An FHWA study reveals the safety benefits of restricted crossing U-turn intersections and proposes a model for estimating their potential for reducing crashes.

Today’s traffic volumes and travel demands can lead to safety problems that are too complex for conventional junction designs to handle properly. Take, for example, traditional two-way STOP-controlled intersections on rural multilane divided expressways. Generally, these intersections feature two lanes of expressway traffic in each direction, with a minor road intersecting them at STOP signs.

These intersections are among the deadliest types of roadway junctions in the United States. According to the National Highway Traffic Safety Administration’s crash database, the Fatality Analysis Reporting System (FARS), more than 70 percent of crashes at these intersections involve fatalities and injuries. What’s more, nearly 80 percent of these types of crashes occur in the lane farthest from the STOP sign; that is, when a driver on the minor road, attempting to make a left turn, crosses over the first pair of lanes to the median and then is struck by a vehicle traveling in the outside lane of the second pair of lanes. Engineers refer to this type of incident as a far-side angle crash.

The causes of such crashes could be a combination of factors. For starters, the driver entering the expressway (or major road) from the minor road might misjudge the gaps between vehicles on the major road. The driver on the minor road also might expect that the drivers on the major road will slow down upon seeing the new vehicle entering the intersection from the minor road. A third factor could be that drivers on the major road expect that the driver entering the expressway will yield the right-of-way upon noting the high speed of traffic approaching the intersection on the major road.

Regardless of the cause, an effective solution to minimize far-side angle crashes is the restricted crossing U-turn (RCUT) intersection. With the RCUT design, all traffic on the minor road must turn right at the main intersection. To make the movements for a left-turn or to cross the major highway, drivers would first turn right on the main road and then make a U-turn some distance downstream where designated. With this design, the conflicts that can lead to far-side angle crashes are eliminated. Field experiences show that properly designed RCUTs can reduce fatal and injury crashes by 70 percent to 80 percent. (See also Field Evaluation of a Restricted Crossing U-Turn Intersection, publication number: FHWA-HRT-11-067.)

The RCUT design, also known as the J-turn intersection or superstreet, originated in Alabama, but most RCUTs are in Maryland and North Carolina. Recently, Minnesota, Missouri, Tennessee, and Wisconsin have started to implement them strategically at high-crash locations. To most other States, though, the concept of the RCUT design is new.

Researchers at the Federal Highway Administration (FHWA) are in the process of developing guidance at the national level for designing RCUT intersections. Until recently, computer models did not exist for predicting the safety benefits associated with this intersection design. To fill that gap, researchers at FHWA’s Turner-Fairbank Highway Research Center in McLean, VA, recently completed a study to develop crash prediction models derived from data associated with 35 rural RCUTs in Maryland, Minnesota, Missouri, and North Carolina.

Determining the Median U-Turn Offset

“A major challenge with implementing innovative concepts like the RCUT is determining appropriate geometry when design guidance is limited,” says Will Stein, P.E., a safety engineer with FHWA’s Minnesota Division, “particularly, U-turn spacing and whether acceleration lanes are needed.”

In fact, for the RCUT design, the median U-turn offset is perhaps the most important design parameter affecting the operation and safety of the intersection at rural high-speed locations. Defined as the distance between the main intersection and the U-turn opening, the median U-turn offset has critical implications for the safety performance of the RCUT design. Engineers determine the offset based on traffic demands and the ability of drivers entering from the minor road to execute multiple tasks (acceleration, lane change, and deceleration into the median U-turn) within a limited time and space.

Consequently, the offset parameter must be reflected in any crash prediction model for RCUTs. Engineers typically follow one of two philosophies toward designing rural RCUTs:

1. Angle the minor road’s right-turn lane in a way that enables the right-turning traffic to enter the inner lane of the expressway after entering into an acceptable gap in traffic at the intersection; or
    
2. Provide an acceleration lane for the right-turning traffic entering from the minor road, so drivers on the minor road can increase their speeds in that acceleration lane before merging into traffic on the expressway.

In a relative sense, the median U-turn offset required by the former design approach (number 1) is shorter. Approach number 1 may operate safely under lower traffic demands.

To develop crash prediction models, the researchers compiled data on the geometric layouts, average annual daily traffic (AADT), speed limits, and crash records related to 35 RCUT intersections--12 in Maryland, 5 in Minnesota, 5 in Missouri, and 13 in North Carolina. For each RCUT, the researchers calculated the U-turn offsets by consulting Google Maps™ for the geometric layouts and cross-checking their findings with construction plans (if available). Crash and AADT data came from the respective State departments of transportation (DOTs). To determine the speed limits at each location, the researchers consulted the street views in Google Maps™. Then they compiled the dataset and associated statistics into tables for comparison.

Next, the researchers used the crash rates and the median U-turn offsets computed for and measured from each RCUT site to develop the crash prediction statistical models. Then they generated a variety of charts from the crash prediction models to help users quickly estimate the crash rates for given RCUT designs. Users planning an RCUT also can employ the charts to help determine the appropriate range of U-turn offsets under given traffic conditions.

“To be able to examine the predicted crash reductions for the various roadway types will be most useful toward determining the most cost-effective measure,” says Jeffrey Wentz, assistant district engineer for traffic with the Maryland State Highway Administration. “Although there are always measures that we have available in our toolbox, having something based on factual data will make it easier to support, both by those who fund such measures as well as by the members of the public who have to use the new turns.”

