A little smoothing would help here...

HOV, or “carpool” lanes, are, like everything, the subject of much controversy in the world of traffic. Do they make congestion worse for everyone while only aiding a few? Do they cause more crashes? Needless to say, I could have written a book only on the complexities of the Diamond Lanes (though I’m not sure who would want to read it).

A new paper, “The Smoothing Effect on Freeway Bottlenecks: Experimental Verification and Theoretical Implications,” from Michael J. Cassidy, Kitae Jung, and Carlos F. Daganzo at the University of California-Berkeley, presented at last week’s TRB (I’ll be data-mining the huge trove of research from that conference over the coming weeks), suggests that HOV lanes can provide overall benefits to highway traffic flow — even when the lane itself is underutilized. As with many things about traffic flow, this is beyond the grasp of the average driver, who may simply look over at the HOV lane, see that fewer cars are in it than his own (even if, of course, they’re carrying more people), and begin to grouse about misguided government policy.

The authors looked at a particular stretch of California freeway with a regularly occurring bottleneck, essentially a “merge bottleneck” resulting from increased volumes of entering traffic. A carpool lane becomes active during the morning and evening rushes; and one might be led to think the activation of the lane somehow causes the bottleneck. But the authors note, “previous analysis established that the carpool lane did not contribute to the bottleneck formation and capacity drop. Instead, and as is typical of merge bottlenecks without carpool lanes, the queue first formed in the shoulder lane and then spread to all lanes.” But even as the capacity of the carpool lane began to drop, the researchers observed an interesting pattern: The “discharge” of vehicles from the bottleneck in other lanes actually began to increase.

They looked at video footage, taken from a pedestrian overpass, to figure out what was going on. Interestingly, it was all about changing lanes. In lane 2, as pictured above (the one next to the HOV), drivers made fewer changes in and out of it when the carpool lane was activated (some of those drivers may have previously been jumping back and forth between lane 1, the ‘fast lane,’ and 2).

And here’s where it gets really interesting:

The same phenomenon was observed a few minutes earlier in lane 3, as drivers started
anticipating the impending carpool restriction. The video data reveal that: (i) drivers’ tendencies to avoid the median (carpool) lane as its activation time approached created crowded conditions in adjacent lane 2, starting at about 14:52 hrs; and (ii) although this crowding temporarily induced some drivers to migrate to lane 3, its more significant impact was to dampen the entries made into lane 2 from lane 3; (see Cassidy, et al, 2008). The net result: lane-changing maneuvers between lanes 2 and 3 diminished at 14:52 hrs, as revealed by the boldfaced oblique cumulative curve in Fig. 5. As in the case of lane 2, this reduction in lane changing was accompanied by a sudden and sustained increase in lane 3’s discharge flow.

The authors observed what they called a “smoothing effect.” Drivers were less tempted to change lanes, because there were fewer options available, and the “discharge flows” actually increased. When the carpool lane wasn’t activated, lanes saw anywhere from 9% to 13% worse performance in VPH (vehicles per hour). When considered in terms of “people hours traveled,” the activation of the carpool lane provided a benefit on the order of 30%. The authors note that if this “smoothing effect” is not observed and quantified, long highway delays might be incorrectly attributed to the carpool lane.

What’s interesting about this is that for many of those individual drivers, they were presumably changing lanes to try and improve their own position. But in doing so they were actually reducing the overall performance of the highway.

As the authors note, it’s worth investigating how signage and striping might reduce “disruptive” lane changing. “Disruptive lane changing,” they add, “might also be reduced in some cases by sorting drivers (and vehicle classes) across lanes according to their preferred travel speeds; or in other cases by inducing a more even distribution of flows across lanes.”) We already have seen “variable speed limits” to help smooth out flow in a linear sense; maybe someday we’ll have “variable lane assignment” to smooth it out across the highway.

This entry was posted on Monday, January 19th, 2009 at 11:24 am and is filed under Cars, Congestion, Traffic Engineering, Traffic safety, Traffic Signs, Traffic Wonkery, Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.