Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home
Public Roads
Featuring developments in Federal highway policies, programs, and research and technology.
This magazine is an archived publication and may contain dated technical, contact, and link information.
Federal Highway Administration > Publications > Public Roads > Vol. 69 · No. 1 > ArticleTitle

Jul/Aug 2005
Vol. 69 · No. 1

Publication Number: FHWA-HRT-05-006

ArticleTitle

by Ann H. Do, Joseph E. Hummer, Jennifer L. Toole, and Nagui M. Rouphail

A new tool will help designers evaluate the level of service on shared-use paths.

“If you build it, they will come,” says Chuck Flink, president of Greenways Incorporated in Durham, NC. Originally uttered by the actor Kevin Costner in the movie, “Field of Dreams” (1989), Flink used the quote to describe the growing demand for new trails and footpaths in the United States. Shared-use paths, defined as paved offstreet travel ways for nonmotorized traffic, attract bicyclists, pedestrians, inline skaters, roller skaters, skateboarders, equestrians, wheelchair users, and others for purposes ranging from commuting to leisure and recreation.

A woman with young children is easily passed by a cyclist going the opposite direction on the new waterfront trail system in Portland, OR.
A woman with young children is easily passed by a cyclist going the opposite direction on the new waterfront trail system in Portland, OR.

As the popularity of shared-use paths grows, trail usage and mileage are increasing steadily. In fact, some urban trails attract hundreds or even thousands of users per hour during peak periods, and many are experiencing morning rush hours on weekdays and traffic jams on weekend afternoons. According to a survey conducted in 2000, the Capital Crescent Trail in the Washington, DC, metropolitan area handles an average of 400 to 500 users per hour during peak periods. And the Pinellas Trail, stretching from Tarpon Springs to St. Petersburg in Florida, attracts an estimated 90,000 users per month, according to the Florida Department of Environmental Protection.

“The potential for conflict, overuse, and misuse is growing as more and more people are using the trails,” says Dan Cross, project manager of Trailnet, a nonprofit organization based in St. Louis, MO, that is dedicated to promoting bicycle and pedestrian activities. “When you have a diverse group of users all using the same facility at the same time, it’s almost like having a tennis game going on in the middle of a baseball game. The walkers don’t like the bike riders because they go too fast, inline skaters get upset with pedestrians for walking two abreast, and the bike riders complain because inline skaters use too much of the trail when swinging their arms back and forth with each glide.”

When creating a shared-use path, designers and planners often begin with the question, how wide should the pathway be? That question nearly always raises more questions: Will the path ever need to be widened? Should different types of users be separated from one another? If designers specify trails that are wider than future use justifies, they potentially waste money that could have been used to construct more trail miles. If the trails are too narrow, user conflicts and the potential for collisions may lead to dissatisfaction among users and the need to consider expensive trail-widening projects.

In 2000 researchers at the Federal Highway Administration (FHWA), North Carolina State University, the Toole Design Group, and the University of North Carolina initiated a study, Evaluation of Safety, Design, and Operation of Shared-Use Paths, to develop a simple tool and guidance to help trail managers and planners optimize their designs for shared-use paths. The research involved four primary efforts. The first task was to develop theoretical equations with which to estimate users’ passing and meeting events on shared-use paths. Passing events occur when a bicyclist either overtakes or is overtaken by other path users while traveling in the same direction. Meeting events are when a bicyclist encounters trail users traveling in the opposite direction. Next, the research team collected field data on path operations to calibrate and validate the equations for conditions in the United States. The third task involved collecting data on the perceptions of path users and developing a model relating those perceptions to operational and geometric variables. Finally, the researchers developed a computer-based tool to estimate the level of service (LOS) on new or existing shared-use paths.

A separate structure carries this shared-use path across the Allegheny River in Pittsburgh, PA. While a pair of bicyclists takes a beak to enjoy the view, a family of pedestrians passes behind them. The generous path width and broad curvature improve sight distance and enhance safety.
A separate structure carries this shared-use path across the Allegheny River in Pittsburgh, PA. While a pair of bicyclists takes a beak to enjoy the view, a family of pedestrians passes behind them. The generous path width and broad curvature improve sight distance and enhance safety.

The Need for a New Model

The fourth edition of the Transportation Research Board’s Highway Capacity Manual 2000 (HCM) represented the first attempt to offer a procedure model for analyzing LOS for shared-use paths on a large scale in the United States. The model estimates the number of passing and meeting events encountered by a test bicyclist traveling at the mean speed of bicyclists on a test trail. Collecting data from the perspective of a bicyclist is analogous to collecting data on traffic volumes on a highway using a moving vehicle.

