Statistical Analysis of High-Flow Traffic States

Abstract

The relation between the fundamental observables of traffic flow (i.e., vehicle density, flow rate, and average velocity) is of great importance for the study of traffic phenomena. Probably the most common source of such data are inductive loop detectors, which count the number of passing vehicles and measure their speed. We will present an analysis of detector data collected by more than 3000 loop detectors during the past three years on the motorway network of the state of North Rhine-Westphalia. Besides presenting some general aspects of traffic flow, our analysis focuses on the characteristics of so-called high-flow states, i.e. traffic states where the flow rate exceeds 50 vehicles per minute and lane (3000 veh/h/lane). We investigate the duration, frequency and other statistics of such states, the viability of the data and we study the conditions under which they occur. The factors that influence the existence of high-flow states in traffic are, for instance, the fraction of slow vehicles (namely trucks), the motorway's general topology (e.g. number of lanes), the hour of the day and day of the week. This information is directly accessible from the detector data.