Safe and Aesthetic Design of Urban Roadside Treatments (2008)

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.

37 The goals of this research effort are to develop design guide- lines for safe and aesthetic urban roadside treatments and ultimately to develop a toolbox of effective treatments that balance the needs of all roadway users while accommodating community values. Of particular interest is the design of urban roadways that carry substantial volumes of traffic and are designed for higher operating speeds, thus often raising additional roadside safety concerns. To accomplish these goals, the research team performed two specific tasks. These two research tasks were as follows: • Developing a systematic analysis approach (referred to as an Urban Control Zone Assessment) to enable jurisdictions to better target hazardous urban roadside locations and • Developing before-after case studies for a variety of urban roadside treatments. These two tasks are described in detail in the sections that follow, with specific case information included in Appen- dixes A and B, available online at http://trb.org/news/blurb_ detail.asp?id=9456. Appendix C (included herein) includes a toolbox that generally summarizes the safe application of roadside elements in an urban environment. A supplemental product of this research effort is draft language for possible in- clusion in the urban roadside chapter of the AASHTO Road- side Design Guide. This document is included in Appendix C. Urban Control Zone Assessment Experimental Design Many urban roadside environments are crowded with potential hazards. The task of identifying which objects pose the greatest risk to users of the road can be daunting for a jurisdiction with limited resources. In 1999, the Florida De- partment of Transportation (FDOT) developed a document called the Utility Accommodation Manual (117). This docu- ment’s purpose was to provide direction for ways to reasonably accommodate utilities in state transportation facility rights- of-way. FDOT included a concept in this document called “Control Zones” for consideration at facilities with limited or no access control. Although the emphasis of FDOT’s doc- ument was utility pole placement, the concept of control zones can be expanded and is a promising approach for evaluating an urban roadside environment in its entirety. FDOT’s Utility Accommodation Manual defines control zones as the following: Areas in which it can be statistically shown that accidents are more likely to involve departure from the roadway with greater frequency of contact with above ground fixed objects. (117) Example control zones include those that contain objects hit more than two times within three consecutive years, objects located within the return radii and object horizontal offset distance at an intersecting street, objects located within 1 m (3 ft) of a driveway flare, and objects located along the outside edge of a horizontal curve for roads with operating speeds greater than 56 km/h (35 mph). The research team performed a systematic evaluation of crash data to define common control zones for urban road- side environments. To accomplish this task, the research team evaluated urban crash data for four different locations. Study areas included urban corridors located in Atlanta, Georgia; Orange County and San Diego County, California; Chicago, Illinois; and Portland, Oregon. Although not all sites were within the same city limits for a regional study, they were all characterized by urban corridors where fixed-object crashes occurred along the corridor (often in cluster configurations). The sources of the crash data varied. The Georgia Department of Transportation (GDOT) provided Atlanta crash data and road characteristic information. The Oregon Depart- ment of Transportation (ODOT) provided Portland crash data and road information. For corridors located in Illinois and California, the research team used data from the High- way Safety Information System (HSIS) database maintained C H A P T E R 3 Findings and Applications

by FHWA. The research team specifically targeted higher speed urban roads in each of the regions, although some of the corridors included transitions to lower speeds. Comprehensive crash data can be informative, but the re- search team supplemented this information by collecting cor- ridor video data for both directions of travel. The team then used this video data to determine the type and placement of roadside objects, adjacent land use, access density, and so forth. Table 13 shows the actual corridors evaluated for this task. The initial goal developed by the research team was a sample of 16 to 32 km (10 to 20 mi) of urban arterial per city; however, due to select long corridors with frequent fixed crashes, the California and Illinois data collection consider- ably exceeded this initial data goal. As a result, 244.9 km (152.3 mi) of urban arterial from a total of four states are included in this analysis. The goal of this task was to identify urban control zones that can be then applied to other regional analyses for project priority and evaluation. These zones are summarized in the sections that follow. Findings and Recommendations The various corridors the research team evaluated for identification of potential urban control zones included a wide variety of speed limits, physical features, and types of crashes. In Appendix A, each site is described in detail in- cluding observed roadway conditions as well as crash type and crash severity information. In addition, the research team performed a cluster crash analysis to identify loca- tions with an overrepresentation of fixed-object crashes. A spot map for each site is also included in the Appendix A summary. Common fixed-object crash features for each site are further identified in Table 14. As members of the research team evaluated fixed-object crashes at each