Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
12 Observational Field Studies This section describes the observational field studies con- ducted to evaluate the allocation of roadway width on both bicyclistsâ and motoristsâ lateral positioning, taking into consideration various roadway characteristics. The gen- eral methodology of the observational field study involved installing temporary lane line markings to delineate bicycle lanes at midblock locations. After a period of time to observe behaviors of bicyclists and motorists, the temporary lane line markings were removed, and new temporary lane line mark- ings were installed along the same midblock location, varying the width of the bicycle lane and in some cases the width of the parking lane. The behaviors of bicyclists and motorists were then observed under the new condition. This process was repeated such that several bicycle lane widths were evalu- ated at five midblock locations. The scenarios included stan- dard and buffered bicycle lane designs. All of the study sites had level (or nearly level) grades. A supplemental grade study was conducted to determine how much bicyclists sway or wobble while pedaling on moderate to steep upgrades to evaluate the need for different design guidance for bike lanes on grades (see Section 4). This section is organized as follows. Section 3.1 briefly describes the site selection process for the observational field studies and presents a general description of the roadway char- acteristics of the study sites. Section 3.2 describes the study sce- narios evaluated at each site. Section 3.3 describes the general data collection methodology. Section 3.4 presents descriptive statistics, the analysis approach, and analysis results of the observational field studies. Section 3.5 summarizes the primary findings from the observational field studies. 3.1 Site Selection and Site Characteristics The research team contacted representatives in several urban areas throughout the United States to determine if the local transportation agencies/authorities were willing to cooperate in this research and to gather information on potential study sites in the respective cities. The focus was on identifying study locations in urban (and suburban) areas since these areas are where bicycle lanes are most often considered and implemented and a sufficient number of bicyclists are present for data collection and analysis purposes. The nature of this research was highly dependent on finding local trans- portation agencies/authorities willing to work with the research team and finding appropriate study sites in the respective cities. Study sites in Cambridge (MA) and Chicago (IL) were selected for inclusion in the research. Study sites in each city were chosen to be as representative as possible of the range of characteristics at typical sites where bicycle lanes are normally planned or installed. When work- ing with the local highway agencies to identify potential data collection sites, sites where bicycle lanes were already planned for installation or were being considered were identified as high-priority locations for inclusion in the study. The road- way characteristics that factored most into the site selection process were: ⢠Bicycle volume, ⢠Traffic volume, ⢠Vehicle mix (i.e., percent trucks), ⢠Lane width or total roadway width, ⢠Presence/absence of on-street parking, ⢠Posted speed limit, and ⢠Grade. With the exception of bicycle volume (for which it was critical to find locations with a sufficient level of bicyclists for data collection and analysis purposes), it was desirable to find sites covering a range of these roadway characteris- tics to draw conclusions and recommendations about each, but some compromises had to be made. For example, all of the potential sites identified during the site selection pro- cess had a posted speed limit of 30 mph, but speed limits of S E C T I O N 3
13 30 to 35 mph are common on many streets in urban and suburban areas. So although potential data collection sites were not found covering a range of speed limits, the sites that were identified had speed limits common to many streets in urban and suburban areas and were typical of locations where bicycle lanes are installed or are considered for installation. Five sites were included in the observational studyâthree sites in Cambridge (Massachusetts Avenue, Prospect Street northbound, and Prospect Street southbound) and two sites in Chicago (Division Street and Clark Street). Table 5 pres- ents site characteristic information for each site. The traffic volumes of the sites ranged from approximately 15,000 to 29,000 vehicles per day (vpd), and the percentage of trucks in the vehicle mix ranged from 2% to 20%. Three of the sites had on-street parking, and two did not. For sites with on- street parking, the width of the travel lane adjacent to the bicycle lane ranged from 10 to 12 ft; and for the two sites where on-street parking was prohibited, the widths of the travel lanes without any bicycle lanes installed were 16 and 18 ft. The speed limit at each site was 30 mph. All of the sites had a level, or nearly level, grade. 3.2 Study Scenarios At each study site, several scenarios were evaluated by vary- ing the width of the bicycle lane. At study sites with on-street parking, in most cases the vehicle travel lane width was held constant. The longitudinal lane line separating the vehicle travel lane from the bicycle lane was installed using either paint or thermoplastic pavement marking and was not moved. Only the longitudinal lane line closest to the parking lane (or curb) was installed using temporary pavement marking material to vary the width of the bicycle lane. In Chicago, the bicycle lane widths varied from 4 to 6 ft, and the parking lane width varied from 7 to 9 ft. For two scenarios in Chicago, a buffered bicycle lane was also evaluated. On Clark Street there was a 2-ft buffer space between a 7-ft parking lane and a 5-ft bike lane. On Divi- sion Street there was a 2-ft buffer space on either side of a 4-ft bike lane. In Cambridge, the bicycle lane widths varied from 3.5 to 5 ft; the parking lane width was held constant at 7 ft; and for the narrower bicycle lane widths of 3.5 ft and 4 ft, a buffer space separated the bicycle lane from the parking lane. These study scenarios are depicted in Figure 1. The study scenarios are numbered for easy referencing throughout the report. Prospect Street was the only study location without on- street parking. Consistent with the other sites, a longitudinal lane line separating the vehicle travel lane from the bicycle lane was installed using temporary pavement marking material. This lane line was moved to vary the width of the bicycle lane. Bicycle lane widths of 4 ft and 5 ft were evaluated along both directions of Prospect Street. The bike lane width was measured from the center of the longitudinal lane line separating the vehicle travel lane from the bicycle lane to the face of the curb. No gutter pan was present in either direction of travel. In addition, for both directions of travel along Prospect Street, data were collected without any bicycle lane lines present (i.e., simply a wide curb lane). These study scenarios are depicted in Figure 2. City Chicago Chicago Cambridge Cambridge Cambridge Street name Clark St. Division St. Mass. Ave. Prospect St. Prospect St. Direction NB EB WB SB NB Begin cross street W. Shiller St. N. Washtenaw Ave. Wendell St. Hampshire Broadway End cross street W. Burton Pl. N. Rockwell St. Garfield St. Broadway Hampshire Traffic volume (ADT) 14,800 16,600 29,000 15,000 15,000 Percent trucks 16%3 20%3 7% 2% 2% Speed limit (mph) 30 30 30 30 30 Presence of on-street parking (Y/N) Y Y Y N N Average travel lane width (ft) 11 1 121 101 182 162 Number of lanes (directional) 1 1 2 1 1 Curb and gutter (Y/N) Y Y Granite curb, no gutter pan Granite curb, no gutter pan Granite curb, no gutter pan Grade Level Level Level Level Level 1 Average width of the travel lane adjacent to the bicycle lane during the study. 2 Average width of the travel lane without any bicycle lanes installed. 3 Most truck traffic consists of single-unit trucks. Note: NB = northbound, EB = eastbound, WB = westbound, SB = southbound, ADT = average daily traffic. Table 5. Roadway characteristics of data collection sites in Cambridge and Chicago.
14 Figure 1. Study sites and scenarios with on-street parking.
15 Table 6 summarizes the 17 study scenarios evaluatedâ 11 scenarios with on-street parking and 6 scenarios without on- street parking. Table 6 shows the widths of the travel lanes, bicycle lanes, and parking lanes (if applicable) for all study sce- narios. Figure 3 shows illustrations of the buffered bike lanes installed on Clark Street and Division Street in Chicago. For these two scenarios, temporary pavement marking materials were not used. The width of the buffer space was not included as part of the total width of the bicycle lane in the analysis. Also, the spe- cific designs varied depending on the location, as illustrated in Figure 1 and Figure 3. For example, the buffered bike lane on Division Street did not include a longitudinal lane line sepa- rating the diagonal cross hatching from the bicycle lane. On Massachusetts Avenue, there was no diagonal cross hatching within the buffer space. At each study site, temporary pavement markings were installed along one or two city blocks, approximately 300 to 600 ft in length. The temporary pavement marking material was 4-in. wide and white. Two bike lane symbols (and arrows) were painted in the bike lane using a stencil at approximately 10 ft and 200 ft downstream from the beginning cross street. The bicycle lane symbol and arrow were positioned such that they would approximately be in the middle of the narrowest lane. This allowed the symbol to be kept in the same location as the temporary lane lines were moved to vary the width of the bicycle lane. Given the site characteristics and the study scenarios, the ranges in the primary roadway and traffic characteristics analyzed in this research are as follows: ⢠Bike lane width: 3.5 to 6 ft ⢠Parking lane width: 7 to 9 ft ⢠Travel lane width: 10 to 18 ft ⢠Presence/absence of buffer space ⢠Traffic volume: 14,800 to 29,000 vpd ⢠Percent trucks: 2% to 20% Figure 2. Study sites and scenarios without on-street parking.
