Saturation Flow Rate, Start-Up Lost Time, and Capacity for Bicycles at Signalized Intersections

Transportation Research Record 1852 105 Paper No. 03-4180 Saturation Flow Rate, Start-Up Lost Time, and Capacity for Bicycles at Signalized Intersections Winai Raksuntorn and Sarosh I. Khan A review of the literature shows that capacity and saturation flow rate for on-street bicycle lanes at intersections have not been measured on the basis of bicycle discharge at intersections at the start of the green phase. The Highway Capacity Manual 2000 recommends a saturation flow rate of 2,000 bicycles per hour for a bicycle lane at a signalized intersection. However, this recommendation is not based on field studies at the intersection and is not a function of the width of the bicycle lane. A revised estimate is provided of saturation flow rate, and an estimate is provided of start-up lost time for bicycles based on data collected at the stop line of signalized intersections. In addition, the lateral stopped distance of automobiles from bicycle lanes, the lateral stopped distance of bicycles from adjacent lanes, and the lateral and longitudinal stopped distance between pairs of bicycles at a signalized intersections are presented. Bicycles may form more than one queue within a bicycle lane at the stop line. Since bicycles maintain a certain distance from the adjacent lane and the curb, the number of queues formed varies based on the width of the bicycle lane. Therefore, the saturation flow rate for a bicycle lane depends on the number of queues or the width of the bicycle lane. The saturation flow rates for bicycle lanes of varying widths are proposed on the basis of the lateral stopped distance of bicycles. Empirical evidence from intersections in Colorado and California is used to propose a new method to estimate the capacity for a bicycle lane. Bicycles traveling in on-street bicycle lanes at signalized and unsignalized intersections are delayed by traffic control at the intersections. The recent Highway Capacity Manual (HCM 2000) (1) introduces a procedure to evaluate the level of service of a designated on-street bicycle lane at signalized intersections. The measure of effectiveness is control delay and is estimated on the basis of Webster s uniform delay for automobiles at signalized intersections. It assumes no overflow delay for bicycles in on-street bicycle lanes. According to the HCM, bicyclists will not normally accept an overflow situation and will select other routes or ignore traffic regulations to avoid excessive delay. In addition, since bicycle volume on any on-street bicycle lane is very low compared with the capacity of bicycle lanes, bicycles in queue during a red phase pass through the intersection during the subsequent green phase. Unlike automobiles, bicycles may form more than one queue at the stop line depending on available space in the bicycle lane. Bicycles at the front of each queue depart the intersection at the same time. Therefore, the saturation flow rate for bicycles increases as the number of queues formed increases. In other words, the saturation flow rate for bicycles increases as the width of the bicycle lane increases. Colorado TransLab, Department of Civil Engineering, University of Colorado, Campus Box 113, P.O. Box 173364, Denver, CO 80217. Capacity and saturation flow rate of signalized intersections for bicycles have been reported by research studies in many countries. These studies include both on-street and off-street bicycle facilities. A study in Davis, California (2), reports that the capacity of a bicycle lane is approximately 2,600 bicycles per hour per 3.3 ft (1 m). Another study in Davis (3) developed an equation to predict the saturation flow rate for bicycles based on the data collected from a simulation. With this equation, the saturation flow rate for bicycles on a 4-ft (1.2-m) wide bicycle lane is 3,060 bicycles per hour. A Dutch study (4) estimated the capacity for bicycles on a bicycle path in the Netherlands based on headway between bicycles. The reported capacity for bicycles ranged from 3,000 to 3,500 bicycles per hour per 2.6-ft (0.78-m) lane. A Canadian study (5) estimated the capacity for bicycles based on an experiment conducted on a test track as 10,000 bicycles per hour