Towards a Daily Agent-Based Transport System Model for Microscopic Simulation, Based on Peak Hour O-D Matrices

This paper presents the development of a daily, large scale, agent-based microscopic transport simulation integrating diverse data-structures and including the following transport modes: car, bus, bicycle, motorcycle, and pedestrian. The daily simulation is built upon an already calibrated model for the morning peak hour of the city of Bologna, Italy. The transport supply integrates diverse open-source data such as OpenStreetMap (OSM), traffic light schemes and General Transit Feed Specification (GTFS). On the other hand, the transport demand is based on peak-hour Origin-Destination Matrices (ODMs) and uses traffic flow data extracted from detectors throughout the city to scale rush-hour trips accordingly and disperse their departing times over 24 h. The plan choice model is calibrated based on a simple utility function approach allowing to predict the latest city transport mode split. The model successfully distributes departure times of internal and external trips, compatible with absolute daily traffic flow profile from the detectors. A microscopic traffic simulation is executed at a 10% population demand. The validation process is then conducted by comparing the simulated and observed traffic flows at traffic counts by hour. Finally, total daily travel times by mode of individuals are interpreted and compared. The simulation outputs indicate significant differences in total daily travel time by mode. In particular, bus users have the longest travel time followed by cyclists, car drivers, motorcyclists and pedestrians. Therefore, the developed model is able to evaluate impacts of hypothetical scenarios over a day.