Traffic jams sometimes start for no apparent reason and can last for hours. Now, after observing a notoriously choked stretch of autobahn, two German researchers think they understand better why even a minor slowdown on a packed highway can rapidly trigger long-lasting gridlock: Traffic at times appears to act like a substance rapidly changing from, say, a gas to a liquid. Their findings, to appear later this month in Physical Review Letters, could lead to more accurate highway traffic predictions.
Twenty-two years ago, Joseph Treiterer of Ohio State University, Columbus, suggested that traffic flows can resemble phase transitions--changes of state such as the condensing of steam into water. To investigate traffic transitions, Boris Kerner of Daimler-Benz in Stuttgart and Hubert Rehborn of Heusch/Boesefeldt, an intelligent transport systems firm in Aachen, used magnetic induction loops in a three-lane stretch of southbound highway near Frankfurt to monitor the number and speed of cars traveling in each lane.
Based on their observations, the duo describe three states of traffic on busy highways: "free flow," in which traffic is light enough that drivers can freely change lanes and pass other cars; "synchronized flow," in which the roads are so clogged that drivers can't pass, but can still move; and jams. Kerner and Rehborn found that surges of vehicles joining the highway disrupted the traffic flow enough to initiate the jump from free to synchronized flow within minutes. After being established by what might be only a brief disturbance, the synchronized flow could last hours.
Kerner suggests that heavy traffic behaves like supercooled steam, in which it only takes a molecular "seed" to trigger millions of molecules to condense into water. A highway seed could be the sudden rush of drivers on an entry ramp, or drivers already on the highway slowing to admire the scenery, Kerner says. These apparently isolated disturbances "can cause a phase transition to a state with very high density of vehicles which stays for hours on the road," he adds. Once synchronized flow is established, the researchers found, free-flow driving resumed only when the traffic volume had dipped to about 50% of pretransition levels.
The findings could be used to help develop traffic forecasting and intelligent transport systems, Kerner says. Incorporating phase transitions into traffic models will be "a real challenge," warns Stefan Krauss, a traffic researcher at the German Aerospace Center in Cologne, because modeling efforts so far have focused on average traffic levels rather than jumps between traffic states. But the new work, Krauss says, offers the prospect that "simulations [could] become good enough to give a good forecast of things like road capacity."