site, recurring road features emerged at fixed crash locations. Many of these are locations where roadside crashes can be anticipated; however, the information included in Table 14 helps demon- strate the frequency of these road conditions at the study locations. The 6-year crash summaries included in Appendix A pres- ent total crash type information for each study corridor. As is often the case along an urban corridor, a large number of crashes occurred at intersections and driveways. Crashes at these locations are generally angle, head-on, rear-end, and, in some instances, sideswipe crashes. In addition, intersection- related crashes often involve more than one vehicle. As a result, crash severity at each study corridor location is further pre- sented in Table 15, in which crash severity percentages for all crashes are contrasted with crash severity of fixed-object crashes only. The average per state for the study corridors representing the percent injured varied from 22.5 percent to 46.1 percent for all crashes (with an overall average of 34.6 per- cent), while fixed-object injury crashes ranged from 22.2 per- cent to 38.3 percent (with an overall average of 29.2 percent). By contrast, fixed-object crash fatalities at all locations were a larger percentage than for all crashes with a total average of 1.1-percent fatal crashes for all reported fixed-object crashes compared with only 0.3-percent fatal crashes for all crash types (the “all crashes” statistic includes the fixed-object crashes). Table 16 further depicts the total percentage of fixed-object crashes and pedestrian crashes for the study corridors. The urban fixed-object crashes were approximately 6.7 percent of all crashes observed for the four state study corridors. The nature of the roadside crashes at these corridor locations pointed to a common set of frequently hit objects resulting 38 California Sites Georgia Sites Study Corridor Name Length km (mi) Study Corridor Name Length km (mi) S. H. Route 1 11.3 (7.0) Alpharetta Highway 3.5 (2.2) S. H. Route 39 29.3 (18.2) Briarcliff Road 4.2 (2.6) S. H. Route 74 3.2 (2.0) Candler Road 5.6 (3.5) S. H. Route 75 5.3 (3.3) 14th St / Peachtree St 4.8 (3.0) S. H. Route 76 11.3 (7.0) Franklin Road 3.7 (2.3) S. H. Route 78 6.4 (4.0) Moreland Avenue 6.4 (4.0) S. H. Route 90 8.0 (5.0) Roswell Road – 1 (Cobb) 3.2 (2.0) Subtotal: 74.8 (46.5) Roswell Road – 2 (Cobb) 3.2 (2.0) Roswell Road (Fulton) 3.5 (2.2) Subtotal: 38.1 (23.8) Illinois Sites Oregon Sites Study Corridor Name Length km (mi) Study Corridor Name Length km (mi) Route 6 11.9 (7.4) Beavercreek Rd 3.2 (2.0) Route 14 16.6 (10.3) Brookwood Ave 5.5 (3.4) Route 19(Cook) 9.2 (5.7) Cascade Hwy 1.6 (1.0) Route 19(Dupage) 12.1 (7.5) Evergreen Pkwy 4.3 (2.7) Route 25 13.7 (8.5) Farmington Rd 2.7 (1.7) Route 31 14.0 (8.7) Foster Rd 9.3 (5.8) Route 41 16.6 (10.3) McLoughlin Blvd 2.4(1.5) Subtotal: 94.1 (58.4) 185th Ave 8.9 (5.5) Subtotal: 37.9 (23.6) Table 13. Urban control zone sites evaluated in this study.

from these crashes, with varying severity levels. Commonly hit fixed objects included the following: • Poles and posts, • Light standards, • Traffic signals, • Trees and landscaping, • Mailboxes, • Walls and fences, • Barrier or guardrail, and • Embankment. In addition, roadside furniture may have been impacted, but the crash databases classified this type of roadside object as “other unidentified object.” As shown in Table 14, the primary locations for fixed-object crashes were characterized by several common road and road- side configurations. Often, locations with these configurations 39 Case No. Corridor Description L a n e m er ge w ith o bje ct off set s 2 – 6 ft L a n e m er ge w ith o bje ct off set s > 6 f t C h an ne liz at io n isl an d ve ry s m al l w ith p ol es D riv ew ay / in te rs ec tio n w ith o bs ta cl es lo ca te d on fa r s id e M ed ia n at h or iz on ta l c ur ve w ith o bs ta cl e of fs et s 4 –6 ft R oa ds id e at h or iz on ta l c ur ve w ith o bs ta cl es 4 –6 ft N um er o u s ro ad sid e ob sta cl es w ith in 1 –2 ft (o n t an ge nt) R oa ds id e di tc h w ith n on - tr av er sa bl e he ad w al ls U ne v en r o ad sid e w ith o bs ta cl es (o ffs et va rie s) C o rr id or c le ar z on e no t m ai nt ai ne d at ri gh t -t ur n la ne s G ua rd ra il w ra pp ed a ro un d c u rb re tu rn L o ng itu di na l b ar rie r/g ua rd ra il o ffs et 2 –6 ft S c en ic / to ur ist lo ca tio n w ith c l os e ob sta cl es UCZ-CA-1 SH 1, Orange County, CA x x x x UCZ-CA-2 SH 39, Orange County, CA x x UCZ-CA-3 SH 74, Orange County, CA x x UCZ-CA-4 SH 75, San Diego County, CA x x x x UCZ-CA-5 SH 76, San Diego County, CA x x UCZ-CA-6 SH 78, San Diego County, CA x x x UCZ-CA-7 SH 90, Orange County, CA x x x UCZ-GA-1 Alpharetta Highway, Fulton County, GA x UCZ-GA-2 Briarcliff Rd., DeKalb County, GA x x x x UCZ-GA-3 Candler Rd., DeKalb County, GA x x x UCZ-GA-4 14th St./Peachtree St., Fulton County, GA x x UCZ-GA-5 Franklin Rd., Cobb County, GA x x UCZ-GA-6 Moreland Dr., DeKalb County, GA x x x UCZ-GA-7 Roswell Rd. (1), Cobb County, GA x x UCZ-GA-8 Roswell Rd. (2), Cobb County, GA x UCZ-GA-9 Roswell Rd., Fulton County, GA x x x UCZ-IL-1 Route 6, Will County, IL x x UCZ-IL-2 Route 14, Cook County, IL x x UCZ-IL-3 Route 19, Cook County, IL x x x UCZ-IL-4 Route 19, DuPage County, IL x x UCZ-IL-5 Route 25, Kane County, IL x x x x UCZ-IL-6 Route 31, Kane County, IL x x x x UCZ-IL-7 Route 41, Cook County, IL x x x x x x x UCZ-OR-1 Beavercreek Rd., Clackamas County, OR x x UCZ-OR-2 Brookwood Pkwy., Washington County, OR x x x x UCZ-OR-3 Cascade Hwy., Clackamas County, OR x x x x UCZ-OR-4 Evergreen Pkwy, Washington County, OR x x UCZ-OR-5 Farmington Rd., Washington County, OR x x x UCZ-OR-6 Foster Rd., Multnomah County, OR x x x x UCZ-OR-7 McLoughlin Blvd., Clackamas County, OR x x x UCZ-OR-8 185th Ave., Washington County, OR x Table 14. Urban control zone corridor overview.

