"Every highway intersection is obsolete," thundered Norman Bel Geddes—the designer and showman perhaps most noted for the Futurama exhibit at the 1939 World's Fair—in his 1940 tract Magic Motorways."The intersection is the chief stumbling block for highway designers and the chief headache for the traffic police," he noted. "Why should the crossroads most heavily traveled today be the ones that are least adapted to the safe flow of the vehicles that use them?"
The question resonates today. In 2007, for example, the Federal Highway Administration reported there were 2.4 million crashes at intersections, representing 40 percent of all crashes, and one-fifth of all fatal crashes. Most intersection crashes fall under the category of "crossing paths," and the most common path-crossing crashes, according to federal statistics, involve left turns.
Left turns are the bane of traffic engineers. Their idea of utopia runs clockwise. (UPS' routing software famously has drivers turn right whenever possible, to save money and time.) The left-turning vehicle presents not only the aforementioned safety hazard, but a coagulation in the smooth flow of traffic. It's either a car stopped in an active traffic lane, waiting to turn; or, even worse, it's cars in a dedicated left-turn lane that, when traffic is heavy enough, requires its own "dedicated signal phase," lengthening the delay for through traffic as well as cross traffic. And when traffic volumes really increase, as in the junction of two suburban arterials, multiple left-turn lanes are required, costing even more in space and money.
And, increasingly, because of shifting demographics and "lollipop" development patterns, suburban arterials are where the action is: They represent, according to one report, less than 10 percent of the nation's road mileage, but account for 48 percent of its vehicle-miles traveled.
So what, perBel Geddes, can be done? What can you do when you've tinkered all you can with the traffic signals, added as many left-turn lanes as you can, rerouted as much traffic as you can, in areas that have already been built to a sprawling standard? Welcome to the world of the "unconventional intersection," where left turns are engineered out of existence. This is not necessarily a new idea: The "Jersey Jughandle" and "Michigan Left" were early iterations of this concept; rolled out widely in the 1960s, both essentially require drivers to first make a right turn, then either looping back or U-turning their way onto the road onto which they had wanted to turn left.
One brute-force response is to simply build over an intersection. This is what happened at one intersection I know, the corner of Summerlin Road and Gladiolus Drive, in Ft. Myers, Fla. For years, it was a relatively innocuous four-way intersection with a single left-turn lane. My family and I used to stop every year at a small produce stand, Nancy's, on the corner. But as the population of Lee County grew and traffic volumes increased, the intersection became increasingly congested. You were no longer assured of making it through on the first green "arrow"—and perhaps not even the second. And then, one year, as I approached the intersection, I saw a huge concrete loop—reminding me of the monolithic sandworm on the old paperback cover of Frank Herbert's Dunerising out of the ground. There was no longer a left-turn signal—you simply whisked up the ramp, up and over the opposing traffic. There was just one drawback: It was absolutely hideous, like some looming relic of a phantom elevated superhighway. It was totally out of scale, even to the huge arterial roads that surrounded it. Nancy's was a dim memory, the frontage it occupied presumably condemned for right of way.
"Grade separation" is the most extreme way to eliminate traffic conflicts. But it's not only aesthetically unappealing in many environments, it's expensive. There is, however, a cheaper, less disruptive approach, one that promises its own safety and efficiency gains, that has become recently popular in the United States: the diverging diamond interchange. There's just one catch: You briefly have to drive the wrong way. But more on that in a bit.
(Get more stories like this one delivered to your inbox. Please sign up here for Slate's daily newsletter.)
The "DDI" is the brainchild of Gilbert Chlewicki, who first theorized what he called the "criss-cross interchange" as an engineering student at the University of Maryland in 2000. (He eventually changed the name for fear of potential confusion with the singer of "Sailing.") Inspired by a similar (and at the time, exceedingly unusual) design in Versailles, France, at the intersection of the Autoroute de Normandie and Boulevard de Jardy *, Chlewicki introduced his concept in a soberly titled paper, "New Interchange and Intersection Designs: The Synchronized Split-Phasing Intersection and the Diverging Diamond Interchange" (PDF) at an engineering conference in 2003.
