Thursday, May 28, 2009

Adding roads doesn't solve traffic problems

I've noted before that increasing roadway capacity creates induced demand, in which additional traffic absorbs the new capacity and creates more problems in streets feeding the new capacity. There's more to the story. Traffic engineers are finding that more roadway choices causes more congestion because of something called Braess' paradox. One day I hope to understand it well enough to summarize it. Until then, there's this excellent explanation.


Nathan Phillips said...

Great post Sean. The blog you linked to - what if Langley Rd. were shut on a trial basis between Center and Beacon? What if Grove St. was disconnected from I-95? Might traffic improve?

Most but not all said...

The reasons for the improvements noted recently by the folks who are championing the phrase "Price of Anarchy" are much more complex than Braess' Paradox can explain. In the transportation research community, Braess' Paradox (not really a paradox, but a lack of intuitiveness between expectation and theory) has only been theoretically discussed (since the late 1960s), but never actually observed. It is more likely that the complex effects of induced and discretionary demand are at work here. For example, if most of the trips using a certain route are discretionary, closing a link on that route could produce an improvement in the system if the increase in travel time for those trips makes them unacceptable, and they are cancelled.

Steve Runge said...

Timely! Just last night, Noah Budnick of NY's Transportation Alternatives said one of the biggest hurdles to bike lanes is making this counter-intuitive relationship between road-share and traffic clear to civic leaders. Repeat, repeat, repeat: more roadway for cars = more congestion. Less roadway=less congestion. And the corollary: people who arrive other than by SOV spend more money in downtown areas: bikers, walkers, public transportation riders. We're big spenders.

Pedestrianist said...

One crucial element is the quality of the road choices. If a road that is perceived to be faster than any other is added to a grid of streets, then everyone is inclined to choose that one road, rather than dispersing through the grid.

There is experimental and empirical evidence that providing more road connections and equalizing the desirability of roads makes traffic flow more smoothly on the whole.

Douglas A. Willinger said...

I guess you have never been on highways such as I-87 in Westchester County in N.Y. between I-287 and the Cross County Parkway.

Perhaps you read something that left out other factors as development and interchanges?

gravityandlevity said...

Hi Sean,

I'm a bit late for the discussion, but I'm glad you found the post interesting.

One point that I should have made more carefully, and that I'm glad to see is being discussed here, is that "Braess's Paradox" doesn't guarantee that closing a road will make traffic worse. Adding a new road may very well improve traffic. I just wanted to show that it is possible to make traffic worse by creating a new "best" option for drivers.