Design Guidance for High-Speed to Low-Speed Transitions Zones for Rural Highways (2012)

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.

23 S e c t i o n 3 The objective of the field studies was to evaluate the effectiveness of a treatment (or combina- tion of treatments) in reducing speeds through a transition zone and through the community. Results of the field studies are used to supplement information gathered during the literature review on effectiveness estimates of treatments implemented within transition zones, and the combined results are used to develop the design guidelines presented in Section 4. The effectiveness of the following three treatments (and combinations of treatments) in reducing speeds through a transition zone and through the community are assessed: • Roundabouts (RAs), • Transverse pavement markings (TPMs), and • Welcome signs (WSs) at community entrances. In addition to evaluating the effectiveness of these treatments in reducing speeds through a transition zone and through the community, crash data are also presented to document the safety experience associated with the treatments and sites. 3.1 Treatment and Site Selection The three treatments were selected for evaluation by first conducting a survey of state and county highway agencies within the United States and combining the results with the survey from NCHRP Synthesis 412. The purposes of combining results from the two surveys were as follows: 1. To develop a catalog of speed reduction treatments that appear most reasonable for imple- mentation within transition zones and have the greatest likelihood for achieving the desired results. 2. To identify potential sites for the field data collection. After organizing the results of both surveys, 11 treatments were identified for potential fur- ther investigation (Table 3-1). The treatments were also prioritized for evaluation based upon the reliability of effectiveness estimates of the treatments to reduce speeds and improve safety, expected benefits (e.g., reducing speed), and availability of data collection sites. Working with contacts at state highway agencies to identify potential study locations, three treatments from the high-priority list—roundabouts, transverse pavement markings, and welcome signs—were selected for further evaluation. Field data were collected at 15 treatment and seven non-treatment sites. A treatment site is defined as a location at which one or more speed reduction treatments was installed within the transition zone leading to the lower speed environment (i.e., a community), and a non-treatment Field Studies

24 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways site is defined as a location at which no particular speed reduction treatment was installed near the transition zone with the exception of reducing the posted speed limit. Where possible, field data were collected on multiple approaches to a community (see Figure 3-1). All sites included in the study had at least a 15-mph reduction in the posted speed limit going from the high-speed to the low-speed environment. In addition, study locations were chosen with few or no intersections or driveways in close proximity to the transition zones. This was important in order to keep to a minimum the number of vehicles that may enter or exit the stream of traffic between speed data collection points. Limiting the number of inter- sections/driveways in the transition zone also helped to ensure that observed accelerations and decelerations were a result of compliance to posted speed limits and/or in response to the speed reduction treatment, rather than to accommodate vehicles entering and exiting the roadway. Table 3-2 lists the locations of the data collection sites and the types of treatments imple- mented within the transition zones. At several locations, multiple treatments were installed in combination. At these locations, the primary treatment (i.e., the treatment that likely has the greatest impact on speed) is listed first and the secondary treatment is shown in parentheses. Table 3-3 presents the site characteristics of the study locations within the transition zones. The speed limit columns indicate the posted speed limit at the beginning and end of the tran- sition zone and the posted speed limit through the community. The reductions in speed limits through the transition zones ranged from 15 to 35 mph. All of the roadways through the transition zones were two-lane undivided facilities (2U). Lane widths ranged from 11 to 13 ft, and shoulder widths ranged from 0 to 10 ft, both paved and unpaved. The horizontal alignment column indicates whether the study location was a tangent (TAN) section of roadway or if there was some degree of curvature (Curve) in the vicinity of the transition zone. The vertical alignment column indicates whether there was rolling terrain, a constant grade (i.e., level or downgrade), or vertical curvature Table 3-1. Prioritized list of treatments for further evaluation. Priority for further evaluation Treatment(s) High priority Roundabouts Transverse pavement markings Raised medians Center islands Welcome signs Landscaping Medium priority Transverse rumble strips Roadway narrowing Bicycle lanes Low priority Speed-activated feedback signs Pedestrian crossings Figure 3-1. Study locations on multiple approaches to a community. Study Locations: Treatment(s) in place No treatment(s) in place Town A Town B TownD Town C

Field Studies 25 Table 3-2. Study locations and treatment types. Site no. Treatment Community (state) Route (direction) KS01 TPM Rossville (KS) US 24 (WB) KS02 None Rossville (KS) US 24 (EB) KS03 Welcome sign McLouth (KS) SR 92 (WB) KS04 Welcome sign McLouth (KS) SR 92 (EB) KS06 TPM (welcome sign) Silver Lake (KS) US 24 (EB) KS09 Roundabout (rumble strips) Fredonia (KS) K 47/US 400 (WB) KS10 Welcome sign Burden (KS) US 160 (EB) KS11 Welcome sign Burden (KS) US 160 (WB) KS12 None Rock (KS) US 77 (NB) KS13 None Rock (KS) US 77 (SB) KS14 TPM Meriden (KS) Rt 4 (SB) KS15 TPM Meriden (KS) Rt 4 (NB) NE01 Roundabout (welcome sign) Blair (NE) US 30/Rt 133 (EB) NE02 None Blair (NE) US 75 (SB) NE03 Roundabout Winnebago (NE) US 77 (NB) NE04 None Winnebago (NE) US 77 (SB) IA01 TPM (welcome sign) Roland (IA) Rt 77 (NB) IA02 TPM (welcome sign) Union (IA) Co Rd D65 (EB) IA03 TPM Union (IA) SR 215 (SB) IA05 None McCallsburg (IA) Co ED E18 (WB) VA01 Roundabout Amherst (VA) US 60 (EB) VA02 None Amherst (VA) US 60 (WB) Note: KS = Kansas, NE = Nebraska, IA = Iowa, VA = Virginia, WB = westbound, EB = eastbound, NB = northbound, and SB = southbound. Table 3-3. Site characteristics of study locations within the transition zones. 30 Site no. Treatment Speed limit (mph) ADT Number of lanes Lane width (ft) Shoulder Horiz. align. Vert. align. Begin End Comm. Width (ft) Type KS01 TPM 65 45 2 4,540 2U 12 4 paved Tan Level KS02 None 65 30 30 4,310 2U 12 3 paved Tan Level KS03 Welcome Sign 55 35 35 2,940 2U 12 0 none Tan Rolling KS04 Welcome Sign 55 35 35 2,980 2U 12 0 none Tan Rolling KS06 TPM (welcome sign) 65 30 30 4,240 2U 12 3 paved Curve Level KS09 Roundabout (rumble strips) 65 30 30 2,200 2U 12 3 paved Tan Level KS10 Welcome Sign 65 40 302 1,700 2U 12 1.5 unpaved Tan Crest KS11 Welcome Sign 65 30 30 870 2U 12 0 none Tan Sag KS12 None 65 45 45 2,480 2U 12 5 unpaved Tan Level KS13 None 65 45 45 2,800 2U 12 5 unpaved Curve Level KS14 TPM 65 45 45 6,700 2U 13 2 paved Tan Downgrade KS15 TPM 65 45 45 7,890 2U 13 4 paved Tan Downgrade NE01 Roundabout (welcome sign) 60 45 45 15,150 2U 12 8 paved Tan Downgrade NE02 None 60 45 45 5,685 2U 12 8 paved Curve Level NE03 Roundabout 60 30 30 4,465 2U 12 8 paved Tan Downgrade NE04 None 60 30 30 6,235 2U 12 10 paved Tan Level IA01 TPM (welcome sign) 55 25 25 670 2U 11 2 unpaved Tan Rolling IA02 TPM (welcome sign) 55 25 202 470 2U 11.5 0 none Tan Rolling IA03 TPM 55 30 30 1,120 2U 12 2 unpaved Tan Downgrade IA05 None 55 45 252 790 2U 11 1 unpaved Tan Crest VA01 Roundabout 55 35 35 2,400 2U 12 2 paved Curve Rolling VA02 None 55 35 35 6,200 2U 12 2 paved Curve Rolling Note: Comm. = community, ADT = average daily traffic, horiz. = horizontal, align. = alignment, and vert. = vertical. 1Highest approach volume through the roundabout. 2At these locations, the speed limit at the end of the transition zone was not the same as the speed limit through the entire community. This occurred either because the speed limit changed within the community zone downstream of where the transition zone should have ended and/or because of unique site characteristics that limited the data collection location.