per 8.2 ft (2.5 m). The HCM 2000 (1) reports a base saturation flow rate as high as 2,600 bicycles per hour on the basis of research conducted by Opiela et al. (6). However, 2,000 bicycles per hour is recommended as an average achievable at most intersections. It may be mentioned that the study by Opiela et al. examines arrival distribution and gap acceptance of bicycles with opposing vehicles. However, saturation flow rate or saturation headway was not reported. The studies so far suggest a saturation flow rate between 1,500 and 5,000 bicycles per hour per 3.3 ft (1 m). This saturation flow rate, however, was measured on the basis of the data collected from sections of bicycle lanes, sections of bicycle paths, the author s assumptions, or a test track. No report of saturation flow rate and start-up lost time for bicycles measured at signalized intersections was found in the literature. At the stop line, bicycles and automobiles normally stop within their own lanes. Automobiles maintain a certain distance from the bicycle lane for safety. Bicycles, on the other hand, not only maintain a distance from automobiles in adjacent lanes and from a curb, but also maintain a comfortable clearance distance from bicycles to the side and in front. Kroll and Ramey (7) studied how motorist and bicyclist behavior is affected by the presence of bicycle lanes. They found that the mean separation distance between bicycles and automobiles is a function of the available travel space, not of the presence of a bicycle lane. In another study Harkey and Stewart (8) examined the lateral locations of bicycles and automobiles on three types of facilities: wide curb lanes, paved shoulders, and bicycle lanes. The study reported mean bicyclist distances to roadway edges as 1.41 ft (0.43 m), 2.59 ft (0.79 m), and 2.59 ft (0.79 m) for the wide curb lane, paved shoulder, and bicycle lane, respectively. They also found that the mean distances between bicycles and

106 Paper No. 03-4180 Transportation Research Record 1852 roadway edges for paved shoulders and bicycle lanes were not significantly different from each other. Moreover, the separation distance between bicycle and automobile depended on the type of facility. The reported separation distances were 6.39 ft (1.95 m), 6.19 ft (1.89 m), and 5.89 ft (1.80 m) for the wide curb lane, paved shoulder, and bicycle lane, respectively. These two studies, however, only report the location of the bicycle and the separation distance between bicycle and automobile on a street section or midblock. So far, no studies on the stopped distance of bicycles and automobiles and the separation distance between bicycles and automobiles at the stop line have been reported. OBJECTIVE The objective of this paper is to report on the saturation flow rate for bicycles. Bicycles may form more than one queue at the stop line. Since bicycles maintain a certain distance from the adjacent lane and the curb, the number of queues formed varies on the basis of the width of the bicycle lane. Therefore, the saturation flow rate for a bicycle lane depends on the number of queues or the width of the bicycle lane. The saturation flow rates for bicycle lanes of varying widths are investigated on the basis of the lateral stopped distance of bicycles. From empirical evidence, a method to estimate the capacity for a bicycle lane is also proposed. DATA COLLECTION Several intersections were surveyed in four U.S. cities as part of a research project funded by the Transportation Research Board s Intelligent Transportation Systems Innovations Deserving Exploratory Analysis (ITS-IDEA) program. Of seven intersections selected for the IDEA project in four cities Boulder, Denver, and Fort Collins, Colorado; and Davis, California only two met the needs of the study presented in this paper. Video recordings of bicycles and automobiles were collected at two locations: (a) the south approach of the intersection between Sycamore Lane and Russell Boulevard in Davis, California, and (b) the south approach of the intersection between Folsom Street and Arapahoe Avenue in Boulder, Colorado. At the south approach of Sycamore Lane, an 8-ft-wide bicycle lane is located between left- and right-turn lanes, and a 6-ft-wide bicycle lane is located next to the curb of the south