a useful tool for ordinary commerce space design. We also refer to the ASSA (Agent Simulator of Shop Around) project for implementing much higher functions.

Toshiyuki Kaneda, Takumi Yoshida, Yanfeng He, Masaki Tamada, Yasuhiro Kitakami

“FlowTech” and “EvaTech”: Two Computer-Simulation Methods for Evacuation Calculation

FlowTech and EvaTech are software tools for evacuation simulation, representing two different approaches: macroscopic and microscopic. FlowTech was developed due to the need that arose in Russia in a software tool for evacuation calculation which would correspond to methods and correlations enlisted in the present Russian State Standard of evacuation calculation “GOST 12.1.004-91*. Fire Safety. General Requirements” (IPK Standards, Moscow,


), based on the work of V.M. Predtechenskii and A.I. Milinskii (“Planning for Foot Traffic Flow in Buildings” (Amerind, New Delhi,


). FlowTech supports now different “hand-calculation” methods of evacuation calculation such as “GOST 12.1.004-91*”, the Moscow city standard “Moscow City Construction Regulations” (MGSN 4.19-2005, Moscow Government, Moscow,


) and the method recommended by the Society of Fire Protection Engineers (SFPE) (Nelson and Mowrer in The SFPE Handbook of Fire Protection Engineering, SFPE, Bethesda, MD,


). FlowTech approaches modeling of flow movement as a representation of evacuation process which consists of formation, movement and merging of flows. Since separate persons are not singled out from the flow, this approach is called a macroscopic approach. In spite of its simplicity it allows to predict evacuation times and congestion points in buildings quite accurately and can be widely used as a fast means of evacuation calculation with little effort of a building model construction. For cases when a representation of individuals is needed (e.g. in case of evacuation from a room with a complex planning, or when human behavior is an important factor which will affect the simulation) the software tool EvaTech is under development.

Ilya Karkin, Vladimir Grachev, Andrey Skochilov, Vladimir Zverev

Large Scale Microscopic Evacuation Simulation

The evacuation of whole cities or even regions is an important problem, as demonstrated by recent events such as evacuation of Houston in the case of Hurricane Rita or the evacuation of coastal cities in the case of Tsunamis. A robust and flexible simulation framework for such large-scale disasters helps to predict the evacuation process. Existing methods are either geared towards smaller problems (e.g. Cellular Automata techniques or methods based on differential equations) or are not microscopic (e.g. methods based on dynamic traffic assignment). This paper presents a technique that is both microscopic and capable to process large problems.

Gregor Lämmel, Marcel Rieser, Kai Nagel

Numerical Optimisation Techniques Applied to Evacuation Analysis

A common problem faced by fire safety engineers in the field of evacuation analysis concerns the optimal design of an arbitrarily complex structure in order to minimise evacuation times. How does the engineer determine the best solution? In this study we introduce the concept of numerical optimisation techniques to address this problem. The study makes use of the buildingEXODUS evacuation model coupled with classical optimisation theory including Design of Experiments (DoE) and Response Surface Models (RSM). We demonstrate the technique using a relatively simple problem of determining the optimal location for a single exit in a square room.

Rodrigo Machado Tavares, Edwin R. Galea

A Multi-Method Approach to the Interpretation of Pedestrian Spatio-Temporal Behaviour

The development of mobile spatial-information technologies requires a profound understanding of pedestrian spatio-temporal behaviour. In a currently ongoing project we use several empirical methods following the concept of “across-method” triangulation to comprehensively study human spatial behaviour. In this contribution we will introduce a multi-method approach including a combination of localisation and tracking techniques (GPS, Bluetooth, unobtrusive observation) as well as enquiries concerning intentions, lifestyle attributes, socio-demographic characteristics, route quality preferences, and preferred way-finding strategies. The combination of qualitative-interpretative and quantitative-statistical data will lead to the determination of a typology of lifestyle-based pedestrian mobility styles, which can serve as a basis to customise navigational and environmental information to individual needs, and to create pedestrian interest profiles in ubiquitous environments. We present experimental results of the first of two consecutive empirical phases based on a data set containing of more than 100 trajectories of people observed by path following in an indoor and outdoor environment.