experienced clustered crashes while much of the roadside along the corridor remained free of crashes. The road and roadside configurations that are often involved in fixed-object crashes can be generally grouped into the following: • Obstacles in close lateral proximity to the curb face or lane edge; • Roadside objects placed near lane merge points; • Lateral offsets not appropriately adjusted for auxiliary lane treatments; • Objects placed inappropriately in sidewalk buffer treatments; • Driveways that interrupt positive guidance and have objects placed near them; • Three kinds of fixed-object placement at intersections; • Unique roadside configurations associated with high crash occurrence; and • Roadside configurations commonly known to be hazardous. Each of these potential Urban Control Zones is discussed in the sections that follow. Obstacles in Close Lateral Proximity to the Curb Face or Lane Edge Historically, a lateral offset (referred to as an operational offset) of 0.5 m (1.5 ft) has been considered the absolute minimum lateral (or perpendicular) distance between the edge of an object and the curb face. As previously indicated, this offset value enabled vehicle access, that is, a person could open a car door if the vehicle were stopped adjacent to the curb. This operational offset was never intended to represent an acceptable safety design standard, although it was sometimes misinterpreted as being one. The urban environment limits lateral offset distances simply because of the restricted right-of-way widths common to an urban setting. 40 All Crashes Fixed-Object Crashes Case No. Percent No Injury or Unknown Percent Injured Percent Fatal Percent No Injury or Unknown Percent Injured Percent Fatal UCZ-CA-1 57.27 42.48 0.26 77.66 21.32 1.02 UCZ-CA-2 55.28 44.13 0.59 74.66 24.80 0.54 UCZ-CA-3 70.30 29.70 0.00 64.29 35.71 0.00 UCZ-CA-4 66.39 33.33 0.28 62.30 37.70 0.00 UCZ-CA-5 29.79 69.09 1.12 56.52 39.13 4.35 UCZ-CA-6 31.72 66.42 1.87 56.41 41.03 2.56 UCZ-CA-7 62.64 37.21 0.14 76.62 23.38 0.00 Average (CA) 53.34 46.05 0.61 66.92 31.87 1.21 UCZ-GA-1 79.48 20.47 0.05 75.00 23.61 1.39 UCZ-GA-2 81.46 18.48 0.06 73.86 26.14 0.00 UCZ-GA-3 75.63 24.25 0.12 74.63 24.63 0.75 UCZ-GA-4 80.23 19.72 0.05 64.02 35.37 0.61 UCZ-GA-5 72.76 27.16 0.08 76.79 23.21 0.00 UCZ-GA-6 75.46 24.43 0.11 71.69 28.31 0.00 UCZ-GA-7 78.33 21.52 0.15 86.49 13.51 0.00 UCZ-GA-8 73.11 26.78 0.11 84.21 15.79 0.00 UCZ-GA-9 80.56 19.31 0.13 72.09 27.91 0.00 Average (GA) 77.45 22.46 0.10 75.42 24.28 0.31 UCZ-IL-1 72.75 26.95 0.30 79.07 18.60 2.33 UCZ-IL-2 79.09 20.64 0.27 78.57 19.05 2.38 UCZ-IL-3 71.63 28.12 0.25 75.68 22.52 1.80 UCZ-IL-4 74.37 25.51 0.12 75.81 24.19 0.00 UCZ-IL-5 68.69 30.78 0.54 72.09 25.58 2.33 UCZ-IL-6 74.35 25.51 0.14 84.00 15.00 1.00 UCZ-IL-7 73.20 26.32 0.49 67.28 30.51 2.21 Average (IL) 73.44 26.26 0.30 76.07 22.21 1.72 UCZ-OR-1 55.91 44.09 0.00 60.00 40.00 0.00 UCZ-OR-2 55.26 44.74 0.00 70.59 29.41 0.00 UCZ-OR-3 46.20 53.80 0.00 50.00 50.00 0.00 UCZ-OR-4 53.90 45.35 0.74 50.00 44.44 5.56 UCZ-OR-5 61.49 38.30 0.21 69.23 30.77 0.00 UCZ-OR-6 58.40 41.13 0.47 44.44 51.85 3.70 UCZ-OR-7 56.79 43.21 0.00 73.33 26.67 0.00 UCZ-OR-8 61.82 38.03 0.15 66.67 33.33 0.00 Average (OR) 56.22 43.58 0.2 60.53 38.31 1.16 Average (4 States) 65.11 34.59 0.30 69.74 29.17 1.10 Table 15. Summary of crash severity distributions for study corridors.