approach of Folsom Street. The automobile lanes at Sycamore Lane are 12 ft wide and at Folsom Street are 10 ft wide. Data were collected each day for 4 h on June 5 and June 6, 2000, at Sycamore Lane and for 4 h each day on August 29 and August 30, 2000, at Folsom Street. The flow rate for bicycles at Sycamore Lane ranges from 100 to 300 bicycles per hour and at Folsom Street ranges from 20 to 70 bicycles per hour. A study examining the normal speed of bicycles at these intersections (9) reports that the average normal speed of bicycles is significantly higher at Folsom Street than at Sycamore Lane. Normal speed at Folsom Street is 20.44 ft/s (22.44 km/h) with a standard deviation of 4.21 ft/s (4.62 km/h) and at Sycamore Lane is 13.61 ft/s (14.95 km/h) with a standard deviation of 2.59 ft/s (2.84 km/h). Stopped Distance of Bicycles and Automobiles To collect the stopped distances of bicycles and automobiles, video recordings from the intersections in Davis and Boulder were analyzed. Four types of the stopped distances were measured: (a) between an automobile and a bicycle lane, (b) between bicycles and the adjacent lane or curb, (c) laterally between a pair of bicycles, and (d) longitudinally between a pair of bicycles (Figure 1). Saturation Flow Rate for Bicycles Several intersections were surveyed to collect data on saturation flow rate for bicycles; intersections with a minimum of five bicycles at the stop line during any red phase were required. The initial observations were made at seven intersections in four cities: Davis, California; and Boulder, Denver, and Fort Collins, Colorado. However, only the intersection at Sycamore Lane and Russell Boulevard in Davis, California, met this requirement. The intersection of Sycamore Lane and Russell Boulevard is a signalized intersection with an 8-ft-wide bicycle lane between the left- and right-turn lanes for automobiles. Since the bicycle volume at this intersection is fairly high (100 to 300 bicycles per hour), there is an exclusive signal phase for bicycles. Bicyclists stop at the stop line during a red phase and wait for a green phase for bicycles. The bicycle traffic at this intersection was videotaped using a camcorder, which captured bicycle movements from about 50 ft upstream of the intersection. During the data collection, a maximum of 10 bicycles was in queue at the stop line during a red phase. Typical bicycle traffic at a signalized intersection is shown in Figure 1. DATA REDUCTION All videotapes were reviewed and captured into an audio-videoimage (AVI) format using a video capture card. This process breaks the bicycle movements down to 10 frames per second. Positions of bicycles and automobiles were determined from the video images using a vehicle videotaping data collector (10). These positions or screen coordinates were converted into roadway coordinates based on a coordinate transformation technique (11). Stopped Distance of Bicycles and Automobiles From the positions of bicycles and automobiles at the stop line, lateral and longitudinal distances between bicycles and automobiles were determined. The distances were categorized into four groups based on type of stopped distance. Saturation Flow Rate for Bicycles Only the AVI files with at least five bicycles in queue at the stop line during the red phase were reviewed frame by frame. The headway between bicycles departing the intersection at the beginning of the green phase was measured as the bicycles crossed the stop line at the intersection. The first headway is the elapsed time in seconds between the start of the green phase and the crossing of the front wheel of the first bicycle over the stop line. The second headway is the elapsed time between the crossing of front wheels of the first and the second bicycles over the stop line. Subsequent headways were measured similarly.

