« Rage Against the (Hybrid) Machine | Main | Driving With Alcohol »

April 10, 2006

Hitting the Sweet Spot

Here's a cool graph from the Puget Sound Regional Council that illustrates the "sweet-spot" for highway speeds.  Apparently, traffic throughput is maximized at about 1,800-2,000 cars per highway lane (the horizontal axis) when vehicles are moving somewhere between 40 and 50 miles per hour (the vertical axis).

As the graph shows, when speeds are lower than that, or higher than that, then highways aren't operating as efficiently as they might.

So it would seem (to me at least) that a key ingredient in reducing demand for new highways is to keep traffic on existing roads flowing at somewhere between 40 and 50 miles an hour, even at times of peak demand.  How to do that?  Metered on-ramps help; so would tolling the most congested highways.

Posted by ClarkWD | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Hitting the Sweet Spot:


Would lowering speed limits help or am I misreading the chart? I find it rather confusing.

As I understand it, most people drive between the speed limit and speed-limit+5 mph more than it. So setting the speed limit at 45 during rush hours on urban highways would (in addition to setting off a political riot) might actually increase hourly traffic throughput.

(I'm reminded of Ivan Illich's long-ago argument that the speed limit should be set at 15 mph everywhere -- to eliminate any reason for owning something with more horsepower than a bicycle.)

Posted by: Alan Durning | Apr 10, 2006 3:41:54 PM

Alan -- Confusing, yeah. I just like the shape.

As I understand it, each dot represents one observation (over the course of an hour, perhaps) of avg. mph and vehicles per lane.

I don't think that reducing the speed limit would help. As I read the graph, people travel at >60 mph mostly when the traffic is light, or at least light enough. When traffic gets a bit heavier, people slow down (regardless of what the legal speed limit is) but throughput actually increases. When traffic gets heavier still, speeds drop farther -- the number of cars on any mile of pavement goes up, but the number of cars moving past an arbitrary point on the road over any given hour goes down.

So a lower speed limit might help smooth out the traffic by keeping people from slamming on the brakes, which slows down traffic behind the stopping car. But it doesn't have much benefit (other than increasing car fuel efficiency) when traffic is light.

Posted by: Clark Williams-Derry | Apr 10, 2006 4:07:55 PM

The PCRC article that Clark links to also notes that throughput increases with the addition of HOV lanes, as they relieve some capacity pressure on the roadway.

The point is not about max/min speeds (as Clark says) but roadway design and capacities; the 520 graph illustrates this issue, as the LW bridge creates a bottleneck (the scattering of dots and the notation that max throughput doesn't occur). The topography around here also precludes some design solutions to alleviate some of this congestion.

Posted by: Dan Staley | Apr 10, 2006 4:30:46 PM

I don't like being stuck in congestion any more than someone else, but I often wonder why we think we can and should solve congestion.

Every time you improve flow, it increases the incentive to drive, which then increases congestion. It seems to always move towards an equilibrium point of filling up to the point of unbearability.

And it also seems that this phenomenon has been a major driver (mind the pun) in the urban renaissance of the recent past, particularly by bringing in greater income diversity than the hollowed-out years of suburban flight. My girlfriend just mentioned to me today that she thinks she'd turned on ayoung co-worker to carsharing, since he recently moved from the suburbs to the urban core. And it has a lot to do with him tiring of being in the car and missing out on nightlife (which includes stumbling home drunk sometimes).

"Solve" congestion, especially for the people with the resources to afford high-end urban real estate, and it seems a lot of damage to the environment could be done as a consequence.

With respect to the flow model itself, I wonder if the speeds are particular to that roadway or more generalizable.

Posted by: Joseph Willemssen | Apr 10, 2006 7:07:07 PM

I think that this chart is trying to tell us that we shouldn't allow people onto the road if the speed drops below ~50 mph because more people actually reduces the number of people it can handle. The metering is good, but perhaps it should be gauged to highway speed, not merely during certain hours: the slower the road is going, the fewer cars are permitted on the highway.

You have to preserve the ability to get advantage by individual action: HOV should always be able to bypass the meter. Sitting at the meter because the highway is running too slowly, rather than at a particular time, might very well inspire people to think of alternatives. After all, many people travel during a time for which there isn't much option except waiting for when the time period ends. (Waiting is just the price to pay, either way.) However, if it's because the freeway is too slow, it becomes a teachable moment.

Also: hybrids shouldn't be permitted to access this because they are not actually improving throughput.

Posted by: Brian Sayrs | Apr 11, 2006 10:10:57 AM