U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590

Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

This report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-11-067    Date:  June 2012
Publication Number: FHWA-HRT-11-067
Date: June 2012


Field Evaluation of A Restricted Crossing U-Turn Intersection



Data were collected by scoring video collected at a Maryland RCUT intersection and a nearby conventional intersection on the same highway, U.S. 15, in Frederick County. Recordings were made at the RCUT on two weekdays. Six digital cameras, three on each of two masts, were used to record operations. On one day, operations on the southbound side of the main intersection were observed with attention focused on the right-turn movement from the minor road. Seven days later, recordings were made on the northbound side, with attention focused on U-turn movements at the southern end of the RCUT.


Traffic conflicts, acceleration lane use, weaving maneuvers, merge lags, and travel times were extracted from the digital video recordings.


In safety analyses, traffic conflicts are often used as surrogates for crash data. Conflicts are recorded when a crash is avoided as a result of evasive maneuvers by one or more vehicles. Conflict severity is usually assessed by minimum time to collision during the conflict event. [4] Time to collision was not used in the analysis of conflicts in this study for the following reasons: (1) potential conflicts could occur anywhere along the more than 2,000 ft of roadway under observation, (2) video was captured from only two observation areas at which the cameras were only 30 ft above the roadway, and (3) the maximum frame rate for the video was 16 frames per second. These factors made the calculation of time to collision infeasible. Instead, conflict severity was judged subjectively. Lag was used as a quantifiable substitute for time to collision, as was the duration of braking to avoid collision.

Acceleration Lane Use

The Maryland State Highway Administration has considered eliminating acceleration lanes from the U-turns of future RCUT designs because of the observation that passenger car drivers do not use these lanes.[3] To quantify acceleration lane use, this study classified each merge at both the right- and U-turn areas of the RCUT intersection. For right turns, the classifications were as follows:

The classifications for the right-turn merge are shown in figure 4.

Figure 4. Illustration. Merge location classifications for RCUT right turn. This illustration shows the location and extent of the zones used for classifying the merging behavior of drivers who turn right from the minor road and are destined to make a U-turn. The drawing indicates that drivers who cross the first 140 ft of gore line that separates the right turn acceleration lane from the through traffic are classified as crossing the gore. Drivers who enter the travel lanes beyond the first 140 ft of the gore line and before the end of the dotted line are classified as entering the through lanes midway. Drivers who enter the through lanes beyond that point are classified as entering the through lanes at the end of merge.

Figure 4. Illustration. Merge location classifications for RCUT right turn.

Because the U-turn acceleration lane also served as a left-turn deceleration lane, the lane use classifications were slightly different from those used for the right turn merge. There was no taper or defined end to the merge area. Therefore, the deceleration lane classifications were as follows:

The classifications for the U-turn merge are shown in figure 5.

Figure 5. Illustration. Merge location classifications for RCUT movements. This illustration shows the location and extent of the zones used for classifying the merging behavior of drivers who turned right from the minor road as the complete a restricted crossing U-turn (RCUT). The “cross gore” area is short, perhaps three car lengths. The “end of gore” area covers the remainder of the length of the solid gore line, which extended several hundred feet. The “halfway” classification area extends approximately three times the distance of the “end of gore” extent. The entire extent of the “end of gore” area is marked by a dotted line.

Figure 5. Illustration. Merge location classifications for RCUT movements.


Lag is defined as the time between the crossing of any part of the subject vehicle over the rightlane edge line and the arrival at the crossing point of the next vehicle in that lane or the next vehicle that had been in that lane when the edge line was first crossed. If the next vehicle did not change speed, lag and time-to-collision would be the same. However, if the next vehicle decelerated in response to the vehicle ahead entering its lane, then lag would be greater than time-to-collision at the time of the lane entry.


At an RCUT, the entire distance between the right- and U-turn areas is a weaving area for leftturn and through movements from the minor road. Also, the entire distance from the U-turn back to the main intersection is a weaving area for left and right turns. This study did not attempt to characterize all weaving in these areas. Rather, it focused on the extremes of the weaving areas.

At the right turn, the study focused on whether drivers who turned right and were destined to use the U-turn turned directly into the left lane (i.e., traversed the right lane as part of a continuous turning movement) or merged into the right lane and then changed to the left lane after traveling some distance in the right lane parallel to the lane’s direction of travel.

