Characterizing and distinguishing free and jammed traffic flows from the distribution and correlation of experimental speed data

Abstract

From a physics point of view, the emergence of a traffic jam is considered to be a dynamical phase transition. To verify this, we performed a series of circuit experiments. In previous work, Tadaki et al (2013 New J. Phys 15 103034), we confirmed the occurrence of this phase transition and estimated the critical density between free and jammed flows by analyzing the fundamental diagram. In this paper, we characterize and distinguish free and jammed flows, beyond the analyses of fundamental diagrams, according to the distribution and correlation of experimental speed data. We find that the speed in free flow does not correlate and its distribution has a narrow single peak at the average. The distribution of speed in jammed flow has two peaks or a single broad peak. The two peaks indicate the car speeds inside and outside of jam clusters. The broad single peak appears as a result of the appearance and disappearance of jam clusters. We also find that the formation of jam clusters induces a long correlation in speed. We can identify the size of jam clusters and the relative distance between coexisting jam clusters from this speed correlation.