Alexandra Millonig, Georg Gartner

The Microscopic Model and the Panicking Ball-Bearing

Prominent microscopic models simulate panic (which has been described as a myth) allowing unwarranted simplifying assumptions that people are irrational, non-deliberative and interchangeable. While these assumptions can be remedied by increasing the behavioral repertoire of modelled individuals, large cognitive architectures would stifle a model’s power to explain emergent crowd effects. We propose the microscopic human factor (MHF) approach that increases behavioral repertoire without compromise to the elegant simplicity from which the models derive their explanatory power.

Colin Marc Henein, Tony White

Design of Decision Rules for Crowd Controlling Using Macroscopic Pedestrian Flow Simulation

Crowd control mechanisms such as temporary access restrictions allow affecting pedestrian flows in public transport facilities in order to avoid overcrowding. Such access restrictions can be based on decision rules depending on measured pedestrian density. In order to design these decision rules, simulations of pedestrian flows are a valuable tool. In this paper we describe the design of decision rules for a real case study constituted by a subway station next to the main soccer stadium in Vienna. The simulations use a macroscopic model which (1) includes dynamic elements (like e.g. arriving and departing trains) and (2) integrates the implementation of decision rules based on real time measurements of real people flows. The model also takes into account measurement errors. We discuss the simulation results for the case study with the resulting decision rules in place.

Stefan Seer, Norbert Brändle, Dietmar Bauer

3-Tier Architecture for Pedestrian Agent in Crowd Simulation

After extensive investigations and in-depth studies of crowd dynamics with Chinese characteristics, especially in stations of Urban Mass Transit, we proposed a new architecture for agent oriented pedestrian simulation. It’s designed to simulate the movement of thousands of individual pedestrians through large, geometrically complex 2D space.

Our PSS (Pedestrian Simulation System for Urban Mass Transit Station) based on this architecture has been developed. The results of the case studies show that the architecture is feasible and effective in simulating intelligent human behaviors. Characteristic by openness and scalability, it can be easily applied to other situations, such as Olympics and Expo, etc.

Gao Peng, Xu Ruihua

Optimising Vessel Layout Using Human Factors Simulation

Evaluating ship layout for human factors (HF) issues using simulation software such as maritimeEXODUS can be a long and complex process. The analysis requires the identification of relevant evaluation scenarios; encompassing evacuation and normal operations; the development of appropriate measures which can be used to gauge the performance of crew and vessel and finally; the interpretation of considerable simulation data. In this paper we present a systematic and transparent methodology for assessing the HF performance of ship design which is both discriminating and diagnostic.

Steven J. Deere, Edwin R. Galea, Peter J. Lawrence

Agent-Based Animated Simulation of Mass Egress Following an Improvised Explosive Device (IED) Attack

We achieved a new degree of scale and realism in developing agent-based simulation models of mass egress from large facilities, following one or more detonations of improvised explosive devices (IED), to evaluate some proposed ways to facilitate evacuation and reduce casualties. We modeled two venues: a sports stadium and a subway station. Our models offer ease of input and animated realism of output that make them much more suitable than traditional discrete-event models as aids to decision-makers. The development and analysis completed to date, while far from exhaustive, suffice to demonstrate the utility of models such as these for evaluating proposed countermeasures, for indicating policy and technology issues that should be analyzed further, and for response planning.

Douglas A. Samuelson, Matthew Parker, Austin Zimmerman, Stephen Guerin, Joshua Thorp, Owen Densmore

A Novel Kinetic Model to Simulate Evacuation Dynamics

We present the Kinetic Model (KM) as an alternative to combat the computational barrier posed by state-of-the-art evacuation modeling and simulation approaches. Instead of tracing individual agents, the KM simulates a coarse measure, namely the local occupant density. The two-phased approach tracks two distinct kinds of traffic, namely individual motion in “rarefied” (i.e. non-congested) regions and vacancies left by individual motion in “dense” (i.e. congested) regions. The dynamics of the phase boundary is governed by the balance of individuals moving from the “rarefied” region and vacancies moving from the “dense” regions across the boundary, making the approach intrinsically computationally efficient. The paper presents the theoretical formulation of the KM and comparisons of the model predictions of occupancy during evacuation with that from higher fidelity ABM. The KM computations are nearly 3 orders of magnitude faster than the ABM simulations with minimal degradation of prediction accuracy.