The DDI is the sort of thing that is easier to visualize than describe (this simulation may help), but here, roughly, is how a DDI built under a highway overpass works: As the eastbound driver approaches the highway interchange (whose lanes run north-south), traffic lanes "criss cross" at a traffic signal. The driver will now find himself on the "left" side of the road, where he can either make an unimpeded left turn onto the highway ramp, or cross over again to the right once he has gone under the highway overpass.
Perhaps counter-intuitively, this complicated approach is actually safer—and more efficient. What makes the DDI work is that it reduces the number of "conflict points" where traffic streams cross each other. There would usually be 26 such points in an intersection like this, but the DDI has only 14 (because, for example, drivers turning onto ramps no longer have to turn across oncoming traffic). But, as Chlewicki explained to me, not having those left-turn movements adds another advantage. In a standard "diamond" interchange, where traffic entering the highway has to turn across traffic, the two sets of traffic signals, because they have to account for the left-turn phase, are difficult to synchronize—which means cars wait in longer queues. But with the DDI, Chlewicki told me, "each signal in the interchange is only two phases, not three. And each of these two phases have some unique characteristics. The left turn from either ramp gets the same green phase as the arterial thru movement that does not conflict with that turn. It's as if the design doesn't need a separate ramp phase since it is built into the design."
The DDI didn't just go from Chlewecki's presentation into asphalt. The concept was extensively tested on driving simulators by the FHWA's Human Centered Systems team, who worried that "human errors due to unfamiliarity" might lead to more crashes (one suggested design element was a special "Keep Left" arrow, rather than the standard R4-8 "Keep Left" sign, which is more commonly used to signify "Keep Right").
But drivers by and large got it, and in 2009, the nation's first DDI went up in Springfield, Missouri. There are now any number of DDIs built or in the works, in states from Tennessee to Minnesota to Utah. Chlwecki admits they aren't always appropriate; they don't work well when traffic is equally heavy from opposing directions. Some truckers have pointed out that the design doesn't allow one to exit the highway and then quickly re-enter by proceeding straight across the intersecting street to the next on-ramp (if one has to check calls, or has gotten off at the wrong exit). And aesthetically, like many large-scale engineering works, they are perhaps best appreciated from above. While the intersections are avowedly built with access for pedestrians and cyclists in mind, as this rather involved walk-through video of a DDI reveals, it doesn't really feel like a human-scaled environment.
In a terrain that has been largely ruled by designs like the cloverleaf, the conventional diamond interchange, or, most recently, the early 1970s vintage single-point urban interchange—which was the only one to tackle the left-turn problem (though not as effectively, in cost or reduced conflict points, as the DDI)—there is presumably no end to the amount of intersection innovation still to come. One new solution, the "Continuous Flow Intersection," is almost as common as the DDI itself. Chlewecki notes nearly three dozen alternatives to the traditional four-way junction, ranging from the "echelon" to the "windmill interchange" to the modern roundabout. (Like the DDI, the roundabout is often initially the subject of some vitriol from drivers; it could theoretically be used in place of a DDI, but doesn't seem to perform as well for large interchanges).
But one wonders whether these new alternatives will eventually run up against the same problems faced by intersections with traditional left-turn lanes. In one Missouri study, a model that forecast traffic volumes at a DDI interchange in the year 2035 found that "two of the DDI models had slightly lower corridor travel times than those of the typical diamond interchange." But here too is the limitation in the "predict and provide" mindset of traffic modeling: It presumes that traffic volumes will be much greater in 2035 because we will keep building houses with three-car garages, towns without sidewalks, eight-lane suburban arterials, and things like CFIs and DDIs to try to keep pace with traffic, rather than enabling and effecting more efficient pairings of land-use and transportation. While a fascinating engineering solution in its own right, the DDI is haunted by a lingering question: Can you ever truly design your way out of congestion?
Correction s, Aug. 1, 2011: This article originally misidentified the Boulevard de Jardy. ( Return to corrected sentence.)A photo caption on this article originally placed the Dorsett Road interchange in Springfield, Mo., and stated that it was the nation's first diverging diamond interchange. The Dorsett Road interchange is near St. Louis, and it is not the first.