Raksuntorn and Khan Paper No. 03-4180 107 (a) (b) (c) (d) FIGURE 1 Bicycles and automobiles at intersection: (a) bicycle lane in center, (b) bicycle lane in rightmost lane, (c) bicycles in queue during red phase, (d) bicycles departing intersection at start of green phase. DATA ANALYSIS AND RESULTS Stopped Distances Four types of distances were measured at intersections between bicycles and automobiles as they stopped at the stop line: 1. Lateral stopped distance between an automobile and a bicycle lane, 2. Lateral stopped distance between a bicycle and an adjacent lane or a curb, 3. Longitudinal stopped distance between a pair of bicycles, and 4. Lateral stopped distance between a pair of bicycles. The lateral stopped distances between automobiles and bicycle lanes were measured as the distance from right edge or left edge of the automobile to the bicycle lane. The lateral stopped distances between bicycles and adjacent lanes or curb were measured similarly. However, it was difficult to identify the right or left edge of a bicycle. Therefore, the stopped distance between bicycle and adjacent lane or curb was measured from the bicycle s center (center of the tire). As bicyclists arrive at signalized intersections during a red phase, they stop within the bicycle lane and form groups in two ways: side by side beside another bicycle or one behind the other. These stopped distances will be referred to as the lateral stopped distance, measured as the distance between a pair of bicycles as they stop side by side, and the longitudinal stopped distance, measured as the clearance distance between a pair of bicycles as they stop one behind the other. These four types of stopped distances are illustrated in Figure 2. Lateral Stopped Distance Between Automobile and Bicycle Lane Because a bicycle lane is an exclusive lane for bicycles, motorists usually do not use it. During a red phase, motorists stop in their own lane and maintain a certain distance from the bicycle lane. At Folsom Street, the average lateral stopped distances of automobiles from the bicycle lane were 0.72 ft (0.22 m) and 0.82 ft (0.25 m), with and without the presence of bicycles, respectively. At Sycamore Lane, the average lateral stopped distances were 2.71 ft (0.83 m) and 2.87 ft (0.88 m), with and without the presence of bicycles, respectively. The comparison of the means in Table 1 reveals that the average stopped distances between automobile and bicycle lane, with and without the presence of bicycles, were not significantly different from each other. Therefore, at both locations, the motorist s

108 Paper No. 03-4180 Transportation Research Record 1852 D C E B Bicycle lane Bicycle A Automobile Automobile lane FIGURE 2 Stopped distances measured at intersections. A = lateral stopped distance between automobile and bicycle lane. B = lateral stopped distance between bicycle and left adjacent lane. C = lateral stopped distance between bicycle and right adjacent lane or curb. D = longitudinal stopped distance between bicycles. E = lateral stopped distance between bicycles. position within an automobile lane is not affected by the presence of a bicycle in the bicycle lane. The average lateral stopped distance of an automobile from the bicycle lane, however, is significantly greater at Sycamore Lane than at Folsom Street at a significance level of 0.05 (Table 2). The automobile lane at Sycamore Lane (12 ft wide) is wider than the automobile lane at Folsom Street (10 ft wide). The lateral stopped distance of an automobile from a bicycle lane depends on the width of the automobile lane. Therefore, the stopped distance may be a function of the width of the automobile lane. However, this relationship was not explored as a part of this study. From the data collected at two locations, the average stopped distance of an automobile from the bicycle lane in a 10-ft-wide automobile lane (Folsom Street) was 0.75 ft (0.23 m) with a standard TABLE 1 Lateral Stopped Distance Between Automobile and Bicycle Lane With and Without Bicycle Present with bicycle without bicycle with bicycle without bicycle present present present present Mean 0.72 0.82 2.71 2.87 Variance 0.81 0.41 1.62 0.76 79 47 76 30 Hypothesized Mean Difference 0.00 0.00 df 120 77 t Stat -0.71-0.70 P(T t) two-tail 0.48 0.49 t Critical two-tail 1.98 1.99

