Introduction to Modern Traffic Flow Theory and Control: The Long Road to Three-Phase Traffic Theory ,
In the past two decades, sensor-rigged highways in Germany and other countries have provided researchers with a considerable amount of data on rates of traffic flow (vehicles per hour), vehicle speed, and vehicle density (vehicles per kilometer). Those data are a boon for traffic researchers such as Boris Kerner at German automaker Daimler AG. A former plasma and solid-state physicist, Kerner has virtually overturned the conventional theory of highway traffic flow and advanced the field to an extent not seen since the 1950s, when Robert Herman, a physicist at General Motors, introduced statistical physics to traffic modeling.
Introduction to Modern Traffic Flow Theory and Control: The Long Road to Three-Phase Traffic Theory is Kerner’s account of how he and his research group analyzed the highway data and developed their theory. They identified three distinct traffic phases: free flow, when vehicles move uninhibited; synchronous flow, when vehicles travel at speeds below those characteristic of free flow but still maintain high rates of flow; and the jammed phase, when vehicles come to a standstill in spots and the flow rate drops to zero. The last two phases are called congested states. The researchers discovered that free flow does not spontaneously transform into the jammed phase but rather goes initially into the synchronous phase. Only from the second phase can jams nucleate. As far as I know, that finding had not been noted in previous research.
Although Kerner’s three-phase theory has been controversial, he appears to have successfully refuted all major criticisms by showing that it works. One of the theory’s novel features is the two-dimensional region of flow rate versus vehicle density that comprises the congested states. Previously, traffic engineers collapsed those data onto a single curve—the venerable fundamental diagram of traffic. Kerner and his colleagues have introduced three-phase models—which feature vehicle dynamics—that accurately reproduce observed traffic patterns. They have successfully applied the theory to identify congested phases in real-time traffic data and to predict travel times. And they have also developed a new algorithm for on-ramp flow control that reduces congestion and improves throughput.
The data collected from highways and subsequent theories developed from those data have made older books such as Adolf May’s popular Traffic Flow Fundamentals (Prentice Hall, 1990) partially obsolete. Introduction to Modern Traffic Flow Theory and Control is a shorter and more readable description of traffic research than Kerner’s previous book, The Physics of Traffic: Empirical Freeway Pattern Features, Engineering Applications, and Theory (Springer, 2004). Most of the mathematics is at the level of introductory calculus. Concepts are clearly illustrated with figures, and the book’s useful glossary of traffic terminology should make the material accessible to graduate students in physics, mathematics, and engineering. Kerner describes and critiques the major traffic models going back more than 50 years and includes an extensive list of references. Much of his work has been published in Physical Review E and other refereed physics journals. In the last chapter, Kerner briefly discusses his ideas on the future of theoretical traffic research.
I have a few minor criticisms. The author starts with empirical data and draws the conclusions that lead to his three-phase theory—a suitable way to begin. But he abandons that approach in chapter 3, in which he uses simulations to illustrate certain features of the theory before introducing the computational models that are required to do the simulations; that approach makes it seem as though empirical data were unavailable. Additionally, I found incorrect grammar usage that should have been spotted in the editorial process. Nonetheless, I highly recommend Introduction to Modern Traffic Flow Theory and Control. Three-phase theory must be taken seriously, and traditional analyses by traffic engineers should be revised. I hope the book will encourage the traffic research community to employ the concepts and methods that Kerner has so convincingly presented.
Craig Davis is a PhD physicist and former research scientist at Ford Motor Co’s Scientific Research Laboratory. Following his retirement, he conducted traffic research as an adjunct professor at Michigan State University in East Lansing and the University of Michigan in Ann Arbor.