Traffic Flow Modelling and Optimal Control of Large-Scale Motorway Networks

Kotsialos Apostolos

ABSTRACT

This thesis is concerned with the problem of modeling and control of the traffic flow in a large-scale motorway network. More specifically, we focus on the problem of coordinated ramp metering. Ramp metering strategies aim at determining the flow that is allowed to enter the mainstream during one discrete control time step.

            The macroscopic modeling tool METANET has been used in order to model the motorway network around Amsterdam. METANET also provided the modeling part for the design of a model-based optimal control strategy for coordinated ramp metering. The resulting optimal control tool is AMOC (Advanced Motorway Optimal Control). AMOC delivers an optimal open-loop solution to the problem of coordinated ramp metering in a motorway network. These solutions are characterized by their efficiency and their equity. Because open-loop solutions are based on system disturbance predictions, which inevitably contain errors, they become suboptimal when they are applied. In order to conserve the positive characteristics of the open-loop solutions, the rolling horizon technique is used within a hierarchical control structure. This hierarchical control strategy has been applied to the Amsterdam ring-road for a number of scenarios.

            Based on the results obtained, the proposed strategy retains the efficiency and the equity properties when confronted with errors in the disturbance predictions, model parameter mismatches and capacity reducing incidents in the network. Furthermore, the required computation time is quite small something which allows the strategy’s real-time application.