U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

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

SUMMARY REPORT
This summary report is an archived publication and may contain dated technical, contact, and link information
Publication Number: FHWA-HRT-08-063
Date: September 2008

Two Low-Cost Safety Concepts for Two-Way Stop-Controlled, Rural Intersections on High-Speed Two-Lane, Two-Way Roadways

FHWA Contact: Joe Bared,
HRDS-05, (202) 493-3314, joe.bared@fhwa.dot.gov

PDF Version (1.75 MB)

PDF files can be viewed with the Acrobat® Reader®

This figure provides an example of an implementation of concept 1. The photo was taken from the major approach at an intersection of 2 two-lane, two-way roadways with stop-control on the minor approach. There are paved shoulders with a white edge line and yellow center line. Concept 1 includes the introduction of a painted yellow median island on the major road approaches. The painted median begins as a double yellow centerline and gradually widens to a width of approximately 1.22 m (4 ft); the full median width is then continued to the intersection. Yellow cross-hatching is installed in the painted median. Rumble strips are also introduced on the outside shoulders and in the painted median on the major road approaches.

Figure 1. Concept 1.

This figure provides a schematic of concept 1. A two-lane, two-way major roadway is oriented in the east-west directions. A two-lane, two-way, stop-controlled minor roadway intersects the major road and is oriented in the north-south directions. The illustration shows where a painted median island is installed on the major approaches prior to the intersection. The lane width on the major approaches in narrowed from 3.66 m (12 ft) prior to the median island to 2.75 m (9 ft) at the narrowest point. Rumble strips are depicted with a solid green line along the shoulders of the major road as well as within the painted median. A legend is provided to denote the symbols for the STOP signs and the rumble strips.

Figure 2. Illustration of concept 1.

OVERVIEW

The Federal Highway Administration (FHWA) Office of Safety has identified intersections as one of its safety focus areas. As part of the FHWA efforts to reduce intersection crashes and the related injuries and fatalities, two concepts have been identified: (1) rumble strips on outside shoulders and in a painted yellow median island on major road approaches and (2) channelizing separator islands on side road approaches with supplemental STOP signs. Specifically, the strategies are low-cost countermeasures for intersections at two-lane, twoway roadways with two-way STOP-control (TWSC). The lane narrowing concept (concept 1) features the introduction of rumble strips on the outside shoulders and in a painted yellow median island on the major road approaches (figures 1 and 2).The objective of this first concept is to induce drivers on major roads to reduce approach speeds at intersections by effectively reducing the lane width. The minor road splitter island concept (concept 2) features channelizing separator islands on the side road approaches on which supplemental STOP signs are installed (figures 3 and 4). The objective of the second concept is to provide redundancy of the STOP sign and increase driver-compliance with the STOP sign. A third concept includes the combination of concepts 1 and 2 (figure 5). The concepts have greater potential for effectiveness on intersections of high-speed roadways. However, they can also be applied to intersections with lower posted speed limits. With the cooperation of several transportation agencies, these two strategies were deployed at a limited number of sites in the United States. This paper documents an evaluation of the operational and safety effectiveness of these strategies.

INTRODUCTION

Intersections represent a significant safety issue from several perspectives. In 2005, more than 50 percent of all crashes in urban areas and over 30 percent in rural areas were intersectionrelated.(1) Crash severity at intersections is also a concern for certain States; over 60 percent of fatal intersection crashes in Minnesota occur at rural intersections.(2) In addition, specific driver populations, including older and younger drivers, are over-represented in these fatalities.

This figure provides an example of an implementation of concept 2. The photo was taken from the minor approach at a three-legged intersection. The major and minor roads are two-lane, two-way roadways with stop-control on the minor approach. The minor approach includes paved shoulders, white edgelines, a double yellow center line, and a STOP bar at the intersection with the major road. Concept 2 introduces a channelizing separator island on the minor approach, on which, a supplemental STOP sign is installed. The double yellow center line separates and a single yellow line continues around each side of the separator island; cross-hatching is installed between the yellow center lines where they split before the island.

Figure 3. Concept 2.


This figure provides a schematic of concept 2. A two-lane, two-way major roadway with 3.66-m (12-ft) lanes, and a double yellow center line is oriented in the east-west directions. A two-lane, two-way, stop-controlled minor roadway intersects the major road and is oriented in the north-south directions. The illustration shows where a separator island is installed on the minor approaches at the intersection. A STOP sign is installed at the intersection on the right side of each minor approach. A secondary STOP sign is also installed on the separator island on each minor approach. A legend is provided to denote the symbols for the STOP signs and the separator island.

1 ft = 0.305 m

Figure 4. Illustration of concept 2.

This figure provides a schematic of concept 3, which is a combination of concepts 1 and 2. A two-lane, two-way major roadway with 3.66-m (12-ft) lanes, and a double yellow center line is oriented in the east-west directions. A two-lane, two-way, stop-controlled minor roadway intersects the major road and is oriented in the north-south directions. The illustration shows where a painted median island is installed on the major approaches prior to the intersection. The lane width on the major approaches in narrowed from 3.66 m (12 ft) prior to the median island to 2.75 m (9 ft) at the narrowest point. Rumble strips are depicted with a solid green line along the shoulders of the major road as well as within the painted median. The illustration also shows where a separator island is installed on the minor approaches at the intersection. A STOP sign is installed at the intersection on the right side of each minor approach. A secondary STOP sign is also installed on the separator island on each minor approach. A legend is provided to denote the symbols for the STOP signs, rumble strips, and separator islands.

1 ft = 0.305 m

Figure 5. Illustration of concept 3.

Driver compliance with the STOP sign is a major contributing factor of intersection crashes. Statistics for crashes reported at STOP-controlled intersections on rural highways in Minnesota indicate that 26 percent of drivers who were involved in multivehicle crashes ran the STOP sign at the intersection.(3) A recent publication by the National Highway Traffic Safety Administration (NHTSA)(4) indicates that drivers failed to obey the STOP sign for 21 percent of fatal crashes at STOPcontrolled intersections. An additional 23 percent failed to yield to crossing main road traffic.

Intersections present several hazards that are different from road segments. First, drivers on the minor road must identify the STOP sign and bring their vehicle to a complete stop at the appropriate location. Each driver's ability to identify the STOP sign can be affected by a number of factors, including the placement, size, and retroreflectivity of the sign. At TWSC intersections, drivers on the minor approach must also identify and select a safe gap for entering the traffic stream on the major road. A driver's ability to judge gaps may be influenced by speeds on the major road (i.e., if speeds on the major road are higher than expected, gaps that appear sufficient may, in fact, be too small). Therefore, strategies for reducing speeds on the major approach and for increasing driver compliance on the minor approach have the potential to enhance safety. Concepts 1 and 2 address the issues of speed on the major road and driver compliance on the minor road, respectively.

Concept 1 focuses on reducing speeds on the major approaches by narrowing the lane width and will hereafter be referred to as the lane narrowing concept. While there was no literature related to this exact concept, other studies have shown that the roadway environment can affect drivers' speed. For example, one study indicated that speed perception was greater when the roadway was lined with trees, creating a narrowing effect.(5) With respect to safety, the effects of lane-narrowing techniques have not been studied thoroughly. Regarding the use of rumble strips, shoulder and centerline rumble strips have been shown to enhance safety by reducing run-off-road and cross-over crashes. However, there is a concern that centerline rumble strips could have a negative impact on safety. The shoulder rumble strip application is more common than the centerline rumble strip, and drivers may have a general a priori expectancy to steer left when encountering rumble strips while unaware of their lane position. A recent study(6) verified this concern using a simulator experiment and concluded that some drivers (about 27 percent) initially steered left when encountering centerline rumble strips while unaware of their lane position. While this concern may hold for applications of the lane narrowing concept, the rumble strips are more of a lanenarrowing device than a warning for unaware motorists. For those drivers that are aware of their lane position, the use of rumble strips as a lane-narrowing device should not be a safety issue. It is likely that the concern would only hold for those drivers who cross the centerline while unaware of their lane position at the approach to the intersection.

Concept 2 focuses on increasing intersection awareness by adding a supplemental STOP sign on the minor approaches via a separator island, hereafter referred to as the minor road splitter island concept. Similar implementations were found to produce a 30 percent reduction in total crashes in New Zealand (7) and a 30 percent reduction in angle and crossing crashes in France.(8)

PARTICIPATING STATES

Several local and State agencies were solicited for participation in the deployment of the proposed concepts. Favorable responses were received from agencies in many States including Maryland, Virginia, New Mexico, Illinois, Pennsylvania, Kentucky, Missouri, Florida, and California. For this study, the lane narrowing concept was implemented at 10 sites, and speed changes were evaluated at 9 sites, including sites in Florida, Kentucky, Missouri, and Pennsylvania. The speed change results were consistent between the first 9 sites; therefore, the tenth site was not included in the speed study. The minor road splitter island concept was implemented and evaluated at one site in Lorton, VA. Concept 3 was implemented at one site in Cumberland, MD. Participating agencies selected the sites for deployment of the proposed concepts; however, the research team worked closely with the participating agencies to provide guidance during the site selection process. Selected intersections exhibited one or more of the following characteristics:

  • The presence of the intersection is difficult to detect for approaching drivers.
  • Speeding was identified as an issue at the intersection.
    • Measured speeds are higher than established criteria (e.g., estimated safe intersection approach speed).
    • The patterns of crashes indicate speed-related causation.
  • Lack of compliance at STOP signs was observed frequently at the intersection.