16 1 4-ft bicycle lane; 1-ft buffer area. 2 3.5-ft bicycle lane; 1.5-ft buffer area. 3 5-ft bicycle lane; 2-ft buffer area. 4 2-ft buffer area; 4-ft bicycle lane; 2-ft buffer area. Note: NB = northbound, SB = southbound, BL = bike lane. City, State Street Scenario Width (ft) Travel Lane Bike Lane Parking Lane Sites with On-Street Parking Cambridge, MA Massachusetts Ave. Y-01 10 5 7 Y-02 41 Y-03 3.52 Chicago, IL Clark St. Y-04 11 6 7 Y-05 5 8 Y-06 4 9 Y-07 10 Buffered3 7 Chicago, IL Division St. Y-08 12 6 7 Y-09 5 8 Y-10 4 9 Y-11 10 Buffered4 7 Sites without On-Street Parking Cambridge, MA Prospect St. (NB) N-1 11 5 N/A N-2 12 4 N-3 16 No BL Cambridge, MA Prospect St. (SB) N-4 13 5 N/A N-5 14 4 N-6 18 No BL Table 6. Location and description of study scenarios. Clark Street (Scenario Y-07) Division Street (Scenario Y-11) Figure 3. Buffered bike lanes in Chicago.
17 3.3 Data Collection Methodology For each study scenario, a video camera was positioned to record cyclist and motorist lateral position along the mid- block portion of the study section. Figure 4 through Figure 8 show the perspectives from the camera for the Massachu- setts Avenue, Clark Street, Division Street, Prospect Street (northbound), and Prospect Street (southbound) study sites, respectively. Cyclist and motorist behaviors were recorded during morning and afternoon peak periods when bicyclist exposure level was expected to be highest. Video data were collected from April into December during calendar years 2011 and 2012. No crashes were observed at any of the sites during the study. Reference markings were placed on the pavement within the bicycle lane (or near the curb on Prospect Street for the study scenario without a bicycle lane present). The reference markings were placed near the midblock portion of the study section and were used during video data reduction to ascertain cyclist and motor vehicle lateral position within the roadway cross section. The video camera was placed approximately 100 ft down- stream of the reference markings. The camera was positioned such that the reference markings and the cyclists passing them could be seen in the recorded video. The position and zoom of the camera were also such that the right tires of a vehicle passing a cyclist in the adjacent travel lane could be seen. During data collection, sketches were made of the proj- ect site, noting camera position, reference marking locations, 5-ft Bike Lane (Scenario Y-01) 4-ft Bike Lane (Scenario Y-02) 3.5-ft Bike Lane (Scenario Y-03) Figure 4. Camera perspective for observational field study scenarios on Massachusetts Avenue in Cambridge.
18 and lane widths (i.e., parking lane, bicycle lane, buffer space, and adjacent travel lane, as applicable). Motor vehicle speed, volume, and classification data were collected during the first scenario at each study site using traffic classifiers. For sites with on-street parking, the following measure- ments were taken hourly along the study location to gather parking data while video was being recorded: ⢠The distance between the curb face and the front right tire (i.e., passenger side) of each parked vehicle ⢠The distance between the curb face and the rear right tire (i.e., passenger side) of each parked vehicle ⢠The width of the rear bumper of each vehicle Empty parking spaces were also noted. Following video data collection, the recordings were viewed to collect the following measurements, based on the known lateral positions of the reference markings within the cross section of the roadway: ⢠Cyclistâs lateral position: The distance from the front tire of the bicycle to the curb face (at the instant the cyclist passed the reference markings). ⢠Lateral position of the nearest passing vehicle (in time) in the adjacent travel lane: The distance from the right tire (i.e., passenger side) of the passing vehicle to the curb face (at the instant the motor vehicle passed the reference mark- 6-ft Bike Lane (Scenario Y-04) 4-ft Bike Lane (Scenario Y-06) 5-ft Bike Lane (Scenario Y-05) Buffered Bike Lane (Scenario Y-07) Figure 5. Camera perspective for observational field study scenarios on Clark Street in Chicago.
19 ings). Note: because of the perspective angle and zoom of the camera, it was not feasible to measure the distance from the left tire (i.e., driver side) of the passing vehicle to the curb face to accurately gather data on passing vehicle encroachment into adjacent (motor vehicle) travel lanes. A final database was assembled that included the relative lateral positions of parked vehicles, bicyclists, and passing vehicles within the roadway cross section. The database was used to analyze the effect of critical roadway characteristics on lateral positions of the respective vehicles (i.e., parked vehicles, bicycles, and passing vehicles) within the parking lane, bicycle lane, and travel lane. 3.4 Data Analysis The data collected at the various sites under various striping scenarios were analyzed to determine whether selected roadway characteristics affect the placement of bicyclists and vehicles within the cross section of the roadway. This section presents basic descriptive statistics of the measurements collected in the field; the statistical analysis approach, including the definition of the dependent variables used for analysis; and the analysis results. 3.4.1 Descriptive Statistics Prior to analysis, the data underwent basic quality checks such as removing outliers and unreasonable field measurements 6-ft Bike Lane (Scenario Y-08) 4-ft Bike Lane (Scenario Y-10) 5-ft Bike Lane (Scenario Y-09) Buffered Bike Lane (Scenario Y-11) Figure 6. Camera perspective for observational field study scenarios on Division Street in Chicago.
20 (e.g., vehicles parked in the travel lane, bicyclist riding in the far left of the travel lane); in total, fewer than 2% of cyclist, passing vehicle, and parked vehicle records were excluded. The final database used for analyses included records for 4,965 bicyclists, 3,163 passing vehicles, and 994 parked vehicles. Of the field measurements collected at each site, the most relevant for the analysis, in addition to the roadway charac- teristics described in Table 5, were: ⢠Total parked vehicle displacement from curb (sites with on- street parking only). This is equivalent to the distance of the left side (i.e., driver side) of the parked vehicle from the curb, calculated as the average distance of the front and rear right tires (i.e., passenger side) to the curb face plus the width of the parked vehicle. ⢠Distance of bike from curb. ⢠Distance of passing vehicle from curb, nearest in time to each cyclist measured. Motor vehicle speed data were also collected at each site but were not included in the analysis. Overall Relative Positioning of Vehicles and Cyclists. The raw data collected in Cambridge and Chicago were plotted separately for each of the scenarios described in 5-ft Bike Lane (Scenario N-1) 4-ft Bike Lane (Scenario N-2) Wide Curb Lane (No Bike Lane) (Scenario N-3) Figure 7. Camera perspective for observational field study scenarios on Prospect Street (northbound) in Cambridge.
21 Table 6. Figure 9 through Figure 13 show the position of parked vehicles, cyclists, and passing vehicles within their respective lanes. From left to right, where the origin indi- cates the curb, each plot shows the individual measure- ments, in feet, of: ⢠The average distance of the front and rear right tires (i.e., passenger side) to the curb face of each parked vehicle; ⢠The total parked vehicle displacement from the curb of each parked vehicle; ⢠The cyclistâs lateral position, based on the distance from the front tire of the bicycle to the curb and an assumed physical width of the bicycle of 2.5 ft (i.e., the middle point of the envelope represents the lateral position of the front bicycle tire, and the outside points of the envelop represent the posi- tions of the left and right ends of the handlebar for a typical adult bicyclist); ⢠The distance from the right tire of the passing vehicle to the curb; and ⢠The distance from the left tire of the passing vehicle to the curb, assuming a vehicle width of 7 ft based on the dimen- sions for a passenger car design vehicle in AASHTOâs A Policy on Geometric Design of Highways and Streets (com- monly referred to as the Green Book; AASHTO, 2011, Table 2-1b). 5-ft Bike Lane (Scenario N-4) 4-ft Bike Lane (Scenario N-5) Wide Curb Lane (No Bike Lane) (Scenario N-6) Figure 8. Camera perspective for observational field study scenarios on Prospect Street (southbound) in Cambridge.
