The data needs for the BCI model are limited and, for the most part, include data that are traditionally collected by states and municipalities for other purposes. However, there will always be locations for which some of the data will not be available. In these cases, the practitioner must make judgments about appropriate values to use within the BCI model. It will also be the case that the available data are not in a form that can be directly input into the model. In that case, specific computations must be made to convert the data into the appropriate format. Described below are the variables required for the model and, where appropriate, computations and assumptions that can be used should the data be either not available or in the incorrect format. It should also be noted that the Microsoft Excel workbook on the enclosed diskette and described in the next section makes many of these computations for the user and incorporates some of the assumptions as default values.
As with any applied model, the output is only as good as the input. Therefore, it is very important that the user of the BCI model understand the variable definitions and assumptions provided below, and that there will always be specific situations requiring their best judgment as to what would be most appropriate for the model. For example, one of the decisions that must be made by the user of the BCI model is which hour of the day to use for evaluating bicycling conditions. It has been assumed throughout this document that the peak hour will be the hour of choice. However, depending on the route being examined, the operational conditions may change with time of day. For example, while traffic volumes may be significantly greater during the peak hour compared with the rest of the day, travel speeds may be significantly lower due to the volumes. On other streets, on-street parking may be prohibited during the peak hour. Thus, the off-peak parking lane becomes the peak-hour curb lane for motor vehicle and bicycle travel. While in most cases the peak-hour analysis will be the "worst-case" scenario and will serve as a good measure of bicycle compatibility for a given roadway irrespective of time of day, the user of the model should be aware that differences in operating conditions such as those described here can significantly change the outcome and can result in different levels of compatibility on the same route. It is recommended that, for those routes or segments where dramatic changes in operating conditions are expected at different times of the day, the analysis be conducted for all scenarios that apply.
Defined below are the variables required for the BCI model:
· Lane Configuration - number of through motor vehicle lanes in one direction and the presence or absence of a bicycle lane or paved shoulder. The number of lanes is used in the workbook to determine lane volumes from the average annual daily traffic (AADT).
when gutter pan is present
when no gutter pan is present
Figure 2. Curb lane width measurement when there is no bicycle lane, paved shoulder, or on-street parking lane.
· Curb lane width - width of the motor vehicle travel lane closest to the curb, measured to the nearest tenth of a meter. If there is no bicycle lane, paved shoulder, or parking lane present, this distance is measured from the center of the lane line or center line to the joint or seam between the pavement edge and the gutter pan as shown in figure 2. If no gutter pan is present, the curb lane width is determined by measuring the distance from the center of the lane line or center line to the curb face and then subtracting 0.3 m from that distance. The 0.3-m value accounts for the space bicyclists will typically leave between themselves and a curb (i.e., the "shy" distance). This value also reflects the difference in bicycle lane design widths recommended by the American Association of State Highway and Transportation Officials (AASHTO), i.e., 1.5 m when no gutter pan is present versus 1.2 m when a gutter pan exists.6 This scenario is also illustrated in figure 2.
when no gutter pan is present
when gutter pan is present
Figure 3. Curb lane and bicycle lane (paved shoulder) width measurements when there is no on-street parking
When there is a bicycle lane or paved shoulder, the curb lane width is measured from the center of the lane line or center line to the center of the edge line as shown in figure 3. If there is a marked parking lane present, the curb lane width is measured in a similar manner as shown in figure 4. If the parking lane is unmarked, the curb lane width can be determined by measuring from the center of the lane line or center line to the curb face (including the gutter pan if present), and then subtracting 2.4 m from this distance (see figure 4). The 2.4-m value accounts for the fact that vehicles occupy, on average, approximately 2.1 m of space when parallel parking and typically park within 0.15 to 0.3 m of the curb.7
when parking lane is marked
when parking lane is not marked
Figure 4. Curb lane width measurement when there is a parking lane present
The other scenario common on residential streets is to have no lane markings at all. In this case, the total cross section width can be measured from curb to curb (or gutter pan seam to gutter pan seam) and divided by the number of lanes (typically two) to determine the curb lane width. If parking is also present on this type of unmarked street, the parking lane widths (usually 2.4 m) should be subtracted from the total cross-section width prior to dividing by the number of lanes.