26 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways (i.e., crest or sag) in the vicinity of the transition zone. In addition to the characteristics listed in Table 3-3, none of the transition zones had sidewalks, and all of the roadsides had grass ditches for drainage. Table 3-4 lists the site characteristics of the study locations within the communities, down- stream of the transition zones. At a few locations, multilane roadways were present through the communities (i.e., three-lane roadway [3T] including a center two-way left-turn lane [TWLTL]; four-lane undivided roadway [4U]; and five-lane roadway [5T] including a center TWLTL), and in some instances, the travel lanes were wider through the communities than within the transi- tion zones. On-street parking, sidewalks, and curbs were present in several of the communities. The number of signalized and unsignalized intersections through the community, the length of the community, the land area, and population size are several other indicators of the level of activity through the community that could potentially affect speeds. 3.2 Speed Study A speed study was conducted to evaluate the effectiveness of a treatment (or combination of treatments) in reducing speeds through the transition zone and maintaining that reduction through the community. 3.2.1 Speed Data Collection Methodology Speed data were collected using traffic classifiers and laser guns. Traffic classifiers were used to collect information about the flow of traffic as it approached the transition zone, exited the transition zone, and entered or passed through the community. Key information obtained from the classifiers included the following: • Spot speeds of vehicles, • Vehicle classifications, and • Vehicle spacing. Traffic classifiers were used to collect data at three locations at each study site. Figure 3-2 illustrates the conceptual layout of the traffic classifier locations relative to the transition zone and community. The locations of the traffic classifiers at each study site were determined using the following protocols: • Location C—Transition Zone Perception/Response Speed: The speed at this location rep- resents the speed of vehicles along the high-speed portion of roadway prior to the transition zone. At this location, there should be no effect on vehicle speed either due to a transition zone treatment intended to reduce speed or due to a speed limit sign that indicates a reduction in the speed limit. – This traffic classifier was placed approximately 200 to 400 feet upstream of the first speed limit sign to indicate a reduction in the speed limit, or the first treatment intended to reduce vehicle speeds, whichever was first encountered by vehicles. – Placement of this traffic classifier was not influenced by the presence/location of “Reduced Speed Limit Ahead” warning signs. • Location B—Transition Zone Exit Speed: The speed at this location represents the speed of vehicles slightly downstream of the transition zone. – This traffic classifier was placed approximately 3 to 5 seconds downstream of the final speed reduction treatment or the first speed limit sign to indicate the speed limit through the community, whichever was located further downstream (i.e., closest to the community).

Table 3-4. Site characteristics of study locations within the community. Site no. Number of lanes Lane width (ft) Shoulder Horiz. align. Vert. align. On-street parking Sidewalks Curbs Number of intersections Length of comm. (mi) Area of comm. (mi2) Pop. Width (ft) Type Signal Stop KS01 2U 12 13 paved Curve Level N Y Y 0 10 2.5 0.61 1,151 KS02 2U 12 11 paved Curve Level Y N Y 0 10 2.5 0.61 1,151 KS03 2U 12 0 N/A Tan Level N N N 0 11 0.9 0.59 880 KS04 2U 12 0 N/A Tan Level N N N 0 11 0.9 0.59 880 KS06 2U 12 2 paved Curve Level Y Y Y 0 15 1.3 0.60 1,439 KS09 2U 12 1 paved Tan Level N N N 0 18 1.5 2.45 2,482 KS10 2U 21.5 0 N/A Tan Level N Y Y 0 7 0.7 0.54 535 KS11 2U 21.5 0 N/A Tan Crest N Y Y 0 7 0.7 0.54 535 KS12 2U 12 2 paved Tan Level N N N 0 3 0.4 5.53 222 KS13 2U 12 2 unpaved Tan Level N N N 0 3 0.4 5.53 222 KS14 2U 13 4 paved Curve Upgrade N N N 0 5 2.3 0.79 813 KS15 2U 13 3 paved Curve Level N N N 0 5 2.3 0.79 813 NE01 5T 12 0 N/A Curve Level N Y Y 5 17 3.0 5.51 7,990 NE02 2U 13 4 paved Curve Downgrade N Y Y 5 17 3.0 5.51 7,990 NE03 3T 15 0 N/A Tan Crest N Y Y 0 10 1.1 0.20 774 NE04 3T 15 0 N/A Tan Crest N Y Y 0 10 1.1 0.20 774 IA01 2U 11 0 unpaved Tan Level N Y N 0 7 0.7 1.08 1,284 IA02 2U 26.5 0 N/A Tan Downgrade Y Y Y 0 9 0.5 0.55 397 IA03 2U 12 8 paved Tan Level Y Y Y 0 12 0.7 0.55 397 IA05 2U 11 0 N/A Tan Level N Y N 0 8 0.6 0.53 333 VA01 4U 12 3 paved Tan Level N N N 0 3 0.4 4.92 2,231 VA02 4U 12 2 paved Tan Level N N N 0 3 0.4 4.92 2,231

28 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways • Location A—Community Speed: The speed at this location represents the speed of vehicles through the community. – This traffic classifier was placed approximately 750 to 1,000 feet downstream of the final speed reduction treatment or the first speed limit sign to indicate the speed limit through the community, whichever was located further downstream. – This traffic classifier was placed prior to any signal or stop-controlled intersection that would have the potential to affect the speeds of vehicles. Placements of the classifiers at the study sites were adjusted as necessary to account for unique field conditions. For sites with no treatment, the traffic classifiers were placed in locations analo- gous to Locations A, B, and C. In general, at least 6 to 8 hours of speed, classification, and head- way data were collected at each study site on a weekday. Speed data were also collected using laser guns to assess vehicle speed profiles through portions of the transition zones. Speed data from the laser guns were collected for a sampling of free-flow vehicles to supplement the information collected from the traffic classifiers and to provide more detailed information on the deceleration characteristics of vehicles. While speed data were being collected from the traffic classifiers and a laser gun, traffic opera- tions within the transition zone were also recorded using a video camera positioned on the roadside. In general, 2 to 3 hours of vehicle operations were recorded at each study location. The video recordings enabled the research team to go back and review driver behavior, as necessary, within the transition zone. During reduction of the speed data from the traffic classifiers, an effort was made to track individual vehicles traveling through the transition zone and community from classifier Loca- tions C to B to A based upon speed, headway, and time stamp (i.e., internal clocks) data from the respective classifiers. Being able to track individual vehicles through the study area would have created a more robust dataset, but because of the distances between the classifiers and due to some vehicles entering from intersections/driveways within the study area, it was too difficult to track individual vehicles reliably using the classifier data. Therefore, the speed data presented below and the subsequent analyses are based upon speeds of individual free-flow vehicles mea- sured at each data collection location grouped together as a whole, rather than representing speed profiles of individual vehicles through the study area. 3.2.2 Descriptive Speed Statistics Speed data were collected at a total of 22 sites. Of the 22 sites, 15 sites represent a total of six unique treatment combinations, and seven sites had no treatment. The sites, their treatment Figure 3-2. Setup of field equipment. Transition Zone Low-Speed RegionHigh-Speed Region Transition Zone Perception/Response Speed Transition Zone Exit Speed Community Speed Drawing Not To Scale C B A

Field Studies 29 combinations, and reduction in posted speed limits from Location C to B and Location B to A are shown in Table 3-5. Summary speed statistics from the classifiers (box plots) and speed profiles from the laser gun (blue distributions) used concurrently at each site are graphically displayed in Appendix A, separately for each site; an example diagram is shown below in Figure 3-3. Selected site details such as locations of posted speed limits, intersections, and treatment(s) are also indicated in each figure. Other details about these figures are given in Appendix A. In all analyses, tables, and figures, only speeds recorded between 6 a.m. and 10 p.m. were considered. Table 3-5. Treatment combinations and study sites. Treatment combination (primary and secondary) Site no. Reduction in posted speed limit (mph) From C to B From B to A Roundabout (RA) None NE03 30 0 VA01 20 0 Transverse Rumble Strips (TRSs) KS09 35 0 Welcome Sign (WS) NE01 15 0 Transverse Pavement Markings (TPMs) None IA03 25 0 KS01 20 15 KS14 20 0 KS15 20 0 Welcome Sign (WS) IA01 30 0 IA02 30 5 KS06 35 10 Welcome Sign (WS) None KS03 20 0 KS04 20 0 KS10 25 10 KS11 35 0 None IA05 10 20 KS02 35 0 KS12 20 0 KS13 20 0 NE02 15 0 NE04 30 0 VA02 20 0 Figure 3-3. Vehicle speed profile example.