Sergei Burlatsky, Vladim Atrazhev, Nikolay Erikhman, Satish Narayanan

Egress Route Choice Modelling—Concepts and Applications

Most microscopic evacuation models offer the possibility to determine the individual egress route based on a shortest path criteria. This option has the advantage that the egress route is only depending on the building structure and can thus be calculated in advance. Yet there has to be the possibility for individual and dynamic adjustments depending on egress route load (avoid congestions), attractiveness of route alternatives or external influences like smoke spread if appropriate. However, care has to be taken that these adjustments do not lead to unrealistic counter flows or artificial blocking effects. Human egress behavior is triggered by a very complex interaction of perception, anticipation and fine-tuning of movement that can up to now only be modelled in a very approximative way on a large scale. The basic concepts of egress route choice modelling implemented in the microscopic evacuation model ASERI are presented together with case studies demonstrating the effect of certain egress route choice options on evacuation efficiency and calculated egress time. Emphasis is put on the mechanisms modelling avoidance strategies in case of congestion or in the presence of equivalent egress route options. Some of these mechanisms are dynamic, based on local occupant load and the local building environment and may involve non-deterministic rules.

Volker Schneider, Rainer Könnecke

Architectural Cue Model in Evacuation Simulation for Underground Space

In this paper, a CAVE-based experiment for measuring the evacuees’ preference on architectural cues is introduced. The research method has proven that a paired cue evacuation decision model can be constructed through the Conjoint Analysis method. According to the preference function the nearest exit assumption in other evacuation models is questionable in some circumstances. Moreover, a set of architectural cue attributes and their quantitative utilities on the preference are discovered.

Chengyu Sun, Bauke de Vries, Qi Zhao

Integrating Strategies in Numerical Modelling of Crowd Motion

We propose here to integrate sophisticated strategies in the model of crowd motion presented by Maury and Venel (Proceedings of Traffic and Granular Flow 2007, Springer,


). This model is based on two different velocities: a spontaneous (or desired) velocity and an actual velocity taking congestion into account. In this paper, we propose several strategies to define the spontaneous velocity: follow the shortest path, adapt its own velocity to the one of his neighbors, or avoid jams. Finally, we present numerical results.

Juliette Venel

Small-Grid Analysis of Evacuation Processes with a Lattice Gas Model for Mixed Pedestrian Dynamics

Pedestrian flow out of a hall is investigated by using a small-grid lattice gas (LG) model with different maximum velocities. Some dynamical characteristics of the transition from choking flow to decaying flow are found. In the choking-flow region, the flow rate increases with the increase of the fraction of faster pedestrians and the door width. The evacuation time decreases with increasing fraction of faster pedestrians and door width. In addition, the dependence of the average velocity on the mixture fraction and door width is discussed. Finally, the relationship of evacuation time with mixture fraction and door width are obtained.

Yan-fang Wei, Yu Xue, Shi-qiang Dai

Evacuation Simulation and Human Behaviour Models in Tall Buildings

In tall buildings, staircases are usually the only means for evacuation. It can take more than two hours for an occupant to reach a refuge floor by the stairs owing to congestion and long travel distance. In a non-fire situation, elevators also could be used in evacuation. Where there are several alternative transport devices available for egress, occupants need to decide which one to use. This decision is based on the occupant’s knowledge of the egress routes in the building. Evacuation models consider human behaviour mostly for horizontal movement and low buildings, without considering elevators as a means of egress. In this article, we describe vertical transportation and agent-based human behaviour models of the Building Traffic Simulator. The evacuation of the World Trade Centre is simulated and the results are compared with published data.

Marja-Liisa Siikonen, Janne S. Sorsa

Proof of Evacuation Routes and Safety Exits: Time Data as the Main Criteria for the Evaluation of Escape Routes and Safety Exits?

Building codes specify means of egress mainly in terms of requirements concerning the arrangement of means of egress, egress capacities and maximum travel distances, derived on an empirical basis. Alternative solutions are increasingly substantiated by egress simulations and resulting evacuation times. Indicative values for the appraisal of calculated times are available, however, they refer to flow times only and no systematic evaluation of evacuation times for means of egress complying with code requirements has yet been performed.

Within this project evacuation times are calculated for examples complying with German building code requirements. The intention was to investigate, whether current code specifications render reasonable results when they are translated into the time domain and whether specified evacuation times may be used as superior criteria for a performance based design of means of egress.