Raksuntorn and Khan Paper No. 03-4180 109 TABLE 2 Lateral Stopped Distance Between Automobile and Bicycle Lane Mean 0.75 2.76 Variance 0.66 1.37 126 106 Hypothesized Mean Difference 0.00 df 182 t Stat -14.86 P(T t) two-tail 0.00 t Critical two-tail 1.97 Intersections No. of Median Mean Minimum Maximum Standard Deviation Folsom St & Arapahoe Ave 1 126 0.62 0.19 0.75 0.23-1.01-0.31 3.10 0.95 0.81 0.25 Sycamore Ln & Russell Blvd 2 106 2.86 0.87 2.76 0.84 0.02 0.01 4.74 1.44 1.17 0.36 NOTE: Negative number means automobile stops inside the bicycle lane. deviation of 0.81 ft (0.25 m). At Sycamore Lane (12 ft wide), the average stopped distance was 2.76 ft (0.84 m) with a standard deviation of 1.17 ft (0.36 m). The data in Table 2 also show that some right-turning automobiles may stop inside the bicycle lane if the bicycle lane is to the right of the right-turn automobile lane. This is not the case when the bicycle lane is to the left of the right-turn automobile lane. A statistical summary of lateral stopped distance of automobiles from the bicycle lane is presented in Table 2. Lateral Stopped Distance Between Bicycle and Curb or Right and Left Adjacent Lanes Bicycles are small compared with the width of a bicycle lane. Therefore, they are free to move and stop anywhere within the lane. A bicycle may stop closer to one side of the lane than to the other, that is, stop closer to the left adjacent lane or the right adjacent lane. On average, stopped distances of bicycles from the left adjacent lane and from the curb at Folsom Street were 1.50 ft (0.46 m) and 1.35 ft (0.41 m), respectively. At Sycamore Lane, the stopped distance of bicycles from the left adjacent and the right adjacent lanes was measured as 1.74 ft (0.53 m) and 1.94 ft (0.59 m), respectively. Statistical tests show that the average stopped distances of bicycles from the curb or the right and the left adjacent lanes at each location are not significantly different at a significance level of 0.05 (Table 3). This finding suggests that bicycles do not favor one side of the lane over the other as they stop at an intersection. Further analysis was conducted to test whether the width of the bicycle lane affects the lateral stopped distance of bicycles from the adjacent lane or curb. An analysis of data shows that the average lateral stopped distance of bicycles is significantly greater at Sycamore Lane than at Folsom Street (Table 4). Since the bicycle lane at Sycamore Lane is wider (8 ft) than the bicycle lane at Folsom Street (6 ft), the implication is that bicyclists stop farther from the edge of the bicycle lane if possible. The average lateral stopped distance on the 6-ft-wide bicycle lane was 1.40 ft (0.43 m) with a standard deviation of 0.77 ft (0.23 m). The average lateral stopped distance of bicycles in the 8-ft-wide bicycle lane was 1.84 ft (0.56 m) with a standard deviation of 0.79 ft (0.24 m). A statistical summary of the lateral stopped distance of bicycles from the adjacent lane is presented in Table 4. TABLE 3 Lateral Stopped Distance Between Bicycle and Curb or Right and Left Adjacent Lanes from left from left from right from curb adjacent lane adjacent lane adjacent lane Mean 1.50 1.35 1.74 1.94 Variance 0.96 0.60 0.62 0.62 50 126 30 31 Hypothesized Mean Difference 0.00 0.00 df 74 59 t Stat 1.00-0.99 P(T t) two-tail 0.32 0.33 t Critical two-tail 1.99 2.00

110 Paper No. 03-4180 Transportation Research Record 1852 TABLE 4 Lateral Stopped Distance Between Bicycle and Adjacent Lane Mean Variance Hypothesized Mean Difference df t Stat P(T t) two-tail t Critical two-tail 1.40 0.70 176 0.00 110-3.78 0.00 1.98 1.84 0.62 61 Intersections No. of Median Mean Minimum Maximum Standard Deviation Folsom St & Arapahoe Ave 1 176 1.22 0.37 1.40 0.43 0.05 0.02 2.94 0.90 0.77 0.23 Sycamore Ln & Russell Blvd 2 61 1.89 0.58 1.84 0.56 0.41 0.13 3.35 1.02 0.79 0.24 Longitudinal Stopped Distance Between Bicycles The average longitudinal stopped distance between a pair of bicycles is 4.20 ft (1.28 m) at Folsom Street and 4.39 ft (1.34 m) at Sycamore Lane. From a comparison of means at a significance level of 0.05, these distances are not significantly different (Table 5). This finding suggests that the longitudinal stopped distance of bicycles is not a function of the width of the bicycle lane or its location. From the combined data, the longitudinal stopped distances range from 0.98 to 9.43 ft (0.30 to 2.88 m) with an average of 4.32 ft (1.32 m) and a standard deviation of 1.93 ft (0.59 m). The statistical summary of the longitudinal stopped distances at the two locations is presented in Table 5. Lateral Stopped Distance Between Bicycles The average lateral stopped distance of bicycles