They Might Need the Lane Someday (and Who Cares How Many People Get Hurt in the Meantime)

What will professional engineers do when an intersection can’t be made safe? What happens when all the studies, observations, and data point to an unresolved and dangerous outcome? Does it become a public health emergency, something that prompts a rethinking of our transportation priorities? Or do professionals accept a recurring pattern of crashes as the price society pays for moving enormous volumes of traffic between a middle school and a Starbucks?

Spoiler alert: This story ends with drivers being blamed for failing to navigate a design that engineers have spent more than a decade trying (and failing) to make understandable, and engineers continuing to deprioritize safety, in spite of recent policy directing them to make it their first priority.

(Note: In the video, KARE 11 gets the traffic volume wrong. It’s not 5,000 vehicles per day. It is closer to 45,000.)

The intersection of Kenwood Trail and 185th Street West sits about a mile east of Interstate 35 in Lakeville, Minnesota. It was converted from a signalized intersection to a full two-lane roundabout in the fall of 2015, with two lanes entering, circulating, and exiting in all four directions.

The original signal struggled with growing traffic volumes, creating backups and a pattern of crashes that local residents remember as less frequent, but more severe. The roundabout was intended to keep traffic moving while replacing high-speed, right-angle conflicts with slower, less severe collisions. Because the design was unusual in Minnesota, researchers from the University of Minnesota began observing it almost immediately after it opened.

Now, amid an increasing number of crashes following more than a decade of study and modification, officials are considering everything from still more signs and markings to removing the roundabout and reinstalling the signals it replaced.

Is that really the only option left?

The Second Lane Problem

Roundabouts achieve safety by slowing traffic and simplifying the decisions drivers must make. A single-lane roundabout eliminates high-speed crossing movements and asks drivers to do something relatively intuitive: slow down, yield to anyone already circulating, enter when there is an opening, and exit at the appropriate point.

The Lakeville roundabout compromises that simplicity in the pursuit of capacity. It has two lanes entering, two lanes circulating, and two lanes exiting in every direction. Drivers must choose the correct lane before entering, understand which movements are permitted from that lane, track the vehicle beside them, and anticipate whether a vehicle in the inner circulating lane will continue around or exit across their path. This is all done with generous, sweeping curves, forcing a driver to perform a lot of situational analysis at what are relatively high speeds.

Researchers documented the consequences almost immediately after the roundabout opened. Over 570 hours of observation, they tracked nearly 800,000 vehicles and identified more than 13,000 failure-to-yield events on the approaches their cameras could reliably monitor. The errors followed a consistent pattern: Drivers were much more likely to pull in front of vehicles traveling in the inner circulating lane.

That pattern makes sense. An entering driver sees a vehicle in the inner lane and assumes it will continue around the circle. The circulating driver may instead be preparing to exit across the entering driver’s path. This is a known, common point of failure; the same misunderstanding appeared at other two-lane roundabouts included in the study.

The researchers then examined the standard engineering responses to this repeated failure: signs, pavement markings, extended solid lines, overhead lane designations, and changes intended to make the proper movements more obvious. Some treatments reduced certain improper turns. None solved the central failure-to-yield problem. The MnDOT report ultimately found no successful intervention that meaningfully corrected that behavior.

That should not be surprising. The problem is not simply that drivers lack information. The multi-lane design requires them to process too much information at once, especially when speeds aren’t dramatically reduced.

A single-lane roundabout reduces conflict by removing choices. A two-lane roundabout restores those choices so the intersection can move more vehicles. The additional lane does not merely add pavement. It adds ambiguity about where drivers are going, who must yield, and whether two vehicles can safely exit at the same time. The safety benefit a roundabout provides comes from simplification, but that benefit is lost when more lanes make the intersection complicated again.

Minnesota engineers understood this tradeoff years ago. The MnDOT-sponsored study noted that several full two-lane roundabouts had been converted to partial one-lane configurations just by changing the striping. Reducing the available lanes reduced the number of two-lane conflict areas and made the intersections safer and easier to navigate.

Tellingly, the report rejected that approach for the Lakeville roundabout, but not because it was unsafe or ineffective. It dismissed it as temporary because engineers anticipated that the additional capacity would likely be needed within 20 years.