A number of limitations, however, make it difficult for designers to use the HCM procedure to resolve problems in path design. First, the procedure is based in part on field data from the Netherlands. U.S. paths typically are wider than those in Europe, and bicyclists in the United States tend to ride more often for recreation and not for commuting. Therefore, the team determined that a new model needed to be calibrated and validated for U.S. conditions, especially the weighting used for criteria in passing and meeting events.

Second, the HCM procedure does not consider passive passing events, which occur when a faster path user passes the test bicyclist. Also the old procedure assumes that path users do not impede one another’s movements, meaning that there is always adequate room for one bicyclist to pass another with no change in speed or lateral positioning. This is true only if the path is wide enough or the bicyclist encounters no opposing traffic during the passing maneuver. If passing is restricted, overtaking events will be delayed.

Another limitation is that the current LOS procedure accounts for pedestrians and bicyclists only. Today, shared-use paths accommodate runners, inline skaters, skateboarders, and other users. Further, the current procedure is based on single statistical values of bicyclist and pedestrian mean speeds. Designers working in areas where bicyclists and pedestrians may travel faster or slower need to have the ability to incorporate that information into their LOS estimates.

Finally, the HCM method is limited to analyzing two- and three-lane paths, where two-lane paths are specified as 2.5 meters (8 feet) wide and three-lane paths as 3 meters (10 feet) wide. No guidance exists for designers considering other widths and numbers of lanes. “Our rule of thumb has been to build 12-foot [3.6-meter]-wide trails,” Trailnet’s Cross says, “with no rhyme or reason, or research to back it up. It looked and felt more comfortable for users.”

Joggers like these are frequent users of the Schuylkill River Trail in Philadelphia, PA.
Joggers like these are frequent users of the Schuylkill River Trail in Philadelphia, PA.

Flink, who has been in the business of designing trails for more than 20 years, agrees, adding that a new LOS tool will help remove much of the guesswork. “In the past, we relied on the AASHTO [American Association of State Highway and Transportation Officials] guidance that calls for 10-foot [3-meter]-wide trails,” he says, “but field experience says that’s often not wide enough. Some people go with 12 feet [3.6 meters], while others choose 14 feet [4.3 meters]. But that’s not based on scientific fact, just observation. Anything that applies more science and math to the topic will help us out a lot.”

Considering these limitations, FHWA and its partners set out to develop a new LOS procedure and produce a tool that designers can use to evaluate the operational effectiveness of shared-use paths, given various traffic volumes or forecasts and a few geometric parameters.

This trail features separate paved lanes for users to spread out or segregate themselves according to travel speed.
This trail features separate paved lanes for users to spread out or segregate themselves according to travel speed.

Collecting Data

The theoretical framework for estimating meeting and passing events on a shared-use path is based on the theory of supply and demand from the perspective of a bicyclist traveling at the average speed of bicyclists on the path. The project team developed equations from basic traffic flow principles to estimate the frequency for meetings, active passes (passing a trail user moving in the same direction), passive passes (being passed by another trail user moving in the same direction), and delayed passes (being unable to pass a trail user moving in the same direction because of oncoming traffic or encountering trail users traveling two or three abreast).

Next the project team identified a method to collect data on meeting and passing events to validate the models. The chosen solution involved mounting a video camera to a bicycle helmet to record data as a team member peddled through selected segments of shared-use paths. The video camera recorded the number of meetings, the number of passes accomplished, and the number of delayed passes. At the same time, another team member counted the number of users of each mode in each direction moving past the midway point of the segment. This exercise provided the necessary volume data. The day before conducting the moving bicycle study, the team determined the desired mean speed for bicycles by recording a sample of speed observations manually using a stopwatch.

Signs like this one encourage pedestrians and slower users to stay to the right, allowing faster users to pass them safely on the left.
Signs like this one encourage pedestrians and slower users to stay to the right, allowing faster users to pass them safely on the left.

Selecting Test Sites

The project budget limited data collection to 10 U.S. cities, and the team used a number of criteria to select candidate sites. First, the team sought paths that represent the diversity of regions in the United States. To reduce travel costs, it selected cities with two or more sites. In addition, the sites needed to contain a variety of geometric characteristics, moderate to high traffic levels at least sometimes, and 0.8-kilometer (0.5-mile)-long segments with no intersections or turnouts, assuring uninterrupted flow.

