BICYCLE LEVEL OF SERVICE for URBAN STREETS Prepared by Ben Matters and Mike Cechvala 4/16/14 Page 1
Introduction The methodology used for the Bicycle (BLOS) analysis is from the Highway Capacity Manual (HCM) 2010 (Transportation Research Board). The HCM calculates a level of service with a letter grade A through F for each mode auto, bicycle, transit, and pedestrian. The BLOS analysis contains a component for roadway links (between intersections), intersections, and segments (a combination of links and intersections). This BLOS analysis uses the Link LOS only for a few reasons. First, the data collection effort needed for intersection and segment LOS analysis is beyond the MPO s resources; second, intersection LOS does not account for bicycle delay or special intersection treatments like bike boxes or green pavement; and third, intersection LOS does not account for hazards such as complex turning maneuvers. Likewise, the HCM has an LOS methodology for multi-use paths. However, this was not done because it relies on detailed flow rates of several categories of path users, which is not available. See Appendix A. The link BLOS is estimated to give a useful approximation of the usability of roadways for bicycling; however, it should be noted that this method does not take into account some variables that could affect a bicyclist s perceived level of comfort, including the presence of driveways and busy intersections. The HCM s BLOS analysis is only appropriate for urban roadways. It was created using volunteers who rated their perceived comfort while watching point-of-view videos taken by bicyclists using various roadways; then the formulas were designed to fit the empirical data. The analysis is intended to span the wide range of roads, including busy urban streets, local streets, and suburban arterials. The HCM methodology for determining BLOS is presented in Chapter 17 for roadway links. The basic inputs and structure to the HCM BLOS formulas are shown below in Figure 1. 4/16/14 Page 2
Figure 1 HCM Methodology for Link Traffic Volume Number of Through Lanes Width of Outside Through Lane Parking Utilization Width of Bicycle Lane and/or Shoulder Motorized Vehicle Cross Section BLOS Link Score Traffic Speed Heavy Vehicle Use Motor Vehicle Speed Pavement Condition Pavement Condition Table 1 LOS Scores and Letter Grades LOS Score LOS Letter Grade 0-2 A 2.001-2.75 B 2.751-3.5 C 3.501-4.25 D 4.251-5 E 5.001 and up F The HCM BLOS calculations were carried out using a spreadsheet created by the MPO. Input values (blue text) are columns C-R (columns S-W are for intersection and segment BLOS). Intermediate calculations are done in hidden columns Y-AR). The BLOS link score and letter grade are in columns AU and AV. Comments are included in the header rows describing the variables to assist in user input. All percentages must be inputted as values less than one (i.e., 75% or 0.75). The BLOS spreadsheet makes a few assumptions that are not explicitly included in the HCM formulas, as shown below. 4/16/14 Page 3
Table 2 Assumptions in BLOS Calculations Variable Traffic volume Gutter pan Parking on shoulders Assumption The daily traffic volumes (ADT) are inputted. The spreadsheet assumes a directional split of 56.5% for two-way roads and a peak hour to ADT ratio of 8% for high volume roads and 10% for low volume roads (less than 5,000 ADT) to convert ADT to peak hour directional volume. Per the HCM calculations, 1.5 is removed from shoulders with curbs. Since most shoulders in Madison with curbs are parking lanes and most actual gutter pans are 2 or 1, the curb column should be ignored and all lane widths exclude the gutter pan. The total width of outside through and bicycle lane (W t ) includes the shoulder only if the parking utilization is zero. However, with wide shoulders, the BLOS score drops extremely quickly resulting in a numerical value of less than zero with a letter score of better than A. The interpretation is that wide shoulders should be treated like unused parking lanes, not bike lanes, so shoulder widths greater than 6 automatically have a parking utilization of at least 1%, causing the width to not be included in W t. A simple sensitivity analysis was done to determine the relative effect of the variables. The starting scenario is a fictional two-way roadway with one 11 foot travel lane in each direction, 8,000 ADT, no bicycle lanes or shoulders, 30 mph speed, pavement condition rating of 3, and 2% heavy vehicles. The results of various changes are shown below. Table 3 BLOS Sensitivity Analysis Scenario LOS Score LOS Base scenario 4.21 D Add 5' bike lane 2.61 B Add 5' shoulder 2.61 B Add parking, 25% 4.09 D Add parking, 50% 4.57 E Add parking, 75% 4.79 E Add parking 50% and 5' bike lane 3.37 C Widen driving lane to 14 3.83 D Halve traffic volume to 4,000 ADT 3.97 D Double traffic volume to 16,000 ADT 4.56 E Reduce speed to 20 mph 3.46 C Increase speed to 40 mph 4.43 E Reduce pavement condition to 1 10.49 F Increase pavement condition to 5 3.70 D Reduce heavy vehicles to 0% 3.90 D Increase heavy vehicles to 4% 4.57 E The base scenario with no bicycle lane and moderate traffic volumes produces a reasonably expected LOS score of a low D. Adding bicycle lanes or shoulders improves the score to B, the formulas do not distinguish between bicycle lanes and shoulders. Adding a small amount of parking slightly improves the score, but any more than about 30% occupancy causes the score to gradually drop off towards E, but does not reach F even with 100%. Traffic volumes have a noticeable effect, as do traffic speeds and heavy vehicles. Reducing the pavement condition to below 3 has a major effect, dropping the score to a very low F (A pavement condition rating of 2 also produces an F). 4/16/14 Page 4