TYPICAL DESIGN OF CONCEPTS

The lane narrowing concept features the introduction of a narrow median island formed by pavement markings in conjunction with rumble strips placed between the two travel lanes of a major road. It also includes the introduction of rumble strips outside of the edgelines within the existing pavement width as shown in figure 3. The median island and rumble strips effectively reduce the travel lane width prior to an intersection. The objective of this concept is to induce drivers to slow down as they approach an intersection. The installation of this concept is relatively low cost. While Pennsylvania spent between $50,000 and $70,000, the installations in other States were much less expensive. Based on data from the remaining States, the implementation costs for the lane narrowing concept ranged from $10,000 to $30,000 per intersection, excluding construction costs unrelated to the concept implementation.

A typical design template for the lane narrowing concept is shown in figure 6. For this scenario, lane widths on the major road are reduced from 3.66 m (12 ft) to 2.75 m (9 ft), as measured from the inside edges of the pavement markings. The effective lane width after implementation is 3.05 m (10 ft), as measured from the inside edges of the rumble strips. The design template shows three distinct sections (A, B, and C). Prior to the lane narrowing, appropriate signing is placed at the beginning of section A to warn motorists of the upcoming taper or intersection. The end of section A corresponds with the beginning of section B (i.e., the lane taper). Section B gradually transitions from a median width of 0 m (0 ft) to the full width of the median at the end of the section. Section C carries the full width of the median for 45.75 m (150 ft) up to the intersection, but the rumble strips end 15.25 m (50 ft) prior to the intersection. Table 1 indicates the lengths of each section based on the posted speed of the roadway. For example, if the posted speed of the major road is 88.55 km/h (55 mi/h), then the lengths of sections A, B, and C would be 30.5 m (100 ft), 61 m (200 ft), and 45.75 m (150 ft), respectively.


This figure illustrates the typical design of concept 1. The two-lane, two-way major approach is oriented in the east-west direction, and the figure focuses on the eastern leg. A north arrow is presented to denote direction. At a distance of 137.25 m (450 ft) from the intersection, section A begins and continues for a distance of 30.5 m (100 ft). At the beginning of section A, a W5-1 or W2-1 warning sign is installed and the lane width is 3.66 m (12 ft) throughout the section. Milled shoulder rumble strips are installed in the westbound direction, beginning 15.25 m (50 ft) prior to the end of section A. At a distance of 106.75 m (350 ft) from the intersection, section B begins and continues for a distance of 61 m (200 ft). The westbound shoulder rumble strips are continued through section B. A painted median island is introduced in section B, where the lane width is narrowed from 3.66 m (12 ft) to 2.75 m (9 ft). Milled rumble strips and cross-hatching are installed in the painted median. The cross-hatching is specified as “transverse markings, yellow, 0.61 m (2 ft) wide, 6.1 m (20 ft) spacing, 45-degree angle.” At a distance of 45.75 m (150 ft) from the intersection, section C begins and continues for 45.75 m (150 ft) to the intersection. The 2.75-m (9-ft) lane width, painted median, and cross-hatching continue to the intersection. The width of the painted median is shown as 1.22 m (4 ft) to 1.83 m (6 ft). The westbound shoulder rumble strips and median rumble strips end 15.25 m (50 ft) prior to the intersection. In the eastbound direction, the milled shoulder rumble strips begin 15.25 m (50 ft) past the intersection and end 106.75 m (350 ft) from the intersection, which is the point between sections A and B. The 2.75-m (9-ft) lane is continued in the eastbound direction through section C, tapers back to a 3.66-m (12-ft) lane through section B, and continues as a 3.66-m (12-ft) lane for section A.

1 ft = 0.305 m

Figure 6. Typical design of concept 1.


Table 1. Concept 1 section lengths.

Speed(mi/h)

Section A(ft)

Section B(ft)

Section C(ft)

45-55

100

200

150

60

150

200

150

1 ft = 0.305 m

Milled rumble strips are installed within the median and along the shoulders. The median rumble strips start at the beginning of section B and continue through section C, ending 15.25 m (50 ft) prior to the intersection. Traveling toward the intersection, shoulder rumble strips start 15.25 m (50 ft) prior to the beginning of section B and end 15.25 m (50 ft) prior to the intersection. Traveling away from the intersection, shoulder rumble strips start 15.25 m (50 ft) past the intersection, continuing through sections C and B, and ending at section A. A typical design for rumble strips is shown in figure 7. Transverse pavement markings are shown in the median. The yellow transverse markings are 0.61 m (2 ft) wide and placed at a 45-degree angle with a spacing of 6.1 m (20 ft).

The minor road splitter island concept features channelizing islands on the minor approaches of one-way or two-way STOP-controlled intersections as shown in figures 4 and 8. The objective of this concept is to improve the conspicuity of the STOP sign and increase driver compliance. This concept primarily focuses on increasing driver awareness of the intersection by adding a supplemental STOP sign on the minor approaches via a separator island.

A typical design template is shown in figure 8. For the minor road splitter island concept, the major design considerations are the type of separator island (traversable or nontraversable) and the turning radius of large vehicles. The typical design of the separator island is an oval, 4.27 m (14 ft) in length and 2.135 m (7 ft) in width, constructed of either earth material or concrete. A secondary STOP sign is placed on the separator island at the nose closest to the cross road. The double yellow pavement markings on the minor road are separated prior to the island and extended along the sides of the island to guide motorists as per the Manual on Uniform Traffic Control Devices (MUTCD).(9) Guidance for calculating the minimum taper length for the pavement markings can be found in the MUTCD, Section 3B-10.(9) To accommodate larger vehicles turning from the minor approach, the separator island is offset left of center.

This figure shows a typical rumble strip design. The travel lane is shown from left to right at the top of the diagram with an edge line. The shoulder is shown at the bottom of the diagram on which the rumble strips are installed. The rumble strips start at the outside of the edge line and extend approximately 4.06 m (16 inches) onto the shoulder. The rumble strips are approximately 1.778 m (7 inches) in width, measured parallel to the direction of travel and spaced approximately 1.27 m (5 inches) apart.


1 ft = 0.305 m

Figure 7. Typical rumble strip design.


This figure shows a typical design of concept 2 at a three-legged intersection with stop-control on the minor approach. All approaches are two-lane, two-way roads with 3.66-m (12-ft) lanes, and concept 2 is implemented on the minor approach. The typical design of the separator island is an oval, 4.27 m (14 ft) in length and 2.135 m (7 ft) in width. The side of the oval has a radius of 3.36 m (11 ft) while the nose has a radius of 0.61 m (2 ft). An R1-1 7.62 m by 7.62 m (30 inch by 30 inch) STOP sign is installed on the minor approach, and a secondary STOP sign of the same size is placed on the separator island at the nose closest to the cross road. The double yellow pavement marking on the minor road is separated prior to the island and extended along the sides of the island. To accommodate larger vehicles turning from the minor approach, the separator island is offset left of the center; the center of the island is located 5.34 m (17.5 ft) from the right edge of the traveled way on the minor approach, and the nose of the island is set back 7.17 m (23.5 ft) from the nearest edge of the crossing roadway. A 18.3-m (60-ft) turning radius connects the major and minor roads. The turning path of a WB-50 design vehicle is shown for each possible movement. Different colors are used to differentiate between the four turning paths: eastbound on major road turning onto minor road, westbound on major road turning onto minor road, minor road turning right onto major road, minor road turning left onto major road. Notes are added that the separator island is designed for nonmountable traffic operations but can be made mountable wherever appropriate. Also, the island may be concrete or grass material and adequate sight lines must be maintained.

1 ft = 0.305 m

Figure 8. Typical design of concept 2.

For the design template, the center of the island is located 5.3 m (17.5 ft) from the right edge of the traveled way on the minor approach, and the nose of the island is set back 7.2 m (23.5 ft) from the nearest edge of traveled way on the crossing roadway. These dimensions are, however, dependent on the turning radius of the design vehicle. The minor approach may also need to be flared depending on the existing turning radius to accommodate the design vehicle.

IMPLEMENTATION OF CONCEPTS

Design templates, shown previously in figures 6 and 8, were provided to the States for the applicable treatment. However, the States did not always follow the template exactly (table 2). As per the design templates, lane widths on the major road were reduced from 3.66 m (12 ft) to 3.05 m (10 ft). The 3.05-m (10-ft) effective lane width is measured as the distance between the inside edges of the rumble strips. Field measurements indicated that lane widths actually varied from one point to the next because of limited accuracy in pavement marking techniques and rumble strip milling techniques. For the lane narrowing concept, typical lane widths on the major approaches ranged from 3.36 m (11 ft) to 3.66 m (12 ft) before implementation. After implementation, typical lane widths on the major approaches ranged from 2.75 m (9 ft) to 3.05 m (10 ft) within the narrowed section. Field visits also revealed different installation practices among contractors. For example, contractors measured between the inside edges of the pavement markings when calculating the effective lane widths for the intersection approaches at a few locations. Several other variations of the design were actually implemented as shown in figures 9 through 12.


Table 2. Design variations among sites.

State

Location

Lane width prior
to taper (ft)

Lane width within
narrowed section (ft)

Painted median
width (ft)

Pennsylvania

PA 4

12

10

4

Kentucky

KY 2

12

9.5

3.5

Missouri

MO 1

11

9

3

Missouri

MO 2

11-12

9-10

4-6

Missouri

MO 3

11-12

9-10

4-6

1 ft = 0.305 m


This figure shows an implementation of concept 1 at site 4 in Pennsylvania. The photo was taken from the major approach at an intersection of 2 two-lane, two-way roadways with stop-control on the minor approach. There are paved shoulders with a white edge line and yellow center line. Concept 1 includes the introduction of a painted yellow median island on the major road approaches. The painted median begins as a double yellow center line and gradually widens to a width of approximately 1.22 m (4 ft); the full median width is then continued to the intersection. Yellow cross-hatching is installed in the painted median. Rumble strips are also introduced on the outside shoulders and in the painted median on the major road approaches. A vehicle is shown travelling on the major road within the limits of concept 1.