The research team for this project observed several objects located within inches of the edge of the road for the selected study corridors. In general, these items were utility poles, light standards, street signposts, and trees. Evaluation of the role of trees in crashes was difficult because the types of trees in the selected study corridors varied dramatically and included mature rigid trees as well as small-caliper orna- mental trees. Due to the varying nature of the tree placement along the corridors (and the wide range of their frangible tendencies), the research team often could not identify a specific tree involved in the crashes recorded in the crash database. Actual crash reports were not available for most of the locations, so unless a tree exhibited scars, it was not feasi- ble to determine actual tree types involved in crashes. Poles, posts, and light standards, however, were easier to evaluate. The lateral placement of these items was generally consistent along short corridor segments. The research team, therefore, further evaluated crash locations on the basis of crash records for crashes involving poles, posts, and light standards to determine common lateral offsets and crash frequency. Posts (which could, in some instances, be classified as breakaway) were included in this assessment because these items are often coded as “poles or posts,” so posts could not always be evaluated separately in the analysis. To evaluate fixed-object crashes associated with poles, posts, and light standards, the research team viewed the cor- ridor videos (for both travel directions) and for each location recorded data for the characteristics listed in Table 17. For the study corridors extending over the 6-year period, a total of 503 crashes into poles, posts, or light standards occurred. Of these, 389 occurred during dry weather, 78 dur- ing wet weather, 4 during ice, 4 during fog, 19 during snow, and 9 during unknown weather conditions. Table 18 depicts the distribution of these weather-related crashes on the basis of corridor speed limit. Most crashes occurred during dry and wet conditions on roads with posted speed limits of 48 to 72 km/h (30 to 45 mph). To evaluate lateral offset to the objects that were hit, Table 19 further shows these crashes aggregated by the posted speed limit and lateral distance category. (The lateral offset was the 41 Case No. Percent Fixed-ObjectCrashes Percent Pedestrian Crashes Percent Others UCZ-CA-1 10.1 1.1 88.8 UCZ-CA-2 5.4 2.6 92.0 UCZ-CA-3 8.5 1.2 90.3 UCZ-CA-4 17.1 0.8 82.1 UCZ-CA-5 6.4 0.7 92.9 UCZ-CA-6 14.6 3.0 82.4 UCZ-CA-7 5.5 0.9 93.6 Average (CA) 9.7 1.5 88.9 UCZ-GA-1 1.9 0.5 97.6 UCZ-GA-2 2.6 0.4 97.0 UCZ-GA-3 3.9 1.5 94.6 UCZ-GA-4 2.8 1.0 96.2 UCZ-GA-5 4.7 1.3 94.0 UCZ-GA-6 5.9 1.6 92.5 UCZ-GA-7 5.6 0.8 93.6 UCZ-GA-8 10.7 0.0 89.3 UCZ-GA-9 2.9 1.6 95.5 Average (GA) 4.6 1.0 94.5 UCZ-IL-1 6.4 1.6 92.0 UCZ-IL-2 3.8 1.3 94.9 UCZ-IL-3 4.7 0.8 94.5 UCZ-IL-4 3.6 0.5 95.9 UCZ-IL-5 6.6 2.1 91.3 UCZ-IL-6 4.8 0.4 94.8 UCZ-IL-7 14.8 1.8 83.4 Average (IL) 6.4 1.2 92.4 UCZ-OR-1 2.7 0.0 97.3 UCZ-OR-2 14.9 0.0 85.1 UCZ-OR-3 7.6 0.0 92.4 UCZ-OR-4 6.7 1.1 92.2 UCZ-OR-5 2.8 1.3 95.9 UCZ-OR-6 3.2 1.6 95.2 UCZ-OR-7 9.3 0.0 90.7 UCZ-OR-8 1.8 1.4 96.8 Average (OR) 6.1 0.7 93.2 Average (4 States) 6.7 1.1 92.3 Table 16. Summary of crash type distributions.

distance from the curb face or lane edge [at locations without curb].) Only a small subset of these urban crashes occurred at locations where curb was not present. As a result, Table 20 de- picts the 456 sites where curb was present at the pole, post, or light-standard crash location. The cumulative percentage demonstrates that, for curb locations, 93.4 percent of all fixed-object pole/post/light standard crashes occurred within 1.8 m (6 ft) of the curb face while 82.5 percent of these oc- curred within 1.2 m (4 ft) of the curb. This observation, combined with a tendency for clustered crashes (for poles, these occurred along short road segments with objects laterally positioned close to the road), suggests that in an urban environment the placement of rigid objects should ideally be more than 1.8 m (6 ft) from the curb face and no closer than 1.2 m (4 ft) wherever possible. In addition, the frequency of object impact was greater at locations where objects were in close proximity to the road and located on the outside of a horizontal curve. These crashes oc- curred at locations where objects were placed on the right edge of the road and at locations where objects were placed on me- dians. This suggests that the lateral offset placement of objects at horizontal curves should be increased wherever possible. Roadside Objects Placed Near Lane Merge Points The placement of roadside objects in the vicinity of lane merge points increases the likelihood of vehicle impact with these objects. The research team identified several cluster crashes at these lane merge locations: lane drop lo- cations, acceleration taper ends, and bus bay exit transi- tions. As shown in Table 14, six sites included cluster crashes at taper point locations where fixed objects were laterally located less than 1.8 m (6 ft) from the curb face or lane edge (for locations where curb was not present). Clus- ter crashes occurred at two additional sites where these objects were located more than 1.8 m (6 ft) laterally. Lon- gitudinal placement of objects within approximately 6.1 m (20 ft) of the taper point increased the frequency of these crashes. Figure 18 shows two example crash locations with a pole located at lane merge tapers. This increased likeli- hood of fixed-object crashes at lane merge tapers suggests that an object-free buffer zone at taper points on urban roadways would eliminate or reduce roadside crashes at these locations and allow drivers to focus solely on merging into the traffic stream. 42 Description Available Options Curb: Yes or No Continuous Edge line Present: Yes or No Bike Lane Present: Yes or No Driveway Immediately Upstream of Object Hit: Yes or No Night Hours (10 p.m. to 6 a.m.): Yes or No Weather at Time of Crash: Dry Wet Ice Fog Snow Other or Not Stated Lateral Distance from Curb Face (when present) or Lane Edge (when no curb): Less than 1’ 1’ – 2’ 2’ – 4’ 4’ – 6’ 6’ – 8’ 8’ – 10’ 10’ – 15’ 15’ – 20’ Greater than 20’ Table 17. Pole/post/light-standard variables. Speed Limit km/h (mph) Weather 40 (25) 48 (30) 56 (35) 64 (40) 72 (45) 80 (50) 89 (55) Total Dry 0 72 152 29 104 19 13 389 Wet 1 18 26 7 22 2 2 78 Ice 0 0 3 0 1 0 0 4 Fog 0 2 0 0 2 0 0 4 Snow 1 2 1 0 4 1 0 9 Other or Not Stated 1 8 6 2 2 0 0 19 Total: 3 102 188 38 135 22 15 503 Table 18. Poles/post/light-standard crashes for speed limit thresholds.

