Roundabouts in the United States (2007)

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.

12 This chapter describes the process used to establish an overall inventory of roundabouts in the United States, the selection of sites for data collection, and the various types of data collected at specific sites. This database serves three major purposes: • It supports the development of United States-based safety and operational models completed as part of this project. • It provides a foundation for additional research into topics beyond the scope and budget of this project. • It establishes a baseline of U.S. roundabout performance during 2003, the year during which most field data were col- lected. Future research will be able to compare conditions at that time with those experienced in 2003 to determine trends in various measures over time. The following sections discuss the development of this database, including the overall inventory and collection of the various geometric, operational, speed, and safety elements. Site Inventory One of the products of this research project is an updated site inventory that contains information on as many round- abouts as possible during its development. Table 4 provides a summary of the database, compiled by the project team, of the 310 known roundabouts that existed in the United States as of 2003; the locations of these roundabouts are shown graphically in Figure 1. This database was initially developed by a team led by Rensselaer Polytechnic Institute (RPI) for a project conducted for the NYSDOT and supplemented by information collected for this research project (30). Most (94%) of these roundabouts are located in urban or suburban areas, with more than half located in the western United States. The most common geometric configuration (more than two-thirds of the roundabouts) consists of a one-lane circulating roadway and four legs. Sixty-one percent of the roundabouts were converted from some form of stop control, while nearly a third were newly constructed intersections. Nearly all of these roundabouts were constructed during the past 10 years, with 46% opening between 2000 and 2003. A complete listing and description of the site inventory used for this project is given in Appendix C. Table 5 lists the subset of known sites used for the analysis in this study. Each of these sites was assigned a unique site identification code consisting of a two-letter state abbrevia- tion and a two-digit identification number (e.g., MD01). This code was used in combination with a cardinal direction designation for a given leg (e.g., MD01-N) and/or a video number (e.g., MD01-N2) to identify specific videos for a given leg. Safety Data Safety databases were required for three purposes: • To develop intersection-level crash prediction models • To develop approach-level crash prediction models • To conduct a before-after study of roundabouts converted from signal or stop control For a roundabout to be eligible for inclusion in the sample used for each of the three purposes outlined above, the data available had to meet minimum inclusion criteria that varied based on the model under consideration. For developing intersection-level prediction models, crash and traffic volume data and basic geometric information such as number of legs and number of lanes had to be available for a period of time after the roundabout was constructed. The same information was required for approach-level models; however, the data needed to be available at the approach level, and more detailed geometric data were required. For the before-after study, it was necessary to have, at a minimum, AADT volumes for either the before or after period, the construction dates, C H A P T E R 2 Data Characteristics

Characteristics Number Percentage of total Total number 310 Setting • Urban • Suburban • Rural 103 164 16 36% 58% 6% Number of legs • 6 • 5 • 4 • 3 • 2 4 16 197 70 4 1% 5% 68% 24% 1% Number of circulating lanes • 3 • 2 • 1 5 72 213 2% 25% 73% Previous intersection • One-way stop • Two-way stop • All-way stop • Signal • None 30 49 16 14 46 19% 32% 10% 9% 30% Year created • 2000-2003 • 1995-1999 • 1994 or earlier 70 70 12 46% 46% 8% Geographic location (zip code) • Northeast (0,1) • Mid-Atlantic (2) • South, Southeast (3,7) • Midwest (4,5,6) • Mountain West (8) • Pacific Coast (9) 24 45 32 39 94 76 8% 15% 10% 13% 30% 25% Note: Not all characteristics are available for all sites; this explains why the totals for each characteristic add up to less than 310, the total number of roundabouts in the database. For example, setting data are available for 283 of the 310 roundabouts. The percentages cited for urban, suburban, and rural settings add up to 100% of the sample of sites for which data for this characteristic is available. The number of legs and geographic location data do not add to 100% because of rounding. Table 4. Characteristics of modern roundabouts located in the United States (2003). Figure 1. Geographic distribution of known roundabouts as of 2003.

