“I never once imagined in my wildest imagination when we built it that we were going to have the number of fender bender noninjury crashes increase,” says Washtenaw County Road Commission engineer Mark McCulloch about the two-lane roundabout at State and Ellsworth.

Before, it could take twenty minutes to get through the intersection at rush hour. After the roundabout opened in 2013, the time dropped to about five minutes. McCulloch notes that means “the amount of the idling emissions went way down.”

But with a quarter of the old through time and the accompanying reduction in pollution the reconfiguration brought more than four times as many collisions. The intersection had 149 crashes in the five years before the roundabout went in–and 650 in the five years after.

While aggravating and expensive, the accidents were relatively harmless: they resulted in only four minor injuries and no serious injuries or deaths.

That’s because roundabouts have no right angles. “When you have a high-speed intersection where there’s a traffic light, someone’s in a hurry and they blow through it at fifty-five, sixty miles an hour and they T-bone [another car] we’re talking body bags,” says McCulloch. “State and Ellsworth, even with all its faults with the fender bender, noninjury crashes that we have, is still a safe intersection and it is handling traffic very well.”

But seven years on, he admits that they didn’t really understand what they were getting into. No one did.

“We as engineers thought we are getting such great results with single lane roundabouts when it came to crash statistics and capacity analysis that we just thought automatically it was going to translate over into the multilane roundabouts,” McCulloch says. But at State and Ellsworth they found out that “that’s just not the case.

“None of us, me included or the design engineers, had any idea [that] people were going to have complications with it,” McCulloch says. Yet in a 2016 survey of 4,300 local motorists, “over and over and over again people said it’s not the single-lanes, it’s the State and Ellsworth they hate.”

When a roundabout opens, crash numbers usually drop. “A modern roundabout provides a 39 percent reduction in total crashes and a 90 percent reduction in serious injury and fatality crashes,” according to the Insurance Institute for Highway Safety. And a Michigan Department of Transportation study affirms that’s true for all intersections. “Before-and-after studies identified reductions in injury crashes for all classes of roundabout conversions.”

Local single-lane roundabouts, like the one at Nixon and Huron Parkway, have single-digit annual crash numbers, no injuries or deaths, and greater efficiency. But local two-lane roundabouts are different. Take the pair on Maple north and south of M-14, built in 2007. North of the highway, crashes fell about a third, from thirty in the three years starting in 2004 (the earliest year for which the road commission has data) to twenty in the following three. There was just one minor injury before, and none after.

South of M-14, however, crashes more than doubled in the same periods, from eleven to twenty-six. In recent years, there’s also been one minor and one serious injury.

There were no serious injuries at State and Ellsworth either before or after the change. However, crash reports show all four minor injuries since 2013 had everything to do with the roundabout.

McCulloch says the turbulence caused by the two powerful crossing streams of State and Ellsworth is a contributing factor, but total volume is crucial. With about 66,000 vehicles daily, the roundabout has more than six times the traffic of Maple and M-14’s 10,500–and ten times the number of crashes as the south roundabout there.

That huge volume of traffic is the reason State and Ellsworth has two lanes. “A single lane would not accommodate the traffic to the point of providing efficiency,” McCulloch explains. “It wouldn’t have been worth our time to spend tax dollars to improve the intersection.”

Though they didn’t expect it, McCulloch says “what we’re just finding out is [that] when you add that second lane in there, it makes it not just literally [more] difficult but potentially exponentially more difficult for some drivers.”

There are ways to mitigate the problem, writes Pittsfield Township police chief Matt Harshberger. “The 173 crashes in 2014 really caused us to work with the county road commission to improve driver awareness and education. [We put in] some high-visibility signage changes at the intersection to help instruct motorists [in] the proper methods of entering the roundabout.

“The work paid off in that crashes were reduced,” the chief says. “The most important factor continues to be that serious injury and fatal crashes in this roundabout, as well as roundabouts in general, are pretty much eliminated.”

Beyond two lanes, McCulloch says other design features contribute to the number of crashes: “the size of the center island, the location of the center island, and the entry deflection angle of the car entering the roundabout in relation to the circulating traffic. One degree in deflection can make significant differences in outcomes when it comes to capacity or traffic incidents happening out there. If we were to change the entry deflection by a couple degrees would those crashes go down? You don’t know.”

And you don’t get to find out. “That’s an experiment with a big price tag,” McCulloch says, “and because people aren’t getting hurt [or] dying, it’s hard to justify making that type of an experiment when there’s so many other needs on our roadways.

“I’m optimistic in time with education and people becoming more familiar with roundabouts as more and more get built that that number can decrease. But [will it ever] get to fifty? I doubt it.”

It’s a national and not just a state or local problem. In fact, the infamous intersection has launched a federal case.

“We are doing a pooled fund study that involves 6 states and one city to study the traffic crash problems at 2 x 2 (multi-lane) roundabouts,” emails Wei Zhang of the Federal Highway Administration. It was “initiated after Mark McCulloch mentioned the 2-lane roundabout at the intersection of State-Ellsworth.”

This is the first time the agency has studied the factors contributing to multilane roundabout crashes. Zhang points out that “over 3,300 modern roundabouts have been constructed in the U.S. since the 1990s, and most of them have exhibited good to excellent operational and safety performances. However, an issue is slowly but steadily emerging with some multi-lane roundabouts that are experiencing much higher than expected crash rates.” The feds consider any two-lane roundabout with over fifty crashes annually as high–and State and Ellsworth has 130.

The study will investigate crashes “associated with drivers failing to yield properly at exit–vehicles entering the roundabout from the outside lane collide with vehicles exiting the roundabout from the inside lane.” While drivers can enter the roundabout in two lanes, they can safely exit only from the outer one. If drivers can’t get over in time, they sometimes try to exit from the inside lane–and sideswipe cars that aren’t.

In Zhang’s view, “Urgent action is needed to determine the root cause of such crashes so that proper solutions can be developed to mitigate the problem before it becomes a widespread issue that may jeopardize the implementation of roundabouts around the country.” Possible remedies include changes in signing, striping, and geometric layout.

The federal study will take three to five years, with the results shared with traffic and safety engineers throughout the country. Then, if there is funding for it, a second project could start implementing proposed countermeasures at select multilane roundabout locations. It won’t include State and Ellsworth, but it’s some consolation to know that our pain could be the country’s gain.