Lateral Offsets Not Appropriately Adjusted for Auxiliary Lane Treatments At many of the study corridors, roadside objects, such as utility poles, were placed a considerable distance from the ac- tive travel lane. Often lateral offsets of 3.7 to 4.3 m (12 to 14 ft) existed at mid-block locations; however, at locations with aux- iliary lanes, such as extended-length, right-turn lanes devel- oped for driveway or intersection turning movements, the lateral location of the objects remained unchanged, resulting in an effective lateral offset that was often less than 0.6 m (2 ft). Two of the corridors depicted in Table 14 consistently included cluster crashes at these turn-lane configurations. In addition, crashes at many of the other corridor sites occurred periodi- cally (but not always in clusters) at similar turn-lane locations. This observation suggests that increased lateral offsets should be consistently maintained at extended-length, left-turn lane locations. When a lane is added that functions as a higher speed turn lane or a through lane, the roadside objects should be shifted laterally as well. Other auxiliary lane locations can include bike lanes. At these locations, the higher speed motor vehicles are further separated from the roadside environment, so the width of the clear zone should include the bike lane. This does not, however, modify the recommended minimum lateral offset from the curb face. Objects Placed Inappropriately in the Sidewalk Buffer Treatment The placement of roadside objects immediately adjacent to active travel lanes at some corridor sites increased when a sidewalk was physically separated from the curb by a 43 Speed Limit km/h (mph) Lateral Distance m (ft) 40 (25) 48 (30) 56 (35) 64 (40) 72 (45) 80 (50) 89 (55) Total Percent Cumulative Percent 0–3 (0–1) 0 35 71 2 19 1 1 129 25.6 25.6 3–7 (1–2) 2 29 44 16 50 13 3 157 31.2 56.8 7–13 (2–4) 0 26 32 2 32 2 3 97 19.3 76.1 13–20 (4–6) 1 6 23 8 18 1 0 57 11.3 87.4 20–26 (6–8) 0 3 11 1 11 0 0 26 5.2 92.6 26–33 (8–10) 0 3 4 3 2 4 2 18 3.6 96.2 33–49 (10–15) 0 0 0 3 2 0 6 11 2.2 98.4 49–66 (15–20) 0 0 3 3 1 1 0 8 1.6 100 Total: 3 102 188 38 135 22 15 503 100 Table 19. Lateral distance to objects that were hit for all corridors. Speed Limit km/h (mph) Lateral Distance m (ft) 40 (25) 48 (30) 56 (35) 64 (40) 72 (45) 80 (50) 89 (55) Total Percent Cumulative Percent 0–3 (0–1) 0 35 71 2 19 1 1 129 28.3 28.3 3–7 (1–2) 2 29 44 16 50 13 3 157 34.4 62.7 7–13 (2–4) 0 26 27 2 30 2 3 90 19.7 82.5 13–20 (4–6) 1 6 23 2 18 0 0 50 11.0 93.4 20–26 (6–8) 0 3 10 1 9 0 0 23 5.0 98.5 26–33 (8–10) 0 3 1 2 0 0 0 6 1.3 99.8 33–49 (10–15) 0 0 0 0 0 0 1 1 0.2 100 49–66 (15–20) 0 0 0 0 0 0 0 0 0.0 100 Total: 3 102 176 25 126 16 8 456 100 Table 20. Lateral distance to objects that were hit for corridors with curb only.