14 Si te ID St a te City County Intersection S et tin g Le gs La n es Sa fe ty O pe ra tio na l Sp ee d Pe de st ri a n Bi cy cl ist CA06 CA Davis Yolo Anderson Rd/Alvarado Ave U 4 1 X X CA10 CA Long Beach Los Angeles Pac Coast Hwy/Hwy 19/Los Coyotes Diag. U 4 3 X CA11 CA Modesto Stanislaus La Loma/James St./G St. U 5 1 X CA17 CA Santa Barbara Santa Barbara Milpas St/US 101 NB Ramps/Carpinteria St U 5 2 X CA23 CA Modesto Stanislaus W Rumble Rd/Carver Rd U 4 1 X CO01 CO Eagle Eagle SH-6/I-70 spur/Eby Creek Rd R 4 1 X CO02 CO Golden Jefferson South Golden Road/Johnson Rd/16th Street U 4 2 X CO03 CO Golden Jefferson South Golden Road/Utah St. U 4 2 X CO04 CO Aspen Pitkin SH 82/Maroon Crk/Castle Crk S 4 2 X CO06 CO Avon Eagle Avon Rd./Beaver Creek Blvd. U 4 3 X CO07 CO Avon Eagle Avon Rd./Benchmark Road U 4 2 X CO08 CO Avon Eagle Avon Rd./I-70 Eastbound Ramp U 4 2 X CO09 CO Avon Eagle Avon Rd./I-70 Westbound Ramp U 4 2 X CO10 CO Avon Eagle Avon Rd./U.S. Hwy 6 U 4 2 X CO49 CO Vail Eagle Chamonix Rd/I-70 EB Ramps/S Frontage Rd S 6 2 X CO50 CO Vail Eagle Chamonix Rd/I-70 WB Ramps/N Frntge Rd S 5 2 X CO51 CO Vail Eagle Vail Rd/I-70 EB Ramps/South Frontage Rd S 5 2 X X CO52 CO Vail Eagle Vail Rd/I-70 WB Ramps/N Frntg/Sprddle Cr. S 6 3 X CT01 CT Killingworth Middlesex Rte 80/Rte 81 R 4 1 X CT04 CT N. Stonington New London Rte 2/Rte 184 U 4 1 X FL01 FL Amelia Island Nassau SR AIA/Amelia Island Plantation S 4 1 FL02 FL Boca Raton Palm Beach Cain Blvd/Boca Raton Dr S 4 1 X FL09 FL Bradntn Bch Manatee SR 789/Bridge St S 3 1 X FL11 FL Clearwtr Bch Pinellas SR 60/Coronado/Mandalay/Poinsettia U 5 2 X X X FL14 FL Ft Wltn Bch Okaloosa Hollywood Blvd/Doolittle Blvd U 3 1 X FL15 FL Gainesville Alachua SE 7th Street/SE 4th Avenue U 4 1 X KS01 KS Olathe Johnson Sheridan St./Rogers Rd U 4 2 X KS02 KS Hutchinson Reno 23rd Ave./Severence St. U 4 1 X KS05 KS Lawrence Douglas Monterey Way/Harvard Rd S 3 1 X KS09 KS Manhattan Riley Candlewood Dr/Gary Avenue S 4 1 X KS10 KS Manhattan Riley Kimball Ave/Grand Mere Parkway S 3 1 X KS15 KS Overland Park Johnson 110th St./Lamar Ave. S 4 2 X KS16 KS Paola Miami K-68/Old Kansas City Rd/Hedge Lane R 5 1 X MD01 MD Bel Air Harford Tollgate Rd./Marketplace Dr. S 3 1 X X MD02 MD Leeds Cecil MD 213/Leeds Rd/Elk Mills Rd (Lanzi Cir.) R 4 1 X X MD03 MD Jarrettsville Harford MD 24/MD 165 R 4 1 X X MD04 MD (unincorporated) Baltimore MD 139 (Charles St.)/Bellona Ave U 4 2 X X X MD05 MD Towson Baltimore MD 45/MD 146/Joppa Rd U 5 2 X X X MD06 MD Lothian Anne Arundel MD 2/MD 408/MD 422 R 4 1 X X X MD07 MD Taneytown Carroll MD 140/MD 832/Antrim Blvd S 4 1 X X X MD08 MD Annapolis Anne Arundel MD 450/Spa Rd./Taylor Ave U 4 2 X MD11 MD (unincorporated) Baltimore MD 372/Hilltop Circle (UMBC) U 4 1 X MD12 MD Bel Air Harford MD 7/Holly Oaks Drive S 3 1 X MD13 MD Brunswick Frederick MD 17/A St/B St/Maryland Ave U 5 1 X MD14 MD Cearfoss Washington MD 63/MD 58/Cearfoss Pike R 4 1 X MD15 MD Ellicott City Howard MD 100 EB Ramps/MD 103 S 4 1 X MD16 MD Ellicott City Howard MD 100 WB Ramps/MD 103 S 4 1 X MD17 MD Ellicott City Howard MD 100 WB Ramps/MD 104 S 4 2 X MD18 MD Ellicott City Howard MD 100 WB Ramps/Snowden River Pkwy S 4 1 X MD19 MD Federalsburg Caroline MD 307/MD 313/MD 318 R 4 1 X MD25 MD Lisbon Howard MD 94/MD144 R 4 1 X MD26 MD Lisbon Howard MD 94/Old Frederick Rd R 4 1 X Table 5. Subset of sites used for analysis.