on a wider lane (8 ft) seems to be greater than on a narrower lane (6 ft). However, the com- parison reveals that the means of lateral stopped distances at the two locations are not significantly different at a significance level of 0.05 (Table 6). From the combined data, the lateral stopped distances of bicycles range from 1.39 to 4.80 ft (0.42 to 1.47 m) with an average of 2.38 ft (0.72 m) and a standard deviation of 0.80 ft (0.24 m). The statistical summary of the lateral stopped distances at these locations is presented in Table 6. Saturation Flow Rate for Bicycles and Start-Up Lost Time The saturation flow rate for bicycles at intersections represents the number of bicycles per hour per lane that passes through a signalized intersection if the green signal indication is available for a full hour. Since a full hour of green phase is not provided for any movement, bicyclists come to a complete stop and form a queue at the stop line if they arrive at the intersection during a red phase, the rule of the road applying to both bicycles and automobiles. The TABLE 5 Longitudinal Stopped Distance Between Bicycles at Intersections Mean Variance Hypothesized Mean Difference df t Stat P(T t) two-tail t Critical two-tail 4.20 2.90 38 0.00 89-0.49 0.63 1.99 4.39 4.31 57 Intersections No. of Median Mean Minimum Maximum Standard Deviation Folsom St & Arapahoe Ave 1 38 4.68 1.43 4.20 1.28 0.98 0.30 8.77 2.67 1.70 0.52 Sycamore Ln & Russell Blvd 2 57 3.39 1.03 4.39 1.34 1.87 0.57 9.43 2.88 2.08 0.63 Total 95 4.39 1.34 4.32 1.32 0.98 0.30 9.43 2.88 1.93 0.59

Raksuntorn and Khan Paper No. 03-4180 111 TABLE 6 Lateral Stopped Distance Between Bicycles at Intersections Mean Variance Hypothesized Mean Difference df t Stat P(T t) two-tail t Critical two-tail 2.31 0.53 30 0.00 58-0.60 0.55 2.00 2.44 0.75 31 Intersections Folsom St & Arapahoe Ave 1 Sycamore Ln & Russell Blvd 2 Total No. of Median Mean Minimum Maximum Standard Deviation 30 2.14 0.65 2.31 0.71 1.39 0.42 4.75 1.45 0.73 0.22 31 2.23 0.68 2.44 0.74 1.39 0.42 4.80 1.47 0.87 0.26 61 2.16 0.66 2.38 0.72 1.39 0.42 4.80 1.47 0.80 0.24 saturation flow rate for bicycles is estimated from a procedure similar to that for the saturation flow rate for automobiles, based on the saturation headway: s b = 3600 ( 1) h b where s b is the saturation flow rate for the bicycle lane in bicycles per hour of green per bicycle lane and h b is the saturation headway for bicycles in seconds. The data collected for this study show that the headways between bicycles remain constant after the fifth bicycle in the queue crosses the stop line. As shown in Figure 3, the constant headway h b represents the saturation headway for bicycles, estimated as the constant average headway between bicycles after the fifth bicycle in the queue and continuing until the last bicycle in queue at the beginning of the green clears the intersection. The saturation headway for bicycles for the 8-ft-wide bicycle lane was found to be 0.80 s. Based on Equation 1, the saturation flow rate is estimated as 4,500 bicycles per hour of green for an 8-ft-wide bicycle lane. This saturation flow rate is more than twice the saturation flow rate for automobiles. This difference is because bicycles generally do not stop one behind the other; instead, they stop next to each other and form more than one queue within the lane. Thus, the saturation headway for bicycles is significantly lower than the saturation headway for automobiles at intersections. Both saturation headway and saturation flow rate may be a function of the width of the bicycle lane. At the beginning of each green phase, bicycles in queue depart the intersection, with the first five bicycles experiencing a start-up reaction time and acceleration time before achieving a constant speed. After the fifth bicycle, the effects of start-up reaction and acceleration are minimal. Figure 3 shows that the headways for the first five 6.0 5.5 5.0 individual time headway of i th bicycle in queue average time headway of i th bicycle in queue Time Headway (seconds) 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 h b 0.5 Y = -0.75 LN (X) +2.01 R 2 = 0.23, Standard Error of Y = 0.45 0.0 0 1 2 3 4 5 6 7 8 9 10 i th Bicycle in Queue FIGURE 3 Headways of bicycles in queue for 8-ft-wide bicycle lane.

