Emily Badger is a former staff writer at CityLab. Her work has previously appeared in Pacific Standard, GOOD, The Christian Science Monitor, and The New York Times. She lives in the Washington, D.C. area.
Dispatches from the imagination of transportation engineers.
Geometry tells us that the traditional four-way intersection is inherently dangerous. When you plot all of the potential points of conflict on a diagram – and transportation engineers actually do this – it turns out that vehicles have 32 distinct opportunities to collide into one another at the nexus of two two-lane roadways. Cars can crash into each other while merging or diverging from a given lane. Then the worst action happens right in the middle of the interchange, at that perilous point where vehicles turn left across oncoming traffic.
There, the geometry gets even more gruesome. Cars colliding head-on, or at a right angle to each other, are much more likely to cause serious carnage than two vehicles merging at a shallower angle (as happens on a right-hand turn). This is what a traditional intersection looks like when you analyze it as a potential death trap:
With that geometry in mind, it becomes clear what we need in the holy grail of intersection design: a scheme that would eliminate left-hand turns while still enabling drivers to move in all four directions. The basic roundabout does this, with only eight relatively less harmful points of conflict:
The challenge is to translate that same idea onto much busier roads. You may already be familiar with the “Michigan Left,” also known as the “median U-turn,” which solves this problem by requiring drivers to pass through an intersection, make a u-turn and then re-approach it from the opposite direction, ultimately making a safer right-hand turn. The concept is also intended to improve traffic flow by eliminating that pesky left-turn cycle at the light:
The simple Michigan Left, though, only begins to tap the wilder dreams of transportation engineers. They’ve concocted some much more intricate intersection designs – and in a few cases actually built them – to improve road safety, save states money and generally stall the inevitable expansion of highways. John Sangster, a doctoral candidate at Virginia Tech and an alternative intersection enthusiast, introduced us to several and pointed us to some spellbinding animations from the Applied Technology and Traffic Analysis Program at the University of Maryland (many thanks to researchers there for sharing these clips).
Meet, for instance, the Jughandle:
And the Superstreet (now gaining steam in North Carolina):
And the Diverging Diamond (now in existence outside Springfield, Missouri):
As you can imagine, these designs are not an easy sell. “It’s a two-fold sale that has to happen,” Sangster tells us. “We’re not going to build these if they’re not safe. We’re also not going to build them if they don’t work better.”
In theory – i.e., in conflict-point diagrams – these intersections should be safer than more traditional ones. But there are two caveats to that promise: Sangster is really talking about safer intersections for cars. Pedestrians and bikers aren’t figured into any of these models, and Sangster has yet to encounter designs that do a good job of incorporating them (or transit). There also isn’t much hard data on the safety of these designs because so few of them have been built (and even accurately modeling them on a computer can be tricky and expensive).
These intersections are also clearly not meant for the inner city. They may work best, Sangster says, when a local road outside of town has grown so congested that it needs to be converted into a highway. Some of these designs could delay that prospect, while allowing communities to forgo the expensive construction of ramps, bridges and underpasses, still facilitating heavier traffic flow.
Traffic engineers generally measure the performance of an intersection in the average delay per vehicle sitting at it. This means that if 50 cars on a major road have to wait 30 seconds each at a red light so that a lone car can pass from a side street, you don’t have a very efficient intersection.
“It’s definitely not trying to give equal time to everyone on the road,” Sangster says. “It’s trying to make the whole road work better." And sometimes that means making life extremely inconvenient for a small number of drivers in order to make the whole system work better for the majority. More often, this also means designing increasingly complex geometry, as in this “Continuous Flow” model:
The Maryland researchers have even modeled some three-dimensional solutions that, for instance, elevate drivers making left turns to keep them out of the main roadway:
None of these models are to be confused with some equally mind-boggling (and pedestrian-unfriendly) simulations of intersections for driverless cars. These "alternative" intersections, conversely, would require actual human drivers to learn radically new – and often counterintuitive – traffic patterns. And we all know how that's gone in America with the simple roundabout.
“The biggest problem is that you can educate locally as many people as you can reach,” Sangster admits. “But there’s always going to be somebody visiting from out of town.”