15 Si te ID St a te City County Intersection S et tin g Le gs La n es Sa fe ty O pe ra tio na l Sp ee d Pe de st ri a n Bi cy cl e MD27 MD Millington Kent US 301 NB Ramps/MD 291 R 4 1 X MD28 MD Millington Kent US 301 SB Ramps/MD 291 R 4 1 X MD31 MD Oak Grove Pr. Georges MD 193/Oak Grove Rd U 3 1 X MD33 MD Rosemont Frederick MD 17/MD 180 R 4 1 X MD38 MD Stevensville Queen Annes MD 18/Castle Marina Rd S 4 1 X MD39 MD Temple Hills Pr. Georges MD 637/Good Hope Ave. S 4 1 X MD40 MD Temple Hills Pr. Georges MD 637/Oxon Run Dr. S 4 1 X ME01 ME Gorham Cumberland US 202/State Route 237 U 4 1 X X X MI01 MI Okemos Ingham Hamilton Rd/Marsh Rd S 3 2 X X X MI03 MI East Lansing Ingham Bogue Street/Shaw Lane U 4 2 X MO01 MO Columbia Boone Business Loop/I-70 S 5 1 X MS01 MS Jackson Rankin MS 475/Airport Rd/Old Brandon Rd S 4 1 X NV01 NV Las Vegas Clark Hills Cen. Dr./Vllg. Cen. Cir./Mdw. Hills Dr. S 4 2 X NV02 NV Las Vegas Clark Town Cen. Dr/Hualapai Way/Far Hills Ave. S 4 3 X X NV03 NV Las Vegas Clark Town Cen. Dr./Village Cen. Cir./Lib. Hills Dr. S 4 2 X X NV04 NV Las Vegas Clark Town Cen./Cyn. Run Dr/Banburry Cross Dr S 4 3 X NV05 NV Carson City Carson City 5th St/Edmonds R 4 1 X NV09 NV Las Vegas Clark Carey Ave/Hamilton St U 4 2 X NV10 NV Las Vegas Clark Carey Ave/Revere St U 4 2 X NV16 NV Las Vegas Clark Lake South/Crystal Water Way S 4 1 X NV18 NV Las Vegas Clark Hills Drive/Longspur S 3 2 X OR01 OR Bend Deschutes Colorado Ave/Simpson Dr U 4 1 X X X X OR04 OR Bend Deschutes Century Dr/Colorado Ave/Chandler Ave U 4 1 X OR07 OR Bend Deschutes Mt. Washington Dr/Shevlin Park Rd. S 4 1 X OR09 OR Bend Deschutes Century Dr./14th St./Simpson Ave. U 4 1 X OR15 OR Eugene Lane Barger Dr/Green Hill Rd S 3 1 X SC01 SC Hilton Head Beaufort Whooping Crane Way/Main St S 4 1 X UT02 UT Orem Utah 2000 South/Sandhill Rd U 4 2 X X VT01 VT Manchester Bennington Rte 7A/Equinox(Grand Union) S 4 1 X X VT02 VT Montpelier Washington Main St./Spring St (Keck Circle) U 3 1 X X X VT03 VT Brattleboro Windham RT 9/RT 5 S 4 2 X X X WA01 WA Gig Harbor Pierce SR 16 SB Ramp/Borgen Blvd. S 4 1 X X WA02 WA Gig Harbor Pierce Borgen Blvd/51st S 4 1 X X WA03 WA Bainbridge Is. Kitsap High School Rd/Madison Ave. U 4 1 X X X X X WA04 WA Port Orchard Kitsap Mile Hill Dr. (Hwy 166)/Bethel Ave S 3 1 X X X WA05 WA Sammamish King NE Inglewood Hill/216th Ave NE S 4 1 X X WA06 WA Monroe Snohomish SR 522 EB Ramps/W. Main St./Tester Rd S 5 2 X WA07 WA Lacey Thurston I-5 NB Ramp/Quinault Dr/Galaxy Dr S 4 1 X X X WA08 WA Kennewick Benton 27th Ave/Union St/Union Loop Rd U 4 1 X X X WA09 WA Gig Harbor Pierce SR 16 NB Ramps/Burnham Dr./Borgen Blvd. U 6 2 X X WA10 WA Federal Way King Weyerhauser Way/33rd Pl./32nd Dr. S. S 3 2 X WA15 WA Lacey Thurston Marvin Rd/Britton Pkwy./Willamette Drive S 4 2 X WA16 WA Lacey Thurston College St. SE/45th Ave. SE S 4 2 X WA17 WA Lacey Thurston Marvin Rd./Hawk Prairie Rd. S 4 1 X WA22 WA University Pl. Pierce Grandview Dr/56th St W S 3 1 X WA23 WA University Pl. Pierce Grandview Dr/62nd Court W/Park Entrance S 4 1 X WA24 WA University Pl. Pierce Grandview Dr/Bristonwood Dr/48th St W S 4 1 X WA25 WA University Pl. Pierce Grandview Dr/Cirque Dr S 3 1 X WA26 WA University Pl. Pierce Grandview Dr/Olympic Blvd S 4 1 X WA27 WA University Pl. Pierce 56th Ave./Alameda Ave. W/Cirque Dr. S 4 1 X WI01 WI Howard Brown Lineville Rd (CTH M)/Cardinal Ln S 4 1 X Legend: U = urban, S = suburban, R = rural. Settings are approximate. The complete 2003 site inventory can be found in Appendix C. Table 5. (Continued).