112 Paper No. 03-4180 Transportation Research Record 1852 bicycles are higher than the saturation headway and are expressed as h + t i, where t i is the incremental headway for the ith bicycle due to the start-up reaction and acceleration. The incremental headway decreases as i increases. The sum of the incremental headways is called the total start-up lost time and was found to be 2.5 s per phase. Generally, bicycles have a lower acceleration rate than automobiles, so bicycles need more time to cross the stop line. Therefore, the total start-up lost time of bicycles is slightly greater than the total startup lost time of automobiles [2 s (1)], even though the saturation headway for bicycles is higher. As mentioned in the previous section, bicyclists, on average, maintain distances of 1.40 ft (0.43 m) and 1.84 ft (0.56 m) from the adjacent lane on 6-ft- and 8-ft-wide bicycle lanes, respectively. Moreover, they maintain a lateral stopped distance of 2.38 ft (0.72 m) approximately from each other. According to these estimates, bicyclists can form up to three queues on an 8-ft-wide bicycle lane and up to two queues on a 6-ft-wide lane. In other words, the 6-ft-wide bicycle lane may be effectively used as two sublanes and the 8-ft-wide bicycle lane may be effectively used as three sublanes. Therefore, the saturation flow rate per effective sublane would be s b = 1, 500 bicycles per hour of green per sublane ( 2) From Equation 2, the capacity for a bicycle lane may be estimated as follows: c g s C N g = = 1, 500 C N ( 3) b b b b where c b = capacity of bicycle lane (bicycles per hour of green), s b = saturation flow of bicycle lane (bicycles per hour of green per sublane), g = effective green time for bicycle lane (s), C = signal cycle length (s), and N b = number of sublanes. From Equation 3 and the lateral stopped distance of bicycles, the estimated number of sublanes and the estimated saturation flow rate for bicycle lanes are presented in Table 7. Since bicycles maintain a certain distance from adjacent lanes and the curb, there is some unused travel space on both sides of the bicycle lane. The proportion of this unused travel space and the total width decrease as the width of the bicycle lane increases. Therefore, with a wider bicycle lane, a higher proportion of space is available as effective travel space than with a narrow bicycle lane. This study shows that the saturation flow rate for a bicycle lane should not be estimated on the basis of the width of the bicycle lane alone but on the number of effective sublanes or de facto sublanes observed. TABLE 7 Number of Sublanes and Saturation Flow Rate for Bicycle Lanes Width of Bicycle Lane Estimated No. of Sub-Lanes Estimated Saturation Flow Rate (bicycles per hour of green) 3 0.92 1 1,500 4 1.22 1 to 2 1,500 to 3,000 5 1.53 2 3,000 6 1.83 2 3,000 8 2.44 3 4,500 10 3.05 4 to 5 6,000 to 7,500 CONCLUSIONS The findings of this study as reported here may be summarized as follows: At intersections, motorists on 10-ft- and 12-ft-wide lanes, on average, maintain the same distance from a bicycle lane irrespective of whether a bicycle is present in the adjacent bicycle lane: 0.75 ft (0.23 m) and 2.76 ft (0.84 m), respectively, from the bicycle lane. However, this distance is greater for wider automobile lanes. This finding shows that the lateral stopped distance of an automobile from a bicycle lane depends on the width of the automobile lane. The wider automobile lane provides free space for motorists to make a stop. Therefore, the stopped distance may be a function of the width of the automobile lane. At intersections, bicyclists maintain, on average, the same distance from the adjacent lanes, both to the left and to the right. The data show that the stopped distances from the curb or the right adjacent and the left adjacent lanes at each location were not significantly different. However, bicyclists tend to stop farther from the adjacent lane in wider bicycle lanes. On average, bicyclists maintain a distance of 1.40 ft (0.43 m) from the adjacent lane on a 6-ft-wide bicycle lane and a distance of 1.84 ft (0.56 m) from the adjacent lane on an 8-ft-wide bicycle lane. As bicyclists arrive at signalized intersections during a red phase, they stop within the bicycle lane. They form groups to stop in two ways: side by side alongside another bicycle or one behind the another. Bicyclists, on average, maintain a lateral distance of 2.38 ft (0.72 m) if they stop side by side and a longitudinal distance of 4.32 ft (1.32 m) if they stop behind another bicycle. The time headways between bicycles on an 8-ft-wide bicycle lane are constant after the fifth bicycle in queue crosses the stop line. The saturation headway for bicycles is measured at 0.80 s, which suggests a maximum flow rate of 4,500 bicycles per hour of green per 8-ft (2.4-m) wide bicycle lane. In addition, the total startup lost time of bicycles on an 8-ft-wide bicycle lane is measured as 2.5 s per phase. Bicyclists may form up to two queues on a 6-ft-wide bicycle lane and up to three queues on an 8-ft-wide bicycle lane. Accordingly, the 8-ft-wide bicycle lane may be divided into three sublanes and a 6-ft-wide bicycle lane may be divided into two sublanes. Therefore, the saturation flow rate for bicycles per sublane is 1,500 bicycles per hour of green. The capacity for the bicycle lane at a signalized intersection may be computed and included in Chapter 19 of the HCM as shown in Equation 3. RECOMMENDATIONS The current version of the HCM recommends a saturation flow rate of 2,000 bicycles per hour. However, this recommendation is not based on field studies since insufficient research has been done in this area, as noted in the manual. In addition, no guidance is provided in the HCM on the start-up lost time for bicycles and the effect of bicycle lane width on saturation flow rate. This study proposes a revised estimate of the saturation flow rate, an estimate of start-up lost time for bicycles, and a new methodology to estimate the capacity for a bicycle lane at signalized intersections based on empirical evidence from intersections in California and Colorado.

