Crowd evacuation in high-density scenarios

Abstract: Nowadays, is becoming increasingly frequent holding events in outdoor environments, involving a large number of people, even though these spaces were not designed from a safe prospective, in the way that safety is defined today. In such events, high density conditions is an ongoing issue, as the evacuation becomes very complex, due to the space design and the interaction among the pedestrians. It becomes, therefore, essential to understand the crowd dynamics in high density conditions in order to design crowd management safely. In this context, modelling tools can be used to identify optimal evacuation and crowd management strategies. Such tools are mainly designed for low to middle density scenarios and, as a result, their use for high (intended here as 2-6 persons/m2) and extreme (>6 persons/m2) density scenarios needs to be evaluated. This thesis presents a possible approach to investigate the simulations of high density scenarios based on the field observation of a real event and by applying the findings to a real case study. The crowd dynamics were taken into account as the combination of three hierarchical components, i.e., theory, data and application, which influence each other and cannot be investigated separately. The theoretical background is presented through the state of the art literature review. Data, which is a very sensitive parameter, required an idea about what to collect (theory) and their interpretation. They were collected performing a video analysis and then were used in the application, i.e., to calibrate the simulation models. The observation of the real case study allowed to get important data, e.g., speed, density and flow, in a complex, high density scenario, due to field data available in literature to this topic are scarce. It has been found that high local (dynamic) densities can be reached in critical points, i.e., bottleneck, despite the controlled (static) densities in the area. A set of tests (simulations), using first the default input parameters and then introducing the output from the field observation, were run in a continuous evacuation model (Pathfinder) to investigate how such type of evacuation models should be calibrated in high density scenarios. Earlier, before this it has been investigated the geometrical limit of the tool, i.e., body size of the agents and comfort distance parameters. The objective was to reproduce the local density and the flow recorded during the real event. The results indicated that it has been possible to achieve reasonable values for both, by prescribing the speed, the agent’s body size, the comfort distance and the initial density. This conclusion is important due to more reliable models can be used for crowd management and to identify solutions/critical conditions ahead of the events. Finally, the high density issue is tightly connected to the pressure increase among pedestrians, which represents one of the major factor leading to fatalities in crush events. For this reason a simple model to forecast the force among pedestrian has been introduced. This model is based on the combination of “leaning crowd model”, frictional forces and the drained discontinuous assembly of particles under load, which is a very similar to the high density packing scenario. The pressure values obtained with the novel modelling approach and the data from the literature are comparable.