30 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways Tables 3-6 through 3-9 present basic distribution statistics for all classifier speeds collected at Locations C, B, and A for each type of treatment, respectively. These statistics include the follow- ing for each location at each site: • Posted speed limit. • Number of free-flow vehicles (passenger cars and trucks combined) traveling between 6 a.m. and 10 p.m., for which valid speed data were available; this number represents the number of valid speed measurements recorded by each classifier (i.e., after data cleanup prior to statistical analysis) and as such is not a true representation of traffic volume. Table 3-6. Classifier speed statistics at sites with a roundabout. epyt tnemtaerT .on etiS reifissalC noitacol detsoP deeps timil )hpm( fo rebmuN selcihev )hpm( scitsitats deepS foegatnecreP selcihev deeps gnideecxe hpm X yb timil naeM dradnatS noitaived naideM ht58 elitnecrep ht59 elitnecrep 5 01 51 )AR( tuobadnuoR 30EN C 06 655,1 4.25 1.6 25 95 36 8.0 1.0 0.0 B 03 150,2 7.03 0.4 23 63 63 2.81 2.0 0.0 A 03 303,2 8.72 6.3 82 13 43 8.1 1.0 0.0 10AV C 55 695 7.84 4.01 15 85 16 9.7 0.1 3.0 B 53 188,1 9.13 4.5 23 73 04 8.3 1.0 0.0 A 53 519,1 1.63 0.5 63 14 44 8.81 1.3 2.0 dna )AR( tuobadnuoR elbmuR esrevsnarT )sSRT( spirtS 90SK C 56 843 6.85 8.7 06 56 86 1.1 3.0 0.0 B 03 057 4.23 4.4 23 73 04 3.02 8.4 0.0 A 03 062 1.24 6.8 44 25 65 8.58 4.55 7.23 dna )AR( tuobadnuoR )SW( ngiS emocleW 10EN C 06 271,2 6.55 1.7 65 36 66 4.6 0.1 1.0 B 54 372,4 5.34 3.6 54 94 25 4.01 0.1 0.0 A 54 177,4 2.04 7.7 24 64 84 2.2 1.0 0.0 Table 3-7. Classifier speed statistics at sites with transverse pavement markings. epyt tnemtaerT .on etiS reifissalC noitacol detsoP deeps timil )hpm( fo rebmuN selcihev )hpm( scitsitats deepS foegatnecreP gnideecxe selcihev X yb timil deeps hpm naeM adnatS rd noitaived naideM 58 ht elitnecrep 59 ht elitnecrep 5 01 51 esrevsnarT tnemevaP )sPMT( sgnikraM 30AI C 55 754 9.35 7.8 55 26 66 5.91 9.5 2.2 B 03 664 3.53 8.6 53 24 64 4.94 0.32 2.5 A 03 484 1.23 9.5 23 83 24 5.72 8.6 3.2 10SK C 56 927 9.85 3.6 95 56 86 9.1 3.0 0.0 B 54 759 1.53 1.5 53 14 44 3.0 0.0 0.0 A 03 888 2.03 7.4 03 43 83 0.31 6.2 3.0 41SK C 56 741,2 8.36 5.5 56 96 17 4.7 5.0 2.0 B 54 857,1 2.84 2.6 84 65 06 7.13 1.61 3.3 A 54 541,2 6.34 6.5 34 84 45 2.9 7.3 9.0 51SK C 56 545,2 5.06 5.5 16 66 86 1.1 1.0 0.0 B 54 865,2 6.34 0.5 34 84 25 1.8 2.1 2.0 A 54 126,2 1.44 1.6 54 84 25 4.8 1.2 3.0 esrevsnarT tnemevaP )sPMT( sgnikraM emocleW dna )SW( ngiS 10AI C 55 002 4.84 5.6 94 65 85 0.1 0.0 0.0 B 52 802 9.03 4.6 03 73 24 1.74 0.52 7.6 A 52 422 3.52 4.4 52 92 13 4.9 8.1 4.0 20AI C 55 521 0.15 5.7 15 95 36 2.11 0.0 0.0 B 52 941 0.03 0.7 82 63 44 7.94 2.62 4.5 A 02 432 4.42 7.4 42 03 23 0.14 4.9 9.0 60SK Ca 56 131,1 2.84 7.4 94 35 55 0.0 0.0 0.0 B 03 221,1 8.33 6.5 33 93 44 6.43 9.01 2.3 A 02 158 8.92 7.3 03 33 63 2.5 4.0 0.0 a timil deeps eht woleb hpm 5 naht erom si deeps elitnecrep ht58 itisnart eht fo gninnigeb eht ta dulcxe eb lliw etis eht ;enoz no analysis.deeps eht morf de

Field Studies 31 • Mean speed. • Speed standard deviation. • 50th percentile speed (median). • 85th percentile speed. • 95th percentile speed. • Percentage of vehicles traveling 5 mph over the posted speed limit. • Percentage of vehicles traveling 10 mph over the posted speed limit. • Percentage of vehicles traveling 15 mph over the posted speed limit. Table 3-8. Classifier speed statistics at sites with a welcome sign. epyt tnemtaerT .on etiS reifissalC noitacol detsoP timil deeps )hpm( fo rebmuN selcihev )hpm( scitsitats deepS foegatnecreP selcihev deeps gnideecxe hpm X yb timil naeM dradnatS noitaived naideM ht58 elitnecrep ht59 elitnecrep 5 01 51 ngiS emocleW )SW( 30SK C 55 559 6.74 2.7 94 45 75 3.1 0.0 0.0 B 53 779 0.03 1.7 23 63 04 8.0 2.0 0.0 A 53 632,1 1.92 3.4 92 33 63 3.0 0.0 0.0 40SK C 55 502 4.44 1.7 54 15 55 0.0 0.0 0.0 B 53 577 7.33 2.5 43 83 34 8.8 9.1 5.0 A 53 821,1 5.03 4.5 13 63 93 5.1 1.0 0.0 01SK Ca 56 535 5.74 8.6 84 45 85 2.0 0.0 0.0 B 04 605 7.03 4.6 13 73 04 0.0 0.0 0.0 A 03 704 8.42 4.4 52 92 23 0.1 0.0 0.0 11SK Ca 56 333 1.64 6.7 64 45 85 3.0 3.0 0.0 B 03 792 6.92 9.5 82 63 04 5.12 7.0 3.0 A 03 304 3.52 1.4 52 03 23 5.0 0.0 0.0 a timil deeps eht woleb hpm 5 naht erom si deeps elitnecrep ht58 itisnart eht fo gninnigeb eht ta dulcxe eb lliw etis eht ;enoz no analysis.deeps eht morf de Table 3-9. Classifier speed statistics at sites with no treatment. tnemtaerT epyt .on etiS reifissalC noitacol detsoP deeps timil )hpm( fo rebmuN selcihev )hpm( scitsitats deepS foegatnecreP selcihev deeps gnideecxe hpm X yb timil naeM dradnatS noitaived naideM ht58 elitnecrep ht59 elitnecrep 5 01 51 enoN 50AI C 55 322 0.06 4.8 06 86 57 8.44 0.22 0.9 B 54 122 8.23 8.7 23 14 64 3.2 0.0 0.0 A 52 232 2.23 5.6 23 83 44 8.75 9.82 9.9 20SK C 56 518 5.75 8.6 85 46 76 0.1 1.0 0.0 B 03 148 1.33 8.4 33 83 14 0.13 2.8 3.1 A 03 568 7.92 7.4 03 43 73 2.8 5.1 3.0 21SK C 56 920,1 6.35 0.7 35 26 66 5.0 0.0 0.0 B 54 920,1 1.54 7.5 54 05 55 8.31 9.4 7.1 A 54 408 8.14 1.7 34 84 15 3.6 6.1 2.0 31SK Ca 56 950,1 6.94 2.5 05 55 85 0.0 0.0 0.0 B 54 098 0.34 8.4 44 84 25 3.7 6.1 1.0 A 54 780,1 7.04 2.8 34 74 15 2.5 6.0 1.0 20EN C 06 482,2 2.05 4.5 05 65 95 2.0 0.0 0.0 B 54 872,2 5.64 4.5 64 25 55 6.32 7.5 8.0 A 54 855,2 7.73 7.6 93 44 74 2.1 0.0 0.0 40EN Ca 06 261,2 0.64 3.5 64 15 45 0.0 0.0 0.0 B 03 652,2 9.63 4.8 83 44 84 3.56 9.63 8.01 A 03 733,2 6.92 6.3 03 33 53 9.4 4.0 0.0 20AV C 55 994 4.84 9.6 94 55 85 4.1 2.0 0.0 B 53 476 9.63 3.5 63 04 44 2.31 7.3 3.0 A 53 988 7.63 5.5 73 24 54 9.22 3.4 8.0 a timil deeps eht woleb hpm 5 naht erom si deeps elitnecrep ht58 itisnart eht fo gninnigeb eht ta dulcxe eb lliw etis eht ;enoz no morf de .sisylana deeps eht

