The website for crowd management simulation

Introduction

Since the nineties of the last century there has been a fast growing interest in understanding and modeling pedestrian behavior. In all kinds of environments the importance of analyzing and quantifying pedestrian flows is acknowledged. These range from urban design in public areas to effective product placement in a store to evacuation dynamics. Major reasons for this increased attention is that the quality of pedestrian flow and particularly the safety in pedestrian environments are more important than ever before.

Especially during emergency situations crowd management is an important aspect. Public facilities are getting bigger and bigger and still pedestrians have to be routed and evacuated through the building in a fast and efficient manner. Classical, qualitative manners of flow analysis are no longer sufficient. Therefore, simulation models are used to optimize proposed infrastructural designs and to test and improve the crowd management inside an existing or planned infrastructure.

An very applicable tool to perform this crowd management simulation is Pedestrian Dynamics developed by the company INCONTROL Simulation Solutions.

Crowd Management simulation approaches

The research of pedestrian behavior is mainly based on observations and empirical studies. The focus in this research lies on observing human behavior and capturing pedestrian movements. In this way existing theories are extended and new theories are developed. The basic properties of pedestrian movements and its research are:

  • Speed (m/s)
  • Density (person/m2)
  • Flow rate (persons/s)
  • Throughput time (s)
  • Inter-arrival time (s)

In order to be able to model pedestrian movements the most important challenge is to capture the human behavior, as result of the encountered circumstances. The most characteristic aspects of behavior in pedestrian movements seem to be:

  • People select the quickest route to their destination and dislike taking alternative (slower) routes even if congestion arises on the initial route
  • Each pedestrian has its own desired walking speed. This speed is dependent on both individual properties (e.g. age, gender, physical state, purpose of travel) as environmental properties (e.g. crowdedness, time, temperature)
  • Pedestrians keep a certain distance to other pedestrians, walls and other obstacles. Dependant on the crowdedness in the area this distance between the pedestrian will differ.

These aspects are important to include in simulation models of pedestrian environments and can be found in almost every simulation tool used for research about pedestrian flows.

Approaches to model pedestrian dynamics can be classified into three main levels:

  • Microscopic level: In the microscopic approach, each pedestrian is represented individually. The individual entities have a unique behaviour. Also the mutual behaviour of pedestrians, like collision avoidance, is taken into account. The microscopic models can be described in two main approaches, either continuous or discrete
  • Macroscopic level: This approach describes the flow of pedestrians as a fluid through space. The main subject of this approach lies with the behaviour of the combined pedestrians in a group. The corresponding mathematical models are partial differential sometimes similar to fluid equations (e.g. BAUER et al. 2007)
  • Mesoscopic level: In a mesoscopic approach the individuality of each particle is maintained. During each time step, particles are aggregated to field quantities such as density, the velocities are computed from these densities, and then each individual particle is moved according to these macroscopic velocities

Simulation

The simulation application Pedestrian Dynamics that forms the basis for this study uses a mesoscopic approach and is developed with the discrete event simulation software Enterprise Dynamics. The basic concept of the application is the controlled movement of individual entities (pedestrians) between locations over a node network.

The position of the nodes in this network is determined by the infrastructure (e.g. train station, airport, stadium and mega events) that is analyzed. In this infrastructure all relevant areas with their corresponding sizes are created by the user. Dependant to the infrastructure, the areas are connected by placing nodes in the area and connect them to each other. These nodes can represent either network intersections, passages or links to processes. Stairs and escalators are special types of nodes with adjusted speeds and capacities.

The application’s purpose is to analyze the “performance” of a functional infrastructure, for example a train station, airport, exhibition hall or stadium. Based on the purpose and destination of the pedestrian it will have (multiple) sequential destinations inside the infrastructure. A traveler will for example go to a ticket machine first and then to the train platform. Based on the defined destination sequence of the pedestrian it will follow the shortest and or fastest route to fulfill its purpose within the infrastructure.

Dependant to the properties (e.g. age, purpose, gender) of the pedestrian, it will have a desired walking speed. Based on the density in the area and the desired walking speed the pedestrian will have a certain walking speed, while travelling from one node to the next. Each time the pedestrian enters a node its walking speed is adapted according to the density of the area the pedestrian is in.

The degree of increase or decrease of the travel speed is a result of the speed-density relation

Apart from the network and the walking behavior also functional processes (e.g. ticket sales desks, ID check gates, shops) are taken into account. These processes are modeled as “servers” with corresponding properties (e.g. capacities, cycle times). As a result of the finite capacity, the cycle times and the distributed inter-arrival time of the pedestrians, queuing and congestion can arise.

Event-based simulation tools are very suitable for the mesoscopic simulation approach. Every time a pedestrian enters a process or node, an event is created for determining the current state (e.g. density) of the area. Based on this state, the pedestrian will act in accordance with certain pre-defined behavior. This behavior is extracted from microscopic research of pedestrian flows. For this study the discrete event simulation software Enterprise Dynamics has been applied.

Microscopic models often use forms of continues simulation. In these simulations the surrounding of a pedestrian (also called agent in this manner) is monitored continuously and the behavior of the pedestrian is adapted instantly. Microscopic research is often used to validate a model that represents the exact behavior of pedestrians in small areas. Due to the continuous monitoring of all the pedestrians in the model the required computer processor capacity of these models is very high.

Therefore large-scale microscopic simulation models are rare. Since the discussed mesoscopic simulation application like Pedestrian Dynamics is very well capable of dealing with large numbers of pedestrians and large scale routing networks, it is very suitable for modeling large infrastructures such as train stations, airports and soccer stadiums. Several simulation models have already reached a simultaneous content of over 70,000 pedestrians.

Please visit Pedestrian Dynamics for more information on crowd management simulation