Figure 9. Concept 1 at PA 4.

This figure shows an implementation of concept 1 at site 4 in Pennsylvania. This photo focuses on the painted median and median rumble strips. In this example, a painted median is installed with cross-hatching. Double rumble strips are installed in the median at a width of approximately 4.06 m (16 inches) wide which almost cover the full width of the median at the widest point.

Figure 10. Example 2 of concept 1 at PA 4.

This figure shows an implementation of concept 1 at site 1 in Florida. The photo shows a section of the major road looking across the road from shoulder to shoulder. Shoulder rumble strips are shown, starting approximately 1.5 m (6 inches) from the edge line and continuing toward the edge of the shoulder. A painted median is installed along with median rumble strips that fully cover the width of the median between the inside edges of the pavement markings. A double solid yellow pavement marking is installed along each side of the median rumble strips, and raised pavement markings are installed along these centerlines.

Figure 11. Concept 1 at FL 1.

Figures 9 and 10 illustrate an installation of the lane narrowing concept in Pennsylvania. Prior to implementation of the concept, the major road approaches included 3.66-m (12-ft) lanes. The deployment of the lane narrowing concept effectively narrowed the major approaches to 3.05 m (10 ft). In this example, milled rumble strips were installed beginning at the outside edge of the pavement markings, similar to the design shown in figure 7. A painted median was also implemented at a width of 1.22 m (4 ft) and is clearly visible with cross-hatching. The double rumble strips in the median were each 4.06 m (16 inches) wide, which almost covers the full width of the median at the widest point. The rumble strip layout and cross-hatching within the median are desirable design characteristics.

Figures 11 and 12 provide an example of the lane narrowing concept in Florida. Much of the design is similar to the installation shown in figures 9 and 10; however, the median is slightly narrower at the location in Florida. The median rumble strips fully cover the width of the median between the inside edges of the pavement markings. In addition, a double solid (as opposed to a single solid) yellow pavement marking was used along each side of the median rumble strips, and raised pavement markings were installed along the centerline. Both the shoulder and median rumble strips end 15.25 m (50 ft) prior to the intersection. The rumble strip and pavement marking layouts are desirable characteristics for the lane narrowing concept.

This figure shows an implementation of concept 1 at site 1 in Florida. The photo shows a section of the major road from a driver’s perspective looking away from the intersection. Shoulder rumble strips are shown to the outside of a white edge line, and the rumble strips end prior to the intersection. A painted median is shown with cross-hatching and median rumble strips that fully cover the width of the median between the inside edges of the pavement markings. The median rumble strips end prior to the intersection. A double solid yellow pavement marking is installed along each side of the median rumble strips, and raised pavement markings are installed along the centerline.

Figure 12. Example 2 of concept 1 at FL 1.

Figures 13 through 15 show examples of the lane narrowing concept at two different sites in Kentucky. At both sites, the concept was implemented at intersections with leftturn lanes on the major approaches. This was accomplished by creating extra space between the left-turn lane and the opposing and adjacent through lanes. Rumble strips were then installed between the left-turn lanes and the opposing and adjacent through lanes as well as along the shoulders. A combination of milled and rolled rumble strips were installed. The previous examples show the median rumble strips ending prior to the intersection to accommodate turning vehicles. At both sites in Kentucky, however, the median rumble strips were installed up to the intersection. Additionally, rumble strips were installed along only the right edge of the white painted median that separates the left-turn lane from the through lane in the same direction. Beyond these similarities, there are subtle differences between the two sites.

At the first site (figures 13 and 14):

  • The median is very obvious due to cross-hatching.
  • Rumble strips are installed along both inside edges of median prior to the left-turn lane.
  • Rumble strips are installed along both edges of the yellow painted median that separates the left-turn lane from the opposing through lane.
  • Pavement markings are painted directly on the rumble strips, which are referred to as rumble stripes. This is in contrast to the previous designs where the pavement markings are painted closer to the traffic. Rumble stripes create a vertical edge, providing better visibility at night and in wet weather conditions.
  • White skip-lines are installed across side road approaches to designate the right edge of the travel way for through vehicles.
  • White arrows are painted on the road to designate left-turn and through lanes.

This figure shows an implementation of concept 1 at site 1 in Kentucky. The photo shows a section of the major road looking across the road from the centerline to the shoulder. The centerline is a solid yellow line, approximately 3.05 m (12 inches) wide and rumble strips cover the entire width. A portion of a left-turn lane is shown next to the center line, which is separated from the adjacent through lane by a white painted median with cross-hatching. The through lane has a white painted arrow, indicating the direction of travel and proper use of the through lane.

Figure 13. Concept 1 at KY 1.

This figure shows an implementation of concept 1 at site 1 in Kentucky. The photo shows a section of the major road from a driver’s perspective across the intersection. Looking across the intersection, the photo shows a through and left-turn lane approaching the intersection and a through lane going away from the intersection. All lanes have painted white arrows installed to indicate the direction of travel and proper use of the lane. A solid yellow line, approximately 3.05 m (12 inches) wide, separates the left-turn and opposing through lane. A white painted median with cross-hatching, approximately 1.22 m (4 ft) wide, separates the left-turn lane from the adjacent through lane. Prior to the left-turn lane, the opposing through lanes are separated by a yellow painted median with cross-hatching, which narrows and becomes the solid yellow line near the intersection. White edge lines and paved shoulders are also visible in the picture. To the left of the picture, a portion of the side road is shown with a double yellow centerline and a STOP bar near the intersection. A white dashed line is also shown across the minor road, connecting the white edge lines along the major road.

Figure 14. Example 2 of concept 1 at KY 1.

This figure shows an implementation of concept 1 at site 2 in Kentucky. The photo shows a section of the major road from a driver’s perspective approaching the intersection. Looking across the intersection, the photo shows a through and left-turn lane approaching the intersection and a through lane going away from the intersection in both directions. The left-turn lane has painted white arrows installed to indicate the direction of travel and proper use of the lane; STOP bars are also installed for the left-turn lanes. A painted yellow median, approximately 0.915 m (3 ft) wide, separates the left-turn and opposing through lane. The yellow median does not have cross-hatching. A white painted median, approximately 0.915 m (3 ft) wide, separates the left-turn lane from the adjacent through lane. The white median has cross-hatching installed. Rumble strips are installed along the shoulders and within the medians and extend to the intersection. In the painted medians, rumble strips are installed along left inside edge of the median that separates the left-turn lane from the opposing through lane and along the right edge of the painted median that separates the left-turn lane from the through lane in the same direction. Right-turn ramps with large turning radii are installed on both the major and minor roads.

Figure 15. Concept 1 at KY 2.


At the second site (figure 15):

  • There is a lack of cross-hatching between the left-turn and opposing through lanes which reduces the conspicuity of the median. This is an undesirable design characteristic.
  • Rumble strips are only installed along the left inside edge of the painted median prior to the left-turn lanes, which narrows the lane for drivers leaving the intersection but not for drivers approaching the intersection.
  • Rumble strips are installed along only the left edge of the painted median that separates the left-turn lane from the opposing through lane.
  • Rumble strips are installed along only the right edge of the painted median that separates the left-turn lane from the through lane in the same direction.

A typical design of the lane narrowing concept with left-turn lanes is shown in figure 16. Due to the presence of left-turn lanes, section C from figure 6 was modified slightly to create sections C and D as shown in figure 16. In this design, sections C and D include narrowed lanes with a painted median that extends to the intersection. Rumble strips are installed along both inside edges of all painted median islands and end 15.25 m (50 ft) prior to the intersection. Section C is now 30.5 m (100 ft) rather than 45.75 m (150 ft), but there is the added length of the turn lane (section D) to increase the length of the lane narrowing. Again, the lane width between pavement markings is 2.75 m (9 ft), but the effective lane width (i.e., distance between rumble strips) is 3.05 m (10 ft).


This figure illustrates the design of concept 1 when left-turn lanes are present at the intersection. The two-lane, two-way major approach is oriented in the east-west direction, and the figure focuses on the eastern leg. A north arrow is presented to denote direction. The approach is divided into four sections; A, B, C, and D. Section A consists of two 3.66-m (12-ft) lanes, is 100 ft in length, and is furthest from the intersection. At the beginning of section A, a W5-1 or W2-1 warning sign is installed and milled shoulder rumble strips are installed in the westbound direction, beginning 15.25 m (50 ft) prior to the end of section A. Section B begins at the end of section A and continues for a distance of 61 m (200 ft). The westbound shoulder rumble strips are continued through this section. A painted median island is introduced in section B where the lane width is narrowed from 3.66 m (12 ft) to 2.75 m (9 ft). Milled rumble strips and cross-hatching are installed in the painted median. The cross-hatching is specified as “transverse markings, yellow, 0.61 m (2 ft) wide, 6.1 m (20 ft) spacing, 45-degree angle.” Section C begins at the end of section B and continues for 30.5 m (100 ft). The 2.75-m (9-ft) lane width, painted median, rumble strips, and cross-hatching continue through this section. Section D begins at the end of section C and continues to the intersection. Section D is composed of a through lane and left-turn lane on the approach to the intersection and a single through lane leaving the intersection. The length of section D represents the length of the left-turn lane. The left-turn lane is separated from the adjacent and opposing through lanes by painted medians. The width of the painted median between the left-turn and adjacent through lane is shown as 1.22 m (4 ft) to 1.83 m (6 ft). The width of the painted median between the left-turn and opposing through lane is shown as 0.61 ft (2 ft) to 1.22 m (4 ft). Cross-hatching is installed in both painted medians and continues to the intersection. Rumble strips are also installed in both painted medians but end 15.25 m (50 ft) prior to the intersection along with the westbound shoulder rumble strips. All lane widths within section D are 2.75 m (9 ft), and painted arrows are shown in all lanes representing the direction of travel and proper use of the lane. In the eastbound direction, the milled shoulder rumble strips begin 15.25 m (50 ft) past the intersection and end at the point between sections A and B. The 2.75-m (9-ft) lane is continued in the eastbound direction through sections D and C, tapers back to a 3.66-m (12-ft) lane through section B, and continues as a 3.66-m (12-ft) lane for section A.