buffer strip that contained fixed objects. Interestingly, crashes varied dramatically at these locations. At locations with buffer strips 0.9 m (3 ft) wide or narrower, objects were systematically hit. Wider buffer strips containing ma- ture trees with large-diameter trunks placed within 0.9 to 1.2 m (3 to 4 ft) of the curb showed a significant increase in the number of severe crashes with the trees. At locations with smaller, ornamental trees located in the center of a buffer strip, the number of severe crashes into trees was dramatically reduced. By contrast, utility poles and light standards were frequently hit at locations where these objects were placed in the center of the buffer strip. At sev- eral sites, however, the research team observed smaller, more forgiving objects, such as landscaping with small- caliper trees, positioned near the center of the buffer strip and the more rigid poles and light standards positioned immediately adjacent to the sidewalk and as far from the active travel lane as possible. The crash analysis at these staggered-object-placement buffer strips showed very few roadside crashes. This research suggests that the placement of rigid objects on sidewalk buffer strips 1.2 m (4 ft) wide or narrower should be avoided. For wider buffer strips, placement of the more forgiving roadside items close to the road and place- ment of the more rigid objects at a greater lateral offset of 1.2 m (4 ft) or more from the curb face is recommended. Figure 19 depicts recommended buffer strip object place- ment scenarios. Driveways Interrupt Positive Guidance/Objects Placed Near Driveways Many rural roads have a white edge line delineating the right edge of the travelway. In urban environments, a continuous white line is often not included at locations with a curb as the curb itself functions to delineate the edge of the road. During nighttime or inclement weather conditions, the need for positive guidance along the right edge of the road may be heightened due to reduced visibility. In addition, impaired or fatigued drivers may depend more heavily on this delineation to help them keep their vehicle within the boundaries of the travelway. For a few of the observed corridors, a continuous white line occurred either at the edge of the gutter pan or a few feet from the gutter pan to delineate a separate bicycle lane. When this white line was not present, single-vehicle crashes tended to occur more frequently at driveway locations. In par- ticular, objects positioned on the far side of driveways were hit more often than objects located away from driveways or objects that were located on the near side of driveways. This observation is not a surprise because the curb line may no longer provide positive guidance to vehicles at driveway entry points, and the driveway configuration certainly does not provide a re-direction function when a vehicle drifts from the road. Figure 20 shows an example crash location with a pole located at the far side of a driveway. The place- ment of the pole within the sidewalk is, of course, also not recommended. Of the 456 pole/post/light crashes that occurred at loca- tions with curb (see Table 20), 181 occurred at driveways, so this location accounted for approximately 40 percent of all of these crashes. Of the 181 driveway-associated, fixed-object crashes, 155 occurred at locations with no supplemental positive guidance such as a white edge line, approximately 86 percent of these crashes. This percentage of the crashes associated with driveways that did not provide additional positive guidance could simply be an artifact of how many sites did not have an edge line, but this high a number of crashes certainly warrants future research. Regardless, avoiding 44 Photo by Karen Dixon. Figure 18. Pole placed at lane merge.

the placement of poles on the immediate far side of driveways will help to reduce the number of crashes. It is important to remember that placing poles on the immediate far side of a driveway may sometimes be the result of trying to avoid putting an object on the near side of the driveway in the visibility triangle for drivers of vehicles exiting the drive- way, so relocation of the pole should avoid this critical loca- tion as well. Three Kinds of Fixed-Object Placement at Intersections Crashes at intersections often occur between vehicles; however, several intersection crashes in which vehicles hit roadside objects were also noted in the corridor analysis. In some cases, the crash occurred because a driver attempted to avoid hitting another vehicle; however, several single-vehicle crashes were also observed at intersection locations. In gen- eral, these single-vehicle, fixed-object crashes at intersections fell into one of the following three categories: • Impacted small channelization islands (often these is- lands included signs, traffic signals, or poles). This inter- section crash type occurred at six of the study corridors (see Table 14). • Impacted objects positioned close to the lane edge. These objects often interfered with turning movements when vehicles veered from their turning path. • Impacted objects where pedestrian ramps at intersection corners were oriented in such a way as to direct errant 45 Curb Face 4' minimum from Curb Face to Impact Surface of Rigid Object Si de w al k Frangible Object near Center of Buffer Strip Curb Width Approx. 6" Buffer Strip Width > 4' A dja cen t L an e Landscape and Rigid Object Placement for Buffer Strip Widths > 4’ Curb Face Rigid Objects Placed on Far Side of Sidewalk Si de w al k Frangible Objects Only in Buffer Strip Region Curb Width Approx. 6" A dja cen t L an e Buffer Strip Width < 4' Landscape and Rigid Object Placement for Buffer Strip Widths ≤ 4’ Figure 19. Object placement at buffer strips.

vehicles toward roadside objects. Figure 21 shows one crash location where this occurred. This crash condition is similar to the driveway crash condition with no positive guidance discussed previously. Unique Roadside Configurations Associated with High Crash Occurrence Several of the study corridors were characterized by high crash numbers at a specific location. Often this peak in crash statistics resulted from a physical road feature unique to the site. For example, at corridor UCZ-IL-1 a disproportionately large number of crashes involved an underpass structure. When members of the research team inspected the site, they determined that sometime in its past, a two-way road with one lane under each side of the underpass wall had been converted into a two-lane, one-way road. This modification occurred at two separate locations due to the creation of a one-way pair configuration. As a result, the approach to the underpass required vehicles to shift in an effort to avoid the wall now located between the lanes in the same direction of travel. The crashes occurred when a vehicle did not safely navigate this required lane shift. This crash cause was evident due to a scarred underpass wall. Locations of this type are unique and should be considered individually for crash mitigation treatments. The creation of simple crash spot maps (as shown in Appendix A) can enable an agency to quickly identify cluster crash locations of this nature. Roadside Configurations Commonly Known to Be Hazardous Several roadside crashes occurred at locations where they would be expected. These sites exhibited characteristics known to result in potentially hazardous conditions. For example, locations with roadside ditches, nontraversable headwalls and culverts (often at driveways), or uneven roadside grading were common roadside crash locations. In addition, crashes occurred at high-speed locations where sloping curb delin- eated the roadside edge, but adequate clear zone was not available. Finally, three of the corridors were located in the vicinity of scenic or tourist attractions. At these locations, roadside objects were hit more frequently even when lateral offsets to the objects were similar to those at other crash-free sites. Regardless of the cause, additional lateral offset to objects in these or similar locations seems prudent to mini- mize the risk of hazardous run-off-road crashes for unfamiliar drivers. Case Study Task and Summary of Findings Experimental Design The individual projects selected as part of this case study task were used to identify strategies where safety and aes- thetics were incorporated into the roadway’s design. The research team identified several recent beautification or roadside improvement projects to use as indicators for the 46 Photo by Karen Dixon. Figure 20. Lack of positive guidance at a driveway. Photo by Karen Dixon. Figure 21. Object orientation with access ramp.