22 Figure 9. Measurements taken on Massachusetts Avenue in Cambridge (assumed 7-ft width for passing vehicle).
23 Figure 10. Measurements taken on Clark Street in Chicago (assumed 7-ft width for passing vehicle).
24 Figure 11. Measurements taken on Division Street in Chicago (assumed 7-ft width for passing vehicle).
25 Figure 12. Measurements taken on Prospect Street (northbound) in Cambridge (assumed 7-ft width for passing vehicle).
26 Figure 13. Measurements taken on Prospect Street (southbound) in Cambridge (assumed 7-ft width for passing vehicle).
27 In each plot, the data are sorted by the cyclistâs lateral posi- tion, with the minimum distance from the curb lowest on the y-axis and the maximum highest on the y-axis. This effectively creates a cumulative distribution of the cyclistâs position relative to the curb in each graph. Thus, the measure- ment number on the y-axis is not an indication of increasing measurement but simply the order of the measurement in the database after the data were sorted by the cyclistâs distance from the curb. As such, the maximum number on the y-axis represents the sample size. Assuming a vehicle width (excluding mirrors) of 7 ft for passing vehicles, several of the figures suggest that encroach- ment of passing vehicles into adjacent (motor vehicle) travel lanes to the left may be a concern. Encroachment of pass- ing vehicles into adjacent (motor vehicle) travel lanes to the left was not a performance measure that the research team focused on in the analyses (see Sections 3.4.2 and 3.4.3) for reasons described previously, but it deserves some level of attention here. In particular, Figure 9 (scenarios Y-01, Y-02, and Y-03) shows a high rate of vehicle encroachment into the adjacent travel lane. Note that Massachusetts Avenue was the only study site with two travel lanes in the same direc- tion of travel adjacent to the bike lane. All other study sites were two-lane streets. Also, it is important to note that there were a total of five study scenarios (Y-01, Y-02, Y-03, Y-07, and Y-11) where the travel lane adjacent to the bicycle lane was 10-ft wide. From Figures 10 and 11, assuming a vehicle width of 7 ft for passing vehicles, there was a much lower rate of vehicle encroachment of passing vehicles into adjacent (motor vehicle) travel lanes to the left on Clark Street and Division Street than on Massachusetts Avenue. However, based on the research teamâs field observations, Figures 9 through 13 may overestimate the rate of encroachment of passing vehicles into adjacent (motor vehicle) travel lanes to the left. Therefore, the same data in Figures 9 through 13 are repeated in Figures 14 through 18, this time assuming a vehicle width (excluding mirrors) of 5.67 ft (68 in.). This width is consistent with the average width of parked vehicles measured in the field and dimensions from a sampling of vehicle specifications for pas- senger vehicles for model years 2013 and 2014. The research team believes that Figures 14 through 18 more accurately represent the behaviors of passing vehicles observed during the field studies with respect to encroachment into adjacent (motor vehicle) travel lanes to the left. Total Displacement of Parked Vehicles. Basic statistics for this measurement at sites with on-street parking are pre- sented in Table 7. These include, for each scenario, the num- ber of parked vehicles measured, mean, standard deviation, relative standard deviation (standard deviation/mean, in per- cent), and four percentiles. Percentile values that exceed the parking lane width are highlighted in red. Figure 19 shows the distribution of this measurement in the form of box plots, across all scenarios, but separately for each parking lane width. Since a number of scenarios included buffered lanes of different widths, 7-ft parking lanes were subdivided accord- ing to the width and type of the buffer. Distance of Cyclists from Curb. Basic statistics for this measurement at all sites are presented in Table 8. These include, for each scenario, the number of cyclists measured, mean, standard deviation, relative standard deviation, mini- mum and maximum distances, and five percentiles. Figure 20 through Figure 24 show this measurement in the form of histograms, separately for each scenario. The positions of the parking lane (where present), buffer space (where pres- ent), bike lane (where present), and travel lane are indicated on each plot. Table 9 (left half) shows the spread of bicyclist lateral positions, separately for each scenario. Here, the spread of bicyclist lateral positions is calculated as the distance between the 5th- and 95th-percentile bicyclist positions. For example, for scenario Y-01 (i.e., the 5-ft bike lane on Massachusetts Avenue), the 5th-percentile bicyclist posi- tion is at 9.2 ft and the 95th-percentile bicyclist position is at 11.7 ft. Thus, the spread of bicyclist lateral positions is 2.5 ft (11.7 ftâ9.2 ft). The right half of Table 9 shows the average spread of bicyclist lateral positions calculated: (1) by bike lane width, separately across all sites with or without on-street parking, and (2) by bike lane width across all sites (note that bike lane widths of 3.5 and 4 ft were combined). The overall average across all sites is shown to be 2.7 ft. As expected, narrowing the bicycle lane appears to reduce the variability of bicyclist lateral positions (i.e., the spread of bicyclist lateral positions). Distance of Passing Vehicle from Curb. Basic statistics for this measurement at all sites are presented in Table 10. These include, for each scenario, the number of passing vehicles measured, mean, standard deviation, relative standard devia- tion, 5th and 10th percentiles, and median. Figure 25 through Figure 29 show this measurement in the form of histograms, separately for each scenario. The positions of the bike lane (where present) and travel lane are indicated on each plot. A few facts about the study sites are worth highlighting. First, the narrowest travel lane width included in the research was 10 ft; this is the case for all scenarios on Massachusetts Avenue and the buffered bike lane scenarios on Clark Street and Division Street. Second, Massachusetts Avenue was the only study site that included two travel lanes in the same direction of travel as the bicycle lane. All other study sites had only a single travel lane in the same direction of travel as the bicycle lane. (text continues on page 41)
28 Figure 14. Measurements taken on Massachusetts Avenue in Cambridge (assumed 5.67-ft width for passing vehicle).
29 Figure 15. Measurements taken on Clark Street in Chicago (assumed 5.67-ft width for passing vehicle).
30 Figure 16. Measurements taken on Division Street in Chicago (assumed 5.67-ft width for passing vehicle).
31 Figure 17. Measurements taken on Prospect Street (northbound) in Cambridge (assumed 5.67-ft width for passing vehicle).
32 Figure 18. Measurements taken on Prospect Street (southbound) in Cambridge (assumed 5.67-ft width for passing vehicle).
33 Street (City) Parking Lane Width (ft) Bike Lane Width (ft) Scenario Number of Vehicles Measured Parked Vehicle DisplacementâDistance from Curb (ft) Mean Standard Deviation Relative Standard Deviation (%) Percentiles5 Median 85th 90th 95th Massachusetts Ave. (Cambridge) 7.0 5.0 Y-01 145 5.9 0.7 12.3 5.9 6.6 6.8 6.9 4.01 Y-02 72 5.8 0.5 8.8 5.7 6.3 6.4 6.8 3.52 Y-03 87 6.2 0.5 8.8 6.2 6.8 6.9 7.1 Clark St. (Chicago) 7.0 6.0 Y-04 41 6.2 0.4 7.0 6.3 6.6 6.7 6.9 8.0 5.0 Y-05 126 6.5 0.5 7.6 6.5 7.1 7.3 7.4 9.0 4.0 Y-06 145 6.6 0.7 10.3 6.5 7.2 7.4 7.8 7.0 Buffered3 Y-07 84 6.5 0.4 6.9 6.5 7.0 7.1 7.2 Division St. (Chicago) 7.0 6.0 Y-08 71 6.9 0.4 6.4 6.9 7.1 7.3 7.7 8.0 5.0 Y-09 65 6.9 0.7 10.6 6.9 7.6 8.0 8.3 9.0 4.0 Y-10 90 6.8 0.5 6.8 6.8 7.2 7.4 7.6 7.0 Buffered4 Y-11 68 6.8 0.5 8.0 6.8 7.4 7.4 7.8 1 4-ft bicycle lane; 1-ft buffer area. 2 3.5-ft bicycle lane; 1.5-ft buffer area. 3 5-ft bicycle lane; 2-ft buffer area. 4 2-ft buffer area; 4-ft bicycle lane; 2-ft buffer area. 5 Percentile values that exceed the parking lane width are highlighted in red. Table 7. Descriptive statistics for parked vehicle displacement. White dot = mean; Star = extreme value; Gray box = mid 50% of data; C = Clark St.; D = Division St. 7 7 + 1 7 + 1.5 7 + 2 (C) 7 + 2 (D) 8 9 4 5 6 7 8 9 10 To ta l p ar ke d ca r d is pl ac em en t f ro m c ur b (ft ) Parking lane + buffer width (ft) N 257 Min 5 Mean 6.2 Median 6.2 Max 9.4 Std Dev 0.7 72 5 5.8 5.7 7.1 0.5 87 5 6.2 6.2 8.0 0.5 84 6 6.5 6.5 7.6 0.4 68 6 6.8 6.8 8.5 0.5 191 5 6.6 6.6 8.5 0.6 235 5 6.7 6.6 8.6 0.6 Figure 19. Distribution of total parked vehicle displacement by parking lane width.
