Hysteresis and Capacity Drop Phenomena in Freeway Networks

Abstract

The objective of this study was to characterize hysteresis and capacity drop phenomena in freeway networks with the use of commonly available loop detector data from three networks: Chicago, Illinois; Portland, Oregon; and Irvine, California. For exploration of the effects of variations in network topology and size on the network fundamental diagram, a comparison was made by using the observed flow–occupancy diagrams of the selected freeway networks. The results provide further confirmation that findings from the literature for a limited number of networks are also valid for freeway networks not previously studied. Freeway networks have been found more likely to exhibit an inconsistent hysteretic pattern in shape and size that depends on the spatial distribution of congestion over the network. On the basis of empirical observations, hysteresis loops were characterized by their shape and size. Two shapes of hysteresis loops, H1 and H2, were identified and characterized. The size of each hysteresis loop was believed to be characterized by its width, height, and the area covered by the hysteresis loop. The authors postulated that the capacity drop phenomenon existed in freeway networks in a manner similar to that in individual freeway sections. Two types of capacity drop were identified. Type 1 was associated with the inability of the freeway network to sustain its throughput at its peak value for a relatively long time, and therefore, capacity dropped while demand was still high and the network was loading. Type 2 was associated with the instability of network traffic when the network underwent reloading (e.g., afternoon peak period) after an incomplete recovery from the initial loading (e.g., morning peak period). In some cases, this reloading resulted in a lower capacity in the afternoon than in the morning. Empirical results showed that the observed phenomena were reproducible on different days and for different networks.