32 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways At each site and classifier location, speed differences between vehicle speeds and posted speed limit were calculated (i.e., a positive difference indicates that the vehicle exceeded the posted speed limit). The distribution of these speed differences is shown in the form of box plots in Figures 3-4 through 3-7, separately for each primary treatment type. (The box represents the middle 50 percent of the speed differences [25th to 75th percentiles; the distance between these two percentiles is called the interquartile range, IQR]; the horizontal line within a box repre- sents the median speed difference [50th percentile]; vertical lines connect the box to the point at Figure 3-4. Speed differences from posted speed limit—roundabouts. Treatment: Roundabout Black dot = mean; Open circle = extreme value NOTE: Difference = Speed - Posted Speed Limit C B A C B A C B A C B A -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40 Di ffe re nc e fro m po st ed s pe ed lim it (m ph ) NE03 VA01 KS09 NE01 Site No. Figure 3-5. Speed differences from posted speed limit—TPMs. Treatment: TPM Black dot = mean; Open circle = extreme value NOTE: Difference = Speed - Posted Speed Limit C B A C B A C B A C B A C B A C B A C B A -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40 Di ffe re n ce fro m po st ed sp ee d lim it (m ph ) IA03 KS01 KS14 KS15 IA01 IA02 KS06 Site No.

Field Studies 33 Figure 3-6. Speed differences from posted speed limit—welcome signs. Black dot = mean; Open circle = extreme value NOTE: Difference = Speed - Posted Speed Limit Treatment: Welcome Sign C B A C B A C B A C B A -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40 Di ffe re n ce fro m po st ed sp ee d lim it (m ph ) KS03 KS04 KS10 KS11 Site No. Figure 3-7. Speed differences from posted speed limit—no treatment. Treatment: None Black dot = mean; Open circle = extreme value NOTE: Difference = Speed - Posted Speed Limit C B A C B A C B A C B A C B A C B A C B A -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40)hp m( ti mil deeps dets op m orf ec nereffiD IA05 KS02 KS12 KS13 NE02 NE04 VA02 Site No. 1.5 × IQR below the 25th percentile and the point at 1.5 × IQR above the 75th percentile; and speed differences beyond the lines are represented with open circles.) Horizontal lines are drawn at 0 and 15 mph to highlight speeds at or below the speed limit (i.e., in compliance) and speeds exceeding 15 mph above the posted speed limit. For example, the percentages of data points above the +15-mph line are directly comparable to the corresponding statistic shown in the last column of Tables 3-6 through 3-9. The speed distributions shown in theses plots are “raw” speeds recorded at each location, that is, they summarize speed differences

34 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways without regard to individual site characteristics (aside from posted speed limits); and, therefore, do not allow for a direct assessment of a particular treatment across sites. However, at a particular site, the box plots allow for comparison of speed trends from Location C (upstream of the transi- tion zone), to Location B (transition zone exit), and to Location A (community). Table 3-10 provides a qualitative assessment of deceleration behavior through the transition zone based upon the speed data from the laser gun. The qualitative assessment was made to determine if drivers did the following: • Decelerated early within the transition. • Decelerated at a steady or gradual rate. • Maintained a relatively constant speed through most of the transition zone and decelerated late within the transition zone. In most cases, the laser gun data showed drivers decelerated at a steady or gradual rate through the length of the transition zone, but in a few cases there was a noticeable reduction in speeds (i.e., an increase in deceleration rate) at the downstream end of the transition zone. In particular, greater deceleration occurred towards the downstream end of the transition zone at three of the four sites with a roundabout treatment. Inspection of the speed distributions in Figures 3-4 through 3-7 and Tables 3-6 through 3-9 shows a few sites where speeds upstream of the transition zone (Location C) are relatively low compared to the posted speed limit. A review of these sites (e.g., considering the placement of the classifier relative to an intersection) did not yield a good explanation for why speeds were so low Table 3-10. Laser gun speeds—qualitative summary by site. Treatment Site no. Deceleration behavior in transition zone Comment Early Steady Late Roundabout NE03 VA01 Roundabout and Transverse Rumble Strips KS09 Assumed deceleration after TRS based on classifier data at Location B Roundabout and Welcome Sign NE01 Transverse Pavement Markings IA03 KS01 KS14 KS15 Transverse Pavement Markings and Welcome Sign IA01 IA02 Assumed deceleration after TPM based on classifier data at Location B KS06 Welcome Sign KS03 KS04 KS05 KS10 KS11 None IA05 KS02 KS12 KS13 NE02 NE04 VA02 Site characteristics not conducive to determine deceleration behavior

Field Studies 35 at these upstream locations. It was determined, however, that such sites are not good candidates to include in an analysis to evaluate the effectiveness of transition zone treatments in reducing speeds. It was not considered useful to estimate the effects of transition zone treatments at the end of the transition zone and within the community, when free-flow speeds at the upstream end of the transition zone are already low. The following rule was applied to determine which sites to include or exclude from further analyses to assess the effect of a particular treatment in reducing speeds in the transition zone and through the community: • If the 85th percentile speed at the upstream end of the transition zone (i.e., Location C) is more than 5 mph below the posted speed limit at that location, exclude the site; otherwise, keep the site. Applying this inclusion/exclusion rule, Sites KS06 (TPM), KS10, KS11 (welcome sign) and KS13, and NE04 (no treatment) are excluded from further analyses to assess the effect of a particular treatment in reducing speeds in the transition zone and through the community. 3.2.3 Analysis Approach The primary measures analyzed to assess the effectiveness of a transition zone treatment in reducing speeds include the following: 1. Transition zone exit speed. Speeds collected with the classifier tubes at Location B were used. The percentage of vehicles in compliance with posted speed limits at the end of the transition zone represents whether the transition zone treatment achieved its objective. Similarly, the percentage of vehicles exceeding speed limits by 5, 10, or 15 mph is a measure of the effect of the treatment on the upper portions of the speed distributions. 2. Transition zone speed reduction. Speeds collected with the classifier tubes at Locations C and B were used. Total speed reductions were measured as the difference in mean speeds upstream and downstream of the transition zone. 3. Community speed. The implementation of the treatment is intended to contribute to reduced vehicle speeds through the community. However, whether vehicles maintained a reduced speed through the community can be thought of as a measure of the “halo” effect of the treatment, although many other factors may contribute to speeds through the commu- nity. Specifically, if the transition zone treatment is implemented upstream of the community and not in the community itself, the effect of the treatment is not expected to directly impact the community speed. Using the speeds recorded with the classifier tubes at Location A, the percentage of vehicles in compliance with posted speed limits or speed limits plus 5, 10, or 15 mph was the measure of interest. The statistical approach to estimate the effect of the treatment on speeds depended on the measure of interest and the location along the roadway. In all cases, basic speed distributions— mean, standard deviation, median, 85th, and 95th percentile speeds—were tabulated separately for each site and location (upstream, transition zone exit, and community). Percentages of vehi- cles exceeding the speed limit or the speed limit plus 5, 10, or 15 mph were also tabulated. The individual analysis approaches were as follows: 1. Analysis of transition zone exit speed. A logistic regression model was used to compare the treatments based on the probability that vehicles drive at or below the posted speed limit exit- ing the transition zone. (Note: this probability is one minus the probability that vehicles exceed the posted speed limit.) This analysis was repeated using speed limit + 5 mph as the criterion. The primary factor of interest in the regression model is the treatment type. To account for site differences such as posted speed limits upstream and downstream of the transition zone, the reduction in posted speed limit from C to B was included in the regression model. To