The team assembled a preliminary list of 37 cities with paths that meet some or all of the criteria and submitted a questionnaire to path managers to help determine the suitability of their respective sites. Based on the responses, the researchers identified 10 cities that provided the best possible chances to satisfy the criteria. The final list of 15 trails represents cities in all regions of the United States and many of the best-known trails in those areas. The trails are located in urban and suburban environments that include parks, lakes, beaches, highways, and downtown areas. Trail widths range from 2.4 to 6.1 meters (8 to 20 feet), and some paths were marked with centerlines.

The team collected data primarily on weekends from July 2001 to March 2002. In total, the researchers completed 771 runs with an average of about 60 runs at most trails.

Locations and Trail Names
Location
Trail Name
Marin County, CA
Mill Valley-Sausalito Pathway
Santa Monica, CA
South Bay Trail
Washington, DC
Capital Crescent Trail
Dunedin, FL
Honeymoon Island Trail (Dunedin Causeway)
Dunedin, FL
Pinellas Trail
Chicago, IL
Lakefront Trail
Arlington, MA
Minuteman Bikeway
Boston, MA
Dr. Paul Dudley Bike Path
St. Louis, MO
Forest Park Trail
St. Louis County, MO
Grant’s Trail
Raleigh, NC
Lake Johnson Trail
Dallas, TX
White Creek Trail
Dallas, TX
White Rock Lake Trail
Vienna, VA
Washington and Old Dominion Trail
Redmond, WA
Sammamish River Trail
This table shows the location of the 15 trails in 10 cities where the project team collected data on bicycle passing and meeting events. Source: FHWA.

Analyzing the Data

The two main objectives of the data analysis were to develop average and default values for key parameters that the team could use in a procedure for estimating LOS and to validate the theoretical framework developed earlier. The average and default values collected included speeds, volumes, mode splits (the breakdown of different types of users), a peak hour factor, percentages of user groups occupying two lanes, and distances needed by bicyclists to make passes. To validate the theoretical framework, the researchers compared the number of meeting and passing events recorded on the trails against the numbers predicted using the model.

The team recorded speeds using a stopwatch. The goal was a sample of at least 30 free-flowing bicyclists and as many other path users as possible during the time the researchers were recording the test bicyclist’s speeds.

These four pedestrians are walking side by side down a shared-use path in Santa Barbara, CA. The researchers found that in 36 percent of the cases studies, pedestrians occupied two lanes, potentially causing a faster users, such as a bicyclist, to slow down and delay passing.
These four pedestrians are walking side by side down a shared-use path in Santa Barbara, CA. The researchers found that in 36 percent of the cases studies, pedestrians occupied two lanes, potentially causing a faster user, such as a bicyclist, to slow down and delay passing.

Based on up to 8 hours of observation at relatively busy times on 15 paths, the team estimated an average volume of just over 400 users per hour. Adult bicyclists made up an average of 56 percent of the users, with pedestrians averaging about 18 percent, runners about 13 percent, skaters about 10 percent, and child bicyclists about 3 percent. Although some trail professionals who use the model may have detailed counts or forecasts for the volume of users by mode on their paths, most will need to rely on default values.

The peak hour factor is an important consideration in capacity and LOS calculations to adjust for peaking of traffic within the hour of interest. The team determined the peak hour factor by dividing the hourly volume by four times the volume of users in the peak 15-minute time period within that hour. It is important that designers consider a peak hour factor for trails because pedestrian and bicyclist peaks tend to be higher than motor vehicle peaks. Therefore, a trail may feel more crowded to its users than a raw volume count or forecast would convey, and designers need to account for that.

Another default value necessary as an input in the delayed passing procedure was the proportion of users moving along the path while occupying two lanes. Usually, a group of two users moving together will occupy two lanes. In many cases, however, a group of two or more users occupy one lane because they move in single file, or one user walks or rides off the path, or the users walk or ride side by side but very close to each other. To collect these data, the team reviewed 21 hours of videotape from the helmet camera and observed that pedestrians most often occupied two lanes, followed by runners, skaters, and bicyclists.

The distance needed by a bicyclist to pass another trail user is another key factor. To collect these data, the team reviewed 50 runs for which high-quality video with passing maneuvers was available on five paths. The overall average for the case of the test bicyclist passing a pedestrian was 18 meters (60 feet) when the bike was traveling from 18 to 23 km/h (11 to 14 mi/h).