Street (City) Bike Lane Width (ft) Scenario Number of Cyclists Measured Distance of Cyclist from Curb (ft) Mean Standard Deviation Relative Standard Deviation (%) Minimum Percentiles Maximum5th 10th Median 90th 95th Sites with On-Street Parking Massachusetts Ave. (Cambridge) 5.0 Y-01 280 10.4 0.8 7.6 7.1 9.2 9.5 10.3 11.3 11.7 12.5 4.01 Y-02 530 10.4 0.9 8.4 5.6 9.0 9.3 10.4 11.4 11.7 12.5 3.52 Y-03 327 10.3 0.9 8.3 7.7 8.9 9.3 10.3 11.5 11.8 12.5 Clark St. (Chicago) 6.0 Y-04 134 9.4 0.8 8.6 7.9 8.2 8.4 9.3 10.4 10.9 13.3 5.0 Y-05 259 10.0 0.8 8.0 8.0 8.6 8.9 10.0 11.0 11.2 12.7 4.0 Y-06 399 10.1 0.8 7.6 7.9 8.9 9.2 10.1 11.0 11.2 13.4 Buffered3 Y-07 473 10.6 1.0 9.0 6.8 9.0 9.4 10.6 11.8 12.2 12.9 Division St. (Chicago) 6.0 Y-08 306 10.1 1.1 10.5 7.0 8.6 8.9 9.9 11.6 11.9 13.4 5.0 Y-09 187 10.1 1.1 10.8 7.7 8.4 8.8 9.9 11.6 12.0 13.3 4.0 Y-10 337 10.5 1.0 9.3 7.7 9.2 9.3 10.4 11.9 12.2 13.1 Buffered4 Y-11 109 10.9 1.2 10.7 4.8 9.3 9.5 10.9 12.5 12.6 13.3 Sites Without On-Street Parking Prospect St.âNB (Cambridge) 5 N-1 243 2.6 0.8 29.0 0.2 1.6 1.9 2.5 3.7 4.2 5.3 4 N-2 305 2.4 0.5 21.9 1.2 1.7 1.8 2.3 3.1 3.5 4.5 No BL N-3 215 2.3 0.8 33.3 0.3 1.3 1.6 2.2 3.4 4.0 4.8 Prospect St.âSB (Cambridge) 5 N-4 281 2.3 0.7 29.0 1.0 1.3 1.6 2.2 3.2 3.4 5.3 4 N-5 301 2.2 0.6 27.5 0.0 1.4 1.5 2.1 3.0 3.3 4.1 No BL N-6 279 2.1 0.7 33.1 0.7 1.3 1.4 2.0 3.0 3.3 5.4 1 4-ft bicycle lane; 1-ft buffer area. 2 3.5-ft bicycle lane; 1.5-ft buffer area. 3 5-ft bicycle lane; 2-ft buffer area. 4 2-ft buffer area; 4-ft bicycle lane; 2-ft buffer area. Note: NB = northbound, SB = southbound, BL = bike lane. Table 8. Descriptive statistics for bike position from curb.
35 Pa rk in g La ne Bi ke L an e Pa rk in g La ne Bu ffe r Pa rk in g La ne Bu ffe r Bu ffe r Bi ke L an e Bu ffe r Bi ke L an e Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Figure 20. Distribution of distance of cyclists from curb on Massachusetts Avenue. Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Pa rk in g La ne Bi ke L an e Pa rk in g La ne Bi ke L an e Pa rk in g La ne Bi ke L an e Pa rk in g La ne Bu ffe r Bu ffe r Bi ke L an e Figure 21. Distribution of distance of cyclists from curb on Clark Street.
36 Pa rk in g La ne Bi ke L an e Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Pa rk in g La ne Bi ke L an e Pa rk in g La ne Bi ke L an e Pa rk in g La ne Bu ffe r Bu ffe r Bi ke L an e Bi ke L an e Bu ffe r Bu ffe r Tr av el L an e Figure 22. Distribution of distance of cyclists from curb on Division Street. Ve r ca l C ur b Ve r ca l C ur b Ve r ca l C ur b Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Figure 23. Distribution of distance of cyclists from curb on Prospect Street (northbound).
37 Ve r ca l C ur b Ve r ca l C ur b Ve r ca l C ur b Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Figure 24. Distribution of distance of cyclists from curb on Prospect Street (southbound). Street (City) Bike Lane Width (ft) Scenario Spread (95th-5th Percentile Positions) (ft) Parking Lane Bike Lane Width (ft) Average Spread (ft) Massachusetts Ave. (Cambridge) 5.0 Y-01 2.5 Yes 6.0 3.0 4.01 Y-02 2.7 5.0 3.0 3.52 Y-03 2.9 3.5â4.0 2.8 Clark St. (Chicago) 6.0 Y-04 2.7 No 5.0 2.4 5.0 Y-05 2.6 4.0 1.9 4.0 Y-06 2.3 No BL 2.4 Buffered3 Y-07 3.2 Division St. (Chicago) 6.0 Y-08 3.3 5.0 Y-09 3.6 4.0 Y-10 3.0 Bike Lane Width (ft) Average Spread (ft) Buffered4 Y-11 3.3 Prospect St.âNB (Cambridge) 5 N-1 2.6 4 N-2 1.8 6.0 3.0 No BL N-3 2.7 5.0 2.8 Prospect St.âSB (Cambridge) 5 N-4 2.1 3.5â4.0 2.6 4 N-5 1.9 No BL 2.4 No BL N-6 2.0 Overall average 2.7 1 4-ft bicycle lane; 1-ft buffer area. 2 3.5-ft bicycle lane; 1.5-ft buffer area. 3 5-ft bicycle lane; 2-ft buffer area. 4 2-ft buffer area; 4-ft bicycle lane; 2-ft buffer area. Note: NB = northbound, SB = southbound, BL = bike lane. Table 9. Spread of cyclist lateral positions.
38 Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Figure 25. Distribution of passing vehicle distance from curb on Massachusetts Avenue. Street (City) Bike Lane Width (ft) Scenario Number of Vehicles Measured Distance of Passing Vehicle from Curb (ft) Mean Standard Deviation Relative Standard Deviation (%) Percentiles 5th 10th Median Sites with On-Street Parking Massachusetts Ave. (Cambridge) 5.0 Y-01 162 15.2 1.0 6.6 13.5 14.1 15.2 4.01 Y-02 306 15.5 1.3 8.1 13.5 14.0 15.4 3.52 Y-03 204 15.3 1.0 6.7 13.5 14.2 15.3 Clark St. (Chicago) 6.0 Y-04 111 14.9 0.9 5.7 13.4 13.6 15.1 5.0 Y-05 200 15.3 1.2 7.9 13.3 13.7 15.4 4.0 Y-06 300 15.2 1.0 6.5 13.5 13.9 15.3 Buffered3 Y-07 284 15.4 1.0 6.3 13.7 14.1 15.6 Division St. (Chicago) 6.0 Y-08 25 15.5 0.9 6.1 13.6 14.0 15.5 5.0 Y-09 148 16.0 1.0 6.4 14.2 14.8 16.0 4.0 Y-10 118 15.9 0.8 5.1 14.3 14.9 16.0 Buffered4 Y-11 47 17.1 0.9 5.1 15.9 16.0 17.2 Sites Without On-Street Parking Prospect St.â NB (Cambridge) 5 N-1 207 7.8 1.1 14.6 5.9 6.4 7.9 4 N-2 241 7.7 1.0 12.6 6.0 6.6 7.9 No BL N-3 182 7.4 1.7 23.0 4.6 5.2 7.8 Prospect St.â SB (Cambridge) 5 N-4 185 8.1 1.1 13.8 6.3 6.8 8.2 4 N-5 226 8.1 1.3 16.3 5.9 6.5 8.3 No BL N-6 217 7.8 1.6 20.5 4.8 5.4 7.9 1 4-ft bicycle lane; 1-ft buffer area. 2 3.5-ft bicycle lane; 1.5-ft buffer area. 3 5-ft bicycle lane; 2-ft buffer area. 4 2-ft buffer area; 4-ft bicycle lane; 2-ft buffer area. Note: NB = northbound, SB = southbound, BL = bike lane. Table 10. Descriptive statistics for distance of passing vehicles from curb.