Nathalie Waldau, Marita Kersken-Bradley, Thilo Hoffmann

Dependence of Modelled Evacuation Times on Key Parameters and Interactions

Times and patterns of buildings evacuations involve interactions between many behavioural parameters reflected in the increasing complexity of computer simulations. A combination of detailed GridFlow computer evacuations simulations, calculation models and experimental evacuations have been used to determine the extent to which evacuation patterns and times are largely dependent upon a small number of key parameters and interactions. Cases investigated included a single retail enclosure and multi-enclosure, multi-storey office buildings designed following UK prescriptive guidance. It is concluded that evacuation times are very dependent upon a small number of critical factors (including PTAT distributions, exit choice ratios, maximum flow rates, merge ratios, and densities of stationary and moving groups), and that even the most sophisticated computer simulations can give misleading results if these factors are not adequately represented. It is considered that simple calculation methods can provide a useful first estimate of evacuation times for designers, and a useful check on the performance of more complex simulation models.

David Purser

A Modification of the Social Force Model by Foresight

The motion of pedestrian crowds (e.g. for simulation of an evacuation situation) can be modeled as a multi-body system of self driven particles with repulsive interaction. We use a few simple situations to determine the simplest allowed functional form of the force function. More complexity may be necessary to model more complex situations. There are many unknown parameters to such models, which have to be adjusted correctly to give proper predictions of evacuation times, local densities and forces on rails or obstacles.

The parameters of the social force model can be related to quantities that can be measured independently, like step length and frequency. The microscopic behavior is, however, only poorly reproduced in many situations, a person approaching a standing or slow obstacle will e.g. show oscillations in position, and the trajectories of two persons meeting in a corridor in opposite direction will be far from realistic and somewhat erratic.

One of the reasons why these models are not realistic is the assumption of instantaneous reaction on the momentary situation. Obviously, persons react with a small time lag, while on the other hand they will anticipate changing situations for at least a short time. Thus basing the repulsive interaction not on the momentary situation but on a (linear) extrapolation over a short time (e.g. 1 s) eliminates the oscillations at slowing down and smoothes the patterns of giving way to others to a more realistic behavior. The exact extrapolation time is of little importance, but a combination of long time with linear extrapolation may get unstable. One second anticipation seems reasonable, and while the actual anticipation in peoples mind will most likely not be based on linear extrapolation, the differences will be small.

A second reason is the additive combination of binary interactions. It is shown that combining only a few relevant interactions gives better model performance.

Bernhard Steffen

Models for Crowd Movement and Egress Simulation

This paper lists models currently available for the simulation of crowd movement and egress simulation. The number of models that have been developed in the last decades are numerous and it is therefore neither possible nor useful to describe all the models in detail in a paper. This is even more the case since new models will be developed and existing models will be changed continuously. Therefore, the more appropriate approach for detailed model description is an open and editable website (wiki) which has already been put up:


The outline of this paper is as follows: The first section contains some general remarks on models.

Hubert Klüpfel

Modelling Pedestrian Escalator Behaviour

This paper presents an escalator model for use in circulation and evacuation analysis. As part of the model development, human factors data was collected from a Spanish underground station. The collected data relates to: escalator/stair choice, rider/walker preference, rider side preference, walker travel speeds and escalator flow rates. The dataset provides insight into pedestrian behaviour in utilising escalators and is a useful resource for both circulation and evacuation models. Based on insight derived from the dataset a detailed microscopic escalator model which incorporates person-person interactions has been developed. A range of demonstration evacuation scenarios are presented using the newly developed microscopic escalator model.

Michael J. Kinsey, Edwin R. Galea, Peter J. Lawrence, Darren Blackshields, Lynn Hulse, Rachel Day, Gary Sharp

Introducing a Coupled Model for Simulating Crowd Behaviour

The simulation of human behaviour is a complex and unsolved problem. In the simulation of crowd behaviour due to evacuations, many different approaches are present, which in most cases incidently consider psychological aspects of this phenomenon. We assume human behaviour in crowds to be the result of coupled individual processes. The introduced

perceiving–acting model

follows this assumption and integrates perceiving, thinking and acting as primary factors in the individual human behaviour. The model focuses on evacuation scenarios and therefore some of the intrinsic complexity of the human and its acting can be neglected.