the control type before construction, and crash data for both the before and after periods. Crash data at these roundabouts were gathered by three primary means: • Crash records were gathered from local jurisdictions in the vicinity of all field data collection sites. • Additional data were gathered via phone calls, e-mail, and traditional mailings to jurisdictions that might have round- abouts with significant crash histories (i.e., roundabouts that had been in operation for more than 1 year). • Data were extracted from files created by RPI for the NYSDOT project. The 90 roundabouts in the intersection-level crash dataset were selected based on the availability of crash data (either summaries or detailed crash records), basic geometric infor- mation (e.g., number of lanes, number of legs, and diameter), and total entering daily traffic volumes; all components were needed for the site to be included in the dataset. The majority of these 90 roundabouts were single-lane roundabout sites, in urban or suburban environments. In addition, the round- abouts studied have an average AADT of approximately 16,700 entering vehicles/day (low of 2,668 and a high of 58,800). Fig- ure 2 characterizes the sites used in the intersection-level crash dataset. Figure 3 provides a summary of the frequency of crashes at sites in the intersection-level crash dataset.As shown, the majority of the roundabout crashes per year are occurring at urban, multilane roundabouts.Within the dataset, there was little difference in the frequency of crashes per year at the sin- gle-lane urban, suburban, and rural roundabouts. Tables 6 and 7 characterize the approach-level model dataset, which is a subset of the total intersection-level dataset. A total of 139 legs were included in the approach- level dataset. These 139 legs were selected independently from the 90 roundabouts used for the intersection-level dataset based on different data requirements. As noted previously, for a leg to be included in the dataset, all of the following data were needed: detailed crash records (e.g., police reports), detailed geometry (e.g., entry width, entry angle, approach curvature, etc.), and approaching and circulating daily traffic volumes. Table 8 provides a summary of the geometric data used in the approach-level modeling. The geometric data were developed from a manual review and reduction of data from as-built drawings of the roundabouts. Within the approach-level dataset, Figure 4 provides a summary of the types of crashes occurring. As shown, the majority of the crashes in the dataset occurred at multilane roundabouts and were either exiting-circulating or rear-end crashes. Fifty-five roundabouts were used for the before-after study, as these were the roundabouts where both before- and after- conversion crash records were available. The two sources for the before-after dataset were the previously conducted before-after study for the IIHS (5) and new data collected for this project. A breakdown of this dataset by jurisdiction, control type before conversion, setting, and number of circu- lating lanes is shown in Table 9 and Figure 5. At these sites, before roundabout installation, there were 1,159 crashes; after installation, there were 726 crashes. The average length of time of the crash history before roundabout installation was 3.7 years; the average length of time of the crash history after installation of the roundabout was 3.3 years. Operational Data The overall inventory of roundabouts provided a rich source from which potential sites for the field data collection could be identified. The following criteria were used to iden- tify these sites: • An expectation of queuing on one or more of the round- about approaches, representing capacity conditions 16 0 5 10 15 20 25 30 35 40 45 50 Urban Suburban Rural Setting Si te s Single-Lane Multilane Total 0 2 4 6 8 10 12 14 Urban Suburban Rural Setting A ve ra ge C ra sh es /y r Single-Lane Multilane Figure 2. Summary of roundabout characteristics used for intersection-level safety analysis. Figure 3. Intersection-level roundabout dataset— average crashes per year.

