Floor plan of the ESPRIT Arena used in the simulation with PedGo. Red dots show the initial locations of the (non-moving) spectators. 

Context in source publication

Context 1

... function f is some strictly and monotonically increasing function, i.e. for increasing t, A, or p the value of f also increases. Both conditions (1) and (2) have to be met for approval of the design. The general design process for pedestrian traffic shown in fig. 3, also applies to route elements of an evacuation system. As mentioned above, the design criteria according to the prescriptive codes are the length and width of escape route elements. Therefore, prescriptive codes are static approaches not taking into account the crowd dynamics aspects of an emergency evacuation. This can be taken into account by evacuation simulations as often used in day to day fire engineering work. However, although simulations significantly improve the analysis method, they still only analyse the evacuation with static input data. The users of such software tools take well defined initial data and analyze it with the simulation, thereby trying to identify the worst case scenarios in order to provide a reliable safety level with their results. All possible scenarios which can occur in the day to day operation of e.g. a large stadium cannot be covered by such a method. An evacuation assistant like the PedGo Guardian is designed to tackle this problem. The input it constantly uses reflects the real and actual situation. Thus it supervises reality and delivers results, which helps operators of the stadium, e.g. the crowd management. Based on the forecast of the simulation, the security staff can prepare for events to come and can test different strategies to handle potentially hazardous or very uncomfortable situations. In order to assess the different strategies and their consequences, a level of hazard concept based on the level of service and the crowd density and movement characteristics is used. The service standards for an evacuation differ from the level of service used for pedestrian walkways, which is based on the notion of comfort [3]. The following table shows the level of service for waiting areas which allows higher densities than for walking areas. It also shows a simple, traffic light like, classification in red, yellow, and green. The example of application and test scenario is the ESPRIT Arena in Düsseldorf, Germany. It is a multi-functional arena, where sports (e.g. football) and cultural events (rock concerts, etc.) take place. The general arrangement of the stadium is shown in fig. 1. A lounge area is integrated in the western grandstand. Most persons access the stadium via the tram station located in the south next to the stadium (see fig. 1 , the station is denoted “U” for underground). As shown in Fig. 2, the Guardian uses two kinds of input data. The basic geometry of the building is static. It consists of walls, stairs and doors which form rooms which form the stadium. The danger management system delivers data about the status of doors and rooms. E.g. if smoke is detected in a room, it is assumed, that this room will not be entered. Thus the doors leading into the room are blocked. Thus, the geometry and resulting the route choice is adapted to the real situation. For distributing the population, the guardian connects to a person counting system. This could be a manual fed database, but automatic (and expensive) systems are available as well. Within HERMES, the German company Vitracom implemented their detection system. It consists of specially developed cameras detecting and counting in and outflow of persons. The information from video cameras is used to extract person flows and determine the occupancy of the rooms in the stadium. Those values are stored in a database for further analysis and at the same time fed into the evacuation simulation. Therefore, the simulation can at any time be triggered based on the actual occupancy of the stadium. Another detection method successfully used by the Guardian was the detection system of the British company Crowd Vision. In contrary to Vitracoms method, the Crowd Vision Software uses conventional CCTV video feeds and counts the detected persons. The combination of the PedGo Guardian and Crowdvisions detection method was demonstrated during the British Instinct event, held in London in 2009. Since the costs of automatic detections systems is a factor, only half of the lower ranks and the adjacent promenade could be supervised by the detection system . This resulted in a maximum of approximately 10.000 persons. The architecture is not only used in the PedGo Guardian implementation of the evacuation assistant, but a general result of the Hermes research project. Further details about Hermes can be found in another publication in these proceedings by Holl [18]. Of course, the data obtained from the CCTV can also be used to calibrate other models like macroscopic flow models. This was actually done within Hermes [18]. The role of the operator is central for the successful use of the PedGo Guardian. The operator processes the information from the security staff, the police, and the fire brigade. Based on that information and the information fed in by the building management system (the smoke detectors and other detectors and actors), the operator initiates a simulation showing the overall evacuation time and the ...