Alicia Guadalupe Ortega Camarena, Dominik Jürgens

Evacuation Modelling of Fire Scenarios in Passenger Trains

Fire incidents inside passenger trains constitute a significant risk factor for life safety. Therefore, it is necessary to count on a suitable evacuation strategy, during the instants previous to the rail vehicle halt and the subsequently evacuation. In this paper, the evacuation of passengers from different fire scenarios and several evacuation conditions were investigated. The analysis was divided into two stages of the evacuation process considering two different high speed trains: (1) the movement and behaviour of passengers in fire scenarios inside the vehicle before the train stopped, and (2) the analysis of train evacuation under different conditions. The results, obtained by means of the egress model STEPS (see MacDonald, STEPS Simulation of Transient and Pedestrian Movements User Manual), allowed to determine the influence of the limitations of the different train geometries under different evacuation conditions, give an estimation of the evacuation times and analyse the impact of human parameters considered in the evacuation process.

Jorge Capote, Daniel Alvear, Orlando Abreu, Mariano Lázaro, Arturo Cuesta

Pedestrian Dynamics with Event-Driven Simulation

The social-force model is systematically modified to achieve a satisfying agreement with the fundamental diagram. Furthermore, our modification allows an efficient computation of the simulation. Finally, different simulation-results will be compared to empirical data.

Mohcine Chraibi, Armin Seyfried



The Need for Behavioral Theory in Evacuation Modeling

This paper posits the need for a complete, comprehensive conceptual model about human behavior in fire evacuations. This would be of intrinsic value to improve training, education, and future data collection efforts, but would also allow for a complete behavioral representation to be embedded within simulation tools. This paper begins by discussing the current, separate theories or “behavioral facts” extracted from research on evacuations from building fires. Then, the paper discusses the methods used by current computer evacuation models to simulate these “behavioral facts” and the limitations of these methods. Last, the paper argues for the inclusion of a comprehensive behavioral conceptual model in computer evacuation models, specifically by highlighting the benefits of behavioral theory for evacuation models and providing examples of social theories used to predict whether people will evacuate from disasters in communities.

Erica D. Kuligowski, Steve M. V. Gwynne

NO_PANIC. “Escape and Panic in Buildings”—Architectural Basic Research in the Context of Security and Safety Research

The increasing number of reported fires or other catastrophes (see Table 1) occurring in large events leads to the interesting question of why preventive design-related and organizational measures have not been taken. It also confirms the need to rethink existing building codes and safety concepts, as well as inclusion of new ways to optimize buildings and event sites. This research project “KEINE_PANIK”—“NO_PANIC” deals with planning criteria in regards to orientation in public buildings (such as airports, train stations, meeting halls or areas, concert halls, stadiums, etc.) during high-stress situations and its influence on the right choice of evacuation routes. The potentials of simulation-data optimizing the safety of persons in buildings must be included to the planning process, the same as the results of the different research groups should be applied in the three dimensional procedure of architectural design.

Christa Illera, Matthias Fink, Harry Hinneberg, Karin Kath, Nathalie Waldau, Andrea Rosič, Gabriel Wurzer

Was It Panic? An Overview About Mass-Emergencies and Their Origins All Over the World for Recent Years

Mass-emergencies are very popular in the news, whether we watch news on TV or read a newspaper. In most of these news we are able to read that people were fallen in panic or a mass-panic occured. This is a simple, but often used explanation why people died in such situations. But is that the truth? If we look at selected mass-emergencies like Bergisel-Stadium we can see, that the loss of a shoe was the origin of this phenomenon, where five girls died. One pedestrian lost a shoe while he was walking to the exit. He stopped to put on his shoe, but because of the high density of pedestrians, the other pedestrian were not able to sidestep at this moment, thus they had to stop. The pedestrians behind them did not see the boy putting on his shoe and thus they pressed against the other pedestrians, just for fun. In this case, the phenomenon of behavior was not panic, we will call this crush with very local panic behavior.

Based on 127 mass-emergencies the authors want to show how often a real mass-panic occurs and what are the real origins of these mass-emergencies.

Christian Rogsch, Michael Schreckenberg, Eric Tribble, Wolfram Klingsch, Tobias Kretz

Hierarchical Structure of the Mass and Group-Level Behaviors in Urban Rail Transfer Stations

In Urban Rail Transit (URT) stations, safety problems and delays occur, because of the complicated passenger organization processes, space and facility layouts, mass composition and movement. This paper analyzed the structure of the mass as well as the features and behaviors of passenger groups in stations. First, we explored the passenger organization processes, the hierarchical structure of the space, the passenger service network and the hierarchical of the mass. Second, we showed a hierarchical model of the mass, and studied both individuals and groups in their features, behaviors and impact on safety and efficiencies of the crowd. Third, we built a process model to demonstrate mass behaviors. Last, we developed a simulation framework of URT station, based on the models and taking into considerations of passenger organization, train operation and space layout.