Data Collection and GIS Layers The effectiveness of the BLOS analysis is only as good as the quality of data that is used. This analysis does not have a directional component each roadway segment represents both directions of two-way streets. Most data needed for the analysis traffic volume, number of through lanes, traffic speed, heavy vehicle use, and pavement condition were taken or derived from existing data sources. The basic segmentation of the road network and much of the data are from the Wisconsin Information System for Local Roads (WISLR) and other Wisconsin Department of Transportation (WDOT) sources. Where available, data was augmented by sources from the City of Madison Traffic Engineering Division (CMTE) and City of Madison Engineering Department (CME). Road speed limits are from the Dane County Land Information Office (DCLIO). The only data that was directly collected for the project were lane widths and parking utilization, which were mostly collected from aerial photographs. Table 2 Data from Existing Sources for the BLOS Analysis BLOS Attribute Name Source * Notes WISLR_OVLY_ID WISLR, WISLR_OVLY_ID (WDOT) Unique segment ID OneWay MPO_FNCT_CLS_GRP Study LenMi LanesFull (all lanes) LanesTh (through lanes each direction) ADT TRUCK WISLR, RDWY_OWRST_INDC (WDOT) updated with STREETCENTERLINES, oneway (CMTE) WISLR, FNCT_CLS_GRP (WDOT) updated with ROADS CURRENT, FAC2004 (MATPB) WISLR, DIR_INDC (WDOT) MPO_FNCT_CLS_GRP Aerial photography WISLR, TRLNS_NB (WDOT) updated with 2010 BASE HIGHWAY NETWORK, LANES (MATPB) Indicates whether or not the road is oneway Used to exclude local streets from the study and estimate heavy vehicle use Segments with DIR_INDC = O (Opposite) were excluded in the study Only Collector and Arterial roads were used in the analysis Rural roads were excluded from the study Length of segment, miles Full number of lanes in both directions 0.5 x LanesFull, round down Number of lanes per direction WISLR, AVG_DLY_TRF (WDOT) updated with FLOWMAP, AWT_DIR1_2 (CMTE) and 2013 TRADS (WDOT) and manual interpolation. Truck Route map (CME) and Wisconsin Long Truck Operators map (WDOT) Average daily traffic volume HV TRUCK, MPO_FNCT_CLS_GRP Roads identified as truck routes are assigned assumed heavy vehicle percentages of 0% for collectors, 2% for minor arterials, and 3.5% for principle arterials. 4/16/14 Page 5
SPD ROADCENTERLINE, SPEEDLIMIT (DCLIO) updated with STREETCENTERLINES, SPEED_LIMIT (CMTE) Posted speed limit. Average vehicles speeds for the analysis are assumed to be 7 mph above the posted speed limit. Wot Aerial photography, engineering plans Width of outside through lane. Measurement does not include the width of the gutter pan and does not include the widths of the shoulder, parking, or bike lanes. Wbl Aerial photography, engineering plans Width of bike lane. Measurement does not include the width of the gutter pan. It includes the width of painted or physical barriers used to separate the bike lane from parking and through lanes. Wos Aerial photography, engineering plans Width of outside shoulder or parking lane. Measurement does not include the width of the gutter pan. Width is assumed to be 8 feet minus the width of the gutter pan (generally 2 feet), or generally 6 feet, unless otherwise delineated. ParkNE ParkSW MPO_WISLR_SFRTG_VAL Aerial photography, local knowledge WISLR, MPO_WISLR_SFRTG_VAL (WDOT) and PCI_MAD_MPO_2013, PAVE_DISTR (WDOT) updated with STREETCENTERLINES, PVMT_RATING (CME) State roadways used Estimated weekday daytime parking utilization (percent) for the north/east and south/west direction. PASER pavement condition rating (1-10 rating scale where 1 is failed and 10 is excellent ) BLOSPC MPO_WISLR_SFRTG_VAL Condition rating used for BLOS (0 is poor, 5 is new) BLOS_PC = 0.5 x MPO_WISLR_SFRTG_VAL * Name of data set, Name of attribute (Agency) Lane widths (outside through, bicycle lane, and parking/shoulder lane) were generally measured from aerial photographs (Google Earth or Dane County imagery). In some cases, condition diagrams or construction drawings from the City of Madison were used. The width of gutter pans was not included in the measurement. Parking lanes that are not marked with an edge line were assumed to be 8 feet minus the width of the gutter pan (6 feet in most cases). For asymmetrical two-way roads, the less favorable direction was used. Lane width measurement was by far the most labor intensive part of the analysis. The lane configuration used for the analysis was considered to be what is in place off peak, which represents the conditions the majority of the time. It should be noted that this assumption contradicts the use of peak period traffic volumes the result is expected to yield a level of service that reflects conditions somewhere in between the peak and off-peak roadway conditions. 4/16/14 Page 6