17 Number of legs = 139 (at 39 roundabouts) Mean length of crash history = 3.8 years All Single Lane Multilane Crash Type Incidence Percentage Incidence Percentage Incidence Percentage Entering- Circulating 141 23% 40 29% 101 22% Exit-Circulating 187 31% 10 7% 177 38% Rear-End on Approach 187 31% 42 30% 145 31% Loss of Control on Approach 77 13% 42 30% 35 7% Pedestrian 5 1% 1 1% 4 1% Bicyclist 8 1% 3 2% 5 1% Sum* 605 100% 138 99% 467 100% *Percentages may not add to 100 because of rounding. Number of legs = 139 (at 39 roundabouts) Mean length of crash history = 3.8 years All Single Lane Multilane Crash Type Maximum Crashes Per Year Mean Crashes Per Year Maximum Crashes Per Year Mean Crashes Per Year Maximum Crashes Per Year Mean Crashes Per Year Entering- Circulating 3.03 0.32 3.03 0.22 2.67 0.41 Exit-Circulating 9.09 0.57 9.09 0.57 7.67 0.97 Rear-End on Approach 5.00 0.40 2.00 0.17 5.00 0.64 Loss of Control on Approach 3.03 0.15 3.03 0.18 1.25 0.11 Pedestrian 1.00 0.01 0.14 0.02 1.00 0.03 Bicyclist 3.03 0.05 3.03 0.05 2.00 0.04 Variable Minimum Maximum Mean No. Legs Inscribed Circle Diameter (ft) 85 300 144.1 139 Entry Width (ft) 12 49 22.2 138 Approach Half-Width (ft) 10 49 20.2 130 Effective Flare Length (ft) 0 308 27.5 134 Entry Radius (ft) 26 282 77.8 131 Entry Angle 0 45 19.2 129 Circulating Width (ft) 12 45 26.1 138 Exit Width (ft) 12 51 23.0 128 Departure Width (ft) 10 50 19.3 123 Exit Radius (ft) 21 285 82.0 115 Central Island Diameter (ft) 20 214 77.7 134 Angle to Next Leg 27 180 89.3 135 1/Entry Path Radius (1/ft) -0.0100 0.0200 0.0058 123 1/Circulating Path Radius (1/ft) -0.0300 0.0091 -0.0101 122 1/Exit Path Radius (1/ft) 0.0000 0.0252 0.0053 123 1/Left-Turn Path Radius (1/ft) -0.0400 0.0244 -0.0184 120 1/Right-Turn Path Radius (1/ft) 0.0000 0.0364 0.0102 121 AADT 220 19,593 4,637 139 Table 6. Incidence of approach-level crashes by type. Table 8. Summary of approach-level geometric data used for safety analysis. Table 7. Annual frequency of approach-level crashes by type.