At the U-turn deceleration lane, the study tracked whether vehicles merged into the deceleration lane at the beginning of the taper and whether the vehicles made late entries or crossed directly from the right lane into the deceleration lane.

At the U-turn, the study focused on whether drivers merged into the left lane or crossed the left lane to merge into the right lane in a single continuous maneuver.

Late entries into the deceleration lane for the right turn were also noted.

At the conventional intersection, analogous classifications of right- and left-turning movements were noted. That is, it was noted whether turning movements were completed in the right or left lanes.

Travel Time

From the minor road at the selected RCUT, drivers making left-turn or through movements via a U-turn are required to travel an extra 4,000 ft compared to the same movements at a conventional intersection. However, these turning movements can generally be made without stopping to wait for a gap in traffic, so there may be a reduction in waiting time compared to a stop-controlled conventional intersection that does not have an acceleration lane. Time to complete through and left-turn movements was recorded at both the RCUT and conventional intersections. At both intersections, travel time measurement was initiated when vehicles on the minor road reached a point about 20 ft upstream of the intersection. Measurement began upstream of the intersection so that it would include stop time while vehicles waited for a gap regardless of where the vehicles waited in relation to the stop line or slip lane. For the left-turn movement, measurement continued until vehicles reached a point 466 ft downstream of the main intersection on the northbound side. This downstream location was chosen because at a conventional intersection, vehicles will not have accelerated to highway speed for some distance downstream of the intersection, whereas at an RCUT, left-turning vehicles return to the main intersection travelling at highway speed. Measuring travel time to the main intersection would thus bias travel time in favor of the conventional intersection because it would not take into account the time required for acceleration at the conventional intersection but would include acceleration time for the RCUT.

Through movement travel time was measured in the same way for both conventional and RCUT intersections, with timing started 20 ft upstream of the intersection until the vehicle was on the far side of the minor road and clear of the intersection.

Location Selection

RCUT Intersection

The RCUT selected for observation was in Maryland on U.S. 15, a four-lane divided highway, at the intersection with U.S. 15 Business/Seton Avenue, which is a two-lane rural road. The RCUT is depicted in figure 1. The conversion from a conventional intersection to an RCUT design was completed in 1988. From northern directional U-turn to southern directional U-turn, the intersection covers over 4,500 ft of U.S. 15. The median is 57 ft wide from left edge line to left edge line but narrows to 47 ft to accommodate left-turn deceleration lanes. There are two through lanes in each direction. All lanes are 11 ft wide. Exclusive of acceleration and deceleration lanes, the right shoulder is 11 ft wide, and the left shoulders are about 3 ft wide. From the highway, deceleration lanes are provided for left and right turns from either direction. From the minor road, the southbound acceleration lane extends 550 ft to the beginning of the taper. The distance from the beginning of the acceleration lane taper to the beginning of the taper for the U-turn deceleration lane is 677 ft. Including the taper, the deceleration lane is 760 ft long. The radius of the directional U-turn is 27.5 ft. The radius of the right-turn slip lane from the minor road onto U.S. 15 south is 93 ft.

The minor road junction is yield-controlled for both the right-turn movement from the minor road and the left-turn movement from the highway. The U-turn movements are also both yield-controlled.

The RCUT does not require yield control on the minor road, nor does it require acceleration lanes for vehicles turning right from the minor road onto the highway. In fact, some RCUT intersections on U.S. 301 included in the crash analysis are stop-controlled and lack acceleration lanes.

Conventional Intersection

Observations were made at the intersection of North Franklin Road and U.S. 15, 5 mi south of the Seton Avenue and U.S. 15 RCUT. This intersection is shown in figure 6.

Figure 6. Photo. The conventional intersection observed in this study. This photo shows a satellite view of the conventional intersection observed in this study. Turning movements were observed so that conflicts between vehicles at the restricted crossing U-turn could be compared to conflicts at a nearby conventional intersection. The four-lane divided highway is crossed at a right angle by a two-lane rural road. There is no acceleration lane for vehicles turning right from the minor road. There are deceleration lanes for vehicles turning left off of the highway from either direction. There is a wide opening in the median to accommodate all left-turn and through movements on the minor road.

Source: Google®, U.S. Geological Survey, Data SIO, NOAA, U.S. Navy, NGA, GEBCO

Figure 6. Photo. The conventional intersection observed in this study.