1 ft = 0.305 m

Figure 16. Typical design of concept 1 with left-turn lanes.

This figure shows an implementation of concept 1 at site 1 in Missouri. The photo shows a section of the major road looking away from the intersection. A painted median is implemented by splitting the double yellow center line and continuing the single yellow lines at a distance of approximately 4.06 m (16 inches) apart. Rumble stripes are installed to cover the entire width of the median, but no cross-hatching is present. Narrow paved shoulders, approximately 0.305 to 0.61 m (1 to 2 ft) wide, are present, and shoulder rumble stripes are installed along the shoulder. The painted median and all rumble stripes end approximately 15.25 m (50 ft) prior to the intersection. A single dashed yellow center line extends to the intersection.

Figure 17. Concept 1 at MO 1.


Figures 17 and 18 illustrate installations of the lane narrowing concept in Missouri. At both sites, a painted median as well as milled shoulder and median rumble stripes were installed. Prior to the installation, lane widths ranged from 3.36 to 3.66 m (11 to 12 ft). The deployment of the lane narrowing concept effectively narrowed the major approaches to 2.75 to 3.05 m (9 to 10 ft). There were a few noteworthy differences between the two sites.

At the first site (figure 17):

  • The painted median and rumble stripes end 15.25 m (50 ft) prior to the intersection and only a single dashed yellow centerline extends to the intersection. It is desirable to extend the painted median to the intersection.
  • Rumble stripes cover the entire width of median. However, cross-hatching is not installed, which may have contributed to the increase in rear-end crashes.
  • At the second site (figure 18), there are several desirable characteristics:
  • Median rumble stripes cover nearly the entire width of the median. All rumble stripes end 15.25 m (50 ft) prior to intersection.
  • Cross-hatching is installed in the painted median, which extends to the intersection.

This figure shows an implementation of concept 1 at site 2 in Missouri. The photo shows a section of the major road looking toward the intersection with the minor road. A painted median is implemented by splitting the double yellow center line and continuing the single yellow lines at a distance of approximately 0.915 m (3 ft) apart. Cross-hatching is installed in the pained median, which extends to the intersection. Median rumble stripes nearly cover the entire width of median. A paved shoulder is present and is approximately 1.83 m (6 ft) wide. Shoulder rumble stripes are installed on the white edge line. All rumble stripes end 15.25 m (50 ft) prior to intersection.

Figure 18. Concept 1 at MO 2.

This figure shows an implementation of concept 2 at site 1 in Virginia. The photo shows one approach of the minor road looking from the intersection toward the minor road. The approach is two-lane, two-way, and divided by a double-yellow center line. There are no paved shoulders and no edge lines. A primary STOP sign is installed on the minor approach along with a STOP bar. A concrete separator island is installed on the minor approach at the intersection, and a secondary STOP sign is installed on the island. The yellow center line is continued along both sides of the separator island. The separator island is nontraversable with vertical edges.

Figure 19. Concept 2 at VA 1.


Figure 19 shows an example of the only deployment of the minor road splitter island concept included in this study. The minor road splitter island concept was deployed in Lorton, VA, at a four-legged intersection with STOP-control on the minor approaches. Prior to the installation of the minor road splitter island concept, the intersection was STOP-controlled on the minor approaches with a single STOP sign per approach located to the right of the traveled way. The deployment of the minor road splitter island concept included the installation of a separator island on the minor approaches near the intersection and the placement of a second STOP sign on the separator islands. The separator islands are nontraversable with vertical edges. Since the separator islands are nontraversable, the turning radius of the design vehicle must be accommodated. The lane markings are also separated on the minor approaches prior to the intersection and painted along both sides of the separator islands. Providing cross-hatching between the pavement markings or painting the separator islands would improve the conspicuity of the treatment.

Figures 20 and 21 show an example of concept 3 in Cumberland, MD. This was the only deployment of concept 3 included in this study. The combined concept was implemented at a three-legged intersection with STOPcontrol on the minor approach. Prior to the installation of concept 3, the intersection was STOP-controlled on the minor approach with a single STOP sign to the right of the traveled way. The deployment of concept 3 included the installation of a traversable separator island on the minor approach. Also, a secondary STOP sign was installed on the separator island. The lane markings are separated on the minor approach prior to the intersection and painted along both sides of the separator island. However, in this case, cross-hatching is included between the pavement markings to improve the conspicuity of the concept. The major road approaches were effectively narrowed by installing shoulder and centerline rumble stripes. Unlike the typical implementation of the lane narrowing concept, this deployment did not include a painted median and the rumble stripes continue to the intersection, which is not desirable.


This figure shows an implementation of concept 3 at site 1 in Maryland focusing on the minor approach. The photo shows the minor, stop-controlled approach of a three-legged intersection. The photo is taken from the left side of the minor road approach looking toward the intersection. The major and minor roads are two-lane, two-way roadways with stop-control only on the minor approach. The minor approach includes paved shoulders, white edge lines, a double yellow center line, and a STOP bar at the intersection with the major road. A separator island is installed on the minor approach, on which a supplemental STOP sign is installed. The double yellow center line separates prior to the island and a single yellow line continues around each side of the separator island; cross-hatching is installed between the single yellow lines prior to the island.

Figure 20. Concept 3 at MD 1 (minor approach).

This figure shows an implementation of concept 3 at site 1 in Maryland focusing on the major approach. The photo shows a section of the major road from a driver’s perspective looking toward the intersection. The major road is two-lane, two-way with a double yellow center line, white edge lines, and approximately 2.44-m (8-ft) paved shoulders. There is no painted median in the application, but rumble stripes are installed on the center line and edge lines. A white dashed line is also shown across the minor road, connecting the white edge lines along the major road. Two vehicles appear in the photo traveling in opposite directions.

Figure 21. Concept 3 at MD 1 (major approach).

DATA COLLECTION

Speed and driver behavior data were collected before implementation at 9 sites for the lane narrowing concept and one site for the minor road splitter island concept. Similar data were collected at all 10 sites in the after period. The lane narrowing concept was deployed at a tenth site in Florida; however, speed data were not available in the before period. Data indicated that results were consistent at the other 9 sites where the lane narrowing concept was deployed; therefore, speed data were not collected at the tenth site. Hi-Star traffic sensors with sequential speed profile measuring capabilities were temporarily placed on the road to collect individual driver speeds, vehicle counts, headways, and traffic composition for a 24-hr period both before and after implementation. These automated devices continuously recorded speed data for each individual vehicle that passed through the site on the approaches of interest. Additionally, speeds were captured by radar and used to validate the speed data recorded by the temporary sensors. Post-implementation data collection occurred at least 3 months after implementation to allow sufficient time for possible short-term "novelty effects" of the concepts to fade. All data were collected on weekdays.

This figure shows the location of the data collection points for concept 1. The figure focuses on the western major approach to the intersection, which is oriented in the east-west directions. The major approach is depicted as a two-lane, two-way road with a double yellow center line. The center line becomes a painted median with cross-hatching. This occurs gradually with a taper length of 61 m (200 ft) and then a full painted median width for 45.75 m (150 ft). The four data collection points are shown as black dots on the approach toward the intersection. Point 1 is located 122 m (400 ft) prior to the beginning of the center line taper. Point 2 is located at the beginning of lane-narrowing taper, approximately 106.75 m (350 ft) from intersection. Point 3 is located at the end of the lane taper, approximately 45.75 m (150 ft) from intersection. Point 4 is located at the intersection.

1 ft = 0.305 m

Figure 22. Illustration of concept 1 speed data collection points.

The data collection points were different for the two strategies. For the lane narrowing concept, speed data were collected at four different locations (figure 22) along each of the major approaches to determine whether or not speed reductions were occurring as desired. The locations of the data collection points were as follows:

  1. Control point: Located 122 m (400 ft) prior to the beginning of the taper (Point 1).
  2. Beginning of lane-narrowing taper: Approximately 106.75 m (350 ft) from intersection (Point 2).
  3. End of lane taper: Approximately 45.75 m (150 ft) from intersection (Point 3).
  4. At the intersection (Point 4).

For the lane narrowing concept, speed observations were screened to eliminate vehicles that had been affected by any other factors such as turning maneuvers or vehicle platoons. To obtain the desired dataset, it was necessary to eliminate vehicles with relatively small headways and vehicles traveling below a certain speed. The cut-off speed for the lane narrowing concept dataset was 64.4 km/h (40 mi/h).