influence of improvements on crash conditions. Ideally, a project where only one item is changed (such as moving trees to the far side of sidewalks) would be perfect for this task; however, the research team could not identify projects of this nature because, in general, transportation agencies implement multiple improvements in each project. As a re- sult, the data analysis for this case study task can provide general indications about the safety impacts of beautifica- tion or roadside improvement projects, but cannot be used to explicitly evaluate individual features and their associated hazards. Initially, the research team proposed two levels of case studies (one with comparison sites); however, the NCHRP Project 16-04 panel requested the stand-alone case studies (locations without comparison sites) so as to evaluate as many projects as possible. The stand-alone case studies in- clude crash type summaries, crash severity summaries, and before-after crash analysis for a specific improvement corridor with data developed and provided by local juris- dictions. The research team attempted to solicit candi- date projects from a variety of geographically distributed jurisdictions. Inclusion of a case study project required that the project have a focus on median or roadside improvements as well as having available most, if not all, of the requested data. The research team attempted to collect crash data for 3-year periods before and after the project was implemented. This level of crash information was not available for all sites. In select cases, projects with a minimum of 1 year of data were included; projects with less than 1-year’s worth of post-reconstruction data were excluded. The construction period is indicated in the case study summaries included in Appendix B. More specifically, the research team collected data in the following five areas for the case study analysis: 1. Crash frequency—the absolute number of crashes occur- ring before and after the context-sensitive improvement. 2. Crash rates—raw crash volumes considered in relation to the traffic volume carried by the roadway. 3. Crash severity—the proportion of crashes involving seri- ous injury or death compared with property-damage-only (PDO) crashes. 4. Crash type—changes in specific types of crashes that have occurred as a result of the improvement. 5. Average daily traffic (ADT)—the average number of ve- hicles per day using the roadway. In addition to crash type summary information and crash severity summary information, Appendix B includes a simple before-after crash summary comparison for each site. Often before-after analyses are limited because researchers study corridors where safety issues are prominent and so the re- sulting improvements can be dramatic; however, for the beau- tification and roadside enhancement projects the focus is not on operations and safety but rather on aesthetics and livability. As a result, the before-after analysis can provide a useful in- dication about possible safety implications of a change to the road environment. Table 21 demonstrates the basic type of data included in the before-after analysis for each case study included in Appendix B. Findings and Recommendations Table 22 illustrates the individual case study elements in- cluded in Appendix B as well as the general observed safety trend for each project. The research team attempted to exclude projects in which entire lanes were added as the observed safety results because these types of projects provide con- founding information; nonetheless, a few of the projects had some lane widening (often due to realignment) and are so noted in the table. The crash trends identified in Table 22 show (1) when crash frequency increased by more than one crash per year (or by more than 5 percent), (2) when crash frequency decreased by more than one crash per year (or by more than 5 percent), or (3) when change in crash frequency was minimal (within one crash per year on average or within 5 percent of the original crash rate). These crash trends are a summary of the before-after analysis documented in Table 21 and the individual case studies included in Appendix B. 47 Comparison Analysis Category Candidate Street -- Before Candidate Street -- After Crash Reductions Standard Deviation Crash Frequency Crash Rate Severe and Fatal Crash Frequency Single-Vehicle Crash Frequency ADT Table 21. Evaluation matrix for case study sites.

In Table 22, the before-after crash trends are represented by the four statistics: • Frequency of all crashes at a site, • Crash rate, • Frequency of severe crashes at a site, and • Frequency of single-vehicle crashes. Ideally, a reduction in all four trend statistics would be observed, clearly demonstrating enhanced safety at a site; however, in many cases, an increase occurred for one before- after crash trend statistic while others remained constant or decreased. For all candidate improvement projects, a designer seeks to reduce the number of severe crashes at a site. Severe crashes, for the purposes of the values shown in the case study tables, generally include incapacitating injuries or fatalities. Only three of the case study sites exhibited an increase greater than one additional severe crash per year. All three of these case study sites included sidewalk improvements with buffer strips, but several similar improvement projects resulted in little change to a reduction in severe crashes. Since the focus of this research effort is roadside crashes, and these frequently are single-vehicle crashes, an increase in these kinds of crashes may be of concern. Single-vehicle crashes increased by more than one crash at eight of the sites. In general, these sites included pedestrian enhancement im- provements; however, as was the case with the sites of severe crashes discussed above, there were many pedestrian enhance- ment projects that resulted in reduced single-vehicle crashes. Since inspection of the individual before-after crash trends provides confounding results, a more effective approach may be to examine all four before-after crash trends collectively. 