Modelling and Simulating Indoor Pedestrian Movement Behaviour and Displacement

Annika Laumets-Tättar
Pedestrian evacuation routes are an important part of building's architecture. Knowing before building a structure if the evacuation routes are efficient enough should benefit both the people and the owners of the building. One's benefit is their life, the other's benefit not paying hospital and other fees for the injured or deceased. Three different approaches of pedestrian simulations are mostly used - particle-based approach (social force model), CA-based approach, and autonomous agents. The two first approaches are macroscopic models and the latter is a microscopic model. This thesis gives an overview of an implemented microscopic model, which uses autonomous agents, group modelling and social comparison theory to evaluate building egress design. With microscopic models each agent can have different attributes and ties with other people in the vicinity (e.g. friends and family). People use social comparison theory (SCT) in their daily lives to compare themselves to others. In an evacuation situation similar people form groups by social comparison theory and influence each other. The model is implemented using Python, SUMO and TraCI. In this thesis a case study is done on Ülemiste center in Tallinn, Estonia using the implemented model. Different number of exits and pedestrians are tested to see how they are associated. The results show that the farther apart the exits are, the smaller the evacuation times are. The same results appears with only one exit and all exits. Bottlenecks near the exits closer together are the reason for higher evacuation times. Agents also prefer exits farther apart when analysing the distribution between exits. Speed does not affect the evacuation time as much as expected with higher number of pedestrians because the density of the crowd disallows the agents to reach their high speed. The final outcome is that with speeds 1.4 m/s and 2.5 m/s 90\\% of people out of 200 agents exit the building around 400 seconds. This kind of result happened with all exit configurations. The other 10\\% might be older or confused people who are not familiar with the building's floor plan. Majority of people are, therefore, safe in 6 minutes and 40 seconds, which seems quite realistic for 200 people. The model implemented can be used to assess and evaluate a building's egress design before the structure is actually built. It can help design better evacuation routes for buildings because a user-specified floor plan can be used by the model.
Graduation Thesis language
Graduation Thesis type
Master - Computer Science
Amnir Hadachi
Defence year