Development of a Posted Speed Limit Setting Procedure and Tool (2021)

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.

NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 31 APPENDIX B. RELATIONSHIP AMONG ROADWAY CHARACTERISTICS, SPEED, AND SAFETY FOR HIGH-SPEED HIGHWAYS Several factors are known or suspected to affect safety on rural or high-speed highways, such as horizontal/vertical alignment, shoulder width, passing zone and passing lane presence, and access point density. As documented in the Transportation Research Board (TRB) Modeling Operating Speed Synthesis Report (114), several factors influence operating speed. Most studies focused on how horizontal curvature influences the free-flow speed selected by roadway users. The following sections discuss the findings reported in the research literature on the relationship among crashes, operating speed, and rural highway roadway characteristics including posted speed limit. Note that the goal of this literature review was to identify variables that influence driver speed choice as well as crash potential. These variables were then considered for inclusion in the SLS-Procedure that was the key objective for this research project (NCHRP Project 17-76). TRAFFIC—VEHICLES Average Daily Traffic Relationship with Crashes The amount of AADT present is associated with crashes (115). The relationship between traffic and crashes can be affected by whether the section is undivided or divided (115). Nightingale et al. (116) investigated the influence of factors affecting crash frequency at high- speed rural intersections. The study found that AADT of both major and minor roadways is positively associated with crash frequencies. Average Daily Traffic Relationship with Operating Speed The amount of AADT present may be associated with operating speed (115). For two- lane rural highways, Lamm et al. (117) found that speed increases as AADT increases, while Jessen et al. (118) found lower speeds to be associated with higher AADT. The AADT for the Lamm et al. study ranged from 400 to 5,000 vehicles per hour (vph). The Jessen et al. study is valid for AADTs less than or equal to 5,000 veh/d; the researchers in this study commented that motorists may view increases in volume as a motivation to slow down. Robertson et al. (119) conducted a study on four-lane highways and found that the hourly directional volume was significant for cars during the day, with higher speeds associated with a larger volume. However, the increase was small and the range of volume available was not very large for a highway (average of 379 vph during daytime). In another study, Dong et al. (120) found that AADT is associated with different speed profiles. Percent Trucks Relationship with Crashes Prior research has demonstrated that the percentage of commercial trucks is associated with an increase in observed crashes (115). Percent Trucks Relationship with Operating Speed Robertson et al. (119) and Himes and Donnell (121) identified the percentage of trucks as a relevant factor in their study of rural four-lane highways.

NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 32 Operating Speed Relationship with Crashes A study using 179 roadway sections in Israel explored the relationship between free-flow speeds (obtained from GPS devices) and crashes on rural, two-lane highways with a typical speed limit of 50 mph (9). NB statistical models were developed, and these models considered day hours and night hours separately. The models used speed indicators, section length, traffic volume, and homogeneous road groups, where the road groups reflected various road design conditions. The main finding of the study was that in both day and night hours, the number of injury crashes increased with an increase in the segment mean speed, while controlling for traffic exposure and road infrastructure conditions. Wang et al. (122) reviewed several previous studies to identify factors, especially traffic and road-related factors, related to crashes. They concluded that some studies found increased speed reduces safety and other studies found the opposite. Another observation was that “speed itself may not be a safety problem but speed variation is.” TRAFFIC CONTROL DEVICES Posted Speed Limit Relationship with Crashes For rural high-speed highways, posted speed limits are typically established with consideration of several factors, including the roadway design speed. Vehicular operating speeds along tangent sections of two-lane highways have been shown to be impacted by the posted speed limit, with vehicular speeds tending to increase as the posted speed limit increases (123, 124). However, the magnitude of the increase in operating speed is typically only a fraction of the amount of the actual speed limit increase. For undivided roadways, mean speeds generally increase by 3 to 5 mph for every 10-mph increase in speed limit above 55 mph, with diminishing effects at higher speed limits (123, 125, 126). The research literature generally suggests that the resulting change in operating speeds would likely lead to an increase in the overall crash rate and would shift the severity distribution toward crashes of greater severity (123). Specifically, Kockelman showed that increasing the non-freeway speed limit would increase the total crash rate, and the probability of a fatality would increase. The injury crash probability was also expected to increase with increasing speed limits, while the property damage crash probability was expected to decrease slightly. The study by Park et al. (127) conducted a full Bayesian before-after safety evaluation based on 33 treatment sites and 44 comparison sites to assess the effects of decreasing speed limits on crashes on expressways in Korea. They found that the reductions in speed limit were associated with reductions in crashes. Russo et al. (128) found that increased speed limits led to a statistically significant increase in fatal and injury crashes (around 11 percent). Warner et al. (129) showed that increased posted speed is associated with higher fatal crashes. Gayah et al. (130) examined the operational and safety impacts of setting posted speed limits below engineering recommendations. The exploration on the rural roads in Montana showed a statistically significant reduction in total, fatal and injury, and non-injury crashes at segments with posted speed limits set 5 mph lower than engineering recommendations. Wilmot and Jayadevan (131) used a Louisiana crash database for 1999–2004 to examine crash rates before and after a speed limit change on rural roads. The findings showed that increase in posted speed limit had an impact on crash rate for six out of the 39 cases at the 5 percent level of significance.

NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 33 Posted Speed Limit Relationship with Operating Speed For rural high-speed highways, posted speed limits are typically established by taking several factors into consideration, including the roadway design speed. Vehicular operating speeds along tangent sections of two-lane highways have been shown to be impacted by the posted speed limit, with vehicular speeds increasing as the posted speed limit increases (123, 124, 118). Operating speed has also been found to be related to posted speed on curves (132, 118). The magnitude of the change in operating speed when there is an increase (or decrease) in posted speed is typically only a fraction of the amount of the actual speed limit change (123, 125, 126, 133). For undivided high-speed rural roadways, mean speeds are generally 3 to 5 mph higher for every 10-mph increase in speed limit above 55 mph, with smaller increases at higher speed limits (123, 125, 126). Hu (134) showed that raising the posted speed limit from 75 to 80 mph on rural interstate roadways leads to higher travel speeds and an increased probability of exceeding the new speed limit. The Highway Capacity Manual (HCM), version 6.0 (135), states in Exhibit 12-18 that the base free-flow speed under ideal conditions exceeds the speed limit by 5 mph for freeway segments with a posted speed limit range of 55 to 75 mph as well as for multilane highway segments with a posted speed limit of 45 to 70 mph. The HCM also provides additional information in Chapter 12 about adjusting the freeway free-flow speed using adjustment factors for lane width, right-side lateral clearance, and total ramp density. The adjustment factors for multilane highway segments include lane width, total lateral clearance, median type, and access point density. Dixon et al. (136) reviewed speed data for 12 rural multilane sites in Georgia to evaluate the effects of repealing the 55-mph national speed limit. They found that operating speeds were higher after the increase in the posted speed limit. Himes and Donnell (121) identified the posted speed limit as relevant to their study of rural four-lane highways. Robertson et al. (119) found the daytime posted speed limit to be a significant variable based on data from 36 rural four-lane non- limited-access roadways. The findings from Gayah et al. (130) showed that operating speeds closely comply with the posted speed limit when the posted speed limit is set equal to or 5 mph lower than engineering recommendations. No-Passing Zone Relationship with Crashes In a Michigan study, as the proportion of each segment with no-passing zones increased, the frequency of total and injury crashes also tended to increase (115). Passing Lane Relationship with Crashes While the presence, length, and location of passing zones on two-lane highways likely has an effect on their safety performance, this effect has not been well documented in the previous literature. In fact, the HSM (16) notes the following treatments related to passing zones as having an unknown effect on traffic crashes:  Different passing sight distances.  Presence of access points/driveways around no-passing zones.  Different lengths of no-passing zones.  Different frequency of passing zones.  Passing zones for various weather, cross-section, and operational conditions (16).

NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 34 Passing Lane Relationship with Operating Speed Freedman and Kaisy (137) investigated the passing maneuvers within a passing lane section on a rural two-lane, two-way highway located on a U.S. highway in Montana. This study considered speed differentials for different passing maneuvers. ROADWAY GEOMETRY Horizontal Alignment Relationship with Crashes Horizontal curves have been identified as the geometric variable that is the most influential on driver speed behavior and crash risk (138). Horizontal curves with radii less than 2,600 ft tend to cause highway operating speeds to drop below those of adjacent tangent sections, with substantial speed declines in speed observed for curves with radii less than 800 ft (139). Horizontal alignment is also associated with negatively affecting safety, as shown in the HSM (16). Prior research has shown that crash frequency increases with the length and/or degree of horizontal curvature (115, 123, 140, 141, 142, 143,). On two-lane rural highways, design speed at horizontal curves inconsistent with a driver’s desired speed could create operating speed irregularities by increasing the driver’s work load, which may incur higher crash potential (138, 144), particularly if operating speeds through the curve are reduced by more than 3 mph from the adjacent tangent section (138). Camacho-Torregrosa et al. (145) developed safe operating speed profiles for 33 Spanish two-lane rural road segments (including curved segments) and checked several consistency measurements based on the global and local operating speed. Horizontal Alignment Relationship with Operating Speed Horizontal curves have been identified as the geometric variable that is the most influential on driver speed behavior and crash risk. Two-lane rural highway studies that have found a horizontal curvature measure include Wooldridge et al. (138), Lamm et al. (117), Morrall and Talarico (146), Islam and Senevirantne (147), Krammes et al. (148), Voigt and Krammes (149), Passetti and Fambro (150), McFadden and Elefteriadou (151), Fitzpatrick et al. (152), Gibreel et al. (153), Schurr et al. (132), Figueroa and Tarko (78), and Misaghi and Hasson (154). The measures used in the studies varied and included degree of curve, length of curve, deflection angle, and/or superelevation rate. Horizontal curves with radii less than 2,600 ft tend to cause highway operating speeds to drop below those of adjacent tangent sections, with substantial speed declines observed for curves with radii less than 800 ft (139). Polus et al. (124) used the characteristics of the horizontal curves prior to and following a tangent, along with the tangent length, to predict the 85th percentile speed. A study on rural four-lane highways in Kentucky (155) developed a speed prediction model that considered factors like lane (inside or outside), horizontal curve length or radius, and indicator variables for shoulder type (surfaced), median barrier presence, pavement type (concrete or asphalt), approaching section grade, and curve presence on approach. In a study on rural four-lane highways in Texas (119), the angle of the next downstream horizontal curve influenced the speed of the daytime car drivers on the approach tangent. Bassani et al. (156) showed that an increase in the horizontal alignment results in a decrease in the observed average speed. Llopis-Castello et al. (157) analyzed truck speeds on 105 horizontal curves of rural two- lane roadways. The study showed that the radius of the horizontal curve and the grade at the point of curvature have a significant influence on heavy vehicle speeds.

NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 35 Vertical Alignment Relationship with Crashes Prior research has shown that steeper vertical alignments could induce higher crash potentials (16, 123). Total crash rates typically increased with the degree of vertical alignments (123), mainly in the presence of hidden horizontal curves, intersections, or driveways (158). Safety risks associated with higher speed limits increased on segments with steeper vertical curves (123). Furthermore, Kyte (159) suggested that vertical alignment is associated with higher crash likelihood in wet-weather conditions. Vertical Alignment Relationship with Operating Speed Fitzpatrick et al. (152) conducted a study on two-lane rural highways and determined that passenger car speeds on vertical curves with limited sight distance and horizontal tangent could be predicted using the rate of vertical curvature as the independent variable. Fambro et al. (160) used inferred design speed to predict the 85th percentile speed on vertical curves. Jessen et al. (118) reported that the approach grade affected vehicle speeds at the location with minimum available sight distance along the crest vertical curve. They also commented that the posted speed of the road had the most influence on speed. Schurr et al. (132) found that speed decreases as approach grade increases on horizontal curves. Figueroa et al. (78) found decreasing speed as grade increases on tangents. In another study, Gibreel et al. (153) included several vertical alignment measures in their speed prediction models for horizontal curves combined with vertical curves, such as length of vertical curve and grades. For rural two-lane highway tangent sections with a non-limited crest vertical curve or a sag vertical curve, the recommendation was to assume the desired speed as being the expected 85th percentile speed. A study on rural four-lane highways in Kentucky (155) found the approaching section grade to be related to operating speed. Llopis-Castello et al. (157) showed that the difference between both speed percentiles was lower as the grade increased for the loaded trucks. On the contrary, the speed difference increased as the grade increased for unloaded trucks. Terrain Relationship with Crashes Rolling terrain, which may serve as an indication of vertical curvature, has been associated with an increase in crashes for undivided highways (115). Garach et al. (161) developed SPFs for Spanish two-lane rural highways on flat terrain. Median Relationship with Crashes The presence of a median on multilane, non-freeway facilities could decrease expected crash rates. Kockelman showed that the inclusion of a median on a multilane roadway was associated with an approximate 9 percent reduction in traffic crash rates (123). A Michigan study found that the presence of a TWLTL was associated with a significant increase in total and injury crashes but was also associated with a significant decrease in fatal crashes (115). Median Width Relationship with Crashes Hu and Donnell (162) found that narrower medians are associated with severe injury outcomes in cross-median crashes.

NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 36 Median Relationship with Operating Speed A study on rural four-lane highways in Kentucky (155) found that the presence of a median barrier is related to operating speed. Dong et al. (120) conducted speed studies at 32 sites that had been upgraded from two-lane roadways to four- or five-lane roadways. The study found that median type is associated with different speed profiles. Median Width Relationship with Operating Speed Dong et al. (120) found that median width is associated with different speed profiles. Number of Lanes Relationship with Crashes A Michigan study found that four-lane, undivided facilities tended to observe significantly more crashes across all severity levels compared to two-lane, undivided facilities (115). Lane Width Relationship with Crashes Past research has shown that the width of travel lanes is related to the safety performance of both two-lane and multilane highways (16). Wider lanes in particular have been associated with reductions in single-vehicle run-off-the-road, head-on, and sideswipe type crashes (16). It is important to note that the effect of lane width on safety performance is reduced for multilane highways compared to two-lane highways (163). Lane Width Relationship with Operating Speed For two-lane rural highways, Lamm et al. (117) found that speeds increase with wider lane width, while Figueroa et al. (78) found a similar relationship using pavement width. Bassani et al. (156) found that an increase in pavement width results in both an increase in mean speed and a decrease in speed deviation along tangent segments. Shoulder Width Relationship with Crashes The HSM suggests that the width of the paved shoulder along non-freeways has a similar effect on crashes as travel lane widths (16). This occurs partially due to the increased recovery and vehicle storage space as well as increased separation from roadside hazards. While this effect depends on traffic volumes, the frequency of traffic crashes tends to increase as paved shoulder widths are reduced below 6 ft. Furthermore, safety performance decreases as the paved shoulder width decreases below 2 ft on roadways with greater than 2,000 vehicles per day (veh/d) (16). Shoulder Width Relationship with Operating Speed For two-lane rural highways, Lamm et al. (117) found that speeds increase with wider shoulder width. A study on rural four-lane highways in Kentucky (155) found the presence of shoulders to be related to operating speed. In a study on rural four-lane highways in Texas (119), the left and right shoulder widths influenced the speed of daytime car drivers.

NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 37 Lane Position Relationship with Operating Speed Himes and Donnell (121) measured different speeds in the left and right lanes for rural and urban four-lane highways and identified the following variables as relevant to their study: heavy vehicle percentage, posted speed limit, and adjacent land use. Gong and Stamatiadis (155) also found the factor of whether the lane was inside or outside to be significant in their study on rural four-lane highways in Kentucky. SURROUNDINGS Access Density (Driveways and Intersections) Relationship with Crashes Prior studies have demonstrated that as the density of access points (or the number of intersections and/or driveways per mile of highway) increases, the frequency of traffic crashes also increases (16, 164, 165). This occurs partially due to driving errors caused by intersections and/or driveways that may result in rear-end and/or sideswipe type crashes (16). Specifically, NCHRP Report 420 concluded that an increase in crashes occurs due to the higher number of access points (165). The density of access points significantly affected the safety performance of undivided and divided highways in a Michigan study (115). The following were identified:  Undivided segments that included 5–15 access points per mile tended to observe approximately 20 percent more total crashes than those segments with fewer than five access points per mile. Further, undivided segments that included greater than 15 access points per mile were associated with increases in observed crashes across all severity levels (115).  The number of access points per mile also had a significant impact on the frequency of total and injury crashes. Divided segments that included 10–20 access points per mile observed an increase of approximately 24 percent for total crashes and an increase of approximately 27 percent for injury crashes compared to segments with fewer than five access points per mile. Furthermore, divided segments that included greater than 20 access points per mile observed a 73 percent increase in total crashes and a 96 percent increase in injury crashes compared to segments with fewer than five access points (115). These effects are much more pronounced than those for undivided segments. The findings are in general agreement with prior research, which has shown the number of access points to have a significant impact on safety performance, particularly at densities greater than 20 per mile (165). Access Density (Driveways and Intersections) Relationship with Operating Speed Figueroa et al. (78) found lower speed when an intersection was present for two-lane rural highways. Gong and Stamatiadis (155) conducted a study on four rural four-lane highways in Kentucky and found access point density to have an inverse relationship with vehicular speeds; mean speeds decrease as the density of access points increases. For rural four-lane non-limited-access roadways, Robertson et al. (119) found the number of access points significant for only trucks during the daytime. The researchers hypothesized that drivers changed lanes to avoid the effects of vehicles entering and exiting driveways because there were two lanes and relatively low volumes. The dataset also included only free-flow

NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 38 speeds, so vehicles close to each other, which could happen at a driveway access point, were removed. Urban or Rural Relationship with Crashes Whether the roadway segment is in a rural area or an urban area has been found to be significantly related to crashes (115). School Relationship with Crashes Larger numbers of schools near a roadway segment have been found to be associated with a greater number of crashes (115). Development (Surrounding Land Use) Relationship with Operating Speed Figueroa et al. (78) found lower speed to be associated with residential development for two-lane rural highways. Himes and Donnell (121) identified adjacent land use as relevant to their study of rural four-lane highways. OTHER VARIABLES Time of the Day Relationship with Operating Speed Robertson et al. (119) identified time of day as relevant to their study of rural four-lane highways. OVERVIEW OF VARIABLE RELATIONSHIP WITH SPEED AND CRASHES Based upon information in the literature, several roadway segment factors are known or suspected to affect a driver’s speed choice and the likelihood of crashes. Table 3 summarizes the high-speed highway factors that affect crashes. Table 4 summarizes the high-speed highway factors that affect operation speed.

NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 39 Table 3. Factors of high-speed highways that affect crashes. Category Factor Key Findings Source Traffic control device Passing lane or no- passing zone(s), length or percent of segment Unknown 16, 115, 116 Traffic control device Posted speed limit Mixed effect 123, 124, 125, 126, 127, 128, 129, 130, 131 Traffic AADT, percent trucks or commercial AADT Positively associated 115, 116 Traffic Operating speed Mixed effect 9, 122 Surroundings Access (driveways and intersections) Negatively associated 16, 115, 164, 165 Surroundings Development (e.g. school) Positively associated 115 Roadway geometry Horizontal alignment, terrain Positively associated 138, 139, 16, 140, 141, 142, 143, 115, 144, 145, 161 Roadway geometry Vertical alignment Positively associated 16, 158, 159, 123 Roadway geometry Median type (e.g., undivided, divided) Two-way left-turn lane associated with a significant increase in total and injury crashes but also associated with a significant decrease in fatal crashes 115, 123 Roadway geometry Median width Negatively associated 162 Roadway geometry Number of lanes Positively associated 115 Roadway geometry Lane width Negatively associated 16, 163  Roadway geometry Shoulder (paved) width Negatively associated 16

NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 40 Table 4. Factors of high-speed highways that affect operating speed. Category Factor Key Findings Source Traffic control device Passing lane present Speed differentials for different passing maneuvers 137 Traffic control device Posted speed limit Positively associated 123, 124, 125, 126, 129,,130, 133, 134, 135, 136, 121, 119, 132, 118 Traffic AADT Mixed effect 115, 117, 118, 119 Traffic Percent trucks or commercial AADT Unknown 121, 119 Surroundings Access (driveways and intersections) Negatively associated 78, 119 Surroundings Development Negatively associated 78, 121 Roadway geometry Horizontal alignment (curve radii and length) Negatively associated 78, 139, 124, 138, 117, 146, 147, 148, 149, 150, 151, 152, 153, 132, 154, 155, 156, 157 Roadway geometry Lane width Positively associated 78, 117, 156 Roadway geometry Median type Median type associated with different speed profiles 155 Roadway geometry Median width Positively associated 120 Roadway geometry Number of lanes Positively associated 121, 155 Roadway geometry Shoulder (paved) width Positively associated 117, 155, 119 Roadway geometry Vertical alignment Negatively associated 78, 152, 153, 132, 155, 160, 118, 157  Other factors Time of day Positively associated with nighttime 119