University of Dresden Traffic Modeling Sees the Light at the End of the Tunnel
IVsource.net
18 April 2003

Traffic modelers in Germany are making some bold propositions – they say traffic-dependent speed limits and driver-assistance technology are the keys to making traffic jams a thing of the past.

Traffic modelers in Germany are making some bold propositions – they say traffic-dependent speed limits and driver-assistance technology are the keys to making traffic jams a thing of the past.

This provocative result is based on the results achieved by a computer model built by Martin Treiber and Dirk Helbing of the Technical University of Dresden.  Treiber and Helbing's model predicts traffic congestion developing on particular stretches of road, and suggests how driving restrictions might eliminate them.

In the model, each driver is assumed to accelerate to his or her preferred speed when the road ahead is clear, and to reduce speed if there is a slower vehicle ahead.  In dense traffic, drivers maintain a constant distance from the next vehicle.  Drivers in the model slow down gradually for obstacles spotted well in advance, but brake sharply in an emergency.

The researchers examined a section of the German autobahn A8-East between junctions for Weyarn and Irschenberg.  Pressure sensors monitored the traffic flow along this 12-km section of highway on a single day in late 1998.  During the evening rush hour, the average travel times of vehicles on the road increased and two traffic jams were observed.

This real-world data, representing traffic entering the stretch of highway, was fed into the model, which then accurately generated the traffic jams -- thus providing some level of model validation.

Treiber and Helbing next experimented with driving restrictions that might eliminate the congestion.  One approach was to impose a speed limit of 80 km per hour – this appeared to keep vehicles moving, but it also slowed the average speed of lighter, non-rush-hour traffic unnecessarily.  As a result, the team advocates dynamic speed limits which are invoked depending on traffic conditions.

Ramp metering was also examined for motorways within the model.  In ramp metering cases, flow on the main line must be balanced with waiting times (and vehicle queues) on the entrance ramps.  The researchers say that their optimal regime smoothed out a potential traffic jam without imposing excessive waiting times at the junction.

It is well known that the human driver is a key contributor to traffic congestion, due to factors such as slow reaction times and over-compensation in responding to other vehicle movements.  The resulting 'ripple effect' turns into traffic jams quickly when vehicle density is high.  According to Treiber and Helbing, the Dresden model shows that “fitting just 10% of cars with driver-assistance systems could ease this problem; putting them in 20% makes congestion vanish.”

In driver-assistance systems, as they put it, “automatic controls in the car sense the distance and speed of the car in front, and accelerate or decelerate accordingly.”  This implies some form of adaptive cruise control.  Needless to say, their assumptions about system operation are key to the modeling results.  IVsource hopes to find out more and report more details in the near future.

[Top]