39 Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Figure 26. Distribution of passing vehicle distance from curb on Clark Street. Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Figure 27. Distribution of passing vehicle distance from curb on Division Street.
40 Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Figure 28. Distribution of passing vehicle distance from curb on Prospect Street (northbound). Bi ke L an e Tr av el L an e Bi ke L an e Tr av el L an e Figure 29. Distribution of passing vehicle distance from curb on Prospect Street (southbound).
41 A couple of points are worth noting with regard to the position of passing vehicles within the travel lane and relative to the bicycle lane: 1. At sites with travel lanes ranging in width from 10 to 14 ft that were adjacent to bicycle lanes, very few passing vehicles encroached into the bicycle lane, even from the narrow- est travel lane of 10 ft. The scenario in which the highest percentage of passing vehicles (approximately 5% to 10%) encroached into the bicycle lane involved a 10-ft travel lane. 2. For scenarios with the narrowest travel lane of 10 ft, half of the passing vehicles on Massachusetts Avenue were posi- tioned 3.3 ft or more from the bicycle lane, half on Clark Street were positioned 1.4 ft or more from the bicycle lane, and half on Division Street were positioned 2.2 ft or more from the bicycle lane. The type of vehicle was not recorded when measuring vehicle position of passing vehicles; how- ever, assuming an overall width of 7 ft for a passenger car based on design vehicle dimensions in the Green Book, the data suggest that about half of the vehicles encroached into the adjacent travel lane (in the same direction of travel) on Massachusetts Avenue. Fewer vehicle encroachments into the adjacent travel lane (in the opposite direction of travel) would have occurred on Clark and Division Streets. Possible reasons for the large difference in position of passing vehicles on Massachusetts Avenue as compared to Clark and Division Streets may be that (1) Massachusetts Avenue has two travel lanes in the same direction of travel while the others do not, so consequences may not be as severe when drivers encroach into an adjacent lane of traf- fic traveling in the same direction; and (2) the percent- age of trucks (and buses) on Clark and Division Streets is higher than that on Massachusetts Avenue, so drivers of wider vehicles might position their vehicles closer to the bicycle lane so as not to encroach into the travel lane in the opposite direction of travel. Motor Vehicle Speed Data. Traffic classifiers were used to collect motor vehicle speeds in the travel lanes at each of the data collection sites. Over a period of 1 to 3 days, speed data were collected in the respective direction of travel at a given study site. Table 11 shows the 85th-percentile speeds measured at each site. Due to the lack of variability in speeds, this variable was not included in subsequent data analyses. 3.4.2 Analysis Approach From all the field measurements pertaining to the position of parked and passing vehicles and cyclists relative to each other or the curb for the various scenarios, an appropriate single measurement was derived that could be used for analy- sis. The measurement developed was called âcentral position- ing.â This measurement was derived to reflect the relative position of the bike on the roadway, while accounting for both the presence and position of lane line markings on the roadway and the presence and behavior of parked and passing vehicles. This central positioning measure serves as the depen- dent variable in the statistical analysis discussed in the next section and was defined in a two-step process as follows. Define an Effective Bike Lane. Figure 30 illustrates how an effective bike lane in which the cyclist is positioned was defined. The top portion of Figure 30 is approximately to scale based on average widths of parking, bicycle, and travel lanes and the distribution of left tire displacement of parked cars observed in this study. Note that this effective bike lane is not meant to be a real bike lane nor does it imply a safe zone for the cyclist; it is simply a portion of the roadway defined so as to be able to perform the analysis. For streets with on- street parking, an effective bike lane is defined based strictly on the behavior of parked vehicles and passing vehicles. For streets where on-street parking is prohibited, an effective bike lane is defined based on the behavior of passing vehicles and the position of the curb. The following measurements and dimensions for each scenario were needed: ⢠85th, 90th, or 95th percentile of the total parked vehicle displacement distribution (from Table 7). ⢠Assumed passenger car open door width. (A 45-in. open car door width was selected to represent a typical open door width of a two-door passenger vehicle based on previous studies and field observations.) Street Number of Speed Measurements 85th-Percentile Speed (mph) Massachusetts Ave. 30,773 30 Clark St. 33,625 33 Division St. 26,548 33 Prospect St. â northbound 8,605 28 Prospect St. â southbound 6,738 29 Table 11. Summary of motor vehicle speeds at study sites.
42 ⢠Position of the left lane line of the bike lane (i.e., the lon- gitudinal lane line that separates the travel lane and the bike lane). ⢠5th percentile of the distribution of passing car distance from curb (from Table 10). The left and right edges of the effective bike lane were then defined using the rules shown in Table 12, depend- ing on whether the site had on-street parking or a marked bike lane. Three options were selected to define the right edge of the effective bike lane for sites with on-street parking and marked bike lanes to cover a range of cyclist safety, from conservative (using the 95th-percentile parked car displacement plus a door width) to less conservative (using the 85th-percentile parked car displacement plus a door width). All three options were used and their results compared in the analysis. A 45-in. open car door width was selected for use in defining the right edge of the effective bike lane. This open door width is based on data for two-door passenger cars for model years 1988 to 1990 reported by Pein (2003). The research team collected similar data on open door widths for several two-door passenger cars for model year 2013 and found that passenger car door dimensions have not changed significantly in the past two decades. Thus, 45 in. was a reasonable choice for representing the open door width for a two-door passenger car. Approximately 15% to 25% of passenger vehicles (including passenger cars, pick- ups, and sport utility vehicles) are two-door vehicles, while the majority of passenger vehicles are four-door vehicles (Kahane, 2003). The average open door width for four- door vehicles is approximately 38 in. as reported by Pein (2003) and verified by the research team for 2013 model year passenger vehicles. Therefore, a 45-in. open car door Site Type Scenario Left Edge of Effective Bike Lane Is The: Right Edge of Effective Bike Lane Is The: Without on-street parking Without marked bike lane N-3 and N-6 5th percentile of the distribution of passing car distance from curb Curb Without on-street parking With marked bike lane N-1, N-2, N-4, N-5 Left lane line of the bike lane Curb With on-street parking With marked bike lane Y-01 through Y-11 Left lane line of the bike lane Distance from curb of 85th, 90th, or 95th percentile of the total parked vehicle displacement distribution plus 45 in. to account for a fully opened car door Table 12. Rules used to define an effective bike lane. Figure 30. Illustrating the effective bike lane and central positioning.
43 width for use in the analysis is a conservative (i.e., strin- gent) choice for analysis purposes. The left edge of the effective bike lane was defined based on the left lane line of the bike lane (for sites with a bike lane) for the following reasons: ⢠In the presence of a bike lane, it was assumed (although not fully supported by the data) that bicyclists would not ride to the left of the bike lane, in the adjacent travel lane; nor would cyclists position themselves within the buffer space if a buffer is provided between the bike lane and travel lane. ⢠As illustrated in Figures 25 through 29, rarely did passing vehicles cross over into the bike lane from the adjacent travel lane. In only one scenario (Clark Stâbuffered bike lane) was the position of the 5th-percentile passing vehicle within the bike lane, and for this scenario, the position of the 5th-percentile passing vehicle was only 0.3 ft within the bike lane. If the position of the 5th-percentile passing vehicles was consistently (or even occasionally) within the bike lane, then it would have been reasonable to define the left edge of the effective bike lane based on the position of passing vehicles; however, this was not the case. Define the Central Positioning of the Cyclist. The bot- tom portion of Figure 30 illustrates how the âcentral posi- tioningâ measurement was defined relative to the effective bike lane. For each bike positioned inside or outside the effec- tive bike lane, two distances were defined: D distance between the left edge of effective bike lane and the cyclistâs position, and Left = D distance between the cyclistâs position and the right edge of effective bike lane. Right = From these two measurements, the final dependent variable was simply calculated as: Central positioning min D ,D .Left Right( )= DLeft and DRight can be either (1) both positive (i.e., the cyclist is within the effective bike lane) or (2) one positive and the other negative (i.e., the cyclist is either to the left or right of the effective bike lane, but never (3) both negative. The data will show that a very small percentage of cyclists (about 1%) ride to the left of the effective bike lane (i.e., in the travel lane), while a large percentage of cyclists (up to 45%) ride to the right of the effective bike lane (i.e., in the door zone of parked vehicles) while still riding in the marked bike lane. Statistical Methodology. A number of complemen- tary approaches were used to analyze the data to investigate whether selected roadway characteristics affect the placement of bicyclists and vehicles within the cross section of the roadway. 1. One approach consists of simply calculating the percent- age of cyclists that ride within the effective bike lane and comparing these percentages across the various scenarios. This was done using all three effective bike lane options based on either the 85th, 90th, or 95th percentile of the parked car displacement. While the percentages them- selves are not that relevant to the study conclusions, the objective is to assess whether these percentages are affected by the roadway layoutâthat is, the combination of travel lane width, parking lane width, buffer space, and bike lane width. These comparisons are made without regard to other roadway characteristics such as traffic volume and percent trucks. 2. The effect of parking lane width on the position of parked vehicles relative to the curb is investigated by means of a one-way analysis of variance (ANOVA). The dependent variable considered in the ANOVA is the total parked vehicle displacement, and the single factor is parking lane width (used as a categorical variable at seven levels, as shown in Figure 19). Each scenario (i.e., one of each of the 11 scenarios) is considered a blocking factor in the analy- sis, and each measured parked vehicle provides replication within each scenario. 3. Another complementary analysis consists of a more rig- orous statistical approach. An ANOVA is used to estimate the effect of roadway characteristics, such as traffic volume, percent trucks, presence or absence of a buffer, parking lane width, and travel lane width, on the calculated central posi- tioning (dependent variable). Each scenario is considered a blocking factor in the analysis, and each measured cyclist provides replication within each scenario. Following the ANOVA and depending on whether a factor is statistically significant, a number of relevant comparisons are made to estimate the effect of a particular roadway characteristic on central positioning. In all analyses, a 10% significance level is chosen. Bike lane width, although at first a logical factor to consider in the model, is not included in this model. For any given city block, travel lane width, parking lane width, and bike lane width are highly correlated since their sum is determined by the width of the roadway; therefore, two out of three widths are sufficient to define the roadway width. Additionally, the focus is on establishing the width of the bike lane given a certain situation in the field, and as such, it is preferable to not include bike lane width as a predictor variable in the model. 3.4.3 Analysis Results The analysis results are presented in the order discussed previously.