48 Streetscape Project Elements Before-After Crash Trends* Case No. C u rb & G ut te r Cu rb E xt en sio ns Si de w al k A dd iti on s / Im pr ov em en ts L a n ds ca pe B uf fe r N ex t t o Ro ad & Si de w al k A dd L an ds ca pi ng / St re et T re es T r ee s R em ov al / R el oc at io n Im pr ov e Ro ad sid e G ra di ng /D itc h R em o v al R el oc at e U til ity P ol es A dd o r E nh an ce S tre et L i gh ts B us S to ps / Ba ys E n ha nc ed P ed es tri an C r o ss in gs / A cc es s M ed ia n Is la nd s / R ai se d Is la nd s B ic yc le L an es W id en in g of R oa d > 8’ Fr eq ue nc y of A ll Cr as he s Cr as h Ra te Fr eq ue nc y of S ev er e Cr as he s Fr eq ue nc y of S in gl e- V eh ic le C ra sh es CS-AZ-1 x x x x x x x ⇓ CS-AZ-2 x x x x x x x ⇑ CS-AZ-3 x x x x ⇓ CS-CA-1 x x x x x ⇑ CS-CA-2 x x x x ⇑ CS-CA-3 x x x x x x x x ⇓ CS-MN-1 x x x x x x ⇓ CS-MT-1 x x x ⇑ CS-MT-2 x x x x x ⇑ CS-NC-1 x x x x x ⇓ CS-NC-2 x x x x x x x x ⇓ CS-NC-3 x x x ⇓ CS-NC-4 x x x x x x x ⇓ CS-NC-5 x x x x x ⇓ CS-NC-6 x x x x x x x ⇓ CS-NC-7 x x x ⇓ CS-OR-1 x x x x x ⇑ CS-OR-2 x x x x x x x ⇑ CS-OR-3 x x x ⇓ CS-OR-4 x x ⇓ CS-OR-5 x x x x x x ⇓ CS-OR-6 x x x x ⇓ CS-OR-7 x x x x x ⇓ CS-UT-1 x x x ⇓ CS-UT-2 x x ⇓ CS-UT-3 x x x ⇓ CS-UT-4 x x x ⇑ ⇓ ⇑ ⇔ ⇑ ⇔ ⇓ ⇓ ⇑ ⇑ ⇓ ⇓ ⇓ ⇔ ⇓ ⇓ ⇑ ⇑ ⇑ ⇓ ⇓ ⇓ ⇓ ⇓ ⇓ ⇓ ⇓ ⇑ ⇓ ⇔ ⇑ ⇔ ⇔ ⇔ ⇔ ⇔ ⇔ ⇔ ⇓ ⇓ ⇔ ⇔ ⇔ ⇔ ⇑ ⇑ ⇔ ⇔ ⇔ ⇔ ⇔ ⇔ ⇓ ⇔ ⇔ ⇔ ⇑ ⇔ ⇑ ⇔ ⇔ ⇓ ⇔ ⇑ ⇑ ⇓ ⇑ ⇔ ⇔ ⇓ ⇓ ⇔ ⇔ ⇔ ⇑ ⇔ ⇓ ⇑ ⇔ ⇑ ⇓ ⇔ *Before-After symbols depict the following:  Crash frequencies increased by more than one crash per year; crash rates increased by more than 5 percent. ⇑ ⇓  Crash frequencies decreased by more than one crash per year; crash rates decreased by more than 5 percent. ⇔  Crash frequencies for the “After” condition were within one crash per year of the “Before” condition; crash rates for the “After” condition were within 5 percent of the “Before” condition crash rates. Table 22. Case study project elements versus before-after crash trends.

At 10 of the sites, crashes were reduced or remained constant. At nine additional sites, three of the four crash trends were reduced or remained similar during the “after” period. This results in 19 of the 27 sites having a general trend of crash re- duction. None of the sites exhibited an increase in all four crash trend statistics, and only five sites exhibited an increase in three of the four crash trends evaluated. As a result, specific case study assessments (see Table 22 and Appendix B) pro- vided inconclusive results and should be used simply as indi- cators of the expected outcomes for similar improvement projects. General Recommendations The use of corridor video analysis combined with historic crash statistics provided meaningful insight into urban road- side crash conditions and locations where roadside objects should not be located, if possible. Conversely, the use of roadside improvement or beautification case studies did not directly help to address specific roadside safety issues, but these case studies can be used by an agency proposing similar projects to determine expected overall safety performance of these improvements. This research clearly shows that there are specific locations prone to roadside crashes where agencies should avoid the placement of rigid objects. For jurisdictions with limited roadside safety improvement funds, urban control zones can be used to help agencies establish spending priorities for incremental roadside safety improvement on their urban cor- ridors. The research suggests the following: • Avoid locating rigid obstacles in close proximity to a curb face or lane edge (at curb locations where it is possible, in- crease the lateral offset to rigid objects to 1.8 m [6 ft] from the face of the curb and do not allow the distance of this offset to be less than 1.2 m [4 ft]); • Restrict the placement of rigid objects at lane merge loca- tions (avoid placing rigid objects within 3.0 m (10 ft) lon- gitudinally of the taper point, which will provide a 6.1-m (20-ft), object-free length); • Maintain offsets at selected higher speed auxiliary lane locations, such as extended-length, right-turn lanes (main- tain the lateral offset from the curb face at these locations); • Maintain careful object placement within the sidewalk buffer treatment (avoid rigid objects in buffers 0.9 m (3 ft) in width or less and strategically position objects in wider buffers); and • Avoid placing rigid objects in the proximity of driveways (avoid placing rigid objects on the immediate far side of the driveway and do not place any objects within the required sight triangle for the driveway). In addition, roadside crashes occurred frequently at inter- sections; at unique configurations (e.g., a one-way lane split at an underpass); and known hazardous roadside conditions, such as roadside ditches, non-traversable headways, and so forth. 49