Efficient networks take work: Traffic management and Braess’ paradox

By Chris McCahill
Not all roads are created equal. In fact, adding certain roads to a system can actually slow down traffic under the right circumstances. This is a fairly well known phenomenon called Braess’ paradox, named after the German mathematician who first wrote about it in 1968. Fortunately, researchers have studied this occurrence extensively and developed methods for knowing when it can happen and how to prevent it.
Although it might seem counterintuitive, the reasons for Braess’ paradox aren’t complicated. A high-speed road is often the shortest route for travelers and the most logical choice from their perspective, so they gravitate toward it. As more people do this, however, the connecting roads may not be able to handle the additional traffic, causing bottlenecks in the system. By encouraging some of those travelers to take alternative routes, as agencies sometimes do with digital highway signs, travel times can improve for everyone. A major problem is that it isn’t always clear when and where the effect occurs, especially after the roads are already built.
This problem motivated researchers from China’s Southeast University to develop a way to identify troublesome network links under different traffic demand conditions and understand how restricting traffic on those links can improve overall network flow. Their study and their fairly straightforward methods are published in a new issue of the Journal of Transportation Engineering.
Essentially, they calculate the total travel time for all road users in a system, then experiment by removing one or two links at a time and seeing whether the travel time goes up or down and by how much. They also test different levels of travel demand to see at what point certain links become a problem.
These methods could let agencies monitor how their roads perform and come up with ways to manage flow, thereby improving travel times. In certain cases, temporary road closures might be beneficial. More likely, however, agencies could implement turning restrictions, ramp metering, speed management, or dynamic pricing to control flows. In the future, connected vehicles and managed systems could direct drivers and vehicles onto different routes, eliminating occurrences of Braess’ paradox entirely.
Chris McCahill is an Associate Researcher at SSTI.