The speed cut-off for the lane narrowing concept was based on speed observations in North Carolina. Speed data were collected on road segments with 2.75-m (9-ft) lanes in North Carolina to determine the relationship among speed, lane width, and driver behavior. The road segments selected in North Carolina were two-lane, rural roadway segments with minimal horizontal and vertical curvature. It was observed that drivers could travel at speeds up to 72.45 km/h (45 mi/h) on 2.75 m (9 ft) wide lanes without crossing into the opposing lane. Therefore, the application of the lane narrowing concept is expected to affect vehicles with speeds greater than or equal to 64.4 km/h (40 mi/h). The final dataset for the lane narrowing concept included vehicles with headways greater than 20 seconds and approach speeds of 64.4 km/h (40 mi/h) or greater.

For the minor road splitter island concept, speed data were collected on the minor approaches. There was only one speed data collection point per approach located 45.75 m (150 ft) upstream from the STOP line or pavement marking on each minor approach. The dataset for the minor road splitter island concept included all vehicles on the minor road approach.

Driver behavior data were collected at the intersection for both strategies. A polemounted camera located at the intersection recorded driver behavior during the same periods as the speed data were collected. The videos were then analyzed and driver behaviors were recorded as follows:

  • STOP sign compliance, as measured in terms of the percent of motorists that did not stop, performed a "rolling" stop, or performed a complete stop.
  • Infringement on major road vehicles by vehicles on the minor approach.
  • Percentage of drivers entering intersection from minor approach when gap is less than desirable if one of the following is observed:
    • Start/Stop maneuvers on minor road.
    • Braking/Severe braking on major road to avoid merging/crossing vehicles from minor approach.
    • Erratic maneuvers on either road.

Crash and traffic volume data were collected for each site in the before and after periods. If data were available, at least 5 years of crash data were obtained for the before period, and 2 years of data were obtained for the after period. Crash data included the total number of crashes occurring at the intersection as well as crash severity (i.e., fatal, injury, or property damage) and crash type (i.e., angle, rear end, or other).

OPERATIONAL PERFORMANCE

The operational effectiveness of each concept is based on the selected measure of effectiveness (MOE). For the lane narrowing concept, the primary MOE is the reduction of speed on the major approaches. For the minor road splitter island concept, there are two MOEs: (1) the reduction of speed on the minor approach (i.e., at the upstream data collection point), and (2) increased driver compliance with the STOP sign on the minor approach. For the lane narrowing concept, the analysis was conducted for two subsets of vehicles: all vehicles and trucks only. This was done to identify differential vehicle effects of the concept.

The lane narrowing concept

For each site, the mean and 85th-percentile speeds were calculated for each approach at each of the four data collection points for both the before and after periods. Again, point 4 is closest to the intersection and point 1 is farthest from the intersection. An unpaired T-test was used to determine if the speed reduction from point 1 to point 4 (1 to 4) was statistically different from the reduction from point 2 to point 4 (2 to 4) in the before period. The test indicated that there was no significant difference between 2-4 and 1-4 at a 95-percent confidence level (p = 0.45, where p is the probability of error) Since there was no difference between the 2-4 and 1-4 points, it was decided that the remaining analysis would focus on speed reductions from point 2 to point 4. Speed reduction from the before data showed no statistical difference of speeds between points 2 and 4.

Table 3. Speed reductions for all vehicles.

Sites

Posted speed limit (mi/h)

Approach lane width within rumble strips

Number of observations

All speeds

85th-percentile speeds

Reduction after treatment(mi/h)

95% confidence interval

Reduction after treatment(mi/h)

95% confidence interval

Mean

S.D.

Mean

S.D.

PA 1

55

10'

376

4.6

0.29

(3.88, 5.33)

5.2

0.21

(4.68, 5.73)

PA 2

55

10'

535

4.3

0.31

(3.53, 5.08)

4.9

0.25

(4.28, 5.53)

PA 3

50

10'

487

3.8

0.41

(2.78, 4.83)

4.3

0.33

(3.48, 5.13)

PA 4

55

10.5'

356

3.7

0.61

(2.12, 5.23)

4.4

0.23

(3.83, 4.98)

KY 1

55

10'

352

1.8

0.32

(1.00, 2.60)

4.2

0.23

(3.63, 4.78)

KY 2

50

10'

317

2.6

0.34

(1.75, 3.45)

4.1

0.17

(3.68, 4.53)

MO 1

55

9'

428

3.9

0.42

(2.85, 4.95)

4.1

0.19

(3.63, 4.58)

MO 3

55

10.5'

408

2.9

0.25

(2.28, 3.53)

4.5

0.28

(3.80, 5.20)

FL 1

55

10'

423

3.4

0.33

(2.58, 4.23)

4.5

0.30

(3.75, 5.25)

All    

3,682

3.5

0.36

(2.61, 4.42)

4.5

0.25

(3.88, 5.08)

1 mi = 1.61 km

1 ft = 0.305 m

Table 4. Speed reductions for trucks only.

Sites

Posted speed limit (mi/h)

Approach lane width within rumble strips

Number of observations

All Speeds

85th-percentile speeds

Reductionafter (mi/h)

95%confidenceinterval

Reductionafter (mi/h)

95% confidenceinterval

Mean

S.D.

Mean

S.D.

PA 1

55

10'

65

4.4

0.38

(3.45, 5.35)

4.8

0.25

(4.18, 5.43)

PA 2

55

10'

87

4.5

0.26

(3.85, 5.15)

5.1

0.22

(4.55, 5.65)

PA 3

50

10'

79

4.6

0.29

(3.88, 5.33)

4.9

0.32

(4.10, 5.70)

PA 4

55

10.5'

66

4.2

0.41

(3.18, 5.23)

4.6

0.33

(3.78, 5.43)

KY 1

55

10'

54

3.9

0.44

(2.80, 5.00)

4.2

0.24

(3.60, 4.80)

KY 2

50

10'

73

4.2

0.43

(3.13, 5.28)

4.4

0.13

(4.08, 4.73)

MO 1

55

9'

90

4.8

0.33

(3.98, 5.63)

5.1

0.27

(4.43, 5.78)

MO 3

55

10.5'

55

3.8

0.28

(3.10, 4,50)

4.4

0.38

(3.45, 5.35)

FL 1

55

10'

49

4.5

0.48

(3.30, 5.70)

4.9

0.34

(4.05, 5.75)

All
   

618

4.4

0.36

(3.47, 5.26)

4.8

0.27

(4.08, 5.42)

1 mi = 1.61 km

1 ft = 0.305 m

Table 3 shows the speed reduction from point 2 to point 4 for all vehicles after the deployment of the lane narrowing concept. The mean and standard deviation were calculated for the reduction in all speeds and the reduction in 85thpercentile speeds. For all vehicles, the mean speed was reduced by 5.64 km/h (3.5 mi/h) with a standard deviation of 0.36. The confidence interval does not include zero; therefore, the reduction is statistically significant at the 95-percent confidence level. The mean reduction in the 85th-percentile speed was even greater (7.25 km/h (4.5 mi/h)) with a standard deviation of 0.25. Again, this reduction is statistically significant at the 95-percent confidence level.

Table 4 shows the speed reduction from point 2 to point 4 for trucks only after the deployment of the lane narrowing concept. For trucks, the mean speed was reduced by 7.1 km/h (4.4 mi/h) with a standard deviation of 0.36, and the reduction is statistically significant at the 95-percent confidence level. The mean reduction in the 85th-percentile speed was slightly greater (7.73 km/h (4.8 mi/h) with a standard deviation of 0.27. Again, this reduction is statistically significant at the 95-percent confidence level. Speed reductions were also disaggregated by time of day and lane width. The speed reductions were not significantly different by time of day. Based on the limited sample size, the degree of speed reduction was not linearly proportional to lane width (i.e., narrower lanes were not always associated with greater speed reductions).

Table 5 shows the maximum speed reduction after implementation from point 2 to point 4. For all vehicles combined, a maximum speed reduction of 19.32 km/h (12.1 mi/h) was observed. For trucks, a maximum speed reduction of 16.26 km/h (10.1 mi/h) was observed. The results indicate that the observed speed reduction is attributable to the lane narrowing concept. All reductions are statistically significant at a 95-percent confidence level and the reductions in the 85th-percentile speed appear to be greater than the reductions in the average speed. Furthermore, the speed reduction from point 2 to point 4 is greater for trucks than for all vehicles combined.

Table 5. Maximum speed reductions.

Sites

All vehicles (mi/h)

Trucks only (mi/h)

PA 1

10.4

9.6

PA 2

6.9

6.9

PA 3

9.1

5.4

PA 4

12.1

7.4

KY 1

5.5

4.5

KY 2

6.9

6.2

MO 1

7.1

7.1

MO 3

8.3

8.0

FL 1

10.2

10.1

All

12.1

10.1

1 mi = 1.61 km

Speed reductions were also disaggregated by time of day and lane width. The speed reductions were not significantly different by time of day. Based on the limited sample size, the degree of speed reduction was not linearly proportional to lane width (i.e., narrower lanes were not always associated with greater speed reductions). This may be counterintuitive but may be explained to some extent by the variation in lane widths for the same lane at a given intersection. It is hypothesized that narrower lane widths will produce greater speed reductions; however, a larger sample size and better field implementation of the concept is needed to support or refute the hypothesis.

Based on driver behavior data from the cameras, a large number of vehicles routinely contacted the rumble strips after implementation. This is likely due to the narrower lane widths and relative proximity of the rumble strips to the lane. While rumble strip contact helps to alert the driver of the edge of the travel way, it may have secondary benefits as well. The contact creates a significant noise outside the vehicle, which may help increase awareness for other motorists near the intersection.

