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.
82 7.1 Development Two basic forms were considered for the bicycle LOS for arterials model. The first was an aggregate model using the outputs from existing segment and intersection LOS models to determine the arterial LOS. The other was an agglomerate model considering the independent characteristics of the roadway environment to calculate an arterial LOS for bicy- clists directly. Both forms were preliminarily evaluated dur- ing model development. The aggregate model was chosen for refinement for several reasons. The stepwise approach to an aggregate model is use- ful because it allows the practitioner to address concerns at individual intersections or along specific segments. The ag- gregate model also retains all the terms found both intuitively and mathematically to be significant to bicyclists riding along a roadway. The agglomerate model would not retain all the terms as significant. Consequently, we focused on the aggre- gate model in model development efforts. We considered various functional techniques for model development, including linear regression and ordered probit. We performed linear regression modeling because it is more intuitive than probit modeling in practice and non-modelers better understand the sensitivity of the regression model. These reasons are particularly important in that these mod- els are most frequently used: the development or analysis of specific design options or in the development of bicycle facil- ity community master plans with presentations to interested citizens and public officials. To ensure the validity of the re- sults of the linear regression modeling results, we evaluated the ordered probit model form as well. The results of both the linear regression and ordered probit modeling efforts are de- scribed below. Before starting correlations analysis and modeling, we cre- ated two data subsets from the overall dataset. The total dataset was sorted by city and LOS grade responses. A ran- dom sampling of 20% of the data representing each city and LOS grade response was taken from the overall dataset for model validation. The balance of the data, 80% of the total dataset, was used for model development. SPSS 14.0 was used to conduct Pearson correlation analysis on the extensive array of geometric and operational variables. Subsequently, we selected the following relevant variables for additional testing: ⢠Segment LOSâThe bicycle LOS for roadway segments (see below). ⢠Intersection LOSâThe bicycle LOS for signalized inter- sections (see below). ⢠Conflicts per mileâThe total conflicts per mile represent the motor vehicle conflicts resulting from motorists turn- ing across the bicycle facility at unsignalized locations. ⢠Size of the city in which the data collection took placeâ The Metropolitan Statistical Area (MSA) population was used to represent the size of each city. At the panelâs request, the MSA variable was dropped from further consideration. Other variables were dropped from further consideration because of their poor correla- tion with the dependent variable or because of their colin- earity with more strongly correlated variables. After testing numerous combinations of variables and variable transfor- mations, we determined the aggregate model using two constituent sub-models would be the most theoretically valid. 7.2 Recommended Bicycle LOS Model The recommended bicycle LOS model is a weighted combination of the bicyclistsâ experiences at intersections and on street segments in between the intersections. Two models of the same form were evaluated, but with different parameters: C H A P T E R 7 Bicycle LOS Model
Bicycle LOS Model 1 Bicycle LOS #1 = 0.160*(ABSeg) + 0.011*(exp(ABInt)) + 0.035*(Cflt) + 2.85 (Eq. 29) Bicycle LOS Model 2 Bicycle LOS #2 = 0.20*(ABSeg) + 0.03*(exp(ABInt)) + 0.05*(Cflt) + 1.40 (Eq. 30) Where ABSeg = The length weighted average segment bicycle score Exp = The exponential function, where e is the base of nat- ural logarithms. ABInt = Average intersection bicycle score Cflt = Number of unsignalized conflicts per mile, i.e., the sum of the number of unsignalized intersections per mile and the number of driveways per mile The output of either model is a numerical value, which must be translated to a LOS letter grade. Exhibit 91 provides the numerical ranges that coincide with each LOS letter grade. The first model provides a better fit with the numerical scores given by the video lab participants to the video clips. This model was derived based on a statistical fitting process to the video clip data. However, this first model does not pre- dict LOS A or B for the video clips. Consequently the second model was developed. The second model has an inferior numerical fit with the video lab data (measured in terms of squared error) but pro- duces the full range, LOS A through F, for the video clips. The second model was derived from the first model by reducing the constant so that the second model would predict LOS A for video clips #328 and #330. The other parameters in the model were then manually adjusted until the second model could produce LOS F for one or more of video