44 Percentage of Cyclists Riding Within the Effective Bike Lane. Using the three selected percentiles (i.e., 85th, 90th, and 95th) from the distribution of total vehicle displacement and assuming a 45-in. open car door width, the percentage of cyclists riding within each of the effective bike lanes was calculated. Naturally, the higher the percentile from the distribution, the fewer cyclists will ride within the effective bike lane. Table 13 displays the following statistics for each of the 11 scenarios with on-street parking and the six scenarios without on-street parking: ⢠Roadway conditions (columns 2 through 7) ⢠Location from curb to right edge of effective bike lane based on 85th, 90th, or 95th percentile of total parked vehi- cle position plus an assumed 45-in. open car door width (columns 8 through 10) ⢠Location from curb to left edge of effective bike lane (column 11) ⢠Width of effective bike lane based on 85th, 90th, or 95th per- centile of total parked vehicle displacement (columns 12 through 14) ⢠Percentage of cyclists within the effective bike lane based on 85th, 90th, or 95th percentile of total parked vehicle displacement (columns 15 through 17) To more thoroughly define the percentage of cyclists rid- ing within the effective bike lane, the widths of the effective bike lanes (shown in columns 12 through 14 of Table 13 for three percentiles of total parked vehicle displacement) need to be considered in conjunction with the cyclistâs operat- ing space. Figure 31 illustrates the critical dimensions for an upright adult bicyclist (AASHTO, 2012). The physical width of the bicyclist is 2.5 ft and is based on the physical width (95th percentile) of the handlebars. The minimum operating width of 4 ft is greater than the physical width occupied by the bicyclist because of natural side-to-side movement that varies with speed, wind, and bicyclist proficiency. The pre- ferred operating width of 5 ft allows for even more lateral clearance from nearby obstacles. When comparing the least conservative measure of the effective bike lane width (i.e., based on the 85th percentile of total parked vehicle displacement) to the physical width of a bicyclist, two of the scenarios evaluated on streets with on- street parking (i.e., scenarios Y-04 and Y-07) had an effective bike lane width greater than 2.5 ft (i.e., the physical width of a bicyclist), while for the other scenarios evaluated on streets with on-street parking, the effective bike lane width was less than the physical width of a typical adult bicyclist. Conceptu- ally this means that, for the majority of scenarios evaluated on streets with on-street parking, the effective bike lane is not wide enough to accommodate either the operating width (4 ft) or the physical width (2.5 ft) of a bicyclist. Also, the boundaries of the effective bike lane are not delineated on the roadway with pavement markings, so it is difficult for bi- cyclists to envision the effective bike lane and position them- selves in the center of it. As is shown later, most bicyclists position themselves to the left or right of the center of the effective bike lane and, in some cases, outside of the limits of the effective bike lane. For the scenarios evaluated on streets where on-street parking is prohibited, the effective bike lane width was always greater than or equal to the minimum oper- ating space (i.e., 4 ft) of a typical adult bicyclist. Since the effective bike lane widths for the majority of scenarios evaluated on streets with on-street parking were found to be less than the physical dimensions of a bicyclist, the decision was made to calculate the percentage of cyclists riding within the effective bike lane based on the position of the front bicycle tire rather than accounting for the physical, minimum, or preferred operating space of a bicyclist. This approach is consistent with the overall guiding principle of this research, which was to provide guidance on how wide a bicycle lane should be in cases where the decision to include a bicycle lane has been made. It should also be recognized that this approach is part of an effort to develop design guidelines that provide a balanced design to accommodate all roadway users. The primary findings based on the width of the effective bike lane and the percentage of cyclists positioned within the effective bike lane are as follows: ⢠For the majority of scenarios evaluated on streets with on- street parking, the effective bike lane widths were narrower than the physical width of a typical adult bicyclist, and for the scenarios evaluated on streets without on-street park- ing, the effective bike lane widths were always greater than or equal to the minimum operating space of a typical adult bicyclist. ⢠Across the scenarios with on-street parking, Massachusetts Avenue has the highest percentages of bicyclists that position themselves within the effective bike lane. ⢠On Clark and Division Streets, with the exception of the 4-ft bike lane scenario on Division Street, the percentage of bicyclists within the effective bike lane is considerably lower for scenarios without any type of buffer. ⢠In general, on streets with on-street parking, the highest percentages of bicyclists are within the effective bicycle lane when buffers are used. ⢠On Prospect Street (a street without on-street parking), there is very little difference among scenarios in terms of the percentage of bicyclists within the effective bike lane, and the percentage of bicyclists in the effective bike lane is very high (close to 100%). Effect of Parking Lane Width on Position of Parked Vehi- cles. The one-way ANOVA showed that parking lane width
Scenario Street (City) Width (ft) Of: Location (ft) from Curb of Right Edge of Effective Bike Lane Using: Location (ft) From Curb of Left Edge of Effective Bike Lane Effective Bike Lane Width (ft) Using: Percent Cyclists Within Effective Bike Lane Using: Parking Lane Buffer1 Bike Lane Buffer2 Travel Lane 85th Percentilea 90th Percentile 95th Percentile 85th Percentile 90th Percentile 95th Percentile 85th Percentile 90th Percentile 95th Percentile Sites with On-Street ParkingâA 45-in. Car Door Was Assumed Y-01 Massachusetts Ave. (Cambridge) 7 0.0 5.0 0 10 10.3 10.5 10.7 12.0 1.7 1.5 1.3 47.5 37.9 34.3 Y-02 1.0 4.0 10.0 10.2 10.6 12.0 2.0 1.8 1.4 63.6 58.1 38.1 Y-03 1.5 3.5 10.5 10.7 10.8 12.0 1.5 1.3 1.2 36.1 33.0 25.1 Y-04 Clark St. (Chicago) 7 0.0 6.0 0 11 10.4 10.5 10.7 13.0 2.6 2.5 2.3 10.4 7.5 4.5 Y-05 8 5.0 10.8 11.0 11.2 13.0 2.2 2.0 1.8 13.9 8.9 5.8 Y-06 9 4.0 11.0 11.2 11.6 13.0 2.0 1.8 1.4 11.5 5.8 1.8 Y-07 7 2.0 5.0 10 10.8 10.8 11.0 14.0 3.2 3.2 3.0 41.6 40.0 35.7 Y-08 Division St. (Chicago) 7 0.0 6.0 0 12 10.9 11.0 11.5 13.0 2.1 2.0 1.5 21.6 18.0 10.8 Y-09 8 5.0 11.4 11.7 12.0 13.0 1.6 1.3 1.0 12.3 7.0 3.7 Y-10 9 4.0 11.0 11.2 11.4 13.0 2.0 1.8 1.6 30.0 25.8 21.4 Y-11 7 2.0 4.0 2 10 11.1 11.2 11.6 13.0 1.9 1.8 1.4 43.1 39.4 29.4 Sites Without On-Street Parking N-1 Prospect St.â NB (Cambridge) 0 0.0 5.0 0 11 0.0 0.0 0.0 5.0 5.0 5.0 5.0 98.8 98.8 98.8 N-2 4.0 12 0.0 0.0 0.0 4.0 4.0 4.0 4.0 98.7 98.7 98.7 N-3 0.0 16 0.0 0.0 0.0 4.6 4.6 4.6 4.6 99.1 99.1 99.1 N-4 Prospect St.â SB (Cambridge) 0 0.0 5.0 0 13 0.0 0.0 0.0 5.0 5.0 5.0 5.0 99.3 99.3 99.3 N-5 4.0 14 0.0 0.0 0.0 4.0 4.0 4.0 4.0 99.0 99.0 99.0 N-6 0.0 18 0.0 0.0 0.0 4.8 4.8 4.8 4.8 99.3 99.3 99.3 a All percentiles pertain to the distribution of total parked vehicle displacement from curb. 1 Buffer between parking lane and bike lane. 2 Buffer between bike lane and travel lane. Note: NB = northbound, SB = southbound. Table 13. Percentage of cyclists riding within effective bike lane by scenario.