Minor Road Splitter Island Concept

Descriptive statistics were calculated for both approaches on the minor road for the before and after periods. Again, these results are based on just one implementation of the minor road splitter island concept. Figures 23 and 24 show the average speed comparisons by time of day before and after deployment for the northbound and southbound minor road approaches, respectively, 45.75 m (150 ft) from the STOP sign. For the northbound (minor) approach, speed reductions ranged from 8.05 to 33.81 km/h (5 to 21 mi/h) after deployment, averaging 17.39 km/h (10.8 mi/h) with a standard deviation of 4.7. For the southbound (minor) approach, speed reductions ranged from 4.83 to 24.15 km/h (3 to 15 mi/h) after deployment, averaging 15.78 km/h (9.8 mi/h) with a standard deviation of 3.0. The speed reductions on both approaches are significant at the 95-percent confidence limit.

Based on driver behavior data from the camera, there were other benefits associated with the minor road splitter island concept. After implementation, there were minor improvements in STOP sign compliance as well as driver behavior. These results are, however, based on a limited sample at a single site. Further analysis is necessary to verify the operational effectiveness of the minor road splitter island concept.

This figure shows the average speed comparisons by time of day before and after the deployment of concept 2 at site 1 in Virginia. The figure is for the northbound minor road approach only and represents data collected 45.75 m (150 ft) from the stop sign. The x-axis represents a 24-hour period, in hours, starting at 2 pm and increasing in 4-hour increments to 2 pm. The y-axis represents speed in miles per hour, starting at zero and increasing to 50 in increments of 10. The before treatment period is represented by blue diamonds, which are connected by a blue line. The after period is represented by pink squares, which are connected by a pink line. In the before period, the average speed fluctuates in a sinusoidal pattern with most points ranging between 48.3 km/h (30 mi/h) and 64.4 km/h (40 mi/h). A low speed of about 16.7 km/h (29 mi/h) occurs around 9 pm, and a high speed of 74.06 km/h (46 mi/h) occurs around 10:30 pm. In the after period, the speeds are more constant with a dip from 4 pm to 6 pm and a hump from 3 am to 5 am. The speeds in the after period are all below those in the before period and range mostly between 32.2 km/h (20 mi/h) and 48.3 km/h (30 mi/h). A low speed of about 27.37 mi/h (17 mi/h) occurs around 6 pm, and a high speed of 45.08 km/h (28 mi/h) occurs around 5 am.

1 mi/h = 1.61 km/h

Figure 23. "Before-After" average speeds for northbound approach on minor road.

This figure shows the average speed comparisons by time of day before and after the deployment of concept 2 at site 1 in Virginia. The figure is for the southbound minor road approach only and represents data collected 45.75 m (150 ft) from the STOP sign. The x-axis represents a 24-hour period, in hours, starting at 2 pm and increasing in 4-hour increments to 2 pm. The y-axis represents speed in miles per hour, starting at zero and increasing to 60 in increments of 10. The before treatment period is represented by blue diamonds, which are connected by a blue line. The after period is represented by pink squares, which are connected by a pink line. In the before period, the average speed fluctuates with most points ranging between 64.4 km/h (40 mi/h) and 80.5 km/h (50 mi/h). A low speed of about 57.96 km/h (36 mi/h) occurs around 11 am, and a high speed of 78.89 km/h (49 mi/h) occurs around 9 am. In the after period, the speeds fluctuate with most points ranging between 48.3 km/h (30 mi/h) and 64.4 km/h (40 mi/h). The speeds in the after period are all below the respective points in the before period for each time of day. A low speed of about 16.7 km/h (29 mi/h) occurs around 5 pm, and a high speed of 61.18 km/h (38 mi/h) occurs around 12 am.

1 mi/h = 1.61 km/h

Figure 24. "Before-After" average speeds for southbound approach on minor road.

TRAFFIC SAFETY PERFORMANCE

Crash data were analyzed using a simple beforeafter comparison method. Crash rates were calculated for each year as the total number of crashes divided by the total number of vehicles entering the intersection per year. Crash rates were also computed for each level of crash severity and crash type. All crash rates are expressed as the crashes per million entering vehicles (MEV). Due to limited sample size and length of the after period, it was not possible to conduct an Empirical Bayes before-after evaluation—a more statistically robust method for isolating the effects of a treatment. While the sample size and duration of crash data are relatively limited, the simple before-after crash evaluation provides initial insights on the safety effectiveness of these strategies. An Empirical Bayes evaluation is planned in the near future using a longer period of crash data.

Lane Narrowing Concept

Crash data for the lane narrowing concept are shown in table 6 for Pennsylvania, Kentucky, Missouri, and Florida. For one site in Pennsylvania (PA 3), the lane narrowing concept was only implemented for a short period of time (less than 6 months). Due to the relatively short duration of implementation, the site was excluded from the crash analysis. The following summarizes the crash findings for the remaining implementations of the lane narrowing concept:

  • For the three sites in Pennsylvania, there was a reduction in the crash rate for total crashes, fatal/injury crashes, and related crashes (i.e., angle and rear end) after the implementation of the lane narrowing concept.
    • For total crashes, reductions ranged from 30 to 83 percent.
    • For fatal/injury crashes, reductions ranged from 1 to 79 percent.
    • For related crashes, reductions ranged from 18 to 100 percent.
  • The two implementations in Kentucky showed mixed results. For KY 1, there was a general reduction in crash rate for total, fatal/ injury, and angle crashes, all around 30 percent. However, there was a 39-percent increase in rear-end crashes. For KY 2, there was a reduction in fatal/injury (24 percent) and angle (13 percent) crashes but an increase in total (34 percent) and rear-end (87 percent) crashes. While there was an increase in the crash rate for rear-ends at both sites in Kentucky, the crash frequency is relatively low (2 or less), which may skew the results. In addition, the one rear-end crash in the after period at KY 1 occurred during wet roadway conditions. The increase in rear-end crashes at KY 2 appears to be contributing to the increase in total crash rate as well. Based on a review of the crash data, it does not appear that the implementation of the lane narrowing concept is contributing to the increase in rearend or total crash rates.

Table 6. Crash data summary for concept 1 implementations.

Site

Period
(years)

Crash rate (crashes per MEV per year)

Percent change in crash rate
(minus indicates a reduction)

Before

After

Before

After

Totalcrashes

Fatal + Injurycrashes

Anglecrashes

Rear-endcrashes

PA 1

5.25

1.08

3.23

1.02

-69%

-64%

-76%

NA

PA 2

5.25

1.08

1.46

1.01

-30%

-1%

-18%

-100%

PA 3

Lane narrowing removed after 6 months; therefore, PA 3 was not included in crash analysis.

PA 4

6.00

1.92

1.36

0.23

-83%

-79%

-100%

-100%

KY 1

5.75

1.17

2.13

1.48

-30%

-31%

-28%

39%

KY 2

5.00

1.50

1.52

2.04

34%

-24%

-13%

87%

MO 1

6.92

0.99

2.15

2.47

15%

116%

-59%

906%

MO 2

6.92

0.99

1.32

0.90

-32%

-20%

-46%

-100%

MO 3

6.92

0.99

2.31

2.05

-11%

26%

-47%

144%

FL 1

4.50

1.50

1.36

1.05

-23%

11%

68%

4%

Combined

52.51

11.22

1.85

1.27

-31%

-20%

-42%

54%

Note: MEV = million entering vehicles and NA indicates that no crashes were observed in the before period.

  • The three implementations in Missouri also showed mixed results.
    • For MO 1, there was a decrease in the angle crash rate (59 percent), but an increase for total, fatal/injury, and rear-end crashes.
    • For MO 2, there was a general reduction in crash rate for all categories.
    • For MO 3, there was a decrease in the crash rate for total (11 percent) and angle (47 percent) crashes but an increase in the crash rate for fatal/injury and rear-end crashes.
    • While the increase in rear-end crash rates appears to be substantial at two of the sites, the change in crash frequency is relatively minor. For MO 1 there was just one rear-end crash in each of the before and after periods. For MO 3, there were 11 rear-ends in the before period and just 4 in the after period. Also, 3 of the 4 rear-ends in the after period at MO 3 occurred during wet roadway conditions.
    • A more concerning issue is the increase in fatal/injury crashes at MO 1 and MO 3. A detailed review of the crash data indicated that the severity rate fluctuates substantially from one year to the next in the before period. For MO 1, the minimum fatal/injury crash rate was 0.00 crashes per MEV and the maximum was 3.49 crashes per MEV, averaging 1.43 fatal/injury crashes per MEV. The fatal/ injury crash rate in the after period was 2.47 crashes per MEV; however, this was based on less than 1 year of data. For MO 3, the minimum fatal/injury crash rate was 0.00 crashes per MEV and the maximum was 1.54 crashes per MEV, averaging 0.81 fatal/injury crashes per MEV. The fatal/ injury crash rate in the after period was 1.02 crashes per MEV, but, again, this was based on less than 1 year of data. It is possible that yearly variations not captured in the relatively short after period are contributing to the apparent increase. These sites should be monitored to confirm the trend in fatal/injury crashes.
  • For the one implementation in Florida, there was a reduction in total crash rate but an increase in the crash rate for fatal/injury and related crashes. While the increase in crash rates for fatal/injury and related crashes may be concerning, the duration of the after period was relatively short. A review of the police reports did not reveal any evidence that the implementation of the lane narrowing concept was contributing to the increase in crash rates; however, this site should be monitored to confirm the crash trend.
  • Overall, there appears to be a reduction in the crash rate for total, fatal/injury, and angle crashes after implementing the lane narrowing concept, as shown in the combined analysis at the bottom of table 6. There are several common features that should be noted when comparing the installations where crashes were reduced:
    • Rumble strips were installed along both inside edges of the median, which covers the entire width of narrow medians (figures 9, 10, and 18).
    • Shoulder and median rumble strips ended 15.25 m (50 ft) prior to the intersection (figures 9 and 18).
    • Pavement markings were obvious, with a painted median that extended to the intersection (figures 9, 13, and 18).
    • Cross-hatching was installed in the median to increase conspicuity (figures 9, 11, 13, and 18).
  • The crash rate for rear-end crashes increased in the after period. While the after period was relatively short for many of the sites, there was some consistency in the increase in rear ends. The following are possible reasons why rear ends may increase:
    • Speed differential: The lane narrowing concept is intended to reduce driver speeds on the major road. Rear ends could result if some drivers reduce their speed when entering the lane narrowing concept while other drivers maintain their previous speed.
    • Turning vehicles: The lane narrowing concept employs shoulder and median rumble strips to effectively reduce the lane width. Prior to the installation, rightturning drivers may use the shoulder to decelerate, rather than the through lane. Where shoulder and centerline rumble strips extend to the intersection, drivers may avoid the shoulder and decelerate in the through lane. This would lead to speed differentials and could result in rear-end crashes as discussed previously.
    • Passing vehicles: Prior to the installation, drivers may use the shoulder or cross the centerline to pass turning vehicles. Where shoulder and centerline rumble strips extend to the intersection, drivers may abort a passing maneuver after encountering the rumble strips, resulting in a rear-end collision.
  • A Wilcoxon Signed-Rank Test was used to determine whether or not the change in crash rate was significant for total crashes and for fatal/ injury crashes (i.e., fatal plus injury). The null hypothesis postulates that there is no difference between the crash rates in the before and after periods. The total sample size is 9 sites. The critical value is 29 for a sample size of 9 and a 5-percent significance level.
    • For total crashes, the test-value (27) was compared to the critical value (29) at a 5-percent significance level; results are not significant.
    • For fatal/injury crashes, the testvalue (33) was compared to the critical value (29) at a 5-percent significance level; results are significant, which leads to the conclusion that fatal/injury crashes were reduced after the lane narrowing concept was implemented.