• A balance between single-lane and multilane sites so that operational characteristics of both kinds of sites could be studied • A range of other geometric conditions so that the effect of these conditions on operations could be studied • A clustering of sites so that driving time to the sites could be minimized, thus maximizing the number of sites that could be studied Table 10 shows a list of the 31 sites at which field video recordings were made during spring and summer 2003. Included in the table are the date of the site visit, the site ID, the intersection name, and the city and state in which the round- about is located. A total of 34 hours of traffic operations was extracted including the entry flow,conflicting flow,exiting flow, accepted and rejected gaps, turning movement proportions, travel time for different movements, vehicle types, and delay. A video recording system was designed to allow the team to record the movement of vehicles at the roundabouts selected for the operations study. The recording system included the following components: • One omni-directional camera that provided a 360-degree view of the roundabout 18 0 20 40 60 80 100 120 140 160 180 200 En ter ing /Ci rcu lati ng Ex it/C ircu lati ng Re ar- En d Lo ss of Co ntr ol Pe de str ian Bic ycl e Crash Type Cr as he s Single-Lane Multilane Signalized before TWSC before AWSC before U S R U S R U S R State (Total Sites) 1L 2L 1L 2L 1L 2L 1L 2L 1L 2L 1L 2L 1L 2L 1L 2L 1L 2L Colorado (9) 3 6 Florida (4) 1 2 1 Kansas (4) 1 1 1 1 Maryland (17) 2 4 1 2 8 Maine (1) 1 Michigan (1) 1 Mississippi (1) 1 Missouri (1) 1 Nevada (3) 1 1 1 Oregon (4) 1 3 S. Carolina (1) 1 Vermont (2) 1 1 Washington (6) 1 2 1 1 1 Wisconsin (1) 1 2 3 0 4 0 0 12 5 4 6 9 0 7 1 1 0 1 0 5 4 0 17 10 9 8 1 1 TOTALS 9 36 10 Legend: TWSC = two-way-stop–controlled; AWSC = all-way-stop–controlled; U = urban; S = suburban; R = rural; 1L = one-lane; 2L = two-lane 0 2 4 6 8 10 12 14 Single- Lane Multilane Single- Lane Multilane Single- Lane Multilane Signalized Two-Way Stop All-Way Stop Previous Control In te rs ec tio ns Urban Suburban Rural Figure 4. Approach-level crashes by type. Figure 5. Dataset for before-after study by control type before conversion, setting, and number of circulating lanes. Table 9. Dataset for before-after study by state, control type before conversion, setting, and number of circulating lanes.

• Three digital video cameras that focused on individual legs at the roundabout • Two masts, each extendable to 30 ft, to which the video cameras were attached • Four DVD-R recorders to record the video directly from the digital and omni-directional cameras at the site Figure 6 shows an omni-directional camera (on the left) and a digital camera (on the right) mounted on the top of the mast. At the beginning of the field data collection, both masts were used, with two cameras located on the top of each mast. However, one mast and two digital cameras were destroyed during an unexpected windstorm in Colorado in late June 2003. Modifications were made to the remaining mast so that one omni-directional camera and three digital cameras could be mounted on that one mast. Figure 7 shows a typical view of one leg taken by one of the digital cameras. This view shows both circulating vehicles and vehicles queued on the approach. All vehicle movements asso- ciated with this leg are clearly visible. Figure 8 shows a typical view from the omni-directional camera. Vehicles on all four legs are shown, as well as vehicles circulating on the roundabout. This omni-directional view provides an excellent record of all vehicle movements, as well as of the intersection geometry and markings. Using this video recording system, a total of 262 DVDs were recorded at the 31 sites: 166 DVDs of individual roundabout legs and 96 DVDs recorded of entire intersec- tions using the omni-directional camera. The recordings made for the individual legs included 474 hours of traffic operations. Of the 166 legs recorded, 12 were located at three-lane sites, 58 were at two-lane sites, and 96 were at one-lane sites. Geometric Data To support the operational and safety model development, a wide range of geometric data were obtained for each site, as shown in Figure 9. Where possible, these data were collected using definitions consistent with those used for international safety and operational models. In addition, the type of pedes- trian crosswalk, presence or absence of striping on the circu- lating roadway, lane configurations, and type of vertical geometry were noted. 