Computersimulation und Management von Fußgängerströmen bei besonderen Belastungen und kritischen Bedingungen anhand von konkreten Beispielen
Zusammenfassung der Projektergebnisse
Numerous crowd disasters occur every year at large gatherings around the world. Unfortunately, the information about the (spatio-temporal) development of these events tend to be qualitative rather than quantitative. Surveillance videos from the crowd disaster in Mina, Kingdom of Saudi Arabia, on the 12th of January 2006, where hundreds of pilgrims lost their lives during the annual Muslim pilgrimage to Makkah, provided the possibility to scientifically evaluate the dynamics of crowds. Through provision of this video material, it was possible to subsequently analyze the behavior of the crowd under increasing crowd density, leading to the disaster. The analysis of the crowd disaster revealed two phase transitions, from smooth flow to stop-and-go flow, and further to turbulent flow. From the analysis of the crowd turbulence and the following crowd disaster, the new quantity of "crowd pressure" has been introduced, which seems to have predictive power and can potentially be used to anticipate crowd disasters both in space and time. Based on the insights from the analysis of the crowd disaster described above, new tools and measures to detect and avoid critical crowd conditions have been proposed, and some of them have been implemented in order to reduce the likelihood of similar disasters in the future. A detailed video-based study of pedestrian trajectories revealed pedestrian interaction patterns. For example, it turned out that the interaction strength between two pedestrians decayed with increasing distance. The angular-dependence turned out to take the form of a half-circle in front of the pedestrians, i.e. pedestrians do not react to what happens behind their backs. Further, it was found that pedestrians do not interact to «ü other pedestrians, but rather to their 7-8 closest neighbors. This fact enabled an optimization of the social-force model. As a result of these findings, an improved version of the social-force-model was formulated, A second modification was made of the social-force model, which made it possible to reproduce the empirically observed crowd patterns - stop-and-go waves and crowd turbulence. When a pedestrian stream is crossing a vehicle stream and the accepted gap of pedestrians are small (corresponding to aggressive pedestrians), it was found that inefficient oscillations arise. A one-dimensional network-based fluid-dynamic model was designed , which made it possible to simulate 3 million pilgrims faster than real time. This model combines the level of detail of fluid dynamic models with the computational speed of queuing-network rnodels. The model was used to combine the pilgrim flow-density data obtained from video analysis with the generated pilgrim schedules. As a result, the relative suitability of different time schedules was obtained, and potential bottlenecks were identified that had to be avoided