· Bicycle lane (paved shoulder) width - width of the bicycle lane or paved shoulder (if present), measured to the nearest tenth of a meter. Note that a paved shoulder is treated the same as a bicycle lane in the BCI model since recent research has shown that these two types of facilities result in virtually identical operational behaviors by motorists and bicyclists.8 If there is no parking lane present, the bicycle lane (paved shoulder) width is measured from the center of the edge line separating the bicycle lane from the motor vehicle travel lane to the joint or seam between the pavement edge and the gutter pan as shown in figure 3. If no gutter pan is present, the distance is measured from the edge line to the curb face, and then 0.3 m is subtracted from that distance to account for the space bicyclists will typically leave between themselves and a curb (i.e., the "shy" distance). This scenario is also illustrated in figure 3.
If a marked parking lane is adjacent to the bicycle lane, the bicycle lane width is measured from the center of the edge line (separating the motor vehicle travel lane and bicycle lane) to the center of the parking lane line separating the bicycle lane and the parking lane as shown in figure 5. If the parking lane is not marked, as would be the case in a shared parking/bicycle lane, the bicycle lane width can be determined by measuring the distance from the center of the edge line to the curb face (including the gutter pan if present) and then subtracting 2.4 m from that distance to account for the width of the parking lane. This scenario is also illustrated in figure 5.
when parking lane is marked
when parking lane is not marked
Figure 5. Bicycle lane width measurements when there is a parking lane present
As noted in all of the possible configurations described above and shown in the figures, the curb lane width and bicycle lane (paved shoulder) width measurements either did not include gutter pan widths or included them but subtracted a value to account for the "shy distance" of the bicyclist. The BCI model was developed using sites that either had no gutter pan or had gutter pans ranging from 0.3 to 0.6 m in width. Many communities have gutter pans that are wider than 0.6 m and provide space that can be utilized by a bicyclist. In fact, some communities designate this space as a bicycle lane. In those cases, it is recommended that the practitioner determine if the extra wide gutter pan does indeed provide adequate space for the bicyclist to ride. If so, this space should be added to the curb lane width or bicycle lane width as appropriate.
· Motor vehicle speed - 85th percentile speed of traffic, in km/h. This value can be obtained from manual or automated speed data collection efforts; for more information on collecting speed data, refer to the Manual of Transportation Engineering Studies.9 However, if the data are unavailable or the resources to collect speed data do not exist, it is recommended that 15 km/h be added to the posted speed limit as a surrogate measure for the 85th percentile speed. Prior research has shown that 85th percentile speeds for vehicles traveling on many urban and suburban streets (including arterial, collector, and local classifications) generally exceed the speed limit by 10 to 23 km/h.10
· Traffic volume - hourly traffic volume by lane in one direction of travel. While hourly counts may be available in some locations, it is more likely that AADT counts (collected for continuous 24-hour periods) will be the source of traffic volume information. Converting these data into hourly counts requires knowing the percentage of daily traffic traveling on the roadway during the hour of interest. In most cases, the hour of interest will be the peak hour. This volume can be determined using the following equation:
PHV = AADT x K x D
where:
PHV = peak-hour directional volume,
AADT = average annual daily traffic (vehicles per day)
K = peak-hour factor (the proportion of vehicles traveling during the peak hour, expressed as a decimal), and
D = directional split factor (the proportion of vehicles traveling in the peak direction during the peak hour, expressed as a decimal).
The K- and D-factors are usually determined on the basis of regional or route-specific characteristics. Generally, the K-factor ranges from 0.07 to 0.15 while the D-factor ranges from 0.50 to 0.65 in urban and suburban areas.11 If these factors are unknown or cannot be easily determined, a default K-factor of 10 percent may be assumed (expressed as 0.10), and a default D-factor of 55 percent may be used (expressed as 0.55). Note also that for one-way streets, the D-factor becomes 1.0 since 100 percent of the traffic is traveling in the same direction.
Once the directional hourly volume of traffic is determined using the above formula, it is necessary to assign traffic volumes to the curb lane and other travel lanes if it is a multilane facility. The lane distribution on non-freeway facilities depends on a variety of factors, including number and location of access points, the type of development, traffic composition, speed, volume, and local driving habits. These factors result in very little uniformity from site to site with respect to how volumes are distributed across lanes.5,11 If counts are available by lane, the percentage of vehicles traveling in each lane can be easily determined. If such counts are not available and considering the lack of consistency in this variable across sites, it is recommended that the hourly volume be distributed equally across all through lanes using the following equations:
CLV = PHV/N OLV = PHV – CLV
where:
CLV = hourly curb lane volume,
OLV = hourly volume in all through lanes except the curb lane,
PHV = peak-hour directional volume, and
N = number of through lanes in one direction.
