Modeling of pedestrian movement around 90 and 180 degree bends

For the planning of large pedestrian facilities, the movement of pedestrians in various situations has to be modelled. Many tools for pedestrian planning are based on cellular automata (CA), discrete in space and time, some use self driven pargticles (SDP), continuous in space and time. It is common experience that CA have problems with modelling sharp bends in wide corridors. They tend to move the pedestrians to the innermost lanes far too strongly, thereby reducing the capacity of the facility. With SDP, the problem seems to be less pronounced but still present. With CA, we compare the performance of two standard shortest distance based static floors on 90 and 180 degree bends with a newly defined one. For SDP, we demonstrate how variations in the modeling of the momentary destination of the agents influence trajectories and capacity.