36 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways also account for speed-limit compliance by drivers upstream, the percentage of speeds at or below speed limit at Location C was included in the model. The statistical significance of treatment, after accounting for C to B posted speed reduction and for compliance upstream was evaluated, and the results from each treatment were compared to those obtained for the non-treated sites. • A generalized linear model (GLM) with a binomial distribution and a logit link function was used to model the compliance proportions. Least squares mean differences between treatment and no treatment sites were estimated after accounting for the potential effect of the additional factors in the model and tested for statistical significance. All analyses were performed using the GENMOD procedure of SAS (2011). 2. Analysis of transition zone speed reduction. At each site and location, the effect of a par- ticular treatment is confounded with that of the posted speed limit sign. Since the sites vary with respect to their posted speed limits at Locations C and B, the interest is in the estimated effect of a particular treatment compared to no treatment beyond that of the posted speed reduction at each site. Given that speeds were recorded for a population of vehicles rather than individual vehicles, this analysis could not be performed based on individual vehicles’ speed reductions, but on overall speed differences upstream and downstream of the treat- ment. To account for the posted speed limit at each location, the overall speed differences at each location for each site were estimated, and the difference in differences between sites with treatment and no treatment was evaluated and tested for significance. • A generalized linear mixed model with a normal distribution and identity link function was used to model the speed differences, accounting for the random site effect. The reduc- tion in posted speed limit from Location C to Location B was also included in the model. Least squares mean differences between treatment and no treatment sites were estimated after accounting for the potential effect of the additional factors in the model and tested for statistical significance. 3. Analysis of community speed. The analysis approach is similar to that used to evaluate the treatment effect at the transition zone exit (discussed above) with some additional considerations. Since site characteristics, such as on-street parking, presence of sidewalks and curbs, and horizontal roadway alignment may influence driver behavior in the com- munity, these factors were initially considered in the logistic model. Multicollinearity between these additional factors was evaluated prior to their inclusion in the model. As for the transition zone exit speed, the percentage of speeds at or below speed limit at Location B was included in the model. Again, the model was repeated for the speed limit + 5 mph criterion. 3.2.4 Analysis Results Transition Zone Exit—Speed-Limit Compliance Evaluation The effect of treatment type—roundabout, TPM, and welcome sign—on exit speed (Loca- tion B) was estimated as compared to that of no treatment. Two dependent variables were of interest, as follows: • Percentage of vehicles driving at or below speed limit at Location B and • Percentage of vehicles driving at or below speed limit + 5 mph at Location B. The percentage of vehicles driving at or below the speed limit at Location B was calculated for each site. Similarly, the percentage of vehicles driving at or below the speed limit at Location C was calculated for each site. The same calculations were performed using speed limit + 5 mph as the criterion. The relationship between observed compliance at exit and upstream is illustrated in Figure 3-8 (speed limit criterion) and Figure 3-9 (speed limit + 5 mph criterion).

Field Studies 37 A logistic regression model was then developed to investigate the relationship between the percentage of vehicles driving at or below the speed limit at Location B and the treatment type, posted speed limit nearest Location C, and the speed reduction from Location C to Location B. The percentage of vehicles driving at or below the speed limit at Location C was also included in the model to account for speed-limit compliance behavior upstream of the transition zone. Figure 3-8. Observed exit (Location B) compliance versus upstream (Location C) compliance—posted speed limit criterion (TZ = transition zone). ti xE ZT ta ec nailp m oc t necreP 20 30 40 50 60 70 80 90 100 Percent compliance Upstream 20 30 40 50 60 70 80 90 100 Compliance at B vs. Compliance at C Compliance: Speed <=Speed Limit Treatment Roundabout TPM Welcome Sign None Figure 3-9. Observed exit (Location B) compliance versus upstream (Location C) compliance—posted speed limit + 5 mph criterion. Pe rc en t c om pl ia nc e at T Z Ex it 50 60 70 80 90 100 Percent compliance Upstream 50 60 70 80 90 100 Compliance at B vs. Compliance at C Compliance: Speed <=Speed Limit + 5 mph Treatment Roundabout TPM Welcome Sign None

38 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways The analysis showed that of the variables considered in the model in addition to treatment type—speed reduction from C to B, speed limit at C, and percentage of speeds at or below the speed limit at C—only the speed reduction was statistically significant at the 10 percent level. This model also showed that the estimated percentage of vehicles driving at or below the speed limit was considerably higher (by approximately 10 percent) at sites with a welcome sign than at other treated sites. Although it is reasonable to assume that a psychological treatment such as a welcome sign can be effective in reducing speeds, the findings seem illogical. A welcome sign to a community is intended to raise the awareness of drivers that they are entering into a community, but the treatment does not change the overall roadway characteristics extensively enough to sug- gest that at these sites the significant reduction in measured speeds was solely due to this psycho- logical treatment (i.e., the welcome sign). Therefore, the remaining two sites with a welcome sign (KS03 and KS04) were excluded from the analysis and a new logistic model was evaluated based on the remaining 15 sites. The final modeling results are shown in Table 3-11, separately for each speed criterion. Only two treatment types remain—roundabout and TPM. For each criterion, the table summarizes the overall significance of the treatment and speed reduction from C to B. A 10 percent sig- nificance level was used to assess significance. The table shows estimated percentage of speeds exceeding the criterion with upper and lower 90 percent confidence limits. The last column indicates whether either treatment significantly improved speed-limit compliance as compared to no treatment, after accounting for posted speed reduction from C to B. In summary, if speed-limit compliance were defined as driving at or below posted speed limit at the exit of the transition zone, then the study data show the following, accounting for the effect of speed reduction in posted speed limits from Location C to B: • The compliance rate upstream of the transition zone had no significant effect on the compli- ance rate at the exit of the transition zone. • The overall treatment effect on compliance rate was not statistically significant at the 10 per- cent significance level. • On average, the rate of compliance at sites with a roundabout or TPM is higher than that at sites with no treatment by an amount of 15 percent (not statistically significant) and 20 per- cent (statistically significant), respectively. • On average, the rate of compliance for sites with TPM (57 percent) was slightly higher than that for sites with a roundabout (52 percent). Table 3-11. Analysis results of speed-limit compliance at transition zone exit (Location B). Treatment type (number of sites) Percent speeds at or below criterion at Location B Treatment significantly better than none at 90% confidence level? (p-value) Estimate 90% Lower limit 90% Upper limit CRITERION: SPEED LIMIT AT LOCATION B Overall treatment effect: not significant (p = 0.16). Compliance rate at Location C: not included (not statistically significant). Posted speed reduction from C to B: significant (p = 0.10). Roundabout (4) 52 40 64 No (0.11) TPM (6) 57 42 70 Yes (0.06) None (5) 37 23 54 CRITERION: SPEED LIMIT+ 5 MPH AT LOCATION B Overall treatment effect: significant (p < 0.01). Compliance rate at Location C: not included (not statistically significant). Posted speed reduction from C to B: significant (p = 0.09). Roundabout (4) 88 81 93 Yes (0.06) TPM (6) 79 69 87 No (0.41) None (5) 77 65 86