Minor Road Splitter Island Concept

Crash data are shown in table 7 for the one implementation of the minor road splitter island concept in Virginia. For this site, there were 4 years of data in the before period and nearly 2 years after implementation. The following points summarize the crash findings for the one implementation of the minor road splitter island concept.

Table 7. Crash data summary for concept 2 implementation.

Site

Period (years)

Crash rate (crashes per MEV per year)

Difference in crash rate

Before

After

Before

After

Total crashes

Fatal + Injury crashes

Angle crashes

Rear-end crashes

VA 1

4.00

1.92

2.59

0.82

-68%

-74%

-74%

-100%

Note: MEV = million entering vehicles.

  • Based on the single implementation of the minor road splitter island concept, it appears that this concept is effective for enhancing safety.
    • The crash rate for total crashes was reduced by 68 percent after deploying the minor road splitter island concept.
    • The crash rate for fatal/injury crashes was reduced by 74 percent after deploying the minor road splitter island concept.
    • The crash rates for angle and rearend crashes were reduced by 74 and 100 percent, respectively, after deploying the minor road splitter island concept.

IDEAS FOR FUTURE DEPLOYMENTS

The typical design and actual implementation of concepts 1 and 2 have been discussed in detail in the preceding sections. Based on the operational and safety analysis, as well as lessons learned from current deployments and observations in the field, the following points are presented as ideas to enhance future deployments of the treatments. Figures 25 and 26 illustrate the potential enhancements to the lane narrowing concept for both rumble strip and rumble stripe designs, respectively.

This figure illustrates potential enhancements to concept 1, focusing on section C along the major road. The length of section C is shown as 61 to 76.25 m (200 to 250 ft). A painted yellow median is shown with cross-hatching that extends to the intersection. The center line is a solid double yellow line, 1.016 to 1.524 m (4 to 6 inches) in width. The cross-hatching is labeled as “transverse markings, yellow, 0.61 m (2 ft) wide, 6.1 m (20 ft) spacing, 45-degree angle.” Shoulder and median rumble strips are installed, ending 15.25 m (50 ft) prior to the intersection. White edge lines are shown at a width of 4 to 6 inches. The lane width is shown as 2.75 m (9 ft) between the inside edges of the pavement markings and 3.05 m (10 ft) between the inside edges of the rumble strips.

1 ft = 0.305 m

Figure 25. Illustration of enhancements to section C of concept 1 with rumble strips.

This figure illustrates potential enhancements to concept 1, focusing on section C along the major road. The length of section C is shown as 61 to 76.25 m (200 to 250 ft). A painted yellow median is shown with cross-hatching that extends to the intersection. The center line is a solid double yellow line, 1.016 to 1.524 m (4 to 6 inches) in width. The cross-hatching is labeled as “transverse markings, yellow, 0.61 m (2 ft) wide, 6.1 m (20 ft) spacing, 45-degree angle.”.Shoulder and median rumble stripes are installed, ending 15.25 m (50 ft) prior to the intersection. White edge lines are shown at a width of 1.016 to 1.524 m (4 to 6 inches). The lane width is dimensioned as 2.75 to 3.05 m (9 to 10 ft) between the inside edges of the pavement markings, which is the same distance as the inside edges of the rumble stripes.

1 ft = 0.305 m

Figure 26. Illustration of enhancements to section C of concept 1 with rumble stripes.

Lane Narrowing Concept

  • Increase the length of the narrowed section for the lane narrowing concept. Currently, the design template shows a length of 45.75 m (150 ft) for the narrowed section on the major approaches. This may be too short to achieve the desired effect. If the length is increased to 61 m (200 ft) or 76.25 m (250 ft), then drivers will travel a greater distance in the narrowed section, which may induce lower speeds. While current speed reductions on the major approaches are statistically significant, there is an opportunity to further reduce driver speeds. The nearest 15.25 m (50 ft) from the intersection should not have rumble strips.
  • Use a different rumble strip pattern in the median. Concerns have been raised that drivers may react similarly (i.e., steer to the left) when encountering both shoulder and centerline rumble strips with the same pattern. A recent study verified this concern and concluded that some drivers initially steered to the left when encountering centerline rumble strips;(6) therefore, future deployments may consider the use of centerline rumble strips that produce a distinct sensation and noise to avoid confusion with shoulder rumble strips.
  • Some States expressed a concern that 2.75-m (9-ft) lanes are too narrow for the treatment section of the lane narrowing concept. The American Association of State Highway and Transportation Officials (AASHTO) "Green Book"(10) indicates that lane widths of 2.75 m (9 ft) to 3.66 m (12 ft) are generally used, with a 3.66 m (12 ft) lane predominant on most high-type highways. One State indicated that it was required to submit a design exception to implement the lane narrowing concept, while three other States indicated that they were not required to submit a design exception to install 2.75 m (9 ft) or 3.05 m (10-ft) lanes as part of the deployment of the lane narrowing concept. For two-way, two-lane, rural highways, wider lanes provide desirable clearance between heavy vehicles in opposite directions.(10) For the lane narrowing concept, the presence of a painted median should provide adequate clearance for opposing vehicles, even with 2.75 m (9 ft) lanes. For those States that would still prefer wider lanes, the pavement markings (i.e., center and edge lines) could be placed in the rumble strips to increase the lane width without changing the placement of the rumble strip. An alternative method for increasing the effective lane width is to use a wider edge line and place the shoulder rumble strips closer to the shoulder.
  • Rumble stripes create a vertical surface that provides enhanced visibility during nighttime and wet weather conditions. However, the operational and safety benefit of rumble strips (pavement markings on side of rumble strip) versus rumble stripes (pavement markings within the rumble strip) has yet to be determined. At this time, States could deploy either rumble strips or rumble stripes based on their typical applications.
  • Provide cross-hatching in the median for the lane narrowing concept. Cross-hatching will better define the presence and width of the median. In some of the deployments of the lane narrowing concept, cross-hatching was not used in the median (figure 13). The treatment is much more conspicuous when cross-hatching is used in the median (figures 9, 10, 11, and 14).
  • Install rumble strips along both sides of the median and consider installing rumble strips across the entire width for narrow medians. Providing rumble strips across the entire median enhances the conspicuity of the treatment.
  • Add a speed advisory plaque to the warning sign (i.e., intersection ahead or lane narrowing) located prior to point 1 in figure 22. Data from North Carolina indicated that vehicles can travel up to 72.45 km/h (45 mi/h) on highways with 2.75 m (9 ft) lanes without crossing into the opposing lane. Where the posted speed is greater than 72.45 km/h (45 mi/h), it may be appropriate to provide a supplemental speed advisory plaque of 72.45 km/h (45 mi/h) prior to the deployment of the lane narrowing concept.
  • Add signing to warn of slowing vehicles. The crash data from Kentucky and Missouri indicated that rear-end crashes increased after the implementation of the lane narrowing concept. The intent of the lane narrowing concept is to reduce speeds on the major road; however, this may create greater speed differentials and may increase the chance of rear-end crashes. Advance signing could help to mitigate this issue.
  • Install both W5-1 and W2-1 warning signs prior to the treatment. Both signs are applicable to this concept and each sign has a specific meaning; W5-1 indicates that the lane narrows and W2-1 indicates that there is an intersection ahead. The use of a single sign does not convey both messages.