· Presence and density of on-street parking - presence of an on-street parking lane and percentage of spaces occupied. The simple presence of an on-street parking lane may not adversely impact the comfort level of the bicyclist. During the development of the BCI model, it was shown that at least 30 percent of the spaces had to be occupied before the parking lane impacted the bicyclistsÕ comfort level. Thus, it is necessary to collect occupancy data for the hour being evaluated to determine if this 30 percent occupancy threshold is being met.
· Type of development - type of development or land use adjacent to the roadway. For purposes of the model, only two classifications are required, "residential" and "other." The residential development type proved to be significantly different from all other types of development and was shown to positively impact the comfort level of bicyclists.
· Large truck volume - hourly large truck volume in the curb lane. For purposes of the BCI model, large trucks are simply defined as all vehicles having six or more tires. This definition captures most single unit trucks and all combination unit trucks and buses. Most vehicle counters used today provide vehicle classification, and thus the percentage of trucks in the traffic stream is readily available if traffic count data are available. The volume of large trucks in the curb lane can then be determined as follows:
CLTV = PHV x HV x T
where:
CLTV = curb lane truck volume,
PHV = peak-hour directional volume (all vehicles),
HV = the proportion of all vehicles in the traffic stream that can be defined as large trucks (expressed as a decimal), and
T = curb lane truck factor (proportion of large trucks traveling in the curb lane, expressed as a decimal).
On a two-lane roadway (one lane of travel in each direction), the T-factor, or proportion of large trucks traveling in the curb lane, is 1.0 since 100 percent of the trucks will be traveling in the curb lane. On a multilane roadway, however, the T-factor must be calculated or assumed. If traffic counts are collected by lane of travel, the T-factor can be directly determined. If such data are not available, it is recommended that a default value of 0.80 be used for this factor on multilane roadways, indicating that 80 percent of the large trucks on the roadway are traveling in the curb lane. This value is based on collected data for freeways showing that up to 89 percent of the trucks travel in the curb lane.5 While comparable statistics were not available for arterials and other types of surface streets, the distribution of large trucks by lane of travel is believed to be similar.
If classification counts are not available, the user will have to input a truck percentage value (HV) believed to be appropriate for the type of roadway. In general, many urban streets will have very little or no truck traffic because of travel restrictions placed on such vehicles. An analysis of the FHWA Highway Safety Information System (HSIS) confirmed this fact for certain functional classifications. For the States of Illinois, Utah, and North Carolina, the mean percentage of traffic that was classified as trucks on local streets was less than 1 percent. On collectors, the mean truck percentage ranged from 0.4 to 2.6 percent, while on minor arterials, the range of means was 0.5 to 3.9 percent. The largest percentage of trucks was found on non-freeway principal arterials where the means ranged from 1.4 to 5.4 percent.12 On the basis of this analysis, it is recommended that the truck percentages shown in table 4 be used for the various functional classifications when the practitioner does not have the appropriate data and is not able to adequately determine the actual truck percentage.
· Parking time limits - parking time limits for on-street spaces. Vehicles pulling into or out of on-street parking spaces were shown to adversely impact the comfort level of bicyclists. Thus, as the parking turnover along a street increases, the comfort level for bicyclists decreases. Since most locations will not have parking turnover data or the resources to collect such data, a surrogate measure of parking time limit is recommended. It should be noted, however, that there may be cases where the time limit does not adequately reflect the level of parking turnover. For example, a street in front of a local post office may have 60-minute parking stalls, but the people using these spaces may generally be there no more than 15 minutes at a time. In that case, the value for a 15-minute limit parking stall may be more appropriate.
Right-turn volumes - hourly volume of vehicles turning right into all driveways and intersecting streets along the midblock segment being evaluated. For the BCI model, the adjustment factor is only applied when the hourly number of right turns is 270 or more. Knowing this information will assist in accounting for high-volume driveways or minor streets. Once the peak-hour volume is calculated, determining the number of right-turning vehicles can be done as follows:
RTV = PHV x R
where:
RTV = right-turn volume,
PHV = peak-hour directional volume,
R = proportion of vehicles in the traffic stream turning right into driveways or minor streets along the roadway segment, expressed as a decimal.
Knowledge of the proportion of vehicles turning right into driveways and minor intersection streets along a segment of roadway often may not exist. And since the adjustment factor in the BCI model and the relative impact on the overall bicycle LOS are small, it does not warrant spending resources to obtain this information. Instead, it is recommended that the practitioner use his/her judgment as to whether a specific midblock segment contains a high volume of right-turning traffic during the hour being evaluated. Examples of locations where right-turn volumes may be a factor during the peak hour include business and industrial entrances and minor streets used to cut through neighborhoods.