Field Studies 39 If speed-limit compliance were defined as driving at or below the posted speed limit + 5 mph at the exit of the transition zone, then the study data show the following, accounting for the effect of speed reduction in posted speed limits from Location C to B: • The compliance rate upstream of the transition zone had no significant effect on the compli- ance rate at the exit of the transition zone. • The overall treatment effect on compliance rate was significant at the 10 percent significance level. • On average, the rate of compliance for sites with a roundabout (88 percent) was significantly higher than that for sites without a treatment (77 percent) by an amount of 11 percent. • On average, the rates of compliance at sites with TPM (79 percent) and sites with no treatment (77 percent) are the same, for all practical purposes. • The average rates of compliance for sites with a roundabout (88 percent) and sites with TPM (79 percent) were not statistically different from each other (p = 0.17; this comparison is not shown in Table 3-11). Evaluation of Speed Reduction from Upstream to Transition Zone Exit An attempt was made to evaluate the effect of the treatment on the upstream to transition zone exit speed reduction (i.e., C to B reduction). Ideally, this would be done based on paired C to B speed differences if each speed collected at Location C could have been paired with the match- ing vehicle’s speed at Location B. However, since an individual vehicle could not be tracked from Location C to B with classifier tubes, an individual vehicle’s speed reduction was not available (see earlier discussion). As a result, speed reduction from C to B, even at a single site, could only be evaluated for all vehicles as a whole. Table 3-12 shows simple speed reduction statistics for each site—mean speed reduction and its lower and upper 95 percent confidence limits. For comparison purposes, the table shows the drop in posted speed limit from C to B. The last column shows the mean speed difference as a percentage of the speed reduction difference. In all but two cases, average speed reductions at a given site are below the reduction in posted speed limits. This is in line with the fact that at Location C average speeds are below posted speed limit at all but one site (IA05) as shown in Tables 3-6 through 3-9. The sites without treatment exhibit the widest range of relative speed reductions as compared to the posted speed reduction. Table 3-12. Speed reduction from Location C to B by treatment type and site. Treatment type Site no. Posted speed limit reduction C to B (mph) Vehicle speed reduction from C to B (mph) Relative speed reductiona (%) Estimated mean Lower 95% confidence limit Upper 95% confidence limit Roundabout KS09 35 26 25 28 74 NE01 15 14 11 18 93 NE03 30 22 21 22 73 VA01 20 17 16 18 85 TPM IA01 30 18 15 20 60 IA02 30 21 18 24 70 IA03 25 19 17 21 76 KS01 20 22 19 24 110 KS14 20 16 15 16 80 KS15 20 17 16 18 85 None IA05 10 27 24 30 270 KS02 35 22 19 26 63 KS12 20 9 7 10 45 NE02 15 4 1 7 27 VA02 20 12 11 13 60 aRelative speed reduction = 100 × mean speed reduction/speed limit reduction.

40 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways Next, taking into account the posted speed limit at each site and location, the potential addi- tional effect of the treatment on speed reduction from C to B was estimated. This was done by calculating the difference between each vehicle’s speed and the posted speed limit at each site and each location. (Note that the average of the rescaled C to B differences for each site is the same as the difference between Column 3 “Posted speed limit reduction C to B (mph)” and Column 4 “Estimated mean” in Table 3-12.) The effect of the treatment on these rescaled differences was then assessed using a mixed linear model approach. The factors considered in the model were treatment type, location (C and B), and reduction in posted speeds from C to B; site was included in the model as a random factor. The analysis of variance (ANOVA) showed that the reduction in posted speed limit was not statistically significant (p-value = 0.24); this factor was therefore removed from the model. The final analysis results are summarized in Table 3-13 as follows: • For each treatment, Column 2 shows the mean speed reduction from C to B, after adjusting each vehicle’s speed for the speed limit at the respective site and location. Each mean can be expressed as: [Avg(SpeedC – Speed LimitC) – Avg(SpeedB – Speed LimitB)]. • Columns 3 and 4 show the 95 percent confidence limits for the adjusted mean speed reduction. • Columns 5 to 7 show the mean and 95 percent confidence limits for the difference in adjusted C to B speed reductions between treated and no treatment sites. After adjusting for posted speed limit at each location, speed reductions from C to B were comparable for sites with treatment and no treatment, at just over 5 mph, with confidence limits ranging from approximately –2 to +12 mph. In summary, based on the study sites, there is no evidence that either treatment (i.e., roundabouts or TPMs) has a statistically significant effect on reducing average speeds from upstream to the transition zone exit beyond that due to the posted speed limit reduction. Community—Speed-Limit Compliance Evaluation Of the 15 sites in the analysis, 12 had no further speed limit reduction from the transition zone exit to the community; the other three sites had reductions in speed limits of 5 mph (TPM), 15 mph (TPM), and 20 mph (no treatment). Compliance rates in the community were calculated with respect to the two criteria defined earlier. The relationship between compli- ance in the community and at the transition zone exit is illustrated in Figure 3-10 (speed limit criterion) and Figure 3-11 (speed limit + 5 mph criterion). The analysis of the effect of the treatment in the transition zone on speed-limit compliance in the community was performed using the same approach as that for transition zone exit compli- ance with a few exceptions. Speed reduction from B to A (none for 12 of the 15 sites) was not Table 3-13. Analysis results of speed reduction from upstream (Location C) to transition zone exit (Location B), adjusted for speed limit reduction. Treatment type C to B reduction adjusted for speed limit reduction (mph) Treatment versus no treatment difference in C to B reduction adjusted for speed limit reduction (mph) Estimated meana Lower 95% confidence limit Upper 95% confidence limit Estimated mean Lower 95% confidence limit Upper 95% confidence limit Roundabout 5.2 –2.3 12.7 0.0 –10.1 10.0 TPM 5.6 –0.5 11.7 0.4 –8.7 9.5 None 5.2 –1.5 11.9 aMean = [Avg(SpeedC – Speed LimitC) – Avg(SpeedB – Speed LimitB)].

Field Studies 41 considered in the logistic model; however, to account for the potential effect of site characteristics in the community, the inclusion of the following categorical variables was initially considered: • On-street parking—yes (three sites), no (12 sites); • Presence of sidewalks—yes (eight sites), no (seven sites); • Presence of curbs—yes (seven sites), no (eight sites); and • Horizontal alignment—tangent (nine sites), curve (six sites). Figure 3-10. Observed community (Location A) compliance versus exit (Location B) compliance—posted speed limit criterion. Pe rc en t c om pl ia nc e in C om m un ity 10 20 30 40 50 60 70 80 90 100 Percent compliance at TZ Exit 20 30 40 50 60 70 80 90 100 Compliance at A vs. Compliance at B Compliance: Speed <=Speed Limit Treatment Roundabout TPM None Figure 3-11. Observed community (Location A) compliance versus exit (Location B) compliance—posted speed limit + 5 mph criterion. Pe rc en t c om pl ia nc e in C om m un ity 10 20 30 40 50 60 70 80 90 100 Percent compliance at TZ Exit 50 60 70 80 90 100 Compliance at A vs. Compliance at B Compliance: Speed <=Speed Limit + 5 mph Treatment Roundabout TPM None

