Cross-comparison of Macroscopic Fundamental Diagram Estimation Methods

The aim of this paper is to investigate MFD estimation methods from traffic states observations. To eliminate all the experimental bias, traffic situations will be determined from a simulation tool that encompasses an equilibrium model (the LWR model). Three methods will be compared: (i) the analytical method that provides the upper bound of the MFD, (ii) the “production” method that requires vehicles trajectories and (iii) the “loop detector” method that aggregates flow, speed and occupancy observations. This latter method will be thoroughly examined to quantify the impact of loop detector positions (spatial distribution, distance from traffic signals …) and of heterogeneities in traffic states. Recommendations will finally be drawn to improve the global understanding on MFD and to better define methods to calibrate them.