Microscopic Simulation of Pedestrian Traffic in a Station Environment: A Study of Actual and Desired Walking Speeds

Abstract: In order to attract pedestrians to travel with public transport instead of private cars, the layout of interchange stations is important and should be designed in an effective way. Microscopic simulation of pedestrians can be used to evaluate different layout scenarios or a future increase in flow. The simulation software Viswalk was investigated, where the movements of pedestrians are based on a social force model,. The purpose of this thesis was to investigate simulated walking speeds for different flow levels and to investigate the effects of dividing pedestrians into types with different desired speeds. The aim was to find a desired speed distribution that can be used for different flow levels. Field studies have been performed to collect pedestrian traffic data with a video camera at Stockholm Central Station. Two disjoint flow levels were identified and used to investigate if the same desired speed distribution could be used for different flow levels. The average observed walking speed was 1.33 metres per second at the low flow level and 1.25 metres per second at the high flow level. The error was 4.5 percent between the average observed walking speed and the average simulated walking speed when the optimal desired speed distribution at the low flow level was used at the high flow level. Effects of using different desired speed distributions for different pedestrian types have also been investigated. The error between the average of the observed and the simulated walking speeds varies between 2.3 and 4.1 percent when dividing pedestrians into different types when the optimal desired speed distributions at the low flow level are used at the high flow level. A sensitivity analysis of some parameters of the social force model in Viswalk has also been performed. Several adjustments of the parameters show that some parameters had great impact of the simulated walking speeds. The final conclusion is that the parameter configuration and how the pedestrians are divided into different types affect the average simulated walking speed.