19 Date Site ID Intersection name City and State Ju n e 9 10 11 12 17 19 20 30 NV01 NV02 NV03 NV04 CO01 CO02 CO03 MD01 Hills Center Dr./Village Center Cir./Meadow Hills Dr. Town Center Dr./Hualapai Way/Far Hills Ave. Town Center Dr./Village Center Cir./Library Hills Dr. Town Center Dr./Banburry Cross Dr. SH-6/I-70 spur South Golden Road/Johnson Rd/16th St. South Golden Road/Utah St. Tollgate Rd. /Marketplace Dr. Las Vegas, NV Las Vegas, NV Las Vegas, NV Las Vegas, NV Eagle, CO Golden, CO Golden, CO Bel Air, MD Ju ly 1 2 7 8 9 10 14 15 16 18 23 25 28 30 31 MD02 MD03 MD04 MD05 MD06 MD07 VT01 VT02 VT03 ME01 MI01 KS01 CO04 UT01 UT02 MD213 at Leeds Rd./Elk Mills Rd. (Lanzi Circle) MD24 at MD 165 (North Harford) MD139 (Charles St.) at Bellona Ave. MD45 at MD146/Joppa Rd. MD 2 at MD 408/MD 422 MD 140/MD 832/Antrim Blvd. Route 7A/Equinox (Grand Union) Main St and Spring St. (Keck Circle) Route 9/Route 5 US 202/State Route 237 Hamilton Rd/Marsh Rd. Sheridan St./Rogers Rd. SH 82/ Maroon Creek, Castle Creek 1200 South/400 West 1200 South/Sandhill Leeds, MD Jarrettsville, MD Baltimore Co., MD Towson, MD Lothian, MD Taneytown, MD Manchester, VT Montpelier, VT Brattleboro, VT Gorham, ME Okemos, MI Olathe, KS Aspen, CO Orem, UT Orem, UT A u gu st 4 5 6 7 11 12 13 15 WA01 WA02 WA03 WA04 WA05 WA06 WA07 OR01 SR 16 SB Ramp Terminal (near Pioneer at Stinson) Borgen Blvd/51st High School Rd/Madison Ave. Mile Hill Dr. (Hwy 166)/Bethel Ave. NE Inglewood Hill/216th Ave. NE SR 522 EB Ramps/W. Main St./Tester Rd. I-5 off-ramp/Quinault Dr/Galaxy Dr. Colorado/Simpson Gig Harbor, WA Gig Harbor, WA Bainbridge Isl., WA Port Orchard, WA Sammamish, WA Monroe, WA Lacey, WA Bend, OR Notes: 1. Other sites were included in the original field list. Bad weather prevented video recording at these sites. 2. The site ID includes the state in which the roundabout is located and the number of that site within a state. Table 10. List of field sites for operational and speed data collection.

Speed Data Currently, speeds are predicted for a roundabout design by measuring speeds along the “fastest path,”as defined in FHWA’s Roundabouts: An Informational Guide (1). This path is assumed to be the fastest path traversable by a single free-flow vehicle without regard to pavement markings or other traffic. This methodology assumes no acceleration or deceleration between points of measure; as such, the resulting predicted speed rep- resents a reasonable upper limit for the given radius, superele- vation, and side friction factor. Appendix G provides details on the specific definitions for defining vehicle paths. Sixteen single-lane and eleven multilane sites were used for this speed analysis. These sites were chosen to represent a range of geometry, surrounding land use, and volumes found at roundabouts, and the data items collected are summarized in the following paragraphs. Spot speed data were collected for this project during the summer of 2003 at each location visited by the field data col- lection team. The speed data were collected using a radar gun, which recorded speeds of free-flow vehicles on each leg to the nearest 1 mph (1.6 km/h) at the following locations: • At least 200 ft (60 m) upstream of the yield line • At the yield line • At the midpoint of the adjacent splitter island • At the exit point of the roundabout The number of actual observed data varied by location and leg, depending on the quantity of free-flow observations avail- able and the time constraints of the field data collection team. For some legs, few or no data points were obtained; at other legs, the number of data points exceeded 30. Data points for entering, circulating, and exiting speeds were differentiated by turning movement (i.e., left, through, and right).All data were differentiated by vehicle type (i.e., passenger cars, trucks), and only passenger car data were used for this analysis. Pedestrian and Bicyclist Data The sites for the pedestrian and bicyclist observational study were selected from the large number of sites where dig- ital video had originally been recorded as part of the overall field data collection effort. The sites chosen for the analysis were generally those with the greatest number of pedestrians and/or bicyclists. Data were also collected at several additional sites with known high volumes of pedestrians and/or bicy- clists, which increased the number of observations and the range of geometric and operational conditions. Ten specific legs, located at seven roundabouts, were cho- sen for the pedestrian study. Three of these legs were also used for the analysis of bicyclist movements. A total of 14 legs at seven roundabouts was selected for the study of bicyclists. 20 Figure 6. Omni-directional camera and digital camera located at top of mast. Figure 7. View of one leg from digital camera, Site WA03-S. Figure 8.Viewfromomni-directional camera,Site WA03.

