Using empirical travel time distributions for calibration of a model of pedestrian dynamics

Before a model can be used for planning real projects, it must be validated to ensure that the simulation delivers plausible results. The motivation for this study comes from such a validation that revealed unexpected long travel times in a bottleneck simulation based on the social force model. Using a validation approach based on the distribution of individual travel times we show that a calibration with an equal reaction time in the driving force term is the origin of this behavior. Building on this insight, we propose an alternative formulation of the driving force term of the social force model that focuses on the base acceleration. In addition, an analysis of the individual travel times of the participants is carried out, which explains the high variance of travel times at high occupancy rates.