42 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways On-street parking was confounded with treatment type and was therefore not included in the model. Also, the presence of sidewalks and that of curbs are confounded; therefore the presence of sidewalks was not considered in the model. An attempt was made to address pedestrian activity in the community as potentially affecting drivers’ speed compliance. Lacking pedestrian volume data, community population (last column in Table 3-4) was considered as a surrogate measure. Population size ranged from 222 to 7,990 with a highly skewed distribution to the low side [i.e., 13 of the 15 communities had a population below 2,500 (10 below 1,300) and the remaining 2 were at 7,990]. Furthermore, no roundabouts were installed near communities with a population below 500 and no TPMs were installed near commu- nities with a population above 2,200. Thus, the uneven distribution of community population size across treatment types did not warrant including community population as a factor in the analysis. In summary, the independent variables considered for inclusion in the model were: treatment type, posted speed limit at Location A, presence of curbs, horizontal alignment, and compliance rate at Location B. The final modeling results are shown in Table 3-14, separately for each speed criterion. Whether one defines compliance in the community as driving at or below the speed limit or at or below the speed limit + 5 mph, the highlights of Table 3-14 can be summarized as follows: • The compliance rate at the exit of the transition zone had no statistically significant effect on the compliance rate in the community. • Horizontal alignment of the roadway in the community had no statistically significant effect on the compliance rate in the community. • Both posted speed limit in the community and presence of curbs had a statistically significant effect on the compliance rate in the community. Table 3-14. Analysis results of speed-limit compliance in the community (Location A). Treatment type (number of sites) Percent speeds at or below criterion at Location A Treatment significantly better than none at 90% confidence level? (p-value) Estimate 90% lower limit 90% upper limit Criterion: Speed Limit at Location A Overall treatment effect: not significant (p = 0.90). Compliance rate at Location B: not included (not statistically significant). Horizontal alignment: not included (not statistically significant). Posted speed limit in community: significant (p < 0.01). Presence of curbs: significant (p < 0.01). Roundabout (4) 58 47 68 No (0.44) TPM (6) 54 42 66 No (0.33) None (5) 59 45 71 Criterion: Speed Limit + 5 mph at Location A Overall treatment effect: not significant (p = 0.84). Compliance rate at Location B: not included (not statistically significant). Horizontal alignment: not included (not statistically significant). Posted speed limit in community: significant (p < 0.01). Presence of curbs: significant (p < 0.01). Roundabout (4) 89 82 94 No (0.43) TPM (6) 86 76 92 No (0.37) None (5) 88 77 94

Field Studies 43 After accounting for the significant effect of posted speed limit in the community and that of the presence of curbs, it was found that, for the study sites, the type of treatment in the transi- tion zone had no significant effect on compliance rates in the community. These compliance rates range, on average and for all site types, from 54 percent to 59 percent (using the speed limit criterion) and from 86 percent to 89 percent (using the speed limit + 5 mph criterion). Overall Summary of Speed Evaluations The primary findings from the speed evaluations can be summarized as follows: • Compared to non-treatment sites, roundabouts and TPMs do not significantly decrease mean speeds from upstream to downstream of the transition zone beyond the speed reduction due to the change in posted speed limit. A small reduction in mean speed was observed, but this reduction was not statistically significant. However, speed-limit compliance rates were significantly higher at roundabouts and TPMs as compared to rates at non-treatment sites. Roundabouts and TPMs increase the rate of compliance of vehicles traveling at or below the speed limit by amounts of 15 and 20 percent, respectively, as compared to no treatment. Roundabouts also increase the rate of compliance of vehicles traveling at or below the speed limit + 5 mph by an amount of 11 percent when compared to no treatment. The fact that the decrease in mean speeds from upstream to downstream of the transition zone was not significantly larger at treatment sites than at non-treatment sites, while speed- limit compliance rates increased with the treatment in place, indicates that there was a reduc- tion in speed variance from upstream to downstream of the treatments at the study sites. While speed variance data were not provided in the analysis results above, in general, the data show a reduction in speed variance from upstream to downstream of the transition zone at all sites and the speed variance was overall lower at the treatment sites (i.e., roundabouts and TPMs) than at the non-treatment sites. This supports the findings that roundabouts and TPMs do not necessarily decrease means speeds from upstream to downstream of the transition zone any more than does no treatment, but roundabouts and TPMs do increase speed-limit compliance. • Neither roundabouts nor TPMs significantly affect compliance rates of vehicles traveling at the speed limit within the community further downstream from the transition zone exit. These findings support guidance from previous research (Forbes, 2011) for the need to pro- vide additional measures through the community to maintain a speed reduction downstream of the transition zone through the community. • The effect of welcome signs at community entrances on mean speeds and compliance rates to speed limits could not be determined because speeds upstream of the transition zones at several of the data collection sites were inexplicably low, leaving too few sites to conduct a reliable analysis of speeds and/or speed-limit compliance rates at transition zones due to welcome signs. 3.3 Crash Data Analysis Crash data were obtained for all of the sites included in the speed study. The primary purpose of analyzing the crash data was not to conduct a rigorous statistical crash analysis (the number of sites included in the study is not sufficient for such an analysis), but rather to verify that all of the treatments evaluated are operating both safely and efficiently, with no unusual crash history or patterns at the sites, and to identify potential trends in the crash data. The crash data are first summarized by severity and then by collision type. For this analysis, the most recent 5 years of available crash data were obtained for each site, either from 2005 to 2009 or 2006 to 2010. Crash data were obtained for the transition zone, consistent with the boundaries defined for the speed study (i.e., from classifier Location C

44 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways to B), and 0.25 mi downstream of the transition zone into the community. For the sites where speed data were collected on both approaches to the community—and the length of the com- munity (i.e., from classifier Location B on one approach to classifier Location B on the other approach) was less than 0.5 mi—the length of the community was divided in half, and crashes were assigned to the respective study site. Both intersection and roadway segment crashes were included in the analysis, while animal crashes and parking lot/driveway related crashes were excluded. Also, for single-vehicle crashes that occurred along the roadway, the vehicle must have been traveling from the high-speed environment to the low-speed environment to be included in the analysis. Of the 80 crashes (i.e., all severity levels) included in the database for analysis, six crashes could be identified as being speed-related crashes (e.g., based upon data elements such as major cause or contributing circumstances). All six crashes occurred at treatment sites, two prior to installa- tion of the treatment and four after installation of the treatment. 3.3.1 Severity Level For the combined transition zone and community, Table 3-15 summarizes the crash data for total crashes and Table 3-16 summarizes the data for fatal and all injury crashes. An empty cell in the tables indicates that no crash data were available prior to installation of the treatment. Each table presents the combined site length, the number of before and after years of crash data (i.e., crash data for the years before treatment installation and for the years after treatment installation), and the observed crash frequency for each period. The last two columns present the crash frequencies on a per mile per year basis so direct comparisons can be made across sites. Crash data before treatment installation were not available for seven out of the 15 treat- ment sites. Table 3-15. Total crashes by site in the combined transition zone and community. Treatment type Site no. Site length (mi) Number of years Crash frequency Crashes/mi/yr Before After Before After Before After Roundabout KS09 1.31 3 1 1 0 0.25 0.00 NE01 0.73 5 6 1.65 NE03 0.79 3 1 4 3 1.70 3.82 VA01 0.94 5 12 2.55 TPM IA01 0.50 5 0 0.00 IA02 0.57 5 3 1.06 IA03 0.53 5 1 0.38 KS01 0.87 3 1 3 0 1.15 0.00 KS06 0.78 2 1 1 1 0.64 1.29 KS14 0.55 3 1 4 0 2.41 0.00 KS15 0.62 3 1 0 0 0.00 0.00 Welcome Sign KS03 0.65 2 2 5 1 3.87 0.77 KS04 0.37 2 2 1 1 1.33 1.33 KS10 0.48 5 0 0.00 KS11 0.38 5 1 0.52 None IA05 0.58 5 1 0.34 KS02 0.61 5 7 2.29 KS12 0.27 5 0 0.00 KS13 0.28 5 3 2.11 NE02 0.33 5 5 3.07 NE04 0.51 5 3 1.19 VA02 0.97 5 13 2.68