Crash Prediction Models

To create models that predict the expected number of crashes per year, the researchers developed statistical regression models that used the AADT in both directions on the major and minor roads and a median U-turn offset in units of 1,000 feet (305 meters) as variables.

The researchers developed separate crash prediction models for two types of RCUT designs: RCUTs without right-turn acceleration lanes (based on data from Minnesota and North Carolina) and RCUTs with right-turn acceleration lanes (based on data from Maryland and Missouri). Deceleration lanes are required for U-turn traffic. Prediction models were developed for total crashes and for fatal and injury crashes.

In addition to the crash prediction model equations, the researchers produced crash prediction charts based on those equations to make it easier to work with the models. These charts were developed using the most common traffic combinations--AADT on the major road and AADT on the minor road and the common ranges of median U-turn offsets for the two design approaches.

For RCUTs without right-turn acceleration lanes, the U-turn offset range is typically between 800 and 1,800 feet (244 and 549 meters). For RCUTs with right-turn acceleration lanes, the median U-turn offset is typically between 2,000 and 3,000 feet (610 and 914 meters). Using the equation or the charts, engineers can estimate the crash frequencies of existing RCUTs or the desired U-turn offsets for planned new RCUTs to achieve certain safety results.

Observations and Conclusions

“The 35 rural RCUTs sampled from four States cover a wide enough range of traffic demands to provide insights into the typical range of conditions under which engineers might consider installing an RCUT to achieve the desired level of safety,” says Monique Evans, director of FHWA’s Office of Safety Research and Development (R&D).

The crash prediction charts indicate that when the U-turn offset is less than 1,500 feet (457 meters), the expected crash rates for RCUTs with right-turn acceleration lanes are higher than for the RCUTs without right-turn acceleration lanes. However, when the U-turn offset is greater than 2,000 feet (610 meters), the trend reverses. The above observation might seem counterintuitive but actually makes sense. Take, for example, a rural expressway on a flat grade with a speed limit of 60 miles (97 kilometers) per hour. According to Table 10-5 in the American Association of State Highway and Transportation Officials’ A Policy on Geometric Design of Highways and Streets (commonly known as the Green Book), the required length for a deceleration lane from 60 miles (97 kilometers) per hour to 15 miles (24 kilometers) per hour would be 500 feet (152 meters) for any type of rural RCUT design.

If a right-turn acceleration lane is provided, according to the AASHTO Green Book (Table 10-3), the required length for that acceleration lane would be 1,020 feet (311 meters), assuming an initial speed of 22 miles (35 kilometers) per hour. For an RCUT with a right-turn acceleration lane, assuming it is separated from the major road by grass or concrete, after the driver has traveled about 1,020 feet (311 meters) of the acceleration lane in order to join traffic on the major road, a U-turn offset of 1,500 feet (457 meters) will leave only 480 feet (146 meters) within which the driver can find a gap to change lanes and then decelerate to 15 miles (24 kilometers) per hour at the U-turn opening, which is insufficient in most cases.

However, for an RCUT without a right-turn acceleration lane, for the same U-turn offset of 1,500 feet (457 meters), if a sufficient gap exists for the minor road driver to enter the intersection, the driver can shift to the inner lane at any location during the first 1,000 feet (305 meters), thus leaving sufficient room to perform the deceleration task.

As the U-turn offset increases from 1,500 feet (457 meters) to 2,000 feet (610 meters) and beyond, the RCUT design with a right-turn acceleration lane will offer critically needed distance and space for drivers making the U-turn to find suitable gaps and change lanes while driving at high speeds, which will be reflected in their safety outcomes. However, for the RCUT design without a right-turn acceleration lane, once the driver finds a gap to enter the main intersection, he or she will most likely complete the lane-changing task within the first 1,000 feet (305 meters) on the major road. Thus, increasing the U-turn offset beyond 1,500 feet (457 meters) is not nearly as critical. When the traffic volume on the major road increases to the level that diminishes gaps available for minor road drivers to even make the right turn from a complete stop, then if the RCUT design is still desired, a right-turn acceleration lane must be provided to ensure capacity and safety.

With regard to fatal and injury crashes, the researchers found that for the same AADT combinations and median U-turn offsets, RCUTs with right-turn acceleration lanes lead to fewer fatal and injury crashes than RCUTs without acceleration lanes. This follows from the fact that the right-turn acceleration lane reduces the speed differential between merging vehicles and through traffic, which should reduce the severity of crashes.

“FHWA’s predictive model for superstreets [RCUTs] appears to be very similar to the data that we’ve collected in North Carolina,” says James H. Dunlop, P.E., a congestion management engineer with the North Carolina Department of Transportation. “There aren’t many safety treatments that are as much of a slam dunk as superstreets. Based on [these] data and our own experience, there will be very few new full-movement median openings on rural expressways in North Carolina.”

FHWA’s Stein adds, “As we continue to monitor and evaluate our [RCUT] sites in Minnesota, this work adds to the body of knowledge and can inform future design decisions.”

Wei Zhang is the program manager for intersection safety research and development in FHWA’s Office of Safety R&D. He has worked for FHWA for 10 years. In his current position, he develops products and programs that help reduce crashes, fatalities, and injuries at intersections and interchanges. He holds a Ph.D. in geotechnical engineering from the University of Minnesota.

Nopadon Kronprasert is a postdoctoral fellow with the National Research Council. He received his Ph.D. degree in transportation infrastructure and systems engineering from Virginia Tech. He conducts research on improving the operation and safety of alternative intersection and interchange designs.

For more information, contact Wei Zhang at 202–493–3317 or wei.zhang@dot.gov.