After comparing the meeting and passing data collected in the field to the predictions from the models, the team determined that the estimates from the models proved to match the field data reasonably well. The next step was to gather input from trail users.

User Perceptions

Although current LOS criteria are based on the expert opinions of the Transportation Research Board’s Highway Capacity and Quality of Service Committee, the team chose to solicit user perceptions to enhance the credibility of the new model. The purpose was to develop a measure of bicyclists’ perceptions of hindrance in terms of their sense of comfort and freedom to maneuver on the trail.

To gather user perceptions, the team recruited 105 volunteers from bicycle and trail user groups from the Raleigh, NC, and Washington, DC, metropolitan areas. The volunteers viewed 36 video sequences from 10 of the 15 trails and rated their perceptions of the facility conditions based on four characteristics: lateral spacing (side-to-side maneuvering room), longitudinal spacing (front-to-back maneuvering room), ability to pass, and the overall perception of comfort and freedom to maneuver. The 60-second clips represented video segments that spanned the range of geographic locations, trail widths, and geometries. To rank the qualitative responses, the project team created a five-point scale, where 1 is bad, 2 is poor, 3 is fair, 4 is good, and 5 is excellent.

Based on the participants’ feedback, the project team logged a total of 15,000 observations and used the data to determine the trail characteristics (operational, geometric, or contextual) that factored significantly into the bicyclists’ perceptions of the four freedom-to-maneuver characteristics noted above, and to what degree. Analysis of the respondents’ comments showed that the variables related to path operations and trail width strongly affected the overall quality of the users’ experiences. The recommended model for overall rating therefore included variables for path width, the number of meeting and passing events, and the presence of a centerline.

The researchers found that the presence of a solid or dashed centerline stripe, like the one shown here, appears to make bicyclists feel less comfortable overtaking slower users. This finding supports the intent of providing a centerline, which is clearly delineate two opposing travel lanes and encourage increased caution in making passing maneuvers.
The researchers found that the presence of a solid or dashed centerline stripe, like the one shown here, appears to make bicyclists feel less comfortable overtaking slower users. This finding supports the intent of providing a centerline, which is clearly delineate two opposing travel lanes and encourage increased caution in making passing maneuvers.

Developing the LOS Procedure

A review of the literature enabled the project team to evaluate existing LOS tools from the standpoints of ease of use, accessibility of the format, inputs required, nature of the outputs, type of scale used, and other factors. “A primary goal of the project was to produce a tool that is easy for a variety of practitioners to use, even if they do not have a background in transportation or engineering,” says Michael F. Trentacoste, director of the FHWA Office of Safety Research and Development. “Trail designers, landscape architects, and park planners should be able to enter data and interpret the outputs as effectively as a transportation engineer.”

In addition to overcoming the limitations of existing methods, the researchers established a number of other objectives for the bicycle LOS procedure. The tool would use an AÐF grading scale and be applicable to trails with widths ranging from 2.4 to 6.1 meters (8 to 20 feet), the full range of the study trails.

Below is the equation for the final model that predicts the overall LOS rating on a scale of 1 to 5, where “E” means weighted events (meetings plus 10 times the active passing events) per minute, “RW” is the reciprocal of path width (that is, 1 divided by path width, in feet), “CL” is equal to 1 if the trail has a centerline or 0 if the trail has no centerline, and “DPF” stands for an adjustment for the delayed passes factor.

Bicycle LOS Score = 5.446 - 0.00809(E) - 15.86(RW) - 0.287(CL) - (DPF)

The variables in the model correspond to the framework of the existing LOS method in the HCM, except for the centerline variable, a heavier weight on passing events, and the delayed passes factor. The scores generated fall in the same range as the perception ratings provided by the respondents: 1 equates to “bad” with 5 equal to “excellent.” The equation requires only four inputs from the tool user: one-way user volume per hour, mode split percentages for the five common path user groups, path width, and the presence of a centerline.

The researchers programmed the model into a Microsoft® Excel® spreadsheet. A one-page user interface links the necessary inputs with the factors in the equations and displays the equation outputs as LOS results. The user interface provides the LOS calculation based on the volume, mode split, and width of the typical study trail. Four additional rows provide space for developing separate scenarios for comparison purposes. Designers can easily save and print the results.