A summary of all sites used for this analysis is provided in Table 11. A description and image of each leg can be found in Appendix C. The observational data were acquired from DVDs and videotapes for the sites described previously. The data recorded for each event included the information necessary to attempt to answer the questions previously posed in the introduction. For each pedestrian crossing event, the follow- ing information was captured: • Number of pedestrians crossing (to distinguish individuals from groups) • Estimated age (adult versus youth) • Start location (entry side versus exit side) • Crossing type (within or outside the crosswalk) • Arrival time (at the curb and prepared to cross) • Start time (when the crossing was initiated) • Wait time (difference in start time and arrival time) • Splitter arrival/departure 21 Figure 9. Geometric data obtained for each site.

• Splitter time (difference in splitter departure and arrival times) • End time (when the crossing was completed) • Crossing time (difference in end time and start time) • Rejected gaps (time between arrival time and the next vehi- cle reaching the crosswalk and time between each subse- quent pair of vehicles until the pedestrian crosses) • Accepted gap (time between arrival time and the next vehi- cle reaching the crosswalk after the pedestrian crosses or time between the last rejected gap vehicle and the next vehicle reaching the crosswalk after the pedestrian crosses) • Pedestrian behaviors during crossing (none, hesitations, stops, retreats, runs) • Motorist behaviors (yield-slow, yield-stop, does not yield) • Conflicts (requiring either party to suddenly change course and/or speed) For bicyclists, the data elements captured had to be expanded. While there were some bicyclists that crossed the leg like a pedestrian, many of the bicyclists captured were approaching the roundabout in the entry lane, departing the roundabout in the exit lane, or traversing the roundabout within the circulating lane. In addition to this basic event type, the following variables were captured for bicyclists: • Position of bicyclist (within the lane on the leg) • Motor vehicle presence (passing, trailing, leading, none) • Entering bicyclist’s behaviors (yield, did not yield, safe gap, unsafe gap) • Exiting bicyclist’s behaviors (lane/sidewalk position upon exit) • Start location for crossing bicyclist (entry side versus exit side) • Crossing type (within or outside the crosswalk) • Conflicts (requiring either party to suddenly change course and/or speed) • Other behaviors (e.g., wrong-way riding) Appendix D includes more detail about each of the variables captured or derived for pedestrian and bicyclist observations. Conclusion The data described in this chapter are the basis for the safety, operational, and design analysis findings from this research. They also are intended to support future research efforts to further understanding of roundabouts in the United States. Discussion of additional research topics to use and expand upon the current database can be found in Chapter 7. 22 Location Intersection Site/Video ID Observation Period (min) Ped. Events Bike Events Davis, CA Anderson Rd/Alvarado Ave CA06 99 — 89 Santa Barbara, CA Milpas St/US 101 NB Ramps/Carpinteria St CA17 120 — 57 Clearwater Beach, FL SR 60/Gulf Blvd (SR 699) FL11-E1 120 135 19 MD05SW-S1 180 89 — MD05SW-W1 181 65 —Towson, MD MD 45/MD 146/Joppa Rd MD05SW-NW1 180 38 — Las Vegas, NV Town Center Dr/Village Center Cir/Library Hills Dr NV03-S1 220 22 — OR01-N1/N2 233/233 — 59 OR01-S1 233 — 27 Bend, OR Colorado Ave/Simpson Dr OR01-W2 233 — 26 UT02-E1 240 131 12 Orem, UT 2000 South/Sandhill Rd UT02-W1 234 35 — Manchester, VT Main St (Rte 7A)/Grand Union VT01-N1/N2 238/142 94 — VT02-N1 240 — 39 VT02-S1 240 — 58 Brattleboro, VT Rte 9/Rte 5 VT02-W1 240 — 49 Gig Harbor, WA Borgen Blvd/51 st Ave WA02-E1 158 24 — WA03-N2 140 — 29 WA03-S1/S2 /S3 240/140/231 136 112 WA03-E2/E3 141/231 — 84 Bainbridge Island, WA High School Rd/Madison Ave WA03-W1 231 — 30 Totals 5,118 769 690 Legend: Ped. = pedestrian Table 11. Summary of pedestrian and bicyclist observation sites.