On the east side of U.S. 15, the minor road name becomes Roddy Creek Road. The intersection is typical of conventional minor road intersections on U.S. 15; there are no acceleration lanes for either right- or left-turn movements from North Franklin. There is a 472-ft-long by 9-ft-wide deceleration lane for the right turn from southbound U.S. 15 onto North Franklin and another deceleration lane for the left turn from southbound U.S. 15 onto Roddy Creek. There is a 290-ft acceleration lane for right turns from Roddy Creek to northbound U.S. 15. There is a 490-ft leftturn deceleration lane from northbound U.S. 15 to North Franklin. The median opening between northbound and southbound lanes of U.S. 15 is 80 ft, and the refuge area in that opening is 40 ft wide. Exclusive of acceleration and deceleration lanes, the median is 40 ft wide north of the intersection and 30 ft wide south of the intersection. The intersection has two-way stop control with no control in the median and stop controls on the minor road.


The RCUT intersections selected for the crash analyses are listed in table 1. The table also shows the log mile location of the intersection, the date the RCUT conversion was completed, and the nature of the U-turn crossings that were provided for left-turn and through movements from the minor road. U-turn locations are labeled as dedicated directional U-turns (DDUT) if they were channelized to permit U-turns originating from the direction of the main intersection. If the U-turns were made at a conventional intersection at the deployment date, then the U-turn location is labeled "Inter." If through or left-turn movements use another RCUT intersection to make the U-turn, then the "RCUT" label was used.

Before-and-after comparisons of traffic crashes were made for each RCUT intersection, the sections between the RCUT intersection and the U-turn locations, and the U-turn locations. This approach is intended to capture the total impact of the RCUT treatment on crash probability.

Table 1. Maryland RCUT intersections.

Intersection Log Mile* Deployment Date Approaches Southern
U-Turn Location
U-Turn Location
U.S. 15 at Hayward Road 16.180 9/1988 4** DDUT at 15.829 Inter at 16.530
U.S. 15 at Willow Road 17.070 11/1992 4 Inter at 16.530 Inter at 18.020
U.S. 15 at Biggs Road 18.020 11/1992 4 RCUT at 17.070 RCUT at 18.330
U.S. 15 at Sundays Lane 18.330 11/1992 4 RCUT at 18.020 RCUT at 18.870
U.S. 15 at College Avenue 34.210 8/1994 4 DDUT at 33.823 DDUT at 34.619
U.S. 15 at U.S. 15 Business 35.020 9/1988 4 DDUT at 34. 619 DDUT at 35.477
U.S. 301 at Main Street 12.380 1/2003 4 U-turn Inter at 12.880
U.S. 301 at Del Rhodes Avenue 12.880 1/2003 4 Inter at 12.380 DDUT at 13.146
U.S. 301 at Galena Road 43.670 1/2002 4 DDUT at 43.360 DDUT at 43.905

*The log miles are those on Maryland State Highway Administration crash records except where offsets were added at county boundaries to adjust for changes in the way log miles were recorded by various agencies.

**This intersection has since been converted from a four-way to a three-way intersection.

Three approaches were used for before-after comparisons: (1) simple before-after comparisons, (2) before-after comparisons adjusted for annual crash rates at conventional intersections on the same corridors, and (3) empirical Bayes (EB) analysis.

The simple before-after comparison requires fewer assumptions than the other approaches but is susceptible to misattribution of causation to changes other than the safety treatment. The inclusion of comparable intersections that do not undergo the treatment can correct for this weakness in the simple before-after comparison to the extent that the comparison intersections are subject to the same non-treatment changes that occur over time. However, when intersections on the same corridor are selected for comparison with the treatment intersection, there is often a reason that the treatment intersections were selected. For instance, the treatment intersections may have been those that experienced the highest crash rates or had the highest traffic volumes. The EB approach adjusts predicted crash rates based on known crash experience of a wider range of similar sites and takes into account the effects of traffic volume on crash rates. However, the EB approach requires more assumptions than the other approaches, requires volume counts that are not always available, and uses safety performance functions (SPFs) specific to the study site geometry. SPFs are not yet available for U-turn crossings.

Use of three approaches to the crash analysis was intended to provide converging evidence regarding RCUT safety performance and to obtain the benefit from the advantages of each approach.


Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101