Raksuntorn and Khan Paper No. 03-4180 113 This study provides a basic understanding of the behavior of bicyclists departing an intersection at the start of a green phase and the behavior of motorists and bicyclists as they come to a stop at the stop line. The findings are based on the data collected on 6-ftand 8-ft-wide on-street bicycle lanes at two intersections and provide information to develop better analysis and design guidelines for on-street bicycle facilities as well as a comprehensive bicycle microsimulation model. The findings of this paper are based on data collected from two intersections. Further studies based on additional data collected at intersections with high bicycle volumes are recommended to examine the factors affecting the saturation flow rate for bicycles at signalized intersections. REFERENCES 1. Highway Capacity Manual. TRB, National Research Council, Washington, D.C., 2000. 2. Homburger, W. S. Capacity of Bus Routes, and of Pedestrian and Bicycle Facilities. Institute of Transportation Studies, University of California, Berkeley, Feb. 1976. 3. Ferrara, T. C. A Study of Two-Lane Intersections and Crossings Under Combined Motor Vehicle and Bicycle Demands. Civil Engineering Department, University of California, Davis, Dec. 1975. 4. Botma, H., and H. Papendrecht. Traffic Operation of Bicycle Traffic. In Transportation Research Record 1320, TRB, National Research Council, Washington, D.C., 1991, pp. 65 72. 5. Navin, F. P. D. Bicycle Traffic Flow Characteristics: Experimental Results and Comparisons. ITE Journal, Vol. 64, No. 3, 1994, pp. 31 37. 6. Opiela, K. S., S. Khasnabis, and T. K. Datta. Determination of the Characteristics of Bicycle Traffic at Urban Intersections. In Transportation Research Record 743, TRB, National Research Council, Washington, D.C., 1980, pp. 30 38. 7. Kroll, B., and M. R. Ramey. Effects of Bike Lanes on Driver and Bicyclist Behavior. Journal of Transportation Engineering, ASCE, 1977, pp. 243 256. 8. Harkey, D. L., and J. R. Stewart. Evaluation of Shared-Use Facilities for Bicycles and Motor Vehicles. In Transportation Research Record 1578, TRB, National Research Council, Washington, D.C., 1997, pp. 111 118. 9. Raksuntorn, W., and S. I. Khan. Speed of Bicycles at On-Street Bicycle Facilities. Journal of Transportation Engineering, ASCE (submitted). 10. Wei, H., C. Feng, E. Meyer, and J. Lee. Video-Capture-Based Methodology for Extracting Multiple Vehicle Trajectories for Microscopic Simulation Modeling. Presented at the 78th Annual Meeting of the Transportation Research Board, Washington, D.C., 1999. 11. Khan, S. I., and W. Raksuntorn. Accuracy of Numerical Rectification of Video Images to Analyze Bicycle Traffic Scenes. In Transportation Research Record: Journal of the Transportation Research Board, No. 1773, TRB, National Research Council, Washington, D.C., 2001, pp. 32 38. Publication of this paper sponsored by Committee on Highway Capacity and Quality of Service.