Bus, bicycle, and right-turn lanes are treated as a shared travel lane. Lower traffic volumes (5-10%) were assumed to account for the exclusion of regular through traffic. University Avenue between Randall Avenue and Bassett Street has an abnormal configuration with a bus/right-turn lane, bicycle lane, three one-way through lanes, a median, and a counter-flow bicycle lane. The westbound direction was used since it is the less favorable and the street was treated as a one-way street. The bus/right-turn lane was not counted in the measurements and the approximately 8 bicycle lane was adjusted to 6 to account for the thick stripe and shy distance on both sides of the lane with high traffic volumes on one side and buses and right turns on the other these conditions do not appear to be accounted for in the formulae. Parking utilization is measured from aerial photographs (Google Earth or Dane County imagery) and occasionally adjusted with known parking patterns. The percent is defined as the total length of curb occupied by parking divided by the total length of curb (including no parking zones). Parking utilization was estimated for both directions and averaged. It should be noted that this measurement is very approximate and could be substantially different if observed on different days. 4/16/14 Page 7
Conclusion The results of the BLOS analysis are shown in Figure 2. In general the results are reasonable, showing high scores (A and B) for lower volume roads with bicycle lanes and lower scores (E and F) for higher volume roads without bicycle lanes. The sensitivity to bicycle/shoulder lane width can be seen in some segments for instance, Campus Drive (a high volume roadway with limited access and shoulders) scores a C because of its shoulders, even though most cyclists avoid it in favor of the parallel off-street path. The sensitivity to on-street parking and heavy vehicles is illustrated on Gorham Street (a one-way two-lane street with a 12 parking and bicycle lane) it shows a BLOS score of E because of the parking and heavy vehicles despite having dedicated space for bicycling. Several road segments with very low traffic volumes resulted in scores that are less than zero. These segments were assigned scores of >A, or better than A. These results that were somewhat unexpected were not adjusted. Figure 2 The general conclusion from staff involved in the BLOS analysis project is that the methodology is generally sound but could benefit from continued refinement. For instance, staff felt that several high volume, high speed roads that received C s, D s, and E s would be considered solid F s by most cyclists, experienced or not. The formulae may be too sensitive to several factors, such as parking, shoulder width, and pavement condition, but that reaction is not based on formal testing of the model. The HCM methodology calculates delay for intersections, but that calculation is not included in the intersection or segment BLOS. Additionally, the methodology may be improved upon by including other factors to account for treatments like painted bicycle lanes, cycle tracks, sharrows, intersection treatments, topography, and marked bicycle lanes compared to shoulders. 4/16/14 Page 8
Appendix A HCM Methodology for Intersection and Segment Length of Intersection Figure A-1 Intersection Width of Outside Through Lane, Bicycle Lane, and Shoulder Cross Section BLOS Intersection Score Traffic Volume Number of Through Lanes Motor Vehicle Speed Signal Timing Bicycle Volume Bicycle Delay Link LOS Intersection LOS Figure A-2 Segment BLOS Segment Score Number of Access Points (Driveways) 4/16/14 Page 9
Figure A-3 Multi-Use Path Path User Flow Rates (bikes, peds, others) Passing and Meeting Events Delayed Passings BLOS Score Path Width Presence of Centerline 4/16/14 Page 10
Appendix B Pedestrian on Sidewalk Figure B-1 Pedestrian Link Sidewalk Geometry (width, terrace, etc) Outside Lane Geometry Parking Utilization Cross Section PLOS Link Score Traffic Volume Number of Lanes Traffic Speed Motor Vehicle Volume Motor Vehicle Speed Sidewalk Geometry (width, terrace, etc) Ped Flow Rate Pedestrian Space 4/16/14 Page 11
Figure B-2 Pedestrian Intersection Number of Lanes Crossed Corner Geometry Traffic Volumes (turning and adjacent) Traffic Speed Cross Section Motor Vehicle Volume Motor Vehicle Speed PLOS Link Score Signal Timing Pedestrian Delay Corner Geometry Crosswalk Width Pedestrian Space Ped Flow Rate Figure B-3 Pedestrian Segment Intersection Spacing Signal Timing Link Score Roadway Crossing Difficulty Factor PLOS Segment Score Intersection Score 4/16/14 Page 12