Minor Road Splitter Island Concept

  • Design the separator island as a traversable island. The AASHTO "Green Book"(10) indicates that curbed islands generally should not be used in rural areas and at isolated intersections unless the intersection is lighted and curbs are delineated. The policy also indicates that traversable islands may be preferable under certain conditions including: lightly developed areas, intersections where approach speeds are relatively high, areas with little pedestrian traffic, areas without fixedsource lighting, areas requiring significant snow plowing, and areas where signs and supports are not needed.(10) Nontraversable separator islands require additional right-of-way due to the required turning radius for large vehicles. If the island is traversable, the turning radius is reduced along with the required right-of-way. Aside from the reduced right-of-way, the fact that these materials can be snowplowed is an added advantage in the northern States. There are several materials currently available for use as traversable separator islands. Examples include materials similar to those used for traversable traffic circles and speed humps. Where traversable separator islands are used, the maximum height of the island is 76.2 mm (3 inches).
  • Use materials that can be prefabricated and nailed in place on-site. There are several examples of materials that can be prefabricated from recycled materials including some of those used for traffic calming devices. Prefabricated islands will reduce the disruption of traffic during deployment and may reduce implementation costs.
  • Reduce the width of the separator island. The current design template shows a width of 2.135 m (7 ft) for a non traversable separator island. The width could be reduced to a minimum of 1.22 m (4 ft) provided that the surface area exceeds the minimum of 6.975 sq-m (75 sq-ft).(10)
  • Delineate the separator island or construct the island of colored material. The AASHTO "Green Book"(10) indicates that islands are sometimes difficult to see at night because of glare from oncoming headlights. Therefore, appropriate delineation should be provided to increase the conspicuity of the island. A similar 22 effect could be achieved by using colored materials to construct the island such as those used for traffic calming devices.
  • Use cross-hatching to better delineate the approach to the separator island for the minor road splitter island concept. As shown in figure 15, the centerline splits prior to the separator island and continues along either side. The deployment of the minor road splitter island concept in Lorton, VA, did not include crosshatching where the centerline splits prior to the island. Providing cross-hatching will further enhance the conspicuity of the separator island.

Summary

Lane Narrowing Concept

For the lane narrowing concept, the major points of this study can be summarized as follows:

Speed

  • After implementation, the average (5.64 km/h (3.5 mi/h)) and 85th-percentile (7.25 km/h (4.5 mi/h)) speed reductions were statistically significant at the 95-percent confidence level for all vehicles on the major approach with an approach speed of 64.4 km/h (40 mi/h) or greater.
  • After implementation, the average (7.084 km/h (4.4 mi/h)) and 85th- percentile (7.728 km/h (4.8 mi/h)) speed reductions for trucks on the major approach were statistically significant when the vehicles had an approach speed of 64.4 km/h (40 mi/h) or greater.
  • The reductions in the 85th-percentile speed appear to be greater than the reductions in the average speed for all vehicles as well as for trucks only.
  • The speed reduction is greater for trucks than for all vehicles combined.

Driver Behavior

  • Driver behavior data indicated a large number of rumble strip contacts on the major approaches after implementation.

Safety

  • Based on limited crash data, there appears to be a general reduction in the crash rate for total, fatal/injury, and angle crashes after implementing the lane narrowing concept, but there is an increase in the crash rate for rearend crashes.
  • A Wilcoxon Signed-Rank Test indicates that there is no significant difference between the total crash rates in the before and after periods; however, there is a significant reduction in fatal/ injury crashes.

Minor Road Splitter Island Concept

For the minor road splitter island concept, the major points of this study can be summarized as follows:

Speed

  • After implementation, the speed reductions on the minor approaches range from 4.83 to 33.81 km/h (3 to 21 mi/h) comparing speeds by time of day, averaging 17.39 km/h (10.8 mi/h) and 15.78 km/h (9.8 mi/h) for the northbound and southbound approaches, respectively.
  • After implementation, the average speed reductions were statistically significant at a 95-percent confidence level for both minor approaches.

Driver Behavior

  • Driver behavior improved slightly after implementation based on increased STOP sign compliance.

Safety

  • Based on limited crash data, there appears to be a general reduction in the crash rate for total, fatal/injury, angle, and rear-end crashes after implementation.

Conclusions

The general conclusion from this research is that positive operational and safety effects can be expected with the installation of concepts 1 and 2. The lane narrowing concept is shown to significantly reduce speeds on the major road approaches at rural, two-lane, TWSC intersections. The results are consistent across sites, which are combined in the main analysis. The minor road splitter island concept is shown to improve driver compliance as well as reduce speeds on the minor approaches at rural, two-lane, TWSC intersections. While the results for the minor road splitter island concept are based on just one site, the initial indications are promising.

Based on the limited after period, there appears to be a general reduction in crashes associated with the implementation of concepts 1 and 2. For the lane narrowing concept, total, fatal/injury, and angle crashes were reduced in the after period, but rear-end crashes increased at some sites. The apparent increase in rear-end crashes should be monitored as additional after data become available. For the minor road splitter island concept, the crash rate decreased for all categories in the after period. While the simple before-after method used in this study does not account for some issues related to safety analysis (e.g., regression-to-the-mean, temporal effects, and other changes that may occur other than the safety improvements themselves), it does provide a preliminary understanding of the safety effectiveness of concepts 1 and 2. Also, it may not be worth conducting a more rigorous evaluation until sufficient data are available for the after period.

It is likely that the lane narrowing concept will be most effective at rural, TWSC intersections with a relatively high posted speed (64.4 km/h (40 mi/h) or greater) on the major road. However, this concept may be applicable at urban and low-speed intersections. While the limited sample did not indicate a linear relationship between speed reduction and lane width, other research has shown a positive association between the two variables (i.e., speed decreases as lane width decreases). Positive features of the lane narrowing concept include highly visible pavement markings with cross-hatching in the median and rumble strips that cover the entire width of the median. For the minor road splitter island concept, it is likely that the implementation will be most effective at rural, TWSC intersections with STOP sign compliance or conspicuity issues on the minor approach. However, this concept may be applicable at other TWSC intersections as well.

Based on the relative low cost and initial effectiveness of concepts 1 and 2 with respect to operational and safety measures, it is expected that these strategies will prove to be cost-effective methods for improving intersection safety. However, more comprehensive analysis is needed before wide-scale implementation can be recommended. There is also a need to develop guidelines for where these strategies should be implemented.

References

  1. Kuciemba, S.R., and J.A. Cirillo, (1992). Safety Effectiveness of Highway Design Features, Volume V: Intersections. FHWARD- 91-048, Federal Highway Administration, Washington, D.C.
  2. Preston, H., R. Storm, M. Donath, C. Shankwitz, M. Gute, and R. Starr, (2007). Rural Intersection Crashes: Identifying Intersections for Intersection Decision Support, TRB 86th Annual Meeting Compendium of Papers CD-ROM.
  3. Preston, H., R. Storm, M. Donath, and C. Shankwitz, (2004). Review of Minnesota's Rural Intersection Crashes: Methodology for Identifying Intersections for Intersection Decision Support (IDS), Report No. MN/ RC-2004-31, Minnesota Department of Transportation.
  4. Campbell, B.N., J.D. Smith, and W.G. Najm, (2004). Analysis of Fatal Crashes Due to Signal and Stop Sign Violation, Report No. DOT HS 809 779, U.S. Department of Transportation.
  5. Shinar, D., E. McDowell, and T.H. Rockwell, (1974). Improving Driver Performance on Curves in Rural Highways through Perceptual Changes. The Ohio State University, Engineering Experiment Station, Report EES428B.
  6. Noyce, D. and V.V. Elango, (2004). Safety Evaluation of Centerline Rumble Strips: Crash and Driver Behavior Analysis. Transportation Research Record: Journal of the Transportation Research Board, No. 1862, National Research Council, Washington, D.C., pp. 44-53.
  7. Joint Crash Reduction Programme: Outcome Monitoring, (2001). New Zealand Land Transport Safety Authority.
  8. Aubin, H., J. Gambard, P. Lebreton, and M. Ninine. Analyse de la Securite en Carrefour Plan [Safety Analysis for At-Grade Intersections]. Service d'Etudes Techniques Des Routes et Autoroutes (SETRA), Paris, France, September 1984.
  9. United States Department of Transportation (USDOT), (2003). Manual on UniformTraffic Control Devices. Available online: http://mutcd.fhwa.dot.gov. Accessed: July 22, 2006.
  10. American Association of State Highway and Transportation Officials (AASHTO), (2004). A Policy on Geometric Design of Highways and Streets Washington, D.C.

Researchers — This study was performed by Vanasse Hangen Brustlin, Inc. (VHB); Principal Investigators Warren Hughes, Ram Jagannathan, and Frank Gross. For more information about this research, contact Joe Bared, FHWA Project Manager, HRDS-05 at (202) 493-3314, joe.bared@fhwa.dot.gov.

Distribution — This Summary Report is being distributed according to a standard distribution.

Availability — This Summary Report may be obtained from the FHWA Product Distribution Center by e-mail to report.center@fhwa.dot.gov, fax to (301) 577-1421, phone to (301) 577-0818, or online at https://www.fhwa.dot.gov/research/topics/safety/intersections/.

Key Words — Safety, low-cost, intersection, STOP-control, lane-narrowing, splitter island.

Notice — This document is disseminated under the sponsorship of the U.S. Department of Transportation in the interest of information exchange. The U.S. Government assumes no liability for the use of the information contained in this document. The U.S. Government does not endorse products or manufacturers. Trademarks or manufacturers' names appear in this report only because they are considered essential to the objective of the document.

Quality Assurance Statement — The Federal Highway Administration (FHWA) provides high-quality information to serve the Government, industry, and public in a manner that promotes public understanding. Standards and policies are used to ensure and maximize the quality, objectivity, utility, and integrity of its information. FHWA periodically reviews quality issues and adjusts its programs and processes to ensure continuous quality improvement.

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