clips #314, 317, 323, and 324 (which were rated LOS F by the video lab participants). Both models use the same bicycle segment and bicycle in- tersection submodels. Bicycle Segment LOS The segment bicycle LOS is calculated according to the following equation: BSeg = 0.507 Ln (V/(4*PHF*L)) + 0.199Fs*(1 + 10.38HV)2 + 7.066(1/PC)2 â 0.005(We)2 + 0.760 (Eq. 31) Where BSeg = Bicycle score for directional segment of street. Ln = Natural log PHF = Peak Hour Factor (see Chapter 10 for default values) L = Total number of directional through lanes V = Directional motorized vehicle volume (vph). (Note: V > 4 *PHF * L) Fs = Effective speed factor = 1.1199 In(S - 20) + 0.8103 S = Average running speed of motorized vehicles (mph) (Note: S >= 21) HV = Proportion of heavy vehicles in motorized vehicle volume. Note: if the auto volume is < 200 vph, the %HV used in this equation must be <= 50% to avoid unrealis- tically poor LOS results for low volume and high percent HV conditions. PC = FHWAâs five point pavement surface condition rat- ing (5=Excellent, 1=Poor) (A default of 3 may be used for good to excellent pavement) We = Average effective width of outside through lane (ft) = Wv â (10ft à %OSP) (ft) ** If W1 < 4 = Wv + W1 â 2 (10 à %OSP) (ft) ** Otherwise %OSP = Percentage of segment with occupied on-street parking W1 = width of paving between the outside lane stripe and the edge of pavement (ft) Wv = Effective width as a function of traffic volume (ft) = Wt (ft) ** If V > 160 vph or street is divided = Wt*(2 â (0.005 à V)) (ft) ** Otherwise Wt = Width of outside through lane plus paved shoulder (including bike lane where present) (ft) Note: parking lane can be counted as shoulder only if 0% occupied. Bicycle Intersection LOS The intersection bicycle LOS is calculated according to the following equation: IntBLOS = â0.2144Wt + 0.0153CD + 0.0066 (Vol15/L) + 4.1324 (Eq. 32) Where IntBLOS = perceived hazard of shared-roadway environment through the intersection Wt = total width of outside through lane and bike lane (if present) CD = crossing distance, the width of the side street (including auxiliary lanes and median) 83 Exhibit 91. Bicycle LOS Numerical Equivalents. LOS Numerical Score A ⤠2.00 B >2.00 and ⤠2.75 C >2.75 and ⤠3.50 D >3.50 and ⤠4.25 E >4.25 and ⤠5.00 F > 5.00
84 Vol15 = volume of directional traffic during a 15-minute period L = total number of through lanes on the approach to the intersection 7.3 Performance of Bicycle LOS Model on Video Clips Exhibit 92 compares the ability of the existing HCM speed- based LOS model and the proposed bicycle LOS models against the mean LOS response for each video clip. The HCM matched the video clips 15% of the time. The proposed LOS models matched the clips between 27% and 46% of the time. The second model had the highest percentage match because it can predict LOS A and B. The first model is better at pre- dicting the poorer levels of service (E and F) than the second model.
Clip Location Outside Lane (ft) Bike/Shldr Lane (ft) Through Lanes Divided (D/UD) Pk Hr. Vol. (vph) Heavy Veh (%) Spd Lim (mph) Pavement Rate (1-5) % OSP Sig. Int X-Dist (ft) Unsig. Conf Per Mile Video LOS HCM LOS Model #1 LOS Model #2 LOS 328 N Village, Cypress/S Vill.(N) 12 4 1 U 79 0% 30 4.0 0% 0 5.5 A B C A 330 N Village, S Vill./Cypress (S) 12 4 1 U 136 0% 30 4.0 0% 0 6.7 A B C A 306 Alumni at Magnolia (N) 11 4 2 U 717 0% 30 4.0 0% 72 0.0 B B C B 305 Collins at Alumni (W) 12 3.5 2 D 813 8% 30 3.5 0% 65 0.0 B C D B 307 Alumni at Magnolia (N) 11 4 2 U 757 0% 30 4.0 0% 72 0.0 B D C B 304 Collins Blvd at Alumni (E) 12 3.5 2 D 428 0% 30 3.5 0% 65 0.0 B E C B 303 Fowler Ave at North 12 5 3 D 1211 0% 50 4.0 0% 0 26.4 C B D C 319 Fletcher, North Palm (S) 12 5 2 D 2961 0% 45 4.0 0% 53 0.0 C B D D 311 15th St at 7th Ave, (W) 12 8 1 U 631 0% 25 3.5 70% 33 13.2 C D D C 329 Ehrlich, Turner/S Village (S) 12 4 2 D 1261 0% 45 3.5 0% 61 8.8 D B D C 302 Fowler Ave at North 12 5 3 D 2119 0% 50 4.0 0% 0 26.4 D B D C 327 W University at 10th (S) 12 8 2 U 165 0% 30 3.0 40% 40 20.8 D C D C 309 Holly at Laurel , (S) 10 0 2 U 134 0% 20 4.0 0% 52 0.0 D D C B 313 21st St at 7th Ave, (W) 10 0 3 OW 536 0% 30 3.5 0% 33 24.0 D D E D 308 Holly at Magnolia , (N) 10 0 2 U 407 0% 20 4.0 0% 86 0.0 D D D C 320 Fletcher Ave at 50th (S) 12 5 2 D 1898 0% 45 4.0 0% 64 0.0 E B D B 321 Fletcher, 50th to 56th (S) 12 5 2 D 2146 0% 45 4.0 0% 0 15.2 E B D C 318 US 41 at 31st St, (W) 12 0 3 D 182 100% 55 3.5 0% 35 24.0 E B F F 322 56th St at 98th Ave, (W) 12 0 3 D 1544 0% 45 3.5 0% 0 28.7 E B E D 310 Fletcher at Sebring, (S) 11.5 0 2 D 1589 0% 40 4.0 0% 0 37.0 E B F E 301 Fowler, River H./Gillette 12 5 3 D 2549 0% 50 4.0 0% 0 28.9 E B E D 312 7th Ave, 17th to 14th (N) 12 0 1 U 631 0% 25 3.5 0% 49 5.0 E C D C 317 US 41 at Dover St, (E) 12 0 2 D 495 17% 55 3.0 0% 0 11.5 F B E D 314 56th St at Busch (W) 12 0 2 D 638 0% 45 3.5 0% 142 28.7 F B F F 323 56th St at Busch (W) 12 0 3 D 357 0% 45 3.5 0% 142 28.7 F E E F 324 Bullard at 56th St 12 4 3 D 636 0% 45 4.0 0% 87 19.6 F E D C % Exact Match to Video 100% 15% 27% 46% % Within 1 LOS of Video 100% 50% 85% 77% Exhibit 92. Evaluation of Proposed Bike Models and HCM against Video Lab Results.