An Experiment with Perimeter Control in Downtown Seattle

Abstract

Efficient urban traffic control requires managing city-scale networks. Traditional methods involve partitioning a network into independent subnetworks (corridor level mostly) and controlling them separately. However, time-invariant macroscopic fundamental diagrams (MFD) offer new insights for effective macroscopic-level traffic control, introducing concepts like perimeter control methods. While most studies were tested on hypothetical networks, in this study, we aim to apply the MFD-based perimeter control on a real-world city traffic network. We first work on a traffic network simulation of Downtown Seattle.After that, the simulation network is partitioned into homogeneous and practical subnetworks (regions). A macroscopic fundamental diagram (MFD) is then trained for each region to calibrate and validate the homogeneity of the partitioned regions. An MFD-based perimeter control model is applied next to control the flows in/out of each region and balance the congestion (represented by the number of vehicles) in each region. In the end, the results and findings of testing the perimeter control strategy on our real-world regions are presented, and challenges and further improvements on the method are discussed.