46 had no statistically significant effect (p-value of 0.50) on the position of parked vehicles. (Descriptive statistics were shown earlier in Table 7 and Figure 19.) The mean parked vehicle dis- placement from the curb was estimated for each category of parking lane plus buffer space width using the model from the ANOVA. The mean, standard error of the mean, and lower and upper 95% confidence limits of the mean are shown in Table 14. Differences between selected pairs in mean displacement of parked vehicles are shown in Table 15. Three parking lane widths were compared to each other: 7-ft, 8-ft, and 9-ft widths. The estimated mean difference in vehicle displace- ment is shown in column 2 and its 95% confidence interval in the last two columns. The p-values associated with the t-values and degrees of freedom indicate that none of the pairwise differences in parked vehicle displacement is statistically sig- nificantly different from zero. (All p-values exceed 0.3.) Although vehicles seem to park farther away from the curb as parking lane width increases, this increase in displacement from the curb, which ranges from 0.03 to 0.4 ft, is not statisti- cally significant at the 5% or even 30% significance level. It should also be noted that, as shown in Table 7, a higher percent- age of vehicles parked outside the designated 7-ft parking lane, while only a few vehicles parked outside the designated 8-ft parking lane. In the presence of a 9-ft parking lane, all vehicles parked within the boundaries of the designated parking lane. The primary findings related to parked vehicle displace- ment and parking lane widths are as follows: ⢠For parking lane widths of 7, 8, and 9 ft, the width of the parking lane does not significantly affect the position of parked vehicles relative to the curb; however, the trend is in the direction one would expect. The narrower the parking lane width, the closer the parked vehicles are to the curb. ⢠For parking lane widths of 7, 8, and 9 ft, a higher percent- age of vehicles parked outside the boundaries of the desig- nated 7-ft parking lane, only a few vehicles parked outside Width of Parking Lane + Buffer (ft) Estimated Mean Distance from Curb (ft) Standard Error of Estimate (ft) 95% Confidence Limits of Estimate (ft) Lower Upper 7 6.33 0.22 5.72 6.94 7 + 1 5.79 0.38 4.73 6.85 7 + 1.5 6.23 0.38 5.17 7.30 7 + 2 (Clark St.) 6.52 0.38 5.46 7.58 7 + 2 (Division St.) 6.83 0.38 5.77 7.89 8 6.70 0.27 5.94 7.45 9 6.73 0.27 5.97 7.48 Table 14. Estimated total parked vehicle displacement as a function of parking lane plus buffer width. Parking Lane Comparison Estimated Mean Difference in Vehicle Displacement (ft)a Standard Error of Difference (ft) Degrees of Freedom t-Value p-Value 95% Confidence Limits of Difference (ft) Lower Upper 7 ft to 8 ft 0.36 0.35 4.03 â1.04 0.36 â1.33 0.60 7 ft to 9 ft 0.40 0.35 4.00 â1.13 0.32 â1.37 0.57 8 ft to 9 ft 0.03 0.38 3.98 â0.08 0.94 â1.09 1.03 a The difference is calculated as the first in the pair minus the second in the pair shown in the first column. Table 15. Comparison of parked vehicle displacement between selected parking lane widths. Figure 31. Bicyclist operating space (AASHTO, 2012).
47 the boundaries of the designated 8-ft parking lane, and no vehicles parked outside the boundaries of the designated 9-ft parking lane. Effect of Roadway Characteristics on the Calculated Central Positioning of Cyclists. Three sets of ANOVAs were run to estimate the effect of roadway characteristics on the calculated central positioning (dependent variable). Given the site characteristics and the study scenarios, the ranges in the primary roadway characteristics that could be analyzed were: ⢠Bike lane width: 3.5 to 6 ft, ⢠Parking lane width: 7 to 9 ft, ⢠Travel lane width: 10 to 18 ft, ⢠Presence/absence of buffer space, ⢠Traffic volume: 14,800 to 29,000 vpd, and ⢠Percent trucks: 2% to 20%. Given the number of study sites, the ranges of traffic volume (ADT) and percent trucks were such that these two road- way characteristics were dichotomized as follows for analysis purposes: ⢠Low ADT: 15,000 to 17,000 vpd. ⢠High ADT: 29,000 vpd. ⢠Low percent trucks: <10%. ⢠High percent trucks: 16% to 20%. Presence of a buffer was defined as âyesâ if either one or two buffers were present. Travel lane and parking lane widths were used as continuous variables (i.e., covariates) in the models. The first set of three ANOVAs (based on the 85th, 90th, and 95th percentile of the total parked vehicle displace- ment when defining the effective bike lane), in which cen- tral positioning was modeled as a function of ADT, percent trucks, presence of buffer, parking lane width, and travel lane width, showed that neither parking lane width nor travel lane width was statistically significant. (The p-values ranged from 0.65 to 0.71 for travel lane width and from 0.77 to 0.91 for parking lane width.) These two variables were excluded one at a time from the models. The second set of three ANOVAs, in which central position- ing was modeled as a function of ADT, percent trucks, and presence of buffer, showed that all three categorical factors were statistically significant. The Type 3 tests for fixed effects are shown in Table 16, separately for each percentile used in defining the effective bike lane. In all cases, the three factors are highly significant, as indicated by the p-values shown in the last column. The least-squares mean central positioning was then esti- mated for each level of each factor and compared between the two levels of a given factor. To estimate the effect of an assumed worsening of the roadway conditions for the cyclists, the differ- ences in central positioning were calculated as follows for the three roadway characteristics: ⢠Change in ADT from low to high. ⢠Change in percent trucks from low to high. ⢠Change from presence of buffer space to no buffer space. The difference in central positioning can be interpreted as a cyclist displacement to one side or the other within the effective bike lane affected by the change in the factor con- sidered. The results are shown in Table 17. The p-values in column 8 show the statistical significance of the difference between the two central positioning estimates. The last two columns provide a 95% confidence interval for the difference. The interpretation of the results in Table 17 is illustrated using the first row in the table. ⢠At low ADT, the estimated central positioning of the cyclist is on average 0.65 ft; this indicates that the cyclists ride, on Effect Numerator Degrees of Freedom Denominator Degrees of Freedom F-Value p-Value USING 85TH PERCENTILE OF PARKED VEHICLE DISPLACEMENT ADT 1 13 93.7 <.0001 Percent trucks 1 13 309.3 <.0001 Buffer 1 13.1 5.7 0.0328 USING 90TH PERCENTILE OF PARKED VEHICLE DISPLACEMENT ADT 1 13 95.2 <.0001 Percent trucks 1 13 297.2 <.0001 Buffer 1 13.1 8.2 0.0134 USING 95TH PERCENTILE OF PARKED VEHICLE DISPLACEMENT ADT 1 13 87.2 <.0001 Percent trucks 1 13 278.8 <.0001 Buffer 1 13.1 6.0 0.0288 Table 16. ANOVA resultsâType 3 tests for fixed effects.