An accompanying user’s guide provides practitioners with an overview of the research, a discussion of LOS procedures, instructions on how to use the spreadsheet and interpret the results, and two fictional case studies.

Shared-use paths accommodate a variety of users, including bicyclists, pedestrians, inline skaters, strollers, and wheelchair users. These bicyclists and inline skaters are sharing a path on the Schuylkill River Trail in Philadelphia, PA.
Shared-use paths accommodate a variety of users, including bicyclists, pedestrians, inline skaters, strollers, and wheelchair users. These bicyclists and inline skaters are sharing a path on the Schuylkill River Trail in Philadelphia, PA.

Practical Applications

The model is applicable to a variety of problems related to trail planning and design, including overcrowding and accommodating diverse user groups. It can be especially useful for planning and design tasks where trail managers need quantitative measures to augment qualitative criteria to strengthen the basis for making decisions about trail design.

Potential uses include planning appropriate widths and cross sections for new trails, evaluating the LOS provided on existing trails, guiding the design of improvements for existing trails where additional capacity is needed, determining how many additional users a trail may be able to serve given a minimum LOS threshold, and evaluating LOS for specific timeframes, such as weekday morning or evening periods when commuting trips are heaviest. Practitioners also can use the tool to determine LOS at a particular location on a trail, such as at a narrow pinch point, in an area with unusually high volume, or in an area experiencing many reported user conflicts. “This study will come in handy for determining the best planning to accommodate all trail users comfortably,” Trailnet’s Cross says.

Finally, the data gathered on the 15 study trails provide examples of peak user volumes and mode splits from a variety of urban trails. These data may prove helpful to practitioners who want to forecast future use of shared trails.

Locations and Trail Names
Location
Trail Name
Marin County, CA
Mill Valley-Sausalito Pathway
Santa Monica, CA
South Bay Trail
Washington, DC
Capital Crescent Trail
Dunedin, FL
Honeymoon Island Trail (Dunedin Causeway)
Dunedin, FL
Pinellas Trail
Chicago, IL
Lakefront Trail
Arlington, MA
Minuteman Bikeway
Boston, MA
Dr. Paul Dudley Bike Path
St. Louis, MO
Forest Park Trail
St. Louis County, MO
Grant’s Trail
Raleigh, NC
Lake Johnson Trail
Dallas, TX
White Creek Trail
Dallas, TX
White Rock Lake Trail
Vienna, VA
Washington and Old Dominion Trail
Redmond, WA
Sammamish River Trail
This table shows the location of the 15 trails in 10 cities where the project team collected data on bicycle passing and meeting events.
Source: FHWA.

Ann H. Do is a research highway engineer at the FHWA Turner-Fairbank Highway Research Center in McLean, VA. She has served as the program manager for research on pedestrian and bicyclist safety since 2001, specializing in evaluations of safety measures, human factors engineering, and geometric design. She designs and manages research projects and provides technical assistance and guidance to other FHWA offices and State and local transportation agencies on pedestrian and bicyclist safety.

Joseph E. Hummer, Ph.D., P.E. is a professor of civil engineering at North Carolina State University. He has researched and taught transportation operations, safety, and design for 16 years. He was a member of the first class of FHWA graduate research fellows in 1984. Hummer has a Ph.D. from Purdue University and M.S. and B.S. degrees from Michigan State University.

Jennifer L. Toole is the president of Toole Design Group, a planning and engineering firm that specializes in multimodal transportation. Toole has more than 15 years of experience serving as an expert consultant on bicyclist and pedestrian projects throughout the United States. She served three terms as president of the Association of Pedestrian and Bicycle Professionals.

Dr. Nagui M. Rouphail is the director of the Institute for Transportation Research and Education at North Carolina State University. Rouphail’s previous work on pedestrian and bicyclist LOS has been adopted in the 2000 Highway Capacity Manual procedural chapters on signalized intersections, pedestrians, and bicyclists. He currently serves on the team leading National Cooperative Highway Research Program (NCHRP) Project 3-78 on providing crossing solutions for visually impaired pedestrians at roundabouts and high-speed turn lanes.

The final report, Shared-Use Path Bicycle Level of Service Calculator: A User’s Guide, and a spreadsheet containing the LOS calculation tool will be available in fall 2005 at www.tfhrc.gov. For more information, please contact Ann Do at 202-493-3319 or ann.do@fhwa.dot.gov.

ResearchFHWA
FHWA
United States Department of Transportation - Federal Highway Administration