Action Filename Description Size Access License Resource Version
Show more files...


The aim of this thesis is to develop a new methodology to determine dynamic Origin-Destination (OD) matrices for urban networks characterized by a high number of traffic hubs, complex route choice possibilities and a high level of traffic controls. By reviewing existing methods, from static to dynamic OD matrix evaluation, deficiencies in the approaches are identified: mainly, the level of detail of the traffic assignment for complex urban networks and the lack in dynamic approaches. The proposed methodology is comprised of a heuristic bi-level approach. Assignment of the initial demand is performed by mesoscopic simulation based on the Dynamic User Equilibrium to model detailed dynamic traffic patterns without numerous calibration parameters. OD flow adjustment is executed by an efficient least square solution which takes into account dynamic aspects of the flow propagation and traffic counts. For this task, a LSQR algorithm has been selected for its capacities to deal with a large matrix and its ability to constrain outputs. Parallel comparison with the most common approach for OD estimation (sequential static approach) has shown: first, the ability of the method to generate OD flows close to the actual demand, compared to the common practice; second, the utilization of the obtained demand by a dynamic traffic model has established its aptitude to reproduce realistic assignment patterns. Finally, applicability and example of utilization of the proposed method has been presented by solving realistic problems using the simulation software AIMSUN in which the proposed methodology is implemented as a plug-in. This research has shown the importance of input data for the OD estimation process and mainly the detection layout configuration used for traffic count data. Sensitivity analysis has shown that a small number of detectors is usually sufficient for efficient OD estimation in short computation time, if the traffic detectors intercept the most critical flows.