48 average, within the effective bike lane (since the estimate is positive) at 0.65 ft from either its left or right edge. ⢠At high ADT, the estimated central positioning of the cyclist is on average -1.47 ft; this indicates that the cyclists ride, on average, outside the effective bike lane (since the estimate is negative), at 1.47 ft to the right of its right edge. [Remember that only about 1% of cyclists ride to the left of the effective bike lane (i.e., in the travel lane) while a large percentage of cyclists (up to 45%) ride to the right of the effective bike lane (i.e., mostly in the car door area.)] ⢠The effect of changing from low to high ADT is estimated by the difference between the two central positioning esti- mates, that is, 0.65 - (-1.47) = 2.12 ft. Therefore, one can conclude that the effect of the higher ADT displaces the cyclists by an average of 2.12 ft toward the curb. (This aver- age ranges from 1.64 to 2.59 ft, the 95% confidence interval shown in the last two columns of in Table 17.) The primary findings related to the effect of roadway char- acteristics on the calculated central positioning (dependent variable), based on the results shown in Table 17, are sum- marized in the following. Of the five roadway characteristics analyzedâtraffic vol- ume, percent trucks, presence of buffer space, parking lane width, and travel lane widthâthe latter two did not sig- nificantly affect the central positioning of a bicyclist within the roadway cross section. However, traffic volume, percent trucks, and presence of buffer space significantly affected the central positioning of a bicyclist in the roadway cross section, as follows: Factor: Change from _ to _ Estimated Mean Central Positioning (ft) Estimated Mean Displacement of Cyclista (ft) Standard Error of Mean Displacement (ft) Degrees of Freedom t-Value p-Value 95% Confidence Limits of Mean Displacement (ft) 1st in Pair 2nd in Pair Lower Upper USING 85TH PERCENTILE OF PARKED VEHICLE DISPLACEMENT ADT: low to high 0.65 â1.47 2.12 0.22 13 9.68 <.0001 1.64 2.59 Truck %: low to high 0.89 â1.71 2.59 0.15 13 17.59 <.0001 2.28 2.91 Buffer: yes to no â0.20 â0.62 0.42 0.18 13.1 2.39 0.03 0.04 0.81 USING 90TH PERCENTILE OF PARKED VEHICLE DISPLACEMENT ADT: low to high 0.62 â1.71 2.33 0.24 13 9.76 <.0001 1.82 2.85 Truck %: low to high 0.84 â1.94 2.78 0.16 13 17.24 <.0001 2.43 3.13 Buffer: yes to no â0.27 â0.83 0.55 0.19 13.1 2.86 0.01 0.14 0.97 USING 95TH PERCENTILE OF PARKED VEHICLE DISPLACEMENT ADT: low to high 0.48 â2.05 2.53 0.27 13 9.34 <.0001 1.94 3.11 Truck %: low to high 0.74 â2.31 3.05 0.18 13 16.7 <.0001 2.65 3.44 Buffer: yes to no â0.52 â1.05 0.54 0.22 13.1 2.46 0.03 0.07 1.01 a The mean displacement is calculated as the central positioning corresponding to the first in the pair minus the second in the pair. Table 17. Estimated cyclist displacement in effective bike lane as a function of changes in ADT, percent trucks, and presence of buffer. ⢠Traffic volume â When the traffic volume was between 15,000 and 17,000 vpd, bicyclists rode inside the effective bike lane an average of 0.65 ft from the closest edge. â When the traffic volume was 29,000 vpd, bicyclists posi- tioned themselves to the right of the effective bike lane by an average of 1.47 ft. â As traffic volume increased, bicyclists moved away from vehicles in the travel lane and positioned themselves closer to the parked vehicles or the curb. The mean dis- placement of a bicyclist due to increased traffic volume was estimated at 2.12 ft and ranged from 1.64 to 2.59 ft. ⢠Percent trucks â When the truck percentage was below 10%, bicyclists rode inside the effective bicycle lane an average of 0.89 ft from the closest edge. â When the truck percentage was between 16% and 20%, bicyclists positioned themselves to the right of the effec- tive bike lane by an average of 1.71 ft. â As truck percentage increased, bicyclists moved away from vehicles in the travel lane and positioned themselves closer to parked vehicles or the curb. The mean displace- ment of a bicyclist due to increased percent truck volume was estimated at 2.59 ft and ranged from 2.28 to 2.91 ft. ⢠Presence of a buffer â In the presence of a buffer, bicyclists positioned them- selves to the right of the effective bike lane (i.e., within the door zone) by an average of 0.2 ft, regardless of the width of the bicycle lane. â The same held true in the absence of a buffer; how- ever, bicyclists positioned themselves even closer to the
49 parked vehicles within the door zone by an average of 0.62 ft, regardless of the width of the bicycle lane. â The presence of a buffer effectively moved bicyclists away from parked vehicles by an average of 0.42 ft, rang- ing from 0.04 to 0.81 ft. These results all pertain to the top portion of Table 17 (i.e., when the effective bike lane is defined using the 85th percen- tile of parked vehicle displacement). This is the least conser- vative definition of effective bike lane in this study. All the results hold whether using the 85th, 90th, or 95th (most con- servative) percentile of parked vehicle displacement. Average cyclist displacement due to a change in traffic volume or per- cent trucks increases by less than 0.5 ft going from the least conservative to the most conservative definition of effective bike lane. That change is less pronounced (0.13 ft) for the presence of a buffer effect. 3.5 Summary of Key Findings The primary findings based on the descriptive statistics and the analyses from the observational field studies conducted to evaluate the allocation of roadway width on both bicyclistsâ and motoristsâ lateral positioning, taking into consideration various roadway characteristics, can be summarized as follows: ⢠For the majority of scenarios evaluated on streets with on- street parking, the effective bike lane widths were narrower than the physical width of a typical adult bicyclist (i.e., 2.5 ft). For the scenarios evaluated on streets without on- street parking, the effective bike lane widths were always greater than or equal to the minimum operating space (i.e., 4 ft) of a typical adult bicyclist. ⢠The general trend in the data suggests that drivers park their vehicles closer to the curb as the parking lane nar- rows from 9 ft to 7 ft; however, the results are not statisti- cally different. For the same parking lane widths, a higher percentage of vehicles parked outside the boundaries of the designated 7-ft parking lane, only a few vehicles parked outside the boundaries of the designated 8-ft parking lane, and no vehicles parked outside the boundaries of the des- ignated 9-ft parking lane. ⢠For parking lanes 7- to 9-ft wide, based on the 95th-percentile parked vehicle displacement and assuming an open door width of 45 in., the open door zone width of parked vehi- cles extends approximately 11 ft from the curb. ⢠At sites with travel lanes ranging in width from 10 to 14 ft that were adjacent to bicycle lanes, very few passing vehicles encroached into the bicycle lane, even from the narrowest travel lane of 10 ft. ⢠Most bicyclists positioned themselves within the desig- nated bicycle lane, but some bicyclists rode to the left in the travel lane adjacent to the bicycle lane, while others rode to the right of the bicycle lane (i.e., in the parking lane or buffer area) on streets with on-street parking. ⢠On streets with on-street parking, in most cases less than 50% of bicyclists positioned themselves within the effective bike lane, and in general, the percentage of bicyclists posi- tioned within the effective bike lane was low (e.g., between 10% and 20%). ⢠On streets without on-street parking, most bicyclists (i.e., approximately 98% to 99%) were positioned within the effective bike lane regardless of whether a marked bike lane was installed. ⢠In general, on streets with on-street parking, the highest percentages of bicyclists were within the effective bicycle lane when buffers were used. ⢠Traffic volume, percent trucks, and presence of buffer space significantly affected the central positioning of a bicyclist in the roadway cross section: â As traffic volume increased from low (i.e., 15,000 to 17,000 vpd) to high (i.e., 29,000 vpd), bicyclists moved away from vehicles in the travel lane and positioned themselves closer to the parked vehicles or the curb. The estimated mean displacement of a bicyclist due to increased traffic volume was 2.12 ft (based on the 85th percentile of parked vehicle displacement) and ranged from 1.64 to 2.59 ft (95% confidence interval). â As truck percentage increased from low (i.e., <10%) to high (i.e., 16% to 20%), bicyclists moved away from vehicles in the travel lane and positioned themselves closer to parked vehicles or the curb. The estimated mean displacement of a bicyclist due to increased per- cent truck volume was 2.59 ft (based on the 85th per- centile of parked vehicle displacement) and ranged from 2.28 to 2.91 ft (95% confidence interval). â The presence of a buffer effectively moved bicyclists away from parked vehicles by an average of 0.42 ft (based on the 85th percentile of parked vehicle dis- placement) and ranged from 0.04 to 0.81 ft (95% con- fidence interval).
