Microscopic Simulation of Pedestrian Traffic

Sammanfattning: There has recently been a renewed interest in planning for pedestrian traffic, primarily in connection to public transport interchange stations, since these are important for public transport to constitute an attractive alternative to car usage. This thesis concerns microscopic simulation of pedestrian traffic, which is a promising tool for analyzing and predicting the traffic situation in a given pedestrian facility; particularly powerful when the traffic is congested. Important applications of microscopic simulation include comparison of possible infrastructure designs such as proposed interchange stations, and evaluations of various traffic management solutions, for example information systems.The purpose of this thesis is to advance the capabilities of pedestrian microsimulation toward a level at which it can be reliably applied for quantitative analysis by practitioners in the field. The work is based on an established microscopic model of pedestrian dynamics, the Social Force Model (sfm), and the advances are made in a number of different areas.To be able to evaluate and compare simulated traffic situations suitable performance measures are needed. A set of local performance measures are proposed that quantifies the local delay rate density and estimates the discomfort perceived by the pedestrians.The sfm is extended to include waiting pedestrians through the introduction of a waiting model, demonstrated to be stable and free from oscillations. The inclusion of waiting pedestrians in the model is critical for accurate modelling of public transport interchange stations, where large groups of waiting pedestrians may hinder passing pedestrians if the design of the station is poor.The relaxation time of the adaptation to the preferred velocity is an important parameter in force based models of pedestrian traffic since it affects several behaviors of the simulated pedestrians, two of which are linear acceleration and turning movements. A comparison of observations of accelerating pedestrians reported in the literature and new observations of turning pedestrians indicates that no value of the relaxation time can give model behavior consistent with both sets of observations. This indicates that modifications of the model is needed to accurately reproduce the observed behavior.An important input to simulations is the preferred speed of the simulated pedestrians. The common assumption that the preferred speed distribution at a location does not vary during the day is tested through observations of pedestrian traffic at Stockholm Central Station. The results demonstrate that the preferred speeds are lower in the afternoon than in the morning, implying that the preferred speed should be treated as a source of uncertainty when applying pedestrian microsimulation.Finally, a sensitivity analysis of a simulation of the lower hall of Stockholm Central Station is performed to find the most important sources of uncertainty in the model predictions, given the available data. The results indicate that the uncertainty related to calibration is the largest of the considered potential error sources.