In other words, engineers identified a simpler, safer configuration, then rejected it because, within the next couple of decades, they might want those extra lanes.

The Capacity Trap

That statement reveals the hierarchy embedded in a traffic safety analysis. Future traffic growth is treated as a fixed requirement. Capacity must be preserved for traffic that may not arrive for years or even decades, regardless of the current harm. Safety is then pursued through signs, markings, education, and repeated modifications, all constrained by the prior decision that the projected traffic volume must be accommodated.

A safety-first approach would reverse that hierarchy. It would use the simplest configuration that works, accepting less theoretical capacity in exchange for fewer conflicts and more intuitive driver behavior. If congestion later became a genuine problem, officials could measure it under actual conditions and decide whether restoring the additional lane was worth the added risk. In the meantime, engineers would give up some capacity today — capacity that is not currently needed but might be at some point in the future — to obtain greater safety in the present.

The professional response has been to treat that tradeoff as unacceptable.

I wrote about this in Confessions of a Recovering Engineer. The standard traffic-engineering process does not begin by asking what level of safety the public wants or should expect. It begins by selecting a design speed and then determining how much present and future traffic the facility must accommodate. Only after those decisions are made does the engineer consult the design manuals to assemble what the profession defines as a “safe” design. The hierarchy is speed, volume, safety, and then cost. Safety is considered, but only within the constraints established by the first two priorities.

That is the hierarchy on display in Lakeville. The engineers have decided that the intersection must retain enough capacity for traffic volumes projected decades into the future. With that requirement fixed, they are left trying to make a complicated two-lane roundabout safe through signs, markings, and driver instructions. The public is invited to debate the remedies, but not the value judgment that determined the range of remedies available.

This is unacceptable. More to the point: this is a policy decision, not an engineering decision. Engineers should not be able to overrule safety because they prioritize future traffic capacity more.

The Policy Directive is Clear

The Lakeville roundabout was built in 2015. MnDOT did not adopt its Safe System Approach Implementation Plan until 2025. The engineers who designed the intersection were working within the conventional traffic-engineering hierarchy, but the officials deciding what to do now are not.

MnDOT’s new policy is not ambiguous. It describes the Safe System Approach as a paradigm shift in which roadway agencies prioritize safety over operational performance. The purpose of the transportation system, the plan states, is not to maximize vehicle throughput or minimize delay. It is to provide safe mobility for everyone who uses it. 

The plan also acknowledges that humans make mistakes and directs professionals to design, operate, and maintain the transportation system in a way that accommodates those mistakes. It prioritizes changes to the built environment over strategies that require people to exercise greater care. Its roadway design hierarchy begins with removing severe conflicts, followed by managing speeds and separating conflicts in time. Increasing driver attentiveness through signs, markings, warnings, and education comes last.

This inverts the approach Lakeville has spent a decade applying to this site. Officials have changed signs, adjusted striping, clarified lane assignments, and tried to make drivers more attentive to a maneuver that thousands of drivers predictably misunderstand. The Safe System Approach says the professionals should stop demanding better performance from drivers and change the physical environment that produces the mistake. Amen.

If we follow the safety policy, the next step is clearly evident: reduce the roundabout to one circulating lane. Leave the excess pavement in place, stripe it out, simplify the approaches, and remove the lane-choice problem that has generated this repeated pattern of conflict.

There will be a loss of capacity. There may be longer queues at busy times. Some drivers may experience additional delay. Those are not reasons to reject the safer configuration. MnDOT’s plan explicitly says to prioritize safety improvements over increasing capacity, and to prioritize safer speeds and the separation of conflicts over operational efficiency and level of service.

The policy decision has already been made. Safety comes first. When safety and traffic performance conflict, traffic performance is supposed to yield.

Make the roundabout one lane. Make it simple. Make it safe.

Future traffic is a forecast. The crashes are happening now.

[[divider]]

Where else have you seen traffic capacity prioritized over a simpler, safer design? Share your thoughts with Chuck and other Strong Towns members in the Commons, where we'll be discussing this article!