Field Studies 45 Focusing on the crash frequencies summarized in the “crashes/mi/yr” columns of Tables 3-15 and 3-16, it appears that treatment sites are operating as safely as expected. Only one site (NE03) seems to have a significant increase in crashes after installation of the treatment (a roundabout), but it is only based on 1 year of after data. In general, crash frequencies for the treatment sites after installation appear to be consistent with crash frequencies before installation of the treat- ment and consistent with crash frequencies of sites with no treatment, so from a qualitative level it does not appear that the installation of the treatments at the study sites improved or negatively impacted safety at these sites. To assess the potential treatment effect in a more quantitative and scientifically based manner, the following two approaches were used to analyze total crashes: 1. Using a simple analysis where crashes at the 15 treatment sites in the after period were compared to crashes at all sites in the before period (i.e., 8 treatment sites and 7 no treatment sites). A gen- eralized linear model (GLM) with a negative binomial distribution and a log link was used to model crashes/mi/yr; average daily traffic (ADT) (continuous variable) and treatment (at four levels—RA, TPM, WS, and None) were included in the model. For the roundabout sites with data before and after installation (i.e., sites KS09 and NE03), the before period data were not included in this analysis, because these sites were unique in that they both had a major intersec- tion near the entrance to the community compared to the other sites, which had a limited num- ber of intersections within the transition zone. The objective was to estimate the significance of the treatment effect and, if statistically significant, to compare each treatment to no treatment. 2. Using a cross-sectional type of analysis where only the eight treatment sites (two RA, four TPM, two WS) with both before and after treatment installation crash data were included. The data were analyzed in a similar fashion as just described with the addition of the period (before/after) as a categorical variable and site as a random factor to account for the within- site temporal correlation. Table 3-16. Fatal and all injury crashes by site in the combined transition zone and community. Treatment type Site no. Site length (mi) Number of years Crash frequency Crashes/mi/yr Before After Before After Before After Roundabout KS09 1.31 3 1 0 0 0.00 0.00 NE01 0.73 5 4 1.10 NE03 0.79 3 1 3 2 1.27 2.55 VA01 0.94 5 7 1.49 TPM IA01 0.50 5 0 0.00 IA02 0.57 5 1 0.35 IA03 0.53 5 1 0.38 KS01 0.87 3 1 2 0 0.77 0.00 KS06 0.78 2 1 0 0 0.00 0.00 KS14 0.55 3 1 2 0 1.20 0.00 KS15 0.62 3 1 0 0 0.00 0.00 Welcome Sign KS03 0.65 2 2 2 0 1.55 0.00 KS04 0.37 2 2 1 0 1.33 0.00 KS10 0.48 5 0 0.00 KS11 0.38 5 1 0.52 None IA05 0.58 5 1 0.34 KS02 0.61 5 1 0.33 KS12 0.27 5 0 0.00 KS13 0.28 5 1 0.70 NE02 0.33 5 4 2.45 NE04 0.51 5 2 0.79 VA02 0.97 5 4 0.82

46 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways Neither analysis showed any significant treatment type or before/after treatment instal- lation effect. The p-value for the overall treatment effect associated with the first approach was 0.16. The p-value for the overall before/after effect at treated sites associated with the second approach was 0.38. In summary, the quantitative analysis of the crash data did not yield any significant conclusions regarding the safety effect of the treatments considered at the study sites. 3.3.2 Collision Type The 80 crashes (i.e., all severity levels) from the 22 study sites fell into 10 collision types as shown in Table 3-17. Only two crashes involved vulnerable road users (i.e., pedestrians and bicyclists), and a majority (84 percent) of the crashes involved multiple vehicles. Rear-end, angle, and fixed-object crashes accounted for 85 percent of the crashes. These crashes were further categorized by the following parameters: • Crash location (intersection or roadway segment); • Treatment type (RA, TPM, WS, None); • Site number; • Location (transition zone or community); • Time period (before or after treatment installation); and • Crash type (10 types listed in Table 3-17). Table 3-18 shows the distribution of collision types of the 51 total crashes that either occurred at an intersection or were intersection related, across all of the sites. No particular trends in col- lision types were evident at any of the sites going from the before to the after period. Very few sites (four of the 22) had a collision type that was occurring on average at least once per year, in either the transition zone or the community. At three of the four sites, rear-end crashes were the most common (i.e., highest crashes/yr). Table 3-19 shows the distribution of the 29 total crashes across those same categories, related to roadway segments. No particular trends in collision types are evident. 3.3.3 Overall Summary of Crash Analysis The primary purpose of the crash analysis was to verify that all of the treatments evalu- ated at the sites included in this research were operating both safely and efficiently, with no unusual crash history or patterns, and to identify potential trends in the crash data. Based upon the summary crash statistics, no specific trends and/or anything unusual in the data Table 3-17. Collision types at all study sites. Collision type Crash frequency Angle 25 Bicycle 1 Fixed Object 10 Head On 2 Multiple vehicle—Other 4 Pedestrian 1 Rear End 33 Sideswipe—Opposite Direction 1 Sideswipe—Same Direction 2 Single vehicle—Other 1

Field Studies 47 Table 3-18. Collision types by site, location, and before/after treatment at intersections. Treatment Site Location Time period Collision type Total crash frequency Crashes/yr Roundabout KS09 TZ Before Rear End 1 0.33 NE01 TZ After Fixed Object 1 0.20Rear End 1 0.20 NE03 Community After Rear End 1 1.00 Before Angle 1 0.33Bicycle 1 0.33 TZ Before Angle 1 0.33 VA01 TZ After Angle 2 0.40Rear End 6 1.20 TPM IA02 Community After Angle 1 0.20 IA03 Community After Angle 1 0.20 KS14 Community Before Angle 2 0.67 Rear End 1 0.33 Welcome Sign KS03 Community Before Angle 3 1.50 TZ After Rear End 1 0.50 Before Rear End 1 0.50 KS04 Community After Rear End 1 0.50 TZ Before Rear End 1 0.50 None KS02 Community Before Angle 3 0.60 Rear End 1 0.20 TZ Before Multiple Vehicle—Other 1 0.20 KS13 Community Before Angle 1 0.20Rear End 2 0.40 NE02 Community Before Angle 2 0.40 Multiple Vehicle—Other 2 0.40 Pedestrian 1 0.20 VA02 TZ Before Angle 4 0.80 Fixed Object 1 0.20 Rear End 5 1.00 Sideswipe—Same Direction 1 0.20 Table 3-19. Collision types by site, location, and before/after treatment on roadway segments. Treatment Site Location Time period Collision type Total crash frequency Crashes/mi/yr Roundabout NE01 Community After Angle 1 0.80Rear End 1 0.80 TZ After Fixed Object 1 0.42Sideswipe—Same Direction 1 0.42 NE03 TZ After Angle 1 1.85 Fixed Object 1 1.85 Before Rear End 1 0.62 VA01 Community After Head On 1 2.22 TZ After Fixed Object 2 0.47Rear End 1 0.23 TPM IA02 Community After Rear End 2 1.60 KS01 Community Before Fixed Object 1 1.33 TZ Before Fixed Object 1 0.54Rear End 1 0.54 KS06 Community After Rear End 1 4.00TZ Before Fixed Object 1 0.94 KS14 Community Before Rear End 1 1.33 Welcome Sign KS03 Community Before Single Vehicle—Other 1 2.38 KS11 Community After Head On 1 1.00 None IA05 TZ Before Fixed Object 1 0.61 KS02 Community Before Angle 1 0.80TZ Before Sideswipe—Opposite Direction 1 0.56 NE04 TZ Before Angle 1 0.77 Multiple Vehicle—Other 1 0.77 Rear End 1 0.77 VA02 Community Before Rear End 1 2.22TZ Before Rear End 1 0.23

48 Design Guidance for High-Speed to Low-Speed transition Zones for Rural Highways were observed regarding the effects that installation of roundabouts, TPMs, or welcome signs have on safety. Additionally, two statistical methods were used to analyze the crash data and neither analysis approach showed any significant treatment or before/after effect. Based on this study’s crash analysis of a combination of roadway segment and intersection crashes over an extended length of roadway beginning at the upstream end of the transition zone to 0.25 mi downstream of the end of the transition zone, there is no evidence to suggest that the installation of roundabouts, TPMs, or welcome signs in a transition zone either improves or negatively affects safety. This finding is not surprising given the limited crash dataset for this study. Regarding the effects that the installation of roundabouts, TPMs, or welcome signs have on safety near a transition zone, the most reliable information available is for roundabouts and can be found in the Highway Safety Manual (HSM) and research by Rodegerdts et al. (2007, 2010) (see Figure 2-2).