Xiaolei Zou, Ruihua Xu, Peng Gao

The Use of a Structure and Its Influence on Evacuation Behavior

A fire incident is just one event in the life of a building. The procedures employed to address a fire exist in an environment with other procedures. Given these facts, and the complexity of evacuee response, simulation tools must have the ability to simulate the building under a number of conditions and the interaction of the procedures in place. This would reduce the number of assumptions required of the engineer and produce a more reliable understanding of procedural effectiveness.

Steve M. V. Gwynne, Dave Boswell



Inhalation Injury of Lung and Heart After Inhalation of Toxic Substances

The clinical manifestations of acute inhalation of toxic substances vary according to the particular injurious agents, concentration, length of exposure, and underlying pre-existing diseases in the subjects. Responses of lung and heart on acute and chronic irritant gases are discussed. The data mainly stem from occupational and environmental exposures. The effects of acute lung injury are demonstrated with special regard to carbon monoxide, hydrogen cyanide, and related substances observed in acute smoke inhalation. Clinical manifestations are illustrated. Most complications of fire exposures and death are due to inhalation injuries rather than by burn damages to the skin. Prevention can only be performed in fire fighters by means of protection masks. In case of intoxication during fire, e.g. in houses, smoke inhalation victims have to be treated according to symptoms and signs and toxic inhalants to be involved so far known. Oxygen supply, drugs to open airways (bronchodilators) and lung protection by corticosteroids are cornerstones in therapy. Victims have to be hospitalized as fast as possible. Evacuations of people in buildings also depend on physical fitness, age, weight and diseases of joints and bones.

Herbert Löllgen, Dieter Leyk

Quantitative Comparison of International Design Standards of Escape Routes in Assembly Buildings

The design standards for escape routes in assembly buildings are discussed by comparing the building codes and regulations of eight European countries, China, and USA. To quantify the results for the simple scenario of the evacuation of an assembly room, the travel times are calculated by flow capacity equations. For each country the results vary with the area of the assembly room. Among different countries there is a significant scatter of the results which varies by a factor of three. As the travel times are influenced by the design occupant densities which range from 0.5 to 4 p/m


for places of entertainment, a second comparison is made with a normalized density of 4 p/m


. Here the results also scatter by a factor of three. The implications on occupant safety are discussed with regard to potential fire scenarios.

Burkhard Forell, Ralf Seidenspinner, Dietmar Hosser

Visualizing the Human Form for Simulation and Planning

In pedestrian simulation, choosing the right visualization for the depiction of a simulated human is important. Recent taxonomies of visualization techniques for simulation can aid leverage a wide range of possible choices. Our work is to (1) narrow down these choices to the set of those applicable in pedestrian simulation, and (2) give a discussion over possible combinations of these visualization techniques.

Gabriel Wurzer

A Real-Time Pedestrian Animation System

The presented system clarifies possible connections between existing pedestrian simulations and human character animation. Crowd behavior control is separated from the local realization of the animation. A data-driven skeletal animation is indirectly controlled by a high level animation description. A mechanism is specified to transform behavior assignments into a continuous skeletal animation, which is used for an asynchronous visualization of virtual humans. A graphics hardware accelerated implementation shows the efficiency of the selected approach.

Christian Schulz, Michael Schultz, Hartmut Fricke

Modeling of Escape Routes According to Occupancy, Economy, and Level of Safety in Slovak Republic

Fire safety is one of the six primary requirements in civil engineering projects. All of them are controlled by governmental authorities both in phase of design and execution. From the point of fire safety one of the most complicated requirement is to achieve safety evacuation of pedestrians. Due to different point of views between governmental institutions, fire safety engineers, and developers lots of projects get more expensive and complicated. The presented paper is based on an example of a building occupancy (hotel building) and their influence on the type of escape route based on Slovak condition. As a result of design and calculation two escape routes and their mutual comparison from different point of views is presented.

This work was supported by the grant project VEGA 1/0739/08 “Theoretical and experimental analysis of thermo-technical, acoustic and fire safety properties of glass systems for a well-founded construction project of multifunction glass system” funded by the Scientific Grant Agency of the Ministry of Education of Slovak Republic and the Slovak Academy of Sciences.

Martin Lopušniak


Weitere Informationen

Premium Partner