Application of Macroscopic Fundamental Diagrams to Dynamic Traffic Management

Abstract

This thesis project is a part of Dutch project PraktijkProef Amsterdam (PPA), in which the coordinated network traffic management in the regional area is analyzed. One of control levels in the project PPA is sub-networks, where the concept of the macroscopic fundamental diagram (MFD) is applied. Unlike a conventional link fundamental diagram, an MFD relates the output and number of vehicles in a large network, and enables road operators make a real-time control more efficiently. This graduation project makes a first attempt to use MFD in the Netherlands. The study area of the project PPA is the metropolitan area of Amsterdam. Because empirical data are not available, especially on the urban roads, the simulation model is used to generate the traffic variables for deriving the MFDs. In this case, the macroscopic model RBV is firstly used. Compared to the theoretical fundamental diagram as well as the MFDs obtained based on empirical data in previous studies, the MFD derived by using the RBV model shows two special patterns. One is that the flow is always lower when congestion is dissolving than that during the onset of congestion. The other characteristic is that flows keep rising with the increasing of density and the congestion branch in a conventional fundamental diagram is missing. The two patterns are observed in all MFDs for different demand levels and sub-networks. After analyzing the principles of the RBV model, the reasons for two strange characteristics are revealed. The drop of flow between the onset and the resolving of congestion results from the fact that the locations for measuring flow and density are not corresponding in the RBV model. In terms of the missing congestion part, it is due to the assumption of the RBV model that the flow of a link is constant, which is the saturation flow, even during the congestion period. Since the critical density is not visible in the MFDs derived by RBV, they cannot be used for traffic control. Hence, the RBV model is not suited for deriving MFD. Then the microscopic model VISSIM is applied. The congestion branch is observed in the MFDs derived from VISSIM. However, the drop of flow between the onset and the resolving of congestion still exists. The further analysis reveals that the OD matrix used in this project leads to a dramatic decrease of flow when congestion is resolving. The inflow becomes very low when the density is still high on the link. Due to the same problem in RBV with respect to measurement locations, the drop is also observed in the MFDs based on the data from VISSIM. However, this defect does not affect reading the important patterns on MFD such as the critical density and the maximum flow. So the VISSIM is proven a feasible model to derive MFD. Afterwards, two DTM measures, ramp metering and extra lane, are implemented in the VISSIM model. The MFDs in the different networks are derived for each scenario. When ramp metering is applied, the MFD of the whole network almost remains same. But the large decrease of the maximum density is seen in the MFD of the motorways. By contrast, the maximum density on the urban roads increases, implying a worse traffic situation in the urban network after using ramp metering systems. In terms of extra lanes, they are implemented on the different road sections with two speed limits. The simulation results show that only when the A10 west is expanded, the MFDs of the study area experience significant changes. The maximum density decreases greatly and the congestion part disappears from the MFDs. The scenario with a lower speed limit seems a bit better due to a slight smaller maximum density. In addition, this project also investigates the possibility of using the MFD as an evaluation method. A multi-criteria analysis is made to compare the MFD and the conventional method, in which the traffic situation is revealed by travel time and travel speed. By assessing the MFD and the conventional method against the criteria of accuracy, visualization, feasibility and costs, the MFD performs worse than the conventional method, from the viewpoints of two stakeholders in this case, Rijkswaterstaat and the city of Amsterdam. However, using the MFD to evaluate the effects of DTM measures is still possible on the motorways because the traffic data are easily collected. The combination of